Professional Documents
Culture Documents
Preface
New materials, composites, and coatings are being developed at a rapid rate, and
there has been an increase in the substitution or replacement of one class of material
by another. More complex materials are being engineered and used in a new variety
of environments. Many materials are used in composite or coated forms to enhance
performance. Various combinations of metals, intermetallics, ceramics, and polymers are becoming more common. Composites with discontinuous, dispersed
phases within a matrix, and fiber and laminated reinforcements are being developed. Coatings also include combinations similar to those for bulk composite materials. Materials are being pushed to perform in a wider range of environments
than ever before. Aqueous and high-temperature environments, which may contain
varying amounts of corrosive species, are commonly encountered by advanced
materials. In other cases, the effect of environments such as water, solvents, wine,
and food thought to be relatively benign must be understood.
All these developments have made it difficult to locate information on the
effects of environment on the new materials, composites, and coatings. This comprehensive book describes such effects for a broad range of materials and environments,
filling the information gap and providing a comprehensive viewpoint for the scientist
or engineer interested in applying new materials to existing applications or old materials to new applications.
This book would not have been possible without the many hours given by each
contributor. Their effort and dedication are greatly appreciated. Also, the assistance of
B. H. Wardlow at PNNL in coordinating the manuscripts is greatly appreciated.
Russell H. Jones
iii
Contents
Preface
Contributors
iii
vii
I. Metallic Alloys
1. Ferrous Alloys (Ferritic and Martensitic)
Bruce Craig
31
55
75
115
v
vi
Contents
151
7. Aluminum Alloys
N. J. Henry Holroyd
173
8. Magnesium Alloys
Mike J. Danielson
253
275
Ceramics
311
351
IV. Composites
12. Metal Matrix Composites
Russell H. Jones
375
391
419
V. Metallic Glasses
15. Amorphous and Nanocrystalline Alloys
Koji Hashimoto
459
Index
501
Contributors
viii
Contributors
Nathan S. Jacobson Materials Division, NASA Glenn Research Center, Cleveland, Ohio
Russell H. Jones Pacific Northwest National Laboratory, Richland, Washington
George Y. Lai Consultant, Carmel, Indiana
Charles A. Lewinsohn
Washington
1
Ferrous Alloys
(Ferritic and Martensitic)
Bruce Craig
MetCorr, Denver, Colorado
I.
INTRODUCTION
This chapter addresses the corrosion behavior of ferrous alloys, specifically ferritic and martensitic irons and steels. The reason for this designation is to distinguish these alloys from the austenitic alloys that will be discussed in a later
chapter. However, the use of the terms ferritic or martensitic is not intended
to exclude pearlitic or bainitic microstructures, but is only intended as a convenience. Therefore, the discussion in this chapter addresses all low-alloy ferrous
materials and ferritic and martensitic stainless steels.
The largest group of ferrous alloys are steels which will be the emphasis
in this chapter, however, cast irons, several of which can be quite corrosion resistant, will also be mentioned. There are tens of thousands of different steels in
the world; however, they are usually referred to in groups as a function of their
chemical composition. Thus, carbon steels (also referred to as mild steels) contain
little or no alloy elements beyond the Mn, P, S, Si, and Al needed to produce a
good quality structural material. The low-alloy steels are the next group that can
be characterized by small additions of Cr, Mo, and Ni, usually in the range of
about 0.104.0% of each element, but generally less than 5% of the total alloying
elements. Higher additions of these elements form a group of steels referred to
as alloy steels. Generally, the alloying content is equal to or less than 10% (e.g.,
9 Cr1 Mo steel).
The distinction between low-alloy and alloy steels is not well defined nor
even well observed in practice. Often, all of these steels are lumped together
under the term low-alloy steel or alloy steel. As will be seen in this chapter,
1
Craig
this distinction is largely unnecessary from a corrosion standpoint because alloying of less than 10% for many environments is not sufficient to impart significant
corrosion resistance to steels.
In a similar vein, cast irons are used as structural or pressure-containing
alloys that have little natural corrosion resistance. Additions of Cr, Ni, and Si
are most often the primary means for improving corrosion resistance.
Unquestionably the most important alloying element in steels and irons
from a corrosion standpoint is Cr. Steels containing in excess of 11% Cr, will
display stainless (rust-resistant) qualities when exposed to the atmosphere; thus,
this group or, more properly, family of alloys is termed stainless steels. In
this chapter, the ferritic and martensitic stainless steels will also be discussed.
Although the number of alloys that are covered by the categories just presented are myriad, the general performance is relatively easy to address. The
corrosion performance of these ferrous alloys is of major importance not only
because they represent the largest tonnage of metals used by the world but because they represent the benchmark from which corrosion performance of other
alloys is compared.
Ferrous Alloys
Fig. 1 Typical polarization behavior for mild steel under active corrosion.
Craig
this film is not mechanically strong and is easily eroded. Thus, this film is not
suitable for exposure to high-velocity streams, yet it is beneficial from a sulfuric
acid storage standpoint because carbon steel containers can be used to handle
the acid under essentially static conditions.
Other environments produce this same behavior and Fig. 3 illustrates this
Ferrous Alloys
development of passivity in the anodic curve that reflects a decrease in the anodic
corrosion current with the formation of a passive film (1).
Great care must be taken in applying this method of passivity, however,
because many factors in actual service can eliminate or degrade this protective
film, causing significant corrosion to occur. Velocity changes, temperature increases, the presence of impurities (i.e., chlorides), and concentration changes
can produce high-corrosion rates instead. A useful example of this change in
Craig
corrosion rate is carbon steel in 90% sulfuric acid at room temperature. Under
static conditions, the corrosion rate is about 0.5 mm/year. However, at a concentration of less than 50% H2SO4, the corrosion rate exceeds 5 mm/year. Thus, the
stability of the passive film is an important factor in the choice of any material
for a specific environment and that choice may be suitable only over a narrow
range of conditions. As will be discussed in later chapters, the stability of the
passive layer on nickel-based alloys, titanium alloys, and other materials is much
greater than for steels, thus the reason these alloys are more resistant to corrosive
environments. It is the great stability of the air-formed oxide on ferritic and martensitic stainless steels that produce their stainless quality when exposed to the
atmosphere. Yet, this passive film is not stable in all environments and care must
be taken in their application, as the oxide is particularly susceptible to attack by
halides.
B. Localized Corrosion
In addition to the uniform or general corrosion of ferrous alloys, there are numerous forms of localized corrosion that can cause failure of these alloys. Pitting
corrosion is a highly localized attack of the metal, creating pits of varying depth,
width, and number. Pitting may often lead to complete perforation of the metal
with little or no general corrosion of the surface. This can be a considerable
problem in steels and is one of the most common causes of failure for stainless
steels. At this time it is impossible to predict the remaining life of a pitted structure; thus, pitting remains one of the leading causes of failure for ferrous alloys.
Crevice corrosion is similar to pitting corrosion in its localized nature but
is associated with crevices. Stainless steels and some nickel-based alloys are particularly susceptible to this form of corrosion; however, steels are less susceptible
to this form of attack, except in aerated environments.
Intergranular corrosion is the preferential corrosion of grain boundaries in a
metal caused by prior thermal treatments and related to specific alloy chemistries,
especially in stainless steels and nickel alloys. Corrosion of this type is rare in
carbon and alloy steels but can be a problem in ferritic and martensitic stainless
steels.
Dealloying is the selective removal of one element (usually the least noble)
from an alloy by the corrosive environment. Also referred to as selective leaching
or dezincification, denickelification, and so forth, designating the element removed. Steels are not generally attacked by this mechanism, nor are ferritic or
martensitic stainless steels. However, some cast irons, especially gray iron, are
quite susceptible to dealloying. For gray cast iron, the graphite flakes are cathodic
to the surrounding ferritic matrix. Thus, the ferrite is selectively corroded away,
leaving a mechanically weak graphite structure.
Corrosion fatigue is the initiation and extension of cracks by the combined
action of an alternating stress and a corrosive environment. The introduction of
Ferrous Alloys
1. Pitting Corrosion
Pitting corrosion is one of the most common and most insidious types of corrosion
attack on steels. Pitting may rapidly produce perforation of a metal or may take
Craig
many years to develop. Currently, there are no methods to accurately predict the
propagation rates of pits and, therefore, no valid means to estimate the remaining
life of a structure or component once pitting has initiated. There has been some
success in modeling pitting as a stochastic process, but, as yet, there is not an
accepted methodology. Because of this inability to predict pitting and remaining
service life, the primary focus in materials selection for a specific environment
is to choose a material that is either immune to a particular environment or at
least highly resistant to pitting in the first place. This can be a difficult task because often it is not the major component of the service environment that induces
pitting but rather the small concentration of some impurity that does. A good
example of this is shown in Table 1, where increasing the Cl content of H 2 SO4
requires a corresponding increase in Cr to the steel to resist pit initiation (3). As
discussed earlier, steels are generally resistant to concentrated H2SO4, but the
introduction of small amounts of Cl makes the solution particularly corrosive.
This same effect is observed for steels exposed to seawater. Seawater itself is
not very aggressive to ferrous alloys; however, it is the introduction of dissolved
oxygen that causes seawater to become corrosive, producing severe pitting attack.
Figure 4 illustrates the dramatic effect of oxygen, in only the parts-per-billion
(ppb) range, on the corrosion of steel (4).
The mechanism of pitting is well understood in a general sense. Pit initiation begins with the very localized breakdown of the passive film, leading to the
formation of a small pit bottom that acts as the anode and the remainder of the
passive surface as the cathode. Thus, there is a large driving force to continue
development and propagation of the pit. However, the nature of pitting is a selfsustained autocatalytic process that continues pit propagation. During the propagation process, the solution in the pit bottom becomes and remains very acidic,
further enhancing propagation. Moreover, the potential difference between the
steel surface and the pit bottom acts as a driving force for propagation. During
Cl (normality)
0.0003
0.017
0.069
0.1
0.1
1.0
1.0
Ferrous Alloys
Fig. 4 Effect of oxygen concentration on corrosion of mild steel in Pacific Ocean water.
this period, the original steel surface, which has not begun pitting, is effectively
protected from further corrosion by the resulting cathodic polarization.
The difficulty in predicting the remaining life of a structure during pitting
corrosion is due to continual pit initiation, propagation, repassivation, and repropagation. Not all pits in the same structure propagate at the same rate and
propagation is not linear but rather an exponential function.
It is generally recognized that pitting will initiate at microstructural discontinuities on the steel surface. These discontinuities can be grain boundaries,
second-phase particles, and so forth, but they are most often sulfide inclusions.
This latter feature is most commonly the origin for pits in stainless steels. Therefore, it is quite predictable that resulfurized steels, especially the resulfurized
stainless steels, will suffer pitting corrosion in an environment long before and
under less severe conditions than the lower sulfur version of the same steel. For
example, AISI 416 stainless steel, which contains 0.15% S minimum, compared
10
Craig
to its counterpart, AISI 410, which contains 0.030% S maximum, is highly susceptible to pitting corrosion and may pit in environments where AISI 410 does
not.
It is impossible to list all the environments in which steels pit because of
the numerous factors involved and the great variety of possible combinations of
chemicals. Moreover, great care must be taken when selecting an alloy for a
certain application in not simply scanning the large number of corrosion data
references and databases for alloys with low corrosion rates, as most of these
resources do not present pitting data, but rather provide only uniform corrosion
rates that can be very misleading. However, that said, it is often the case that
steels and, for that matter, many other alloys have a great tendency to pit in
environments that contain chlorides or, more generally, halides. Although chlorides are by far the most prevalent species in many environments bromides, iodides and fluorides can also induce pitting.
Therefore, the presence of halides in a process stream should be a signal
that pitting must be considered in the choice of alloys. Yet, the absence of these
species does not necessarily eliminate the possibility of pitting. An example of
pitting in the absence of halides is corrosion from CO2 gas dissolved in water.
This condition produces carbonic acid that can lead to pitting corrosion of carbon
and low-alloy steels. Of course, the situation becomes more complex as a function
of temperature and the introduction of chlorides. Figure 5 shows the envelope
of applicability of AISI Type 420 stainless steel (also referred to as 13 Cr) to a
combined environment of CO2 and chlorides as a function of temperature in the
absence of oxygen (5). Within this envelope, no pitting occurs and corrosion is
minimal but uniform. However, the introduction of small concentrations (ppb)
of oxygen creates a severe pitting attack of the 13 Cr even at ambient temperature,
thereby eliminating the use of this alloy.
Thus, prior experience or laboratory testing is often necessary to confirm
that a particular alloy will not be susceptible to pitting in a specific environment.
2. Environmental Cracking
As previously indicated, environmental cracking (EC) is a general term that encompasses all forms of cracking that are induced or accelerated by the service
environment. The two principal categories within this form of corrosion that are
pertinent to this discussion on ferrous alloys are SCC and HD. An in-depth review
of these types of cracking and their mechanisms can be found elsewhere (6). It
is simplest and consistent with much of the literature to discuss SCC in terms
of an active path corrosion coupled with a tensile stress (often referred to as
anodic cracking) and HD as all those forms of cracking that depend on hydrogen assistance (often referred to as cathodic cracking in aqueous environments,
Ferrous Alloys
11
Fig. 5 The corrosion resistance of 13 Cr (Type 420) stainless steel in CO2 /NaCl environments in the absence of O2.
12
Craig
Fig. 6 Proposed sequence of crack initiation, coalescence, and growth for steels undergoing subcritical cracking in aqueous environments.
7 should not be construed to mean that ferrous alloys do not crack at all below
150 ksi, because quite low-strength steels are very susceptible to EC, just in
different environments. Moreover, the cracking of these high-strength steels in
seawater is thought to involve HD. Thus, it is convenient to further discuss the
EC of ferrous alloys in two groups; high-strength steels (150 ksi) and lowstrength steels (150 ksi).
a. Environmental Cracking of Low-Strength Steels Low-strength steels
(150 ksi yield strength) are quite susceptible to EC in certain specific environments. The yield strength of the steel in this strength range is not particularly
significant to the susceptibility to EC as it is for higher-strength steels. Rather,
other factors such as applied stress, steel composition, pH, solution composition,
potential, and temperature are much more critical. Increasing applied (or residual)
stress and increasing temperature enhances the SCC of low-strength steels as
does decreasing pH. Small concentrations of trace or impurity elements in the
alloy can have a profound effect on SCC of steels.
Some of the more common environments known to cause SCC of lowstrength steels are liquid NH3, CO2 /CO, carbonate/bicarbonate, hydroxide, nitrate
Ferrous Alloys
13
8N
4N
2.5N
1N
NH 4 NO3
Ca(NO3 ) 2
LiNO3
KNO3
NaNO3
2.2
5.6
5.6
6.7
9.0
3.4
7.8
9.0
4.5
9.5
7.8
13.4
21.3 (2N )
15.7
24.7
13.4
25.8
25.8
26.9
29.2
14
Craig
(10). Area C can also be handled successfully with austenitic stainless steels.
This diagram illustrates the important effect of residual welding stresses on SCC
and the ability to extend the range of applicability of steels in OH simply by
reducing the residual stresses.
An area of great concern that has recently received increased attention is
the SCC of low-strength pipeline steels. The external SCC of pipeline steels has
occurred in two distinct environments. Early failures were in soil environments
that produced solutions of carbonate/bicarbonate with a pH of about 9.5 on the
outside of the pipe, causing intergranular SCC. Figure 9 shows the intergranular
SCC of a low-strength pipeline steel that failed in the high-pH environment. More
recently, transgranular SCC has been found to be the cause of several pipeline
failures. The pH in this latter case frequently falls in the range of 68. Many of
Ferrous Alloys
15
the pipeline failures occurred in pipelines that are more than 20 years old that
have yield strengths around 52,000 psi. This long incubation time for crack initiation and propagation is typical for low-strength steels and in sharp contrast to
the often rapid initiation and fracture of high-strength steels. Yet, it would be
misleading to assume that SCC of low-strength steels is always a slow process.
Figure 10 shows that the crack growth rate in many low-strength steels is a strong
function of the solution composition and is directly related to the bare surface
current density (11). This current density, in combination with straining at the
crack tip, is the driving force for cracking and is frequently referred to as active
path or anodic cracking. It is generally believed, though not entirely agreed, that
SCC progresses by the rupture of the oxide film at the crack tip, thereby providing
a bare surface for the peak current to advance the crack tip a certain distance
16
Craig
Fig. 10 Crack velocities and peak current densities at the same potentials for a variety
of systems and alloys.
before the oxide re-forms and the crack arrests. These events may occur over
many cycles or just a few. However, it is important to recognize the difference
in this mode of cracking versus that due to cathodic cracking, where hydrogen
is the primary agent to assist cracking.
In general, martensitic stainless steels are used at higher strengths than
ferritic stainless steels because the former can be strengthened by heat treatment
and the latter cannot. Therefore, the martensitic stainless steels are discussed
under the high-strength section. Although ferritic stainless steels are generally
more resistant to SCC than austenitic stainless steels, especially in chloride solutions, they are not entirely immune. Small additions of Ni and plastic deformation
can each increase the tendency for SCC in chloride environments.
Hydrogen damage of low-strength steels typically occurs in steels that have
yield strengths less than 100 ksi. As with the SCC of low-strength steels, the
yield strength is not an important factor in HD. Moreover, residual and/or applied
stresses have little effect.
Ferrous Alloys
17
The primary cracking modes are stepwise cracking (SWC, also referred to
as HIC) and blistering (HB) or blister cracking. Both types of cracking are the
result of relatively high hydrogen input fugacities compared to the HD of highstrength steels and are often found together in the same steel. Both SWC and
HB are considered to occur by the classical hydrogen-pressure mechanism. According to this mechanism, hydrogen atoms enter the steel and combine at discontinuities (i.e., nonmetallic inclusions) to form molecular hydrogen, which is too
large of a molecule to diffuse back out of the steel. The molecules continue to
accumulate, increasing the local hydrogen pressure until a crack or blister forms.
Figure 11 shows an example of SWC in a low-strength steel exposed to H2S.
Hydrogen damage of steels occurs over the entire strength range of typical engineering applications. Figure 12 shows that regardless of the strength level of the
steel, some form of hydrogen cracking may occur and the only distinction is in
the morphology of cracking (12).
b. Environmental Cracking of High-Strength Steels The EC of highstrength steels (150 ksi) is highly dependent on strength, and in many environments, it is difficult to distinguish between the more classical HD and SCC mech-
Fig. 11 Stepwise cracking from hydrogen in a low-strength steel exposed to H2S. Magnification: 25.
18
Craig
anisms. On the other hand, EC from caustic solutions are not obviously hydrogen
related, but the crack propagation rate is many times faster than for low-strength
steels in caustic.
Seawater and brackish waters do not typically produce EC failures of lowstrength steels, but they do produce the EC of high-strength steels, as demonstrated earlier in Fig. 7. Again, this is most likely a HD mechanism.
From an engineered materials sense, the actual mechanism is not as important as the fact that high-strength steels are so susceptible to EC, and the resulting
crack propagation rate so high that catastrophic failure in many otherwise benign
environments can easily occur. Because of this high risk of steel failure with
increasing strength beyond 150 ksi, it is common practice to select other alloys
and materials that have a greater overall corrosion resistance for high-strength
applications. These materials and their performance are dealt with in the remainder of this book.
Ferrous Alloys
19
20
Craig
C. Galvanic Corrosion
Galvanic corrosion is one of the most common yet least recognized corrosion
problems for ferrous alloys. When two or more dissimilar metals are intimately
connected and placed in a solution, current will flow between them because of
the potential difference of the metals. The metal with the least resistance to corrosion (active metal) in the particular environment will become the anode, and the
more corrosion-resistant metal (noble metal) will become the cathode. Corrosion
of the anode will usually be more severe than if that metal was alone in the
same solution, whereas the cathode will achieve a degree of protection from the
environmentsometimes to the extent that corrosion is completely stopped on
the cathodic metal. These effects can be measured and have been done for couples
of metals in seawater at 25C. Table 3 provides a relative ranking of metals and
alloys regarding their resistance to corrosion in seawater (14). The greater the
distance between two metals in the table, the greater their potential difference and
the higher the probability that the active metal will suffer accelerated corrosion.
Note that some alloys and metals are listed twice in the table: once with
the word active following and once with the word passive. Some metals
and alloys become essentially immune to corrosion in certain environments beTable 3
Active or anodic
Magnesium
Magnesium alloys
Zinc
Galvanized steel
Aluminum 1100
Aluminum 2024
Mild Steel
Wrought Iron
Cast Iron
13% Chromium stainless steel
Type 410 (active)
18-8 Stainless steel
Type 304 (active)
Leadtin solders
Lead
Tin
Muntz metal
Manganese bronze
Naval brass
Nickel (active)
Nickel Alloy 600 (active)
Yellow brass
Admiralty brass
Red brass
Copper
Silicon bronze
7080 Cupro nickel
Gbronze
Silver solder
Nickel (passive)
Nickel alloy 600 (passive)
13% Chromium stainless steel
Type 410 (passive)
18-8 Stainless Steel
Type 304 (passive)
Silver
Graphite
Gold
Platinum
Noble or cathodic
Ferrous Alloys
21
cause of the formation of a surface film so thin that it is impossible to see with
the naked eye or even with an optical microscope. The stability of these films is
paramount to the enhanced corrosion resistance of these alloys. Moreover, corrosion films represent the controlling factor in almost all corrosion (15). The addition of chromium to iron can produce a passive alloy (18-8 stainless steel) of
considerable corrosion resistance compared to the original iron. However, in an
environment in which the passive film is not functional, the active surface becomes far less noble, as indicated in Table 3.
A more active metal in the series will corrode at the expense of a nobler
one. Thus, coupling zinc to steel will cause the zinc to corrode and will protect
the steel. This is the reason for galvanizing steel; when pinholes in the galvanizing
occur, the steel underneath, once exposed to the environment, will be protected
by the zinc. This is also the basis for cathodic protection. Sacrificial anodes are
made of metals or alloys that are more active than steel, which allows for the
consumption of the anode and the protection of the steel structure. However, if
steel is coupled to copper, the distance on the chart is large, so in this case, steel
will be the anode and have a greater tendency to corrode.
The galvanic series is useful for approximating the behavior of coupled
alloys; however, care must be used in its application. Several parameters affect
galvanic corrosion and, as such, may affect the actual behavior of a couple in
service. Two important factors in galvanic corrosion are the temperature and the
relative area of the metals. Increasing temperature in some cases may cause a
reversal in the anodecathode relationship. This reversal has been responsible
for failures of galvanized systems or systems protected with zinc sacrificial
anodes. These effects point to the need to measure the potential of a couple,
especially in cathodic protection, in the actual environment prior to its application. It must always be borne in mind that the ranking of alloys in Table 3 is
strictly true only for seawater and that extending it to other environments may
result in some changes in the position of metals and alloys in the series.
The other factor, the ratio of the area of the anode to the area of cathode,
is of considerable importance. If the anode area is smaller than the cathode area,
the corrosion rate may be increased many orders of magnitude as a function of
this ratio. However, if the anode area is greater than the cathode area, corrosion
of the anode will be less than for a 1: 1 anode/cathode ratio.
22
Craig
Fig. 14 Corrosion rate of steels in wet CO2 as a function of the chromium content of
the alloy.
Ferrous Alloys
23
attack. A similar benefit is observed for cast irons exposed to nitric and hydrochloric acid by alloying with either Si or Ni in excess of about 5%.
V.
In general, the corrosion rate of ferrous alloys increases with increasing velocity.
This can be readily explained by the increased mass transport of ferrous ions
across the fluid boundary layer that is established under flowing conditions com-
Fig. 15 Effect of the tempering on the yield strength and on the corrosion resistance
of quenched-and-tempered 13% Cr stainless steel.
24
Craig
bined with the enhanced transport of corrodents to the metal surface across this
boundary layer. Of course, the picture is actually more complex when the Helmholtz double layer and the presence of a corrosion product film are included.
However, regardless of these issues, at some critical velocity these layers are
essentially stripped away and bare metal is continually exposed to the fluid
stream. At this point, one of two entirely divergent phenomenon may occur. The
erosioncorrosion rate becomes extremely high due to the loss of the ratedetermining mass transport across all of these layers or the erosioncorrosion
rate becomes much lower as a result of the inability of the corrodent to reach
the bare metal surface and have sufficient time to react. Both of these phenomena
are observed for ferrous alloys. Figure 16 illustrates the former case for carbon
steel in distilled water (18). At different pHs, the corrosion product formed can
be resistant to erosioncorrosion (pH 5 and 9), thus minimizing the erosion
corrosion rate, or the corrosion film is unstable (pH 4, pH 68), leading to
high erosioncorrosion rates. The introduction of solid particles such as sand
significantly lowers the erosioncorrosion threshold, making the selection of
alloys resistant to erosioncorrosion much more difficult. In the absence of solid
particles, it has been found that erosioncorrosion resistance is strongly a function of the nature of the oxide. Therefore, more corrosion-resistant alloys such
as stainless steels, nickel-based alloys, and titanium alloys have greater erosion
corrosion resistance than steels, even if the alloy is much softer, because of their
tighter more resilient oxides.
Ferrous Alloys
25
26
Fig. 17
Craig
metal surfaces which act as a catalyst to form SO3 and thus H 2 SO4 in moist
environments or particulates such as dust that can cause local cells to form by
aiding in the absorption of water and chlorides can accelerate corrosion of alloys
that normally would be resistant to atmospheric corrosion in a relatively clean
environment.
Climatic conditions have a variable effect on corrosion rate. For example,
in some regions, winter exposure may be more severe than summer if fuels are
used during cold spells that increase combustion products in the air such as NO x
and SO2. Conversely, if these fuels are not used in the region, then summer may
be worse due to the higher metal temperatures.
Likewise, periodic rainfall may be beneficial causing a rinsing action on
the surface compared to a climate where the surface is continually wet. Thus,
time of wetness and wet/dry cycles can also be quite important, especially if
frequent periods of wet and dry can limit the formation and development of a
protective oxide layer. Moreover, the existence of insoluble corrosion products
can act to entrain water during short dry cycles, keeping the metal surface sufficiently wet to continue corrosion.
Two other external environments that can cause significant deterioration
of ferrous alloys are soils and concrete. The majority of buried structures in the
Ferrous Alloys
27
world are made of cast irons and steels. Soils have a great variability in their
tendency to cause corrosion of ferrous alloys. Some of the more important factors
that contribute to the aggressiveness of soils are resistivity, pH, moisture, oxygen,
bacterial activity, and temperature of the ferrous alloy (i.e., hot pipelines) in contact with the soil. Decreasing resistivity and pH, increasing moisture content,
oxygen content, and bacterial activity all enhance corrosion of ferrous alloys in
soils. The elevated temperature of the steel surface not only can increase the
corrosion rate but also can lead to other forms of more serious attack such as
EC. As shown in Fig. 9, the external SCC of pipelines has become a particular
problem for those pipelines that operate above ambient temperature and the trend
in the future is to operate pipelines at even higher temperatures. Often, coatings
and cathodic protection can be used to limit these problems; however, they can
also exacerbate them if not properly maintained. One of the major reasons older
pipelines have become susceptible to SCC is that the coatings have degraded
over time and the cathodic protection systems cannot effectively limit corrosion.
Corrosion of steel reinforcing bar in concrete has gained attention due to
the widespread problem of the crumbling infrastructure (bridges, highways,
buildings, etc.) in many countries. Typically, the steel rebar corrodes as a result
of the pH of the cement paste in contact with the steel and the diffusion of chlorides and oxygen into and through the concrete. As corrosion products form on
the steel, they represent a larger volume than the original iron in place, thereby
spalling and cracking the concrete. It has been found that temperature and relative
humidity are important factors in rebar corrosion as well as chlorides and oxygen.
Thus, tropical climates that are hotter and more humid than temperate climates
would be expected to have a greater problem with rebar corrosion than cooler
climates. However, even northern climates have had problems for other reasons;
for example, deicing salts used on bridges and roadways to eliminate snow and
ice can cause severe rebar corrosion. Several of the methods currently used to
fight rebar corrosion are organic coated steel, galvanized steel, stainless-steel
rebar, and cathodic protection.
As can be appreciated from the foregoing comments, coatings on ferrous
alloys are an important means of extending the applicability and service life of
these materials. As this is a book on engineered materials, it is beneficial to at
least mention the general types of coatings used on ferrous alloys. Coatings can
be grouped into four general categories: organic, inorganic, conversion, and metallic. Table 4 lists some of the typical coatings under each of these categories.
Under the category for metallic coatings, the specific metal is not listed because
many metals can be applied; rather, the process is provided because it will determine which metal can be applied. The selection and use of a particular coating
is a function of many factors and care must be taken in selecting the right coating
for a ferrous alloy.
It must always be borne in mind that a coating is part of a system that
28
Table 4
Craig
General Categories of Coatings
Organic
Coal tars
Epoxy
Phenolics
Alkyds
Vinyls
Urethanes
Acrylics
Metallic
Galvanizing
Plating
Ion implantation
Cladding
Flame spray
Chemical vapor deposition
Physical vapor deposition
Conversion
Anodizing
Phosphating
Chromate
Molybdate
Inorganic
Silicates
Ceramics
Glass
Fig. 18
Ferrous Alloys
29
includes the steel substrate; therefore, if the coating is damaged, corrosion often
occurs by different mechanisms than if only the steel were involved. For example,
galvanizing (Zn) acts as a sacrificial anode to the underlying steel if the coating
is damaged. Thus, the steel substrate is protected against corrosion. However, a
more noble coating such as Ni on steel can act as a large cathode, thereby accelerating corrosion at a damaged location (referred to as a holiday). Even organic
coatings can display accelerated corrosion at holidays in the coating depending on
the particular environment to which they are exposed (Fig. 18) (19,20). Generally,
coatings show a slow decrease over time in the coating (film) resistance, indicating the gradual permeation of water and other ionic species through the coating.
Thus, coating life in any environment is finite and organic coatings are not truly
barriers, as is so often mistakenly suggested. However, the application of the
correct coating can often double or triple the useful life of a ferrous alloy in
certain environments and is, therefore, an important factor in the selection of
ferrous alloys.
In conclusion, even though ferrous alloys (ferritic and martensitic) are the
most widely used engineered materials, they are also the least corrosion resistantreadily degrading in most environments. Because of this behavior, they
most often require additional means of controlling corrosion (i.e., coatings, cathodic protection, inhibitors) to provide a satisfactory service life.
REFERENCES
1. RL Martin. Application of Electrochemical Polarization to Corrosion Problems. St.
Louis, MO: Petrolite Corp., 1977.
2. ASM Metals Handbook, Vol. 13, Corrosion. ASM International, Materials Park,
OH, 1989.
3. ND Stolica. Pitting corrosion on FeCr and FeCrNi Alloys. Corrosion Sci 9:
455460, 1969.
4. D Wheeler. Treating and monitoring 450,000 b/d injection water. Petrol Eng Int
1975; November: Vol. 38, 6872.
5. BD Craig. Selection guidelines for corrosion resistant alloys in the oil and gas industry.
Technical Publication No. 10073, The Nickel Development Institute, Toronto, 1995.
6. BD Craig, RH Jones. Environmentally induced cracking. In: ASM Metals Handbook, Vol. 13, Corrosion, ASM International, Materials Park, OH, 1989, pp. 145
171.
7. FP Ford. Quantitative prediction of environmentally assisted cracking. Corrosion
51:375395, 1996.
8. EH Phelps. Stress corrosion behavior of high yield strength steels. Proc. Seventh
World Petroleum Congress. Amsterdam: Elsevier, 1967.
9. RN Parkins, R User. The effect of nitrate solutions in producing stress corrosion
30
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
Craig
cracking in mild steel. First International Congress on Metallic Corrosion, London,
1961, p. 289.
Corrosion Data SurveyMetal Selection, 6th ed. Houston, TX: NACE International, 1985, p. 176.
RN Parkins. Predictive approaches to stress corrosion cracking failure. Corrosion
Sci 20:147166, 1980.
E Sato, M Hashimoto, T Murata. Corrosion of Steels in a Wet H2S and CO2 Environment. Second Asian Pacific Corrosion Control Conf., Kuala Lumpur, 1981.
CS Carter. Fracture toughness and stress corrosion characteristics of a high strength
maraging steel. Met Trans 2:16211625, 1971.
Corrosion Basics. LS Van Delinder, ed. Houston, TX: NACE International, 1984,
p. 35.
BD Craig. Fundamental Aspects of Corrosion Films in Corrosion Science. New
York: Plenum Press, 1991.
NG Galindez Ruiz. The effect of crude oil on corrosion of alloys in H2S/CO2 environments. Masters thesis, Colorado School of Mines, 1993.
JL Crolet. Acid corrosion in wells (CO2, H2S): Metallurgical aspects. J Petrol Technol 35:15521558, August 1983.
MG Fontana. Corrosion Engineering, 3rd ed., New York: McGraw-Hill, 1986, p. 94.
BD Craig, DL Olson. Corrosion at a holiday in an organic coated-metal substrate
system. Corrosion 32:316321, 1976.
JR Scully. Evaluation or organic coating deterioration and substrate corrosion in
seawater using electrochemical impedence measurements. Corrosion/86, NACE,
Houston, TX, 1986.
2
Austenitic Stainless Steels
Russell H. Jones, Stephen M. Bruemmer, and Mike J. Danielson
Pacific Northwest National Laboratory, Richland, Washington
Bruce Craig
MetCorr, Denver, Colorado
I.
INTRODUCTION
Austenitic stainless steels derive their stainless properties from the presence
of a very effective passive film. This film forms spontaneously within aqueous
environments and the stability of this film in various environments determines
the corrosion resistance of stainless steels. Alloy composition is also very
significant in determining the stainless character of this family of alloys. Chromium concentration is the one element that directly affects the passive film stability, and Ni, Mn, Mo, C, and N also play a role. Passive behavior begins at about
10% Cr, with increasing film stability occurring with increasing Cr concentration.
Repassivation can occur in aqueous environments and it is the rate at which a
break in the passive film re-forms that also contributes to the stainless character of these materials. Under selected conditions, the passive film is not stable
and this can lead to general or localized corrosion phenomena. Austenitic stainless steels are often prepassivated in an acid bath, but the removal of surface
contaminants that would hinder passivation in aqueous solutions is the primary
purpose of this prepassivation treatment. Given a clean surface in an alloy with
31
32
Jones et al.
sufficient Cr, austenitic stainless steels will passivate upon immersion in an aqueous environment. Also, there will always be an air-formed oxide even without
this prepassivating treatment.
Austenitic stainless steels undergo all the common forms of corrosion, including (1) general, (2) galvanic, (3) pitting, (4) crevice, (5) intergranular, and
(6) stress corrosion. General corrosion occurs when stainless steel is immersed
in an environment in which the passive film is not stable, such as hot sulfuric
acid, boiling MgCl2, or another very aggressive environment. Pitting corrosion
occurs, as in other metals, because of a local discontinuity in the passive film
such as an inclusion or local chemistry change. Halogen ions are the most prevalent cause of pitting in stainless steels, with chloride being the most common
halogen to initiate pitting. Pit growth depends on a variety of factors, with the
localized corrosion environment in the pit being the most significant factor. Surface condition can also contribute to pitting, with the presence of deposits being
a significant factor. Prepassivating treatments are used to clean the surface of
deposits. Intergranular corrosion and intergranular stress-corrosion cracking are
related phenomena which occur when the grain-boundary microchemistry is altered by thermal treatment such as welding or heat treatment. The process that
alters the grain-boundary corrosion resistance is called sensitization and occurs
when chromium carbides precipitate at the grain boundaries.
33
in strength, but they exhibit a high fracture toughness and are used over a wide
range of temperatures. As a class, they have good resistance to hydrogen embrittlement. In recent years, a new class has emerged called superaustenitics. They
are very high in Mo (47%) and nickel, which results in the highest resistance
(within the austenitics) to any localized attack processes in chloride-containing
media.
Table 1
Austenitics
201
202
301
302
304
304L
304LN
316
316LN
316L
321
347
348
384
Superaustenitics
254 SMO
AL-6X
AL-6XN
20Cb-3
442
446
Mn
Si
Ni
Cr
S20100
S20200
S30100
S30200
S30400
S30403
S30453
S31600
S31653
S31603
S32100
S34700
S34800
0.15
0.15
0.15
0.15
0.08
0.03
0.03
0.08
0.03
0.03
0.08
0.08
0.08
5.57.5
7.510.0
2
2
2
2
2
2
2
2
2
2
2
1
1
1
1
1
1
1
1
1
1
1
1
1
16.018.0
17.019.0
16.018.0
17.019.0
18.020.0
18.020.0
18.020.0
16.018.0
16.018.0
16.018.0
17.019.0
17.019.0
17.019.0
3.55.5
4.06.0
6.08.0
8.010.0
8.010.5
8.012.0
8.012.0
10.014.0
10.014.0
10.014.0
9.012.0
9.013.0
9.013.0
0.06
0.06
0.045
0.045
0.045
0.045
0.045
0.045
0.045
0.045
0.045
0.045
0.045
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
S38400
0.08
15.017.0
17.019.0
0.045
0.03
S31254
0.02
0.8
N08366 0.035
N08367 0.03
N08020 0.07
2
2
2
1
1
1
20.022.0
20.022.0
19.021.0
23.525.5
0.03
23.5025.50 0.04
32.038.0
0.045
0.03
0.03
0.035
N08904 0.02
19.023.0
23.028.0
0.035
S40500
S40900
S42900
S43000
S43600
S43035
0.08
0.08
0.12
0.12
0.12
0.07
1
1
1
1
1
1
1
1
1
1
1
1
11.514.5
10.511.75
14.016.0
16.018.0
16.018.0
17.019.0
0.5
0.5
0.04
0.045
0.04
0.04
0.04
0.04
S44200
S44600
0.2
0.2
1
1.5
1
1
18.023.0
23.027.0
0.04
0.04
0.045
0.01
Other
0.25 N
0.25 N
0.100.16 N
2.03.0 Mo
2.03.0 Mo;
2.03.0 Mo
5 % C min Ti
10 % C min Nb
0.2 Co; 10 % C min Nb;
0.10 Ta
0.03
0.100.30 Al
0.045 6 % C min0.75 max Ti
0.03
0.03
0.03
0.03
0.15 Al; 12 % C min
1.10 Ti
0.03
0.03
0.25 N
Jones et al.
904L
Ferritics
405
409
429
430
436
439
UNS
34
Type
S44660
0.025
25.027.0
1.53.5
0.04
0.03
AL 29-4C
S44735
0.03
28.030.0
0.04
0.03
AL 29-4-2
S44800
0.01
0.3
0.2
28.030.0
2.02.5
0.025
0.02
Duplex
329
Uranus 50
S32900
S32404
0.2
0.04
1
2
0.75
1
23.028.0
20.522.5
2.505.00
5.58.5
0.04
0.03
0.03
0.01
S32550
0.04
1.5
24.027.0
4.506.50
0.04
0.03
S40300
S41000
S41400
S41600
0.15
0.15
0.15
0.15
1
1
1
1.25
0.5
1
1
1
11.513.0
11.513.5
11.513.5
12.014.0
1.252.50
0.04
0.04
0.04
0.06
S42000
S42200
0.15 min
0.200.25
1
1
1
0.75
12.014.0
11.513.5
0.51.0
0.04
0.04
440A
S44002
440B
S44003
440C
S44004
Lapelloy
S42300
Precipitation Hardening
138 Mo
S13800
0.600.75
0.750.95
0.951.20
0.270.32
1
1
1
0.951.35
1
1
1
0.5
16.018.0
16.018.0
16.018.0
11.012.0
0.5
0.05
0.2
0.1
12.2513.25 7.58.5
0.07
0.07
0.09
0.070.11
1
1
1
0.51.25
1
1
1
0.5
14.015.5
15.517.5
16.018.0
16.017.0
Ferralium
255
Martensitic
403
410
414
416
420
422
S15500
S17400
S17700
S35000
3.55.5
3.05.0
6.57.75
4.05.0
1.002.00 Mo
2.03.0 Mo; 1.02.0 Cu;
0.20 N
2.004.00 Mo; 1.502.50
Cu; 0.100.25 N
0.03
0.03
0.03
0.15
0.6 Mo(b)
min
0.03
0.03
0.751.25 Mo; 0.751.25 W;
0.150.3 V
0.03
0.75 Mo
0.03
0.75 Mo
0.03
0.75 Mo
0.025 2.53.0 Mo; 0.20.3 V
0.01
0.008
0.04
0.04
0.04
0.04
0.03
0.03
0.04
0.03
35
15-5 PH
17-4 PH
17-7 PH
AM-350
(Type 633)
0.04
0.04
0.04
0.025
Superferritics
Sea-Cure
(SC-1)
36
Jones et al.
the appropriate heat treatment. These alloys can develop extremely high strength
and can also share some of the good corrosion properties of the austenitics, but
hydrogen embrittlement has been identified as a cause of fracture. This grade of
alloy can be martensitic or austenitic.
These five classes of alloys can also be obtained as cast alloys. Microstructure is a key variable in controlling the strength, localized corrosion behavior,
and SCC behavior of all these classes. It needs to be pointed out that the existence
of cast microstructures increase the complexity of choosing the ideal material
and making certain the microstructure is under control.
B. Composition Effects on Corrosion
Stainless steels are vulnerable to all forms of corrosive attack. Their successful
use requires (1) knowledge of the alloying constituents that impact resistance,
(2) detailed information on the chemical composition and temperature of the test
environment, and (3) knowledge of the failure processes to which the particular
alloy or alloy class is subject. Each of the major alloying elements will be briefly
described to give the reader a general knowledge of how to fit the alloy composition to the environment.
1. Chromium
Chromium is the single most important element contributing to the stainless
behavior of these iron-based alloys, and in general, the higher the chromium level
(once it gets above 10%), the better the performance. Additions of chromium
greatly improve the behavior over that of iron in neutral and acidic pH ranges
but, curiously, has little effect in high-pH environments. The chromium appears
in the oxide film and acts to inhibit the transport of corrosion products across it,
leading to the formation of a passive film. In particular, the chromium acts to
decrease the general corrosion rate and improve crevice corrosion resistance.
2. Molybdenum
Molybdenum is perhaps the second most important alloying element from a corrosion standpoint. Small additions have a profound effect by enhancing the passive character of the passive film, particularly in chlorides and reduced sulfur
environments. In particular, the pitting and crevice corrosion behavior are improved relative to similar alloys without the molybdenum. Every class of stainless-steel alloy has members that are very high in Mo, and these are called super austenitics, martensitics, and so forth. The alloys with high Mo levels have
the disadvantage in that certain undesirable phases (sigma, chi, laves) can form
unless care is taken in heat treatment and welding.
37
3. Nickel
Nickel is the most important austenite stabilizer, but it has a complex effect on
the corrosion behavior. It acts to decrease the general corrosion rate in reducing
acidic environments. In sufficiently high levels, it acts to improve the SCC behavior in chloride and caustic solutions, but at intermediate levels, it can decrease the
SCC resistance. The austenite structure results in a high resistance to hydrogen
embrittlement.
4. Manganese
At the low levels used in most stainless steels, manganese can be considered a
substitute for nickel. It controls the solubility of sulfur by precipitating the sulfur
as MnS inclusions. Modern steel-making practice is to keep the sulfur as low
as possible because localized corrosion events such as pitting initiate at MnS
inclusions.
5. Carbon
Carbon is used at low levels as a strengthener in stainless steels, but it also can
render the stainless steels vulnerable to localized attack or SCC from sensitization. In sensitization, the carbon reacts with chromium to precipitate a Cr 23C 6
carbide causing a very local decrease of the chromium concentration in the metal,
making this depleted region vulnerable to localized attack or SCC. In the austenitics, this has led to the L grades, which are low in carbon (less than approximately 0.03 at.%) and fairly immune to this problem.
6. Nitrogen
Nitrogen is used at very low levels and acts as a strengthener and austenite stabilizer. Nitrogen greatly improves the pitting and crevice corrosion performance
of the austenitics and superaustenitics, particularly in concert with Mo. It helps
prevent (slows down) the undesirable chi (and probably sigma, laves) phase from
forming during welding or heat-treating operations with the high-Mo alloys. Its
behavior is mixed in other classes of stainless steels.
C. Corrosion Behavior
Choosing the right material is a multistage process. Once the mechanical properties and certain other properties (e.g., wear resistance, weldability, machinability,
availability in the needed form, etc.) of the metal are defined for the application,
the next level requires a careful definition of the chemical (e.g., pH, chloride
level, sulfide level, temperature, velocity, oxygen, hydrogen gas) and exposure
environment (single or multiple phases present, occasional dryout, differential
38
Jones et al.
aeration, etc.). Once this multistage process is completed, the general class of
alloy can be defined. Because cost is always an important parameter for the
material choice, the tendency will be to use the lowest alloyed material because
nickel and chromium are expensive. As a general rule for iron-based stainless
steels, the higher the alloy is in nickel and chromium (and molybdenum), the
more corrosion resistant the material will be. The task of the materials scientist is to find the least expensive material that will perform adequately. Often,
the exact answer will not be found in the literature; consequently, testing of various representative alloys will be needed. Many test methods are described in
Ref. 1.
From their inception, stainless steels were developed to have low general
corrosion rates. This simplification is still fairly true as long as one avoids certain
environments such as reduced sulfur environments or highly acid environmentsonly highly alloyed stainless steels will perform in these environments.
The surprises with stainless steels are usually due to localized attack such as
pitting, crevice corrosion, and SCC. Because the chemical environment has infinite variability, the corrosion behavior for each class of stainless steel will be
examined using certain defined environments which will act as benchmarks or
models. Examples will be potable water (low chloride), seawater (high chloride),
and so forth. The book by Sedriks (2) is a particularly good reference.
1. Atmospheric Environments
Atmospheric environments are those exposed to the natural elements in the air
at ambient temperatures. City and marine environments present the most difficult
environments because they are contaminated with sulfur compounds and chlorides. The austenitics, particularly those containing molybdenum, give the best
outdoor performance from the standpoint of pitting and crevice corrosion, but it
should be noted that they will not be pit-free. Rather, the pit density and depth
will usually remain low enough that the cosmetic use of the material will not be
impacted. Localized corrosion rates for all classes of stainless steels are highest
where the chlorides and sulfides are highest. The ferritics have also been used
in this environment but not as successfully as the austenitics. Both the austenitics and ferritics have good immunity to SCC. The higher-strength alloys such
as the martensitics and precipitation-hardened alloys are prone to pitting and
SCC. This environment can get extremely aggressive if there is a source of heat
such that condensation and dryout takes place, resulting in very concentrated
solutions. One extreme of this environment might be found under insulation, and,
here, SCC and pitting can occur with all the standard stainless classes, even the
austenitics and ferritics. The high-Mo (super types) alloys provide the greatest
resistance to pitting and SCC when the atmospheric environment is the most
aggressive.
39
5. Seawater
Seawater is the most challenging environment for stainless steels, and the problems are compounded if reduced sulfur species are present due to the action of
40
Jones et al.
bacteria or decaying vegetation. In general, the standard alloys within each grade
will be prone to pitting and SCC with much variability in the reported results. For
maximum safety and reliability, only the high-Mo alloys should be considered for
this environment. A review paper by Streicher (6) on 30- and 60-day ambienttemperature crevice tests clearly shows only the superaustenitics, superferritics,
high-Mo duplex, and nickel-based alloys are suitable. All localized corrosion
and SCC problems worsen at elevated temperatures. Here, the famous boiling
magnesium chloride test can give insight into a materials behavior. The 300
series alloys have completely unsuitable SCC behavior in this test (unless they
are cathodically protected). The nickel level has to be above 20% before the
austenitics show significant improvement in SCC behavior in the boiling magnesium chloride test. Molybdenum is also beneficial in raising the threshold stress
intensity for SCC. Austenitics (high Ni), ferritics, and duplex alloys all can show
resistance to SCC, but it is clear that they must be high in Mo. For the most
difficult applications, titanium or nickel-based alloys will need to be used.
6. Acidic Environments
Chromium is the most important element that imparts resistance to acidic environments; consequently, the highest resistance is associated with the highest Cr
levels within each grade of stainless steel. In general, the 300 series austenitic
stainless steels can be used in nitric acid over a wide concentration range (0
65%), even up to the boiling point. However, stainless steels are completely unsuitable for the HCl environment, and mixed acids containing HCl and other
halides are also very problematic. The alloy 20Cb-3, containing copper, was especially developed for use in sulfuric acid. Clearly, it is important to define the
acid compositions and temperatures and then to utilize the literature for alloy
recommendations. General corrosion and localized attack due to microstructural
problems are the major causes of failure. Much less is known about the SCC
behavior. Organic acids are less corrosive than the mineral acids.
7. Basic Environments
In general, all stainless steels are quite resistant to general corrosion in concentrated caustic solutions, even up to boiling temperatures. However, there is a
severe SCC problem, and the alloys can act in a very brittle manner. Alloys with
the highest nickel content are the most resistant to general attack and SCC. The
threshold for SCC is a function of the temperature, caustic concentration, and
nickel content. The austenitics 304 and 316 are resistant to SCC up to about 60C
in 60% concentrated caustic. At higher temperatures, nickel-based alloys must
be used. Microstructural effects are very important because sensitization makes
the materials more susceptible to SCC. Austenitic structures are more resistant
than ferritic structures.
41
42
Fig. 1
Jones et al.
be significantly changed from the matrix by equilibrium (nonequilibrium) processes resulting in segregation (depletion) of alloying (impurity) elements and
precipitation of second phases. These compositional changes in the grain-boundary region can influence IG crack advance through effects on electrochemical
behavior (e.g., dissolution, repassivation, and hydrogen recombination) as well
as effects on interfacial mechanical behavior (e.g., deformation and cohesive
strength).
B. Precipitation and Grain-Boundary
Composition Changes
The dominant material variable controlling IGSCC susceptibility in austenitic
stainless steels results from the precipitation of Cr-rich M23C6 carbides at highenergy interfaces. This promotes the development of a Cr-depleted region adjacent to carbide precipitates. This depletion is controlled by the thermodynamics
of carbide formation and differences between the diffusivities of Cr and C.
ATEMEDS has enabled Cr-depletion profiles to be routinely measured demonstrating that interfacial Cr concentrations decrease (from 18% to 10%) as the
heat-treatment temperature is decreased due to changes in C and Cr activities. The
width of the depleted zone increases with time after IG carbides are nucleated.
The extent of grain-boundary Cr depletion has been directly linked to the
IG corrosion and SCC susceptibility of austenitic stainless steels (912). The
threshold concentration to promote IG degradation can be quite different for cor-
43
44
Jones et al.
45
46
Jones et al.
V.
47
The factors controlling crack initiation are expected to be similar to those that
control stress-corrosion cracking, namely environment, stress, and material microstructure and microchemistry. However, surface features resulting from fabrication or heat treatment contribute to initiation of cracks but are not a factor in
crack growth. Examples of features contributing to stress-corrosion crack initiation in stainless steels have been given by Jones (25), Clarke and Gordon (26),
and Licina and Giannuzzi (27). Jones showed a surface feature resulting from
surface grinding in preparation for welding of type 304 SS that would allow
crevice/occluded cell conditions to occur. Clarke and Gordon found that interface
fracture between the matrix and titanium carbonitrides and dissolved silicates
and sulfides were potential crack initiation sites in type 304 SS tested in hightemperature water. Licina and Giannuzzi also showed several examples of surface
irregularities on ground surfaces of type 304 SS, which would act as crevices/
occluded cells and hence crack initiation sites. Therefore, surface grinding can
produce defects that may accelerate crack initiation, although crack initiation will
eventually occur at inclusions or grain boundaries if the environment/material/
stress conditions are within a cracking regime.
There have been a number of studies aimed at measuring crack initiation.
Fatigue crack initiation has been studied extensively, with several studies showing cracks emanating from pits (28). Under stress-corrosion conditions, Clarke
and Gordon (26) studied the sites at which cracks initiated in type 304 SS in
high-temperature water. They observed that fractures of titanium carbonitride
matrix interface and dissolved silicates and sulfides were sites for crack initiation
when these sites were on grain boundaries. Stewart et al. (29) have correlated
electrochemical transients with intergranular crack initiation of type 304 SS. They
performed slow-strain-rate tests in dilute thiosulfate and found numerous electrochemical current transients that the authors correlated with intergranular crack
advance prior to the advance of a single dominate crack. The authors estimated
that each current transient was associated with crack advance equal to about one
grain diameter and that current decay occurred because of crack arrest.
Isaacs (30) used an in situ scanning vibrating electrode technique to identify
locations of electrochemical potential variation on the surface of type 304 SS
stressed in tension to an unknown level in 10-ppm thiosulfate solution at room
temperature. The in situ probe was capable of detecting currents emanating from
growing stress-corrosion cracks, and it was shown that several cracks initiated
and repassivated prior to a dominate crack initiated and continued to propagate.
This result is very consistent with those of Stewart et al. and those obtained with
acoustic emission (25). Locci et al. (31) measured the elongation of samples
loaded uniaxially in a simple beam apparatus used for creep tests. They measured
the crack initiation behavior of ferritic stainless steels in chloride solutions using
48
Jones et al.
this apparatus and found the elongation versus time curve correlated with the
growth of corrosion trenches to sharp cracks. The initiation period correlated
with the transition from flat-bottomed corrosion trenches to 1020-m-long sharp
cracks emanating from the corrosion trenches in a time period of 1050 min for
a 25% Cr ferritic stainless steel loaded to 90% of its yield strength in boiling
42% LiCl at a potential of 460 mVSCE.
49
Fig. 6 Stress-corrosion experience for Type 304 and 316 in various plant environments.
ductivity of the solution, the lower the crack penetration rate (33). Factors such
as oxygen content, chloride content, and temperature significantly affect solution
conductivity and are also difficult to consistently maintain in a plant environment.
The need for greater pitting and crevice corrosion resistance of austenitic
stainless steels has led to the development of a new stainless-steel family referred
Fig. 7 Effect of solution conductivity on crack penetration rate in Type 304 stainlesssteel piping.
50
Jones et al.
Grade
S31254
N08366
N08367
S30403
S31603
N08904
254 SMO
AL-6X
AL6XN
Type 304L
Type 316L
Alloy 904L
90.5
63.5
90.5
27.5
27.5
32
32.5
17.5
32.5
2.5
2.5
0
51
Cr and Mo additions have found even wider use in the chemical process industry
in heat exchangers and other equipment. However, one of the major disadvantages of these alloys is their low strength and non-heat-treatability. Even though
the ferritic stainless steels are resistant to SCC from chlorides, they are quite
susceptible to pitting and crevice corrosion in environments containing chlorides,
so this family of alloys more often fails from localized corrosion rather than SCC.
52
Jones et al.
REFERENCES
1. ASTM Volume 3.02, Metals Test Methods and Analytical Procedures. West Conshohocken, PA: American Society for Testing Materials, 1998.
2. AJ Sedriks. Corrosion of Stainless Steels, 2nd ed. New York: WileyInterscience,
1996.
3. G Cragnolino. Cracking of austenitic alloys. Int Metals Rev 3:85135, 1979.
4. G Cragnolino, DD Macdonald. Intergranular SCC of austenitic stainless steel at temperatures below 100CA review. Corrosion 38:406424, 1982.
5. DR McIntyre. Experience SurveySCC of Austenitic Stainless Steels in Water. St.
Louis, MO: Materials Technology Institute, 1987.
6. MA Streicher. Analysis of crevice corrosion data from two seawater exposure tests
on stainless steels. Mater Perform 22:3750, 1983.
7. GS Was, SM Bruemmer eds. Grain Boundary Chemistry and Intergranular Fracture,
Materials Science Forum, 1989, p. 46.
8. PL Andresen, FP Ford, Mater Sci Eng A103:167, 1988.
9. SM Bruemmer. Corrosion 98, 1998, Paper 138.
10. SM Bruemmer, BW Arey, LA Charlot, Corrosion. 48(1):42, 1992.
11. SM Bruemmer, BW Arey, LA Charlot. In: RE Gold, EP Simonen, eds. 6th Int.
Symp. Environmental Degradation of Materials in Nuclear Power SystemsWater
Reactors. The Minerals, Metals & Materials Society, 1993, p. 277.
12. SM Bruemmer, LA Charlot, EP Simonen. In: EP Simonen, RE Gold, DE Cubicciotti.
5th Int. Symp. Environmental Degradation of Materials in Nuclear Power Systems
Water Reactors. American Nuclear Society, 1992, p. 821.
13. RH Jones and SM Bruemmer. In: RP Gangloff, MB Ives. Proc. Environment-In-
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
53
3
Nickel-Based Alloys for Resistance
to Aqueous Corrosion
Paul Crook
Haynes International, Kokomo, Indiana
I.
In the world of metallic materials for (aqueous) corrosion resistance, nickel and
its alloys fill the wide performance gap between the stainless steels and the exotic
materials, such as tantalum. Within this same performance band reside the titanium alloys; however, these have more specific uses.
The advantages of many nickel-based alloys relative to the stainless steels
include the following:
1. Much higher resistance to stress corrosion cracking
2. Higher uniform corrosion resistance, especially in reducing acids, such
as hydrochloric, hydrofluoric, and low to moderate concentrations of
sulfuric
3. Higher resistance to localized attack (pitting and crevice corrosion),
particularly in the presence of chlorides
These advantages stem from three attributes of nickel. First, it is more noble
than iron; second, it exhibits a ductile, face-centered-cubic (fcc) structure at all
temperatures in its solid form; third, it has a high tolerance for useful solutes
(alloying additions), such as chromium and molybdenum.
The nickel-based corrosion alloys may be grouped in several ways. For the
purpose of this chapter, however, they are characterized in terms of the major elemental constituents. Representative, wrought compositions from each group (or
55
56
Crook
family) are given in Table 1. From this table, it is evident that the chief alloying
elements used in the corrosion-resistant nickel alloys are chromium, copper, molybdenum, tungsten, iron, and silicon. Of these, copper, molybdenum, and tungsten
are used to enhance the nobility of nickel; chromium and silicon are used to enhance
passivation; iron is used either to lower the cost of the alloys or to shift positions on
alloy phase diagrams, so that deleterious microstructural precipitates are avoided.
The issue of precipitation within the nickel-based corrosion alloys is extremely complex. However, some discussion of the subject is warranted here. It is
generally accepted that, from a corrosion standpoint, the ideal microstructure is one
that consists of a single phase. A primary concern, therefore, during the design
of the nickel-based alloys has been to avoid phases other than fcc. This has been
accomplished not by restricting alloying additions within the soluble range at room
temperature but by ensuring that a (metastable) single-phase microstructure is possible at room temperature through a process of annealing and quenching. In other
words, the important factors during design have been the high-temperature solubilities and the kinetics of the intermediate-temperature precipitation reactions.
As a result of this approach, it is possible for secondary precipitates to form
during subsequent excursions to intermediate temperatures (e.g., in heat-affected
zones during welding). These precipitates typically form at the grain boundaries
of wrought alloys, as they represent ideal nucleation sites. The extent to which
these precipitates affect the corrosion resistance and the mechanical properties
is related to their atomic structure, shape, composition, and how they influence
the composition of the surrounding fcc solid solution.
To minimize these precipitation reactions, minor elements such as carbon
and silicon are generally held at low levels in the nickel-based wrought alloys.
Silicon is a problem, however, in most nickel-based cast alloys for corrosion
service, because it is necessary for fluidity. An added problem with castings is
elemental segregation (inhomogeneity).
This chapter concerns mostly the performance of the wrought, nickel-based
alloys. These are typically electric-arc melted, then refined by argon-oxygen decarburization. In many cases, they are remelted (e.g., by the electroslag process) for
further refinement and to optimize the structure of the ingot for further processing.
Wrought products such as bars and plates are normally made by hot forging
and/or hot rolling. Sheets are generally made from hot-rolled coils and are typically cold finished prior to final annealing, to achieve tight tolerances and to
control grain sizes.
Ni alloys
Nickel 200
NiCr alloys
Alloy 625
NiCu alloys
Alloy 400
Alloy K-500
NiMo alloys
Alloy B-2
B-3 alloy
NiCrMo alloys
Alloy C-4
C-22 alloy
Alloy C-276
C-2000 alloy
Alloy 59
Alloy 686
NiCrSi alloys
D-205 alloy
NiFeCr alloys
Alloy 825
Alloy G-3
G-30 alloy
Cr
99.0 min.
Cu
Mo
0.25 max.
9.0
Fe
Mn
max.
0.40 max. 0.35 max. 0.35 max. 0.15
5.0 max.
max.
2.50 max. 0.50 max. 2.00 max. 0.3
max. 2.7
2.00 max. 0.50 max. 1.50 max. 0.25
61.0
21.5
66.5
66.5
31.5
29.5
69
65
28
2 max.
1 max.
1.5
1.5
0.2 max. 28.5 3 max.
65
56
57
58.5
Bal.
Bal.
16
22
16
23
23
21
1.6
65
20
2.0
2.5
42
44
21.5
22
2.2
2.5
3
7
43
30
1.7
5.5
Si
16
13 3
16 4
16
15.75
61
3.7
3 max.
3
5
3 max.
1.5 max.
5 max.
0.6
max.
0.03
30
1.5 max. 19.5
0.5 max.
1 max.
1 max.
1 max.
2.5
1 max.
1.5 max.
max.
0.03 max.
0.9
Zr
0.7 max.
0.01 max.
0.01 max.
0.01 max.
0.01 max. 0.5 max.
15
0.50 max. 0.50 max. 0.10max. 0.40 max. 0.40 max. 3.6
Cb Ta
Ti
Al
0.35 max.
0.35 max.
0.5 max.
0.8
Table 1
57
58
Crook
in Table 1. Several other grades, with tighter limits on the residual elements, are
also available for specific uses (e.g., where enhanced physical or mechanical
properties are needed).
A review of the physical and mechanical properties of annealed Nickel 200
in Table 2 reveals that this group of materials is characterized by moderate density, a high melting range, low tensile strength, and moderately high-tensile elongation. Fabrication (forming and welding) of commercially pure nickel presents
no difficulties, and Nickel 200 is covered by the ASME Boiler Code.
A primary use of commercially pure nickel is hardware for handling caustic
soda (sodium hydroxide), as it resists practically all concentrations and temperatures of this compound. For example, Nickel 200 is favored for the construction
of evaporators, shell and tube heat exchangers, pumps, crystallizers, valves, and
fittings used in the concentration and handling of caustic soda (1). Commercially
pure nickel is also suitable for caustic potash (potassium hydroxide) service.
With regard to the performance of commercially pure nickel in sulfuric acid,
the principal applications of Nickel 200 are at room temperature, in unaerated solutions, or where an organic inhibitor is present (2). In hydrochloric acid, Nickel 200
is useful only at room temperature in air-free solutions, up to a concentration of 10
wt%; corrosion rates at higher temperatures are generally unacceptable (3).
B. NiCr Alloys
This group of materials may also be considered a subset within the NiCrMo
family. However, they have significantly lower molybdenum contents, so are not
quite as resistant to nonoxidizing media and localized attack (pitting, crevice
corrosion, and underdeposit corrosion) as the standard NiCrMo alloys.
Alloy 625 was developed for use as both an aqueous corrosion-resistant
material and as an alloy for use in gaseous, high-temperature environments. One
of its attributes is moderately high strength, as compared with the NiCrMo
alloys, allowing the use of thinner structures. The strength level can be controlled,
to some extent, by varying the annealing temperature. The strength of Alloy 625
is a result of the presence of niobium (columbium) in the solid solution. Alloy
625 is also slightly age-hardenable, by virtue of the sluggish precipitation of the
intermetallic Ni3Cb, although the alloy is not normally used in this condition.
Alloy 625 is very popular in the marine and off-shore industries because
of its high resistance to localized attack in seawater, as compared with many
stainless steels. Undersea applications include exhaust ducts for Navy utility
boats and sheathing for communication cables. Alloy 625 is covered by the
ASME Boiler Code.
C. NiCu Alloys
Copper is very soluble in nickel (in fact, they are mutually soluble in all proportions) and enhances its nobility. The basic NiCu alloys, such as Alloy 400, are
Physical and Typical Room-Temperature Mechanical Properties of Representative Corrosion-Resistant Nickel-Based Alloys
Density
Ni alloys
Nickel 200
NiCr alloys
Alloy 625
NiCu alloys
Alloy 400
Alloy K-500a
NiMo alloys
Alloy B-2
B-3 Alloy
NiCrMo alloys
Alloy C-4
C-22 alloy
Alloy C-276
C-2000 alloy
Alloy 59
Alloy 686
NiCrSi alloys
D-205 alloy
NiFeCr alloys
Alloy 825
Alloy G-3
G-30 alloy
0.2% Offset
yield strength
Tensile
elongation
(%)
g/cm3
lb/in.3
8.89
0.321
14351446
26152635
379552
5580
103207
1530
4055
8.44
0.305
12881349
23502460
8271034
120150
414655
6095
3060
8.83
8.47
0.319
0.306
12991349
13161349
23702460
24002460
483621
8961138
7090
130165
172345
586827
2550
85120
3560
2035
9.22
9.22
0.333
0.333
13701418
25002585
896917
862883
130133
125128
400414
400421
5860
5861
5561
5358
8.64
8.69
8.89
8.50
8.60
8.73
0.312
0.314
0.321
0.307
0.311
0.315
13571399
13231371
24752550
24152500
13101360
13381380
23902480
24402516
765807
765800
786793
752779
690 min.
722848
111117
111116
114115
109113
100 min.
105123
338421
359407
359365
345393
340 min.
359421
4961
5259
5253
5057
49 min.
5261
5263
5770
5961
6268
40 min.
5671
7.99
0.288
11711299
21402370
786 (sheet)
114 (sheet)
338 (sheet)
49 (sheet)
57 (sheet)
8.14
8.30
8.22
0.294
0.300
0.297
13711399
25002550
586724
689 (plate)
676689
85105
100 (plate)
98100
241448
310 (plate)
310352
3565
45 (plate)
4551
3050
58 (plate)
5565
MPa
ksi
MPa
ksi
59
Melting range
Table 2
60
Crook
therefore ductile, single-phase materials with excellent resistance to non-oxidizing media, such as dilute hydrochloric acid (at low temperatures), hydrofluoric
acid, and dilute sulfuric acid.
For higher strength levels, age-hardenable NiCu alloys, such as Alloy K500, are available. These contain aluminum and titanium to encourage the controlled precipitation of submicroscopic particles of the intermetallic Ni3(Ti, Al).
As may be deduced from Table 2, Alloy K-500 in the aged condition has about
twice the strength of Alloy 400.
With regard to applications of the NiCu alloys, these are wide and varied,
particularly within the chemical process industries and marine engineering. The
chemical process industry uses include pressure vessels (Alloy 400 is covered
by the ASME Boiler Code), heat exchangers, valves, and pumps. Marine engineering uses include propellers and pumps because the NiCu alloys possess
high resistance to degradation in both seawater and brackish water under highvelocity conditions (4). Typical applications of the age-hardenable Alloy K-500
include pump shafts, blades, and scrapers.
D. NiMo Alloys
The first NiMo alloy (known as Alloy B) was developed in the 1920s for use
in cast form. As such, it was limited in performance by elemental segregation
and by precipitates induced by significant carbon and silicon contents. Alloy B
contained 28 wt% molybdenum and 5 wt% iron, a combination which fortuitously
placed the alloy in a fairly safe phase field (i.e., the precipitation of Ni4Mo was
avoided).
B-2 and B-3 alloys (Table 1) are the modern wrought equivalents. In Alloy
B-2, the carbon, silicon, and iron levels were reduced to enhance corrosion resistance, although the reduction in iron caused the alloy to fall in the phase
field, where is the fcc phase and corresponds to the ordered intermetallic
compound Ni4Mo. This phase is extremely deleterious because it forms rapidly
in the temperature range 550800C (especially in cold-worked microstructures)
and reduces both ductility and resistance to stress-corrosion cracking.
By the deliberate addition of minor elements within specific narrow ranges,
the precipitation of Ni4Mo has been avoided in B-3 alloy while maintaining the
very high corrosion resistance (5). Instead, the phase, Ni3Mo, is the stable precipitate. This ordered intermetallic compound takes considerably longer to form,
as it requires more diffusion of molybdenum. By virtue of this slower precipitation reaction the B-3 alloy is much more forgiving of slow cooling, following
hot forging, hot rolling, and annealing, and it is much more tolerant of elevated
temperature excursions during welding.
A study of the physical and mechanical properties of the NiMo alloys
(Table 2) reveals that they possess moderately high tensile strengths, high ductili-
61
ties, and high melting ranges. With regard to welding and fabrication, care and
knowledge of the precipitation propensities are necessary to achieve good results.
Both B-2 and B-3 alloys are covered by the ASME Boiler Code.
The applications of the NiMo alloys are very specific and related to pure
solutions of nonoxidizing acids (organic and inorganic). B-2 and B-3 alloys, for
example, are resistant to pure sulfuric and hydrochloric acids at nearly all concentrations and temperatures, up to the boiling points. One of the most important
applications, in recent years, has been hardware for making and handling acetic
acid.
E.
NiCrMo Alloys
The NiCrMo alloys are the multipurpose materials of the chemical process
and allied industries. They combine the benefits of molybdenum in nickel with
the advantages of passivity due to chromium. Whereas the NiMo alloys are
unsuitable in the presence of oxidizing contaminants, such as ferric and cupric
ions, and dissolved oxygen, the NiCrMo alloys can cope with such conditions.
They are also extremely resistant to localized attack (pitting, crevice corrosion,
and underdeposit corrosion) in the presence of chlorides.
From Table 1, it is evident that the chromium contents of the modern
wrought NiCrMo alloys range from 16 to 23 wt%, and the molybdenum levels
range from 13 to 16 wt%, in some cases augmented by tungsten. Other deliberate
additions can include iron, to allow the use of less expensive raw materials or
to alter positions within the NiCrMo alloy phase field, copper, to enhance
resistance to nonoxidizing acids, aluminum (for the control of oxygen), manganese (for the control of sulfur), and elements capable of tying up residual carbon,
such as titanium and vanadium.
Although all the NiCrMo alloys are versatile, they each have specific
attributes which are taken into account during alloy selection. The high-chromium alloys (C-22, C-2000, and 59), for example, possess much higher resistance
to oxidizing media than the low-chromium alloys. Also, a high combined molybdenum plus tungsten content (C-276 and 686) is beneficial in nonoxidizing acids,
as is the presence of copper (C-2000). Alloy C-4 has the highest thermal stability
(resistance to sensitization).
From Table 2, it is evident that the NiCrMo alloys possess high ductility,
which is helpful in fabricating components. They are also of moderate strength
in the annealed condition, thus limiting the thicknesses required for pressure
containment. All but the most recently developed NiCrMo alloys are covered
by the ASME Boiler Code, and applications are in process for the newest materials.
With regard to the general performance of the NiCrMo alloys, they are
resistant to a wide range of chemicals, even in the presence of chlorides. They
62
Crook
are generally suitable for use in sulfuric acid up to moderate temperatures [e.g.,
up to 100C in the case of C-2000 alloy (except in the concentration range 75
85 wt%)] (6). The NiCrMo alloys are also suitable for use in hydrochloric
acid, although the concentration and temperature limitations are more severe.
However, several of these alloys provide high resistance to dilute (less than 5
wt%) hydrochloric acid, in some cases up to the boiling points.
The NiCrMo alloys are among those materials able to handle hydrofluoric acid. Also, they withstand organic acids and alkaline media, such as sodium
hydroxide.
With regard to industrial applications of the NiCrMo alloys, these are
many and varied; however, they include reaction vessels, piping, heat exchangers,
tanks, valves, nozzles, and pumps. In the power industry, the NiCrMo alloys
have become the premier materials for lining flue gas desulfurization ducts. Also,
in the oil and gas industry, C-276 is used for pipework in some of the harshest
downhole conditions.
F.
NiCrSi Alloys
D-205 alloy, the composition of which is given in Table 1, is the sole representative of this alloy group. It was developed as an alternate to the high-silicon stainless steels, which also possess outstanding resistance to superoxidizing media,
such as concentrated commercial sulfuric acid.
The advantages of nickel over iron as a basis for a high-silicon alloy include
lower work-hardening rates (which are important in the pressing of heat-exchanger plates) and freedom from sigma formation during elevated temperature
excursions (instead, a less deleterious intermetallic, Ni3Si, which can actually be
used to age-harden the material, forms in D-205 alloy).
In the annealed condition, D-205 alloy is characterized by moderate
strength and high ductility. Unlike the other wrought nickel-based alloys described in this chapter, however, it is not suitable for use in the as-welded condition because of a continuous, brittle, silicon-rich eutectic phase which forms in
the weld metal. The tensile elongation of D-205 weld metal (applied using the
gas tungsten arc process) is only about 3%. Annealing at 1040C (1900F) is
necessary to provide sufficient ductility to these welds for most industrial uses.
Table 2 indicates that D-205 alloy also exhibits a relatively low density, compared
with the other nickel-based alloys, and a low solidus temperature. This low solidus limits the annealing temperature range.
So far, the predominant use of D-205 alloy has been in the form of plate
heat exchangers for the cooling of hot, concentrated sulfuric acid, where it has
replaced cast iron (cascade coolers) and anodically protected stainless steel (shell
and tube heat exchangers).
G.
63
NiFeCr Alloys
Although grouped together because of their major alloying elements, Alloy 825,
Alloy G-3, and G-30 alloy (Table 1) have distinctive characteristics and are suitable for different market segments.
Alloys 825 and G-3 were designed as multipurpose materials to fill the gap
between the high-nickel stainless steels (which are more prone to stress-corrosion
cracking) and the NiCr and NiCrMo alloys. Both 825 and G-3 offer good
resistance to sulfuric acid, by virtue of their combined molybdenum and copper
contents. With its higher molybdenum level, Alloy G-3 is significantly more resistant to pitting and crevice corrosion in the presence of chlorides.
G-30 alloy was designed specifically for service in commercial phosphoric
acid, which is the primary chemical in the agrichemical industries. For this type of
service, a high chromium content was found to be desirable, in addition to copper
and molybdenum. As a result of its high chromium content, G-30 alloy is also
suitable for strong, oxidizing media, such as nitric-acid-based pickling solutions.
A review of their respective physical and mechanical properties (Table 2)
reveals that the NiFeCr alloys are characterized by moderate strengths, in the
annealed condition, and high ductilities. All three alloys chosen as representatives
of this category are covered by the ASME Boiler Code.
Applications of Alloy 825 include equipment for containing and handling
sulfuric acid sludges in petroleum refineries (i.e., tanks, heat exchangers, piping,
valves, and pumps) and for petrochemical processes which employ phosphoric
acid as a catalyst. The main use of G-30 alloy in the agrichemical industries has
been for shell and tube heat exchangers.
III. NICKEL ALLOYS IN SULFURIC ACID
Sulfuric acid is one of the most important industrial chemicals and pervades not
only the chemical process industries but also the mining/metal extraction, metal
finishing, and agrichemical industries.
The performance of the nickel-based alloys in sulfuric acid is very much
dependent on acid concentration, temperature, and the presence of other species
in the solution.
In Ref. 7, three concentration regimes are defined as follows:
1. Low-concentration acid
2. Intermediate-concentration acid
3. High-concentration acid
By compiling and comparing data for many nickel-based alloys and stainless
steels, it was deduced that the most beneficial elements in low-concentration acid
64
Crook
are nickel itself, molybdenum (or tungsten), and chromium. At low concentrations, sulfuric acid is nonoxidizing (reducing), and the cathodic reaction is hydrogen evolution.
The intermediate-concentration range is defined as approximately 2060
wt%. Here too, the cathodic reaction is believed to be hydrogen evolution; hence,
the solutions are nonoxidizing. In this concentration range, the NiMo alloys are
outstanding, so it is presumed that molybdenum is one of the most beneficial
alloying elements. Reference 7 also infers that copper and silicon are beneficial
alloying elements within this concentration range.
The high-concentration acid is complex, both in electrochemical terms (the
redox potential exhibits a steep increase once the concentration exceeds about 70
wt%) and in terms of the influence of the elements on performance. Certainly, the
NiMo alloys continue to resist corrosion at these high concentrations, and silicon
can become very beneficial above about 90 wt%. However, the corrosion performance of the NiCu alloys deteriorates markedly as the redox potential increases.
A review of the data in Ref. 7 leads to the following conclusions:
1. Of the nickel alloys, those based on the NiMo system are the most
resistant to pure sulfuric acid, at concentrations up to about 95 wt%;
however, these should not be used in aerated systems or where other
oxidizing species are present.
2. The NiCrMo and NiFeCr systems constitute the next best choice
at concentrations up to about 95 wt%; these alloys will also resist sulfuric acid in the presence of oxidizing species.
3. In the low- and intermediate-concentration ranges, the NiCu alloys
are also worthy of consideration.
4. At very high concentrations of sulfuric acid, the NiCrSi alloy may
be the most suitable alloy, especially in the presence of impurities.
Iso-corrosion diagrams for B-3 and C-2000 alloys in sulfuric acid are presented in Figs. 1a and 1b, respectively. These indicate the very safe (0 to 5 mpy),
moderately safe (520 mpy), and unsafe (over 20 mpy) concentration/temperature regimes for these alloys in pure sulfuric acid.
For a perspective, it should be stated that the NiCrMo and NiFeCr
alloys are within the same performance band as Alloy 20 (stainless steel) at concentrations up to about 60 wt%; however, they are considerably better than Alloy
20 in the approximate range of 6080 wt%. Relative to 316L stainless steel, the
NiCrMo and NiFeCr possess much higher resistance to sulfuric acid at
concentrations up to about 85 wt%.
A comparison of the NiCrSi alloy (D-205) and a high-silicon stainless
steel (Fe17 Cr20 Ni5 Si) is shown in Fig. 2. This indicates their relative
resistance to commercial-grade sulfuric acid at 130C, in the concentration range
65
(a)
(b)
Fig. 1 (a) Iso-corrosion diagram for B-3 alloy in sulfuric acid; (b) iso-corrosion diagram
for C-2000 alloy in sulfuric acid.
9699 wt%. A plot of corrosion rate versus temperature for D-205 alloy in 99
wt% acid is shown in Fig. 3; this indicates an upper temperature limit of 150C.
66
Fig. 2
Crook
and only a few alloy systems are suitable for use in the acid at temperatures
above ambient. As a result of its volatility, high concentrations are rarely encountered, except in pressurized systems. In a flask/condenser system, for example,
the highest concentration that is stable at the boiling point is 20 wt%.
Many nickel-based alloys can be used in hydrochloric acid at temperatures
Fig. 3 Corrosion rate versus temperature for D-205 alloy in 99% commercial-grade sulfuric acid.
67
(a)
(b)
Fig. 4 (a) Iso-corrosion diagram for B-3 alloy in hydrochloric acid; (b) iso-corrosion
diagram for C-2000 alloy in hydrochloric acid.
at and above ambient. Again, their performance is generally a function of temperature, concentration, and the impurities present. The most useful nickel-based
alloys for hydrochloric acid service are those of the NiMo and NiCrMo systems. As shown in the iso-corrosion diagram for B-3 alloy (Fig. 4a), the NiMo
alloys exhibit low corrosion rates in pure (reagent-grade) hydrochloric acid, in
the concentration range 120 wt%, even up to the boiling points. As in sulfuric
acid, however, the NiMo alloys should not be used in aerated systems or in the
presence of other oxidizing species, such as ferric ions. The same restrictions
68
Crook
apply to the Ni and NiCu alloys, which are useful in low concentrations of
hydrochloric acid at low temperatures.
The NiCrMo alloys not only exhibit low corrosion rates in hydrochloric
acid over reasonably wide concentration and temperature ranges (as shown in
the iso-corrosion diagram for C-2000 alloy, in Fig. 4b), but they also are tolerant
of oxidizing species. However, there are considerable differences in performance
between individual compositions within this alloy system, so attention to existing
corrosion data or field testing are warranted.
V.
Alloy 625
Alloy 400
Alloy B-2
B-3 alloy
C-22 alloy
Alloy C-276
C-2000 alloy
Alloy 825
G-30 alloy
316L stainless steel
904L stainless steel
20Cb-3 alloy
Alloy 255
5% HF
10% HF
20% HF
48% HF
44.7
16.6
11.8
12.4
25.0
15.9
10.8
31.8
34.0
3,877.0
457.0
30.0
802.0
349.0
16.6
13.8
16.3
31.1
19.0
21.5
44.1
89.3
11,043.0
780.0
29.0
1,528.0
934.0
22.4
17.8
20.7
51.6
34.7
20.2
40.8
83.6
17,760.0
2,449.0
37.3
4,682.0
1,576.0
40.4
25.5
30.6
27.0
36.8
19.4
127.0
269.0
21,336.0
59.7
7,247.0
69
steel). From these data, it is evident that the NiCu alloys, the NiMo alloys,
and certain NiCrMo alloys can be used in hydrofluoric acid at this temperature,
but with moderate rates of attack. The NiCr and NiFeCr alloys appear to be
unsuitable at this temperature and within this concentration range, as do the austenitic and ferriticaustenitic stainless steels. At much lower temperatures, the
NiCr and NiFeCr alloys may be used in dilute hydrofluoric acid, as may the
copper-containing stainless steel, 20Cb-3 alloy.
The performance of the NiCu and NiMo alloys in hydrofluoric acid is
strongly influenced by the presence of oxygen, as it is in sulfuric and hydrochloric
acids. Considerable increases in the corrosion rates of Alloy 400 in 38% and
48% hydrofluoric acid (boiling) have been recorded, even in the presence of 500
ppm oxygen (9).
The Ni alloys, such as Nickel 200, are not as resistant as Alloy 400 to
hydrofluoric acid. They are also more strongly affected by oxygen in this acid.
Their use is therefore limited to oxygen-free solutions, at temperatures below
about 79C (175F) (9).
70
Crook
Fig. 5 Corrosion rates for nickel-based alloys and stainless steels in 60% commercial
phosphoric acid at 116C.
71
Fig. 6 Corrosion rates for nickel-based alloys and stainless steels in 65% boiling nitric
acid.
72
Crook
Fig. 7 Critical pitting temperatures for nickel-based alloys and stainless steels in an
oxidizing NaCl/HCl solution.
exhibit moderate resistance to localized attack and are therefore suitable under
mild conditions.
To provide some perspective, pitting and crevice corrosion data for several
nickel-based alloys and stainless steels are presented in Figs. 7 and 8. The values
represent the lowest temperatures at which pitting (Fig. 7) and crevice corrosion
(Fig. 8) occur in a solution of 4 wt% NaCl 0.1 wt% Fe2 (SO4)3 0.01 M HCl.
From these data, it is evident that the temperature required to induce crevice
corrosion is generally lower than that needed for pitting to occur. It is also evident
that the NiCrMo alloys are by far the best nickel-based alloys, with regard to
resistance to localized attack in oxidizing chloride solutions.
73
Fig. 8 Critical crevice corrosion temperatures for nickel-based alloys and stainless steels
in an oxidizing NaCl/HCl solution.
REFERENCES
1. Resistance of Nickel and its Alloys to Corrosion by Caustic Alkalies. Corrosion
Engineering Bulletin 2, The International Nickel Company.
2. Resistance of Nickel and High Nickel Alloys to Corrosion by Sulfuric Acid. Corrosion Engineering Bulletin 1, The International Nickel Company.
3. Resistance of Nickel and High Nickel Alloys to Corrosion by Hydrochloric Acid,
Hydrogen Chloride and Chlorine. Corrosion Engineering Bulletin 3, The International Nickel Company.
4. MONEL NickelCopper Alloys, 3rd ed. The International Nickel Company, 1978.
74
Crook
4
Nickel-Based Alloys for Resistance
to High-Temperature Corrosion
Mark A. Harper
Special Metals Corporation, Huntington, West Virginia
George Y. Lai
Consultant, Carmel, Indiana
I.
INTRODUCTION
76
the underlying alloy are more probable in an oxidizing environment (i.e., high
oxygen activity, excess oxygen) than in a reducing environment (i.e., low oxygen
activity, no excess oxygen). With the formation and growth of oxide scales occurring more slowly under reducing conditions, these types of environments are
usually much more aggressive/corrosive than oxidizing environments.
Regarding the corrosion reaction(s) occurring in an industrial process, it is
important to note that the oxygen activity present in the system is usually an
important influence on the rate of attack. For example, sulfidation attack is controlled by both the sulfur and oxygen activities, with an increase in the oxygen
activity causing a decrease in the sulfidation attack, and vice versa. Carburization
and nitridation behave in a similar manner. However, halogen corrosion, moltensalt corrosion, and molten-metal corrosion behave differently (i.e., oxidizing environments tend to be more corrosive than reducing environments).
Thus, an alloy selection process must take into consideration the nature of
the corrosive environment and the mode of corrosion attack. Once these two
items have been determined, an adequate corrosion database is required in order
for a materials engineer to make an informed decision on the appropriate alloy
selection.
The nominal compositions of several Ni-based wrought alloys used for
high-temperature service are shown in Table 1. A more complete list of alloys
can be found in Ref. 1. As shown in this table, the levels of carbon, and sometimes
nitrogen, in these alloys are relatively high. This is primarily due to the use of
the various metal carbides to provide high-temperature strength. Other elements
such as Mo, Nb, and W are used for solid-solution strengthening. Although not
discussed in this chapter, the high-temperature strength, creep, and thermal fatigue usually play as important a role as the high-temperature corrosion resistance
of the alloy. The remainder of this chapter will discuss the various modes of
high-temperature corrosion and a review of the available data that exist for Nibased alloys in the various types of environment. Using this information, a materials engineer can make a more informed decision on the appropriate alloy selection
for a given application.
II. OXIDATION
Oxidation is the most common and important high-temperature corrosion reaction, and many alloys rely on the protective oxide scale to prevent attack of the
alloy from sulfidation, carburization, and so forth. When the environment contains corrosive impurities (e.g., sulfur, chlorine, etc.), the principal corrosion
mode may no longer be oxidation, even though it may still be part of the overall
corrosion reaction. For this reason, this section considers oxidation primarily associated with air and clean combustion atmospheres generated by using clean
Alloy name
214 alloy
UNS alloy
No.
Cr
Ni
b
Co
Fe
3
0.05
16
75
600 alloy
S alloy
601 alloy
625 alloy
230 alloy
617 alloy
X alloy
263 alloy
HR-160 alloy
HR-120 alloy
RA330 alloy
800HT alloy
556 alloy
N06600
N06601
N06625
N06230
N06617
N06002
N12160
N08120
N08330
N08811
R30556
0.10 a
0.02 a
0.05
0.10 a
0.10
0.07
0.10
0.06
0.05
0.05
0.05
0.08
0.10
15.5
16
23
21
22
22
22
20
28
25
19
21
22
72 c
67 b
Bal
62 b
57 b
Bal
47 b
52 b
37 b
37
35
32.5
20
2a
1a
5a
12.5
1.5
20
30
3a
18
7
3a
14.1
5a
3.0a
1.5
18
0.7 a
3.5 a
33 b
Bal
Bal
31 b
RA85H alloy
S30615
0.20
18.5
14.5
Bal
Mo
Others
3.7 NbTa, 0.5 Mna , 0.5 Sia , 0.4 Ala , 0.4 Tia
2
14
0.5 Mn, 0.4 Si, 0.3 Al, 0.02 La, 0.015 B a
9
1 Al
9
0.6
1 Mn a , 1 Si a , 0.008 B a
6
Table 1
Note: 214, 230, HR-160, HR-120, and 556 are trademarks of Haynes International. RA330 and RA85H are trademarks of Rolled Alloys, Inc. 800HT is a
trademark of the INCO Family of companies.
a
Maximum.
b
As balance.
77
78
fuels (e.g., natural gas or number one and/or number two fuel oil). These fuels
generally contain low levels of contaminants such as sulfur, chlorine, alkali metals, and vanadium.
Many oxidation problems result from using an alloy in a temperature region
that exceeds its protective capability; that is, for a given alloy in service at excessively high temperatures, significant scaling occurs. In response to this problem,
a large database on the oxidation of commercial alloys, ranging from carbon
steels to superalloys, is available in the literature. Although not discussed in this
section, the comparative oxidation resistance of carbon and low-alloy steels, 9
Cr and 12 Cr steels, 17 Cr and 27 Cr stainless steels and austenitic stainless
steels at temperatures between 480C, and 930C is summarized in The Making,
Shaping and Treating of Steel (2). Also, Eiselstein and Skinner (3) summarized
the comparative oxidation resistance of austenitic stainless steels, FeNiCr
alloys, and Ni-based alloys at 980C. Nickel-based alloys performed significantly
better than stainless steels.
Most commercial high-temperature alloys rely on a chromia (Cr 2O3 ) scale
for protection at elevated temperatures and these alloys are often called chromia
formers. However, at temperatures above 1000C, Cr2O3 exhibits significant
volatilization to CrO3 and, thus, diminishes the protective capability of a chromiaforming alloy to the oxidation attack. Several chromia-forming Ni-based alloys
have been found to exhibit good oxidation resistance at 982C (1800F) and
higher. For example, Incos Inconel 601 and 625 alloys and Haynes Internationals Hastelloy S and X alloys, as well as the Haynes 230 alloy, suffered less
than 25 m after 1008 h in air at 982C. For the same conditions at 1093C
(2000F), many alloys (e.g., 600, 230, S, and 617) exhibited less than 50 m of
total attack. Even at 1149C (2100F), several alloys (e.g., 600, 230, S, and 617)
showed less than 100 m of oxidation attack. However, at 1204C (2200F), the
oxidation rate increases significantly for the chromia formers, and only a few
alloys exhibited less than 250 m of attack for 1008 h exposure in air. These
results, along with the results of other Ni-based alloys and a few austenitic stainless steels, are shown in Table 2 (4). At this temperature, alloys that form and
use an alumina (Al2O3) scale for protection are preferred. For example, the 214
alloy exhibited less than 18 m of oxidation attack when exposed to air at 1204C
for approximately 1000 h.
It is important to evaluate an alloys long-term oxidation behavior and potential for breakaway oxidation. Harper et al. (5) conducted long-term oxidation
studies on several Ni-based alloys at elevated temperatures for periods of 1 and
2 years. Figure 1 shows the weight change as a function of exposure time for
the three alloys (HR-120, 800HT, and RA85H) that were exposed to still air at
982C for 720 days. This figure demonstrates the importance of long-term testing
and provides a good example of breakaway oxidation. As shown in this graph,
982C
Alloy
Metal
loss
(m)
Average metal
affected a
(m)
Metal
loss
(m)
Average metal
affected a
(m)
Metal
loss
(m)
214
S
230
617
601
600
X
625
556
800H
RA 330
310
304
316
1.8
4.6
6.4
7.9
13.5
8.1
8.6
8.1
9.9
23.9
10.2
8.9
140.7
314.2
5.1
12.4
18.0
33.3
32.0
22.9
23.9
18.3
26.7
45.5
108.5
28.7
205.7
362.0
2.0
11.2
11.4
16.3
30.7
27.9
37.8
83.1
24.6
136.9
20.8
24.6
Consumed
Consumed
2.0
32.8
32.3
46.5
67.1
41.4
69.1
121.9
65.3
187.7
170.2
57.4
689.4 b
1737.4b
3.8
25.7
58.2
27.4
59.9
43.9
114.3
405.4
236.5
191.0
40.6
75.4
Consumed
Consumed
a
b
1204C
1149C
1093C
Average metal
affected a
(m)
4.1
42.2
87.4
85.1
133.9
72.6
148.1
462.3
295.7
255.0
221.7
112.8
549.9 b
2667 b
Metal
loss
(m)
Average metal
affected a
(m)
5.6
Consumed
115.1
269.5
112.3
129.8
Consumed
Consumed
Consumed
286.3
95.8
201.9
Consumed
Consumed
16.5
805.2 a
201.4
316.7
191.5
213.9
899.2 b
1209 b
3810 b
344.2
209.6
260.4
1725.9 b
3566.2 b
Table 2 Metal Loss and Average Metal Affected for Various Alloys Exposed to Flowing Air for 1008 h and Cycled to Room
Temperature Once a Week
79
80
Fig. 1 Weight change versus time for long-term oxidation in still-air testing of HR-120,
RA85H, and 800HT alloys at 982C for 720 days. (From Ref. 5.)
for exposures up to 180 days (4320 h), all three alloys appear to be equivalent
from a weight change perspective. However, between 180 and 210 days of exposure, the 800HT sample exhibited an accelerated rate of weight loss, characteristic
of breakaway oxidation. The RA85H sample exhibited similar behavior after 360
d of exposure. This breakaway oxidation is a result of the transition from a protective Cr 2O3 scale to a less protective Fe,Ni,Cr-spinel oxide scale (6). Table 3 shows
the results of the testing conducted at 982C for a total exposure period of 720
days. The samples were exposed to still air and were subjected to a thermal cycle
to room temperature once every 30 days. Figure 2 provides an easy comparison
of the alloys exposed in this test and shows the relative amounts of attack that
occur via metal loss (i.e., actual thickness loss of the sample) and internal attack
caused by internal oxidation and/or void formation. Table 1 also shows the results
of similar testing for 360 days of exposure at 1093, 1149, and 1204C, and Figs.
35 show the corresponding graphs of the metal loss plus internal attack. Similar
to the results described above, very few chromia-forming alloys can survive longterm exposure at temperatures approaching 1200C. In contrast, Table 1 and Fig.
5 show the excellent oxidation resistance of an alumina-forming alloy at these
high temperatures.
982C720 days
Alloy
Metal
loss
(mm)
Average metal
affected
(mm)
214
230
617
601
556
HR-160
HR-120
RA85H
800HT
0.00
0.00
0.01
0.02
0.06
0.04
0.16
0.53
0.15
0.24
0.57
0.39
0.42
0.27
1.36
2.03
1093C360 days
Metal
loss
(mm)
Average metal
affected
(mm)
0.05
0.14
0.36
0.09
0.83
0.45
1.13
Metal
loss
(mm)
0.27
1.15
0.54
0.74
0.97
2.04
1.30
Average metal
affected
(mm)
0.28
0.54
0.32
Consumed
0.19
1.11
0.51
1.66
Metal
loss
(mm)
0.86
0.94
1.85
6.29
1.49
1.35
2.41
1.79
Average metal
affected
(mm)
0.02
0.01
1204C360 days
1149C360 days
1.51
2.23
0.70
0.42
Consumed
0.78
Consumed
2.64
6.36
2.95
2.62
6.37
6.39
6.35
Table 3 Metal Loss and Average Metal Affected for Various Alloys Exposed to Still Air and Cycled to Room Temperature Once
Every 30 Days
81
82
Fig. 2 Average metal affected for various alloys exposed to still air at 982C for 720
days. Samples cooled to room temperature once every 30 days.
Fig. 3 Average metal affected for various alloys exposed to still air at 1093C for 360
days. Samples cooled to room temperature once every 30 days.
83
Fig. 4 Average metal affected for various alloys exposed to still air at 1149C for 360
days. Samples cooled to room temperature once every 30 days.
Fig. 5 Average metal affected for various alloys exposed to still air at 1204C for 360
days. Samples cooled to room temperature once every 30 days.
84
Fig. 6 Approximate rate constants as a function of temperature for various oxides. (From
Ref. 7.)
The results of the 214 alloys shown in Table 2 and 3 are not surprising
because alumina formers are known to be more oxidation resistant than chromia
formers. Figure 6 shows a plot of the parabolic rate constants for various oxides
(7). As shown in this figure, the growth of Al2O3 is at least two orders of magnitude slower than the growth of Cr2O3. However, it should be noted that some
studies have shown that these alloys are prone to the formation of internal, randomly distributed voids. In particular, the mechanically alloyed, oxide-dispersion-strengthened (ODS) alloys have been shown to be more susceptible to this
type of void formation than conventional wrought alloys (810).
In addition to the use of alumina-forming alloys at temperatures approaching 1200C, these alloys are finding use in applications/processes where
contamination of the product is a critical issue. An example of this type of application is the furnace equipment used to process electronic components (e.g., semiconductors, capacitors, etc.), glass, and chinaware. In these processes, the major
source of contamination is the oxide spalled off of the furnace components, such
as wire mesh belts, baskets, and fixtures. These components are typically made
from chromia-forming alloys. However, due to the slower growing and more
85
III. SULFIDATION
One of the most common high-temperature corrosion modes responsible for plant
component failures is sulfidation, with two conditions typically responsible for
this type of attack. The first involves attack by a gaseous environment, which
can be either reducing with H 2 S or oxidizing with SO2. The second involves salt
deposits in oxidizing atmospheres with low concentrations of SO2 (less than 1.0%
SO2 ). The first condition will be discussed in this section, and the second type
of attack will be discussed in the section on ash- and salt-deposit corrosion.
One method of approaching the sulfidation problems associated with the
attack by gaseous environments is to segregate the environments into three different types: (1) H 2 H 2S mixtures or sulfur vapor with oxygen activities below the
thermodynamic stability region of Cr 2O3 , thus the sulfides are the stable phases,
(2) reducing, mixed-gas environments containing H2, H2O, CO, CO2 , H2S, and
so forth, and oxygen activities high enough to form Cr 2O3 , and (3) SO2-bearing
environments.
A. Sulfur Vapor and H2 H2S Mixtures
A review of the sulfidation of metals and alloys in sulfur-vapor and H 2 H 2 S
environments has been conducted by Mrowec and Przybylski (12), Mrowec (13),
and Young (14). Most studies have been conducted in sulfur-vapor environments
with sulfur pressures greater than 103 atm and in H 2 H2 S environments with
sulfur partial pressures less than 102 atm. These sulfur potentials and the very
low oxygen activities in the system resulted in the formation of sulfides. Among
the FeCr, NiCr, and CoCr alloy systems studied, a significant difference
in performance was not noted; however, increasing the chromium content within
a particular alloy system generally improved its sulfidation resistance.
One application where the H 2 H 2S mixture is observed is in the gas stream
of hydrotreating units for petroleum refining. Severe corrosion attack of the processing equipment has been reported (1517). The sulfidation behavior of various
alloy systems in H 2 H 2S mixtures has been described by iso-corrosion rate curves
86
87
Metal loss
(mm)
Average metal
affected (mm)
HR-160
HR-120
556
RA85H
253MA
800H
310
0.14
0.59
0.48
0.14
0.44
0.47
Consumed
0.71
1.29
1.30
3.18
3.18
3.18
3.18
800H) (23,24). One example of this behavior is shown in Table 4. These data
were generated from a 500-h exposure to a H 2 25 vol% CH 4 14.8 N 2 4 CO
0.6 CO2 0.6 H 2S atmosphere at 899C (1650F) (25). In iron-based alloys, aluminum has been shown to be beneficial (26) and Santorelli et al. (27) have reported
the sulfidation behavior of two advanced iron-based alumina formers: MA 956
and U.K. Atomic Energys Fecralloy alloy. However, nickel-based alumina formers showed poor sulfidation resistance (28).
C. SO2-Bearing Environments
Most sulfidation studies of SO2-bearing environments have been conducted using
pure SO2 or SO2 O2 mixtures containing high concentrations of SO2. Primarily,
pure metals (e.g., Fe, Ni, and Cr) and binary alloys (e.g., NiCr alloys) have
been studied. An extensive study of NiCr alloys with various amounts of Cr
was conducted by Vasantasree and Hocking (29), with increased amounts of chromium resulting in decreased rates of sulfidation attack. Reviews of the corrosion
behavior of metals and alloys have been published by Kofstad (30). However,
very few data have been published for commercial alloys.
Sulfur furnaces used for the manufacturing of sulfuric acid are the most
typical application where environments containing high levels of SO2 are experienced. In this process, sulfur is used as a feedstock for combustion with excess
air at approximately 1150C. The product gas typically contains about 1015%
SO2, (plus 510% O2, balance N 2 ), which is then converted to SO3 for sulfuric
acid production. One study that examined this type of environment looked at the
behavior of a Type 304 stainless steel and the 556 alloy in an oxidizing environment with and without SO2 present (31). Table 5 summarizes the results of this
work and suggests that oxidation resistance may be at least one criterion to be
used in selecting an alloy for use in an SO2-bearing environment.
88
Table 5 Metal Loss and Maximum Depth of Attack for Type 304 Stainless Steel and
the 556 Alloy Exposed at 982C for 550 h in an Oxidizing Environment With and
Without SO2
Type 304 SS
556 alloy
Test gas
Metal
loss
(mm)
Maximum
depth of attack
(mm)
Metal
loss
(mm)
Maximum
depth of attack
(mm)
Ar5 O2 5 CO2
Ar5 O2 5 CO2 10 SO2
0.31
0.61
0.44
0.61
0.005
0.06
0.056
0.10
The other type of environment where SO2 is usually present, albeit in much
lower quantities than discussed earlier, is in the combustion of sulfur-bearing
fuels such as oil and coal. The combustion flue gas stream produced in a coalor oil-fired boiler typically contains less than 1% SO2. Again, little work has been
published for corrosion occurring in these relatively low level SO2-containing
environments. One study by Viswanathan and Spengler (32) found that a Ni15
Cr alloy suffered more attack in a 0.2% SO2 bal N 2 atmosphere than pure SO2
at 870C, with the addition of oxygen significantly reducing the corrosion rate.
Studies on commercial alloys in environments containing low levels of SO2, particularly those with no excess oxygen, are needed, as this type of atmosphere is
relevant to the localized reducing zones that are frequently developed in some
industrial boilers.
89
tents generally being beneficial also. Steel and Engel (37) studied FeNiCr
alloys with chromium contents between 15 and 35 wt%, and found that chromium
has a definite beneficial effect on alloys containing 225 wt% nickel. Increased
chromium levels were less effective for alloys containing 2645 wt% nickel, and
for alloys containing 4670 wt% nickel, the chromium additions were slightly
detrimental. Nickel is known to reduce the diffusivity of carbon in Fe15 Cr
Ni alloys (38) and increased nickel contents in FeNiCr alloys exhibit improved
carburization resistance (37). Figure 7 shows this effect of nickel on the carburization resistance of FeNiCr alloys. Also, Grabke et al. (39) showed that FeNi
Cr alloys that had a nickel to iron ratio of 4: 1 exhibited maximum carburization
resistance, this being in general agreement with the minimum value for the product of the carbon solubility and diffusivity (40). One other beneficial alloying
element is aluminum, because alumina formers (e.g., the 214 and MA 956 alloys)
have been found to be more carburization resistant than chromia formers (41,42).
An example of this behavior is shown in Table 6, where the mass of carbon
absorbed by the 214 alloy is compared to other chromia-forming alloys when
exposed to a carburizing gas at 1093C for 24 h (43). As shown in this table,
the 214 alloy absorbed approximately one-third the amount of carbon as the best
Fig. 7 Effect of nickel on the carburization resistance of FeNiCr alloys. (From Ref.
37.)
90
Carbon absorption
(mg/cm2 )
214
600
625
230
X
S
304
617
316
800H
330
3.4
9.9
9.9
10.3
10.6
10.6
10.6
11.5
12.0
12.6
12.7
Weight gain
(mg/cm2 )
0.3
10.0
19.0
36.0
91
Fe-, Ni-, or Co-based alloys decomposing into a dust of metal particles, carbon, and sometimes oxides and carbides. Depending on the alloy, maximum attack is observed at approximately 650C, with alloys typically showing signs of
pitting.
Metal dusting has been reported for the following: a 1 Cr0.5 Mo steel in
the waste-heat boiler of a partial oxidation syngas production unit at a chemical
plant (44); stainless steels in the waste-heat boiler of a synthesis gas reactor (45)
and a plant producing gasoline from coal (46); Alloy 800 in the preheater of a
gasifier in a coal gasification plant (47) and in a bypass line in a hydrogen reformer (48); Alloy 600 in the re-former of a natural gassynthetic fuel conversion
plant (49); and for various Fe- and Ni-based alloys in carburizing furnaces (50).
The environments are typically enriched in H 2 and CO.
The fundamentals of metal dusting related to Fe-based alloys have been
explained by Grabke et al. (5153), and a fracture mechanism for metal disintegration during metal dusting has been proposed by Katsman et al. (54). Regarding
Ni-based alloys, Grabke et al. (55) have recently published a study on the metaldusting behavior of several commercial alloys and concluded the following: (1)
any metallic material is susceptible to metal dusting if carbon is possible at an
activity of carbon greater than unity, (2) the mechanism that applies for iron and
low-alloy steels does not apply for nickel and Ni-based alloys, and (3) a protective
oxide scale or surface poison is required for protection. Also, Klower et al. (56)
have found that Ni-based alloys with chromium contents of at least 25 wt%
showed no significant evidence of metal dusting for exposures up to 10,000 h,
thus supporting work that has shown that a protective chromia scale retards a
metal-dusting attack.
Compared to the available data on the carburization resistance of commercial alloys, very little data has been published on their metal-dusting behavior.
Clearly, more studies are needed not only to develop a better understanding of
the mechanism of a metal-dusting attack on Ni-based alloys but also to obtain
engineering data regarding the relative resistance of various alloys in various
metal-dusting-prone environments.
V.
NITRIDATION
All alloys are susceptible, to some degree, to nitridation attack in ammonia-bearing atmospheres at elevated temperatures and this type of environment is common
in the chemical processing industries when ammonia, nitric acid, melamine, and
nylon 6-6 (57,58) are produced. Also, ammonia is widely used in the heat-treating
industry as a nitriding gas for the case hardening of steels.
Similar to carburization, nitridation usually results in the formation of internal nitrides, thus causing an alloy component to become embrittled. When the
92
Alloy
Alloy
base
Nitrogen
absorption
(mg/cm2 )
C-276
230
HR-160
600
625
RA333
601
S
617
214
X
825
800H
556
316
310
304
Nickel
Nickel
Nickel
Nickel
Nickel
Nickel
Nickel
Nickel
Nickel
Nickel
Nickel
Nickel
Iron
Iron
Iron
Iron
Iron
0.7
0.7
0.8
0.8
0.9
1.0
1.1
1.3
1.3
1.5
1.7
2.5
4.3
4.9
6.9
7.4
9.8
Depth of
nitride
penetration
(mm)
0.02
0.03
0.01
0.03
0.01
0.03
0.03
0.03
0.03
0.04
0.04
0.06
0.10
0.09
0.19
0.15
0.21
100% NH 3 in the inlet gas and approximately 30% NH 3 in the exhaust gas.
exposure temperatures are relatively low (e.g., 650C), a surface nitride layer
typically forms (59), with the kinetics of the nitridation process depending on
the usual system parameters (i.e., temperature, ammonia concentration in the gas
phase, and alloy composition). In general, austenitic stainless steels have been
successfully used for the processing equipment in ammonia-bearing environments (58,6062). However, when the environment is too severe for the stainless steels, nickel-based alloys are known to be more nitridation resistant than
iron-based alloys (61). Barnes and Lai (59) studied the nitridation resistance of
a wide range of iron-, nickel-, and cobalt-based alloys, with Table 810 showing the amount of nitrogen absorption and depth of internal nitride penetration
exhibited by the nickel-based alloys when exposed to ammonia for 168 h at
649C, 982C, and 1093C, respectively. Based on this work, Barnes and Lai
constructed a plot showing the amount of nitrogen absorption as a function of
93
Alloy
Alloy
base
Nitrogen
absorption
(mg/cm2 )
214
600
S
601
230
617
HR-160
625
X
RA333
800H
825
316
556
304
310
Nickel
Nickel
Nickel
Nickel
Nickel
Nickel
Nickel
Nickel
Nickel
Nickel
Iron
Nickel
Iron
Iron
Iron
Iron
0.3
0.9
0.9
1.2
1.4
1.5
1.7
2.5
3.2
3.7
4.0
4.3
6.0
6.7
7.3
7.7
Depth of
nitride
penetration
(mm)
0.04
0.12
0.18
0.17
0.12
0.38
0.18
0.17
0.19
0.42
0.28
0.58
0.52
0.37
0.58
0.38
100% NH 3 in the inlet gas and less than 5% NH 3 in the exhaust gas.
nickel plus cobalt in the alloys tested at 982C (see Fig. 8). Clearly, the beneficial
effect of nickel and cobalt on the nitridation resistance of an alloy is evident from
this graph.
Nitrogen-containing atmospheres can also be nitriding. For example, when
exposed to pure nitrogen at 1093C for 900 h, Alloy 600 and Alloy 800 showed
1.85 mm and 3.81 mm of attack, respectively (63). Swaminathan and Lukezich
(64) observed internal nitridation of nickel-based alloys that were exposed to
high-velocity combustion gases generated in a gas turbine. Also, Lai (65) observed internal nitridation in four nickel-based alloys (230, 617, 263, and X) that
were tested in a laboratory dynamic oxidation burner rig at 982C for 1000 h.
Figure 9 shows an optical micrograph of the cross section of each alloy, with
the nitrides highlighted. Certainly, more studies are needed, given that data for
the nitridation attack of commercial alloys in this type of environment are rather
limited and the fact that nitrogen-containing atmospheres are being used more
and more in the production of sintered powder-metallurgy products and other
heat-treating operations.
94
Alloy
Alloy
base
Nitrogen
absorption
(mg/cm2 )
600
214
S
230
617
HR-160
601
625
316
304
X
556
825
RA333
800H
310
Nickel
Nickel
Nickel
Nickel
Nickel
Nickel
Nickel
Nickel
Iron
Iron
Nickel
Iron
Nickel
Nickel
Iron
Iron
0.2
0.2
1.0
1.5
1.9
2.5
2.6
3.3
3.3
3.5
3.8
4.2
5.2
5.2
5.5
9.5
Depth of
nitride
penetration
(mm)
0.00
0.02
0.34
0.39
0.56
0.46
0.58
0.56
0.91
0.58
0.58
0.51
0.58
0.71
0.76
0.79
100% NH 3 in the inlet gas and less than 5% NH 3 in the exhaust gas.
95
Fig. 9 Optical micrographs showing internal oxidation and internal nitridation of four
alloys tested in dynamic oxidation burner rig at 980C for 1000 h with a 30-min thermal
cycling. (a) Alloy 230, blocky nitrides (arrows), fine carbide precipitates were due to aging
during testing; (b) Alloy 617, blocky nitrides (arrows), large needle nitrides and tiny needle
nitrides; (c) Alloy 263, tiny needle nitrides; (d) Alloy X, blocky nitrides. (From Ref. 65.)
96
Chloride
Melting
point (C)
Boiling
point (C)
104 atm
temperature
(C)
FeCl 2
FeCl 3
NiCl 2
CoCl 2
CrCl 2
CrCl 3
CrO2Cl 2
MoCl 5
WCl 5
WCl 6
TiCl 2
TiCl 3
TiCl 4
AlCl 3
676
303
1030
740
820
1150
95
194
240
280
1025
730
23
193
1026
319
987
1025
1300
945
117
268
337
750
137
536
167
607
587
741
611
58
72
11
921
454
38
76
chlorine with the other impurities to form chloride salts. The corrosion that occurs
under these will be discussed in Section VII.
Returning to the subject of volatile halides, the melting and boiling points
of some relevant metal chlorides, as well as the temperatures at which the vapor
pressures of these chlorides reach 104 atm, are shown in Table 11 (71,72). These
data illustrate the relatively high volatility and low melting points of the metal
chlorides.
Nickel and nickel-based alloys are widely used in chlorine-bearing environments. Kane (73) studied the behavior of several commercial iron- and nickelbased alloys in an Ar30% Cl2 atmosphere at temperatures between 400C and
704C and found that higher nickel contents resulted in lower chloridation attack.
Table 12 shows the amount of weight loss experienced by various alloys when
exposed to the above-mentioned atmosphere and temperatures.
The above-discussed data were the result of testing in a chlorine-bearing
environment with no measurable oxygen present. However, in many industrial
environments, Cl 2 and O2 are usually present, and under these conditions, the
formation of both a condensed oxide and the volatile metal halides may occur.
The corrosion behavior of various commercial alloys, especially iron and nickel
based, in O2 Cl 2 environments has been studied by several investigators (74
78). The most notable results of this work has been the observation that alloy
97
400C
500C
600C
704C
Ni-201
600
601
625
617
800
310
304
347
0.2
0.02
0.3
0.7
0.6
6
28
108
215
0.3
5
3
7
7
13
370
1100
Consumed
47101
127180
85200
200270
97
160
215
180
190
890
820
1000
Consumed
Fig. 10 Comparison of the corrosion behavior for the 214, S, and 800H alloys when
exposed to an Ar20 O2 0.25 Cl 2 atmosphere for 400 h at temperatures between 700C
and 1000C. (From Refs. 76 and 77.)
98
Table 13 Metal Loss and Total Depth of Attack for Several Alloys After Exposure to an Ar20 O2 0.25 Cl 2 Atmosphere at 700,
800, 850, 900, and 1000C for 400 h
700C
800C
850C
1000C
900C
Alloy
Metal
loss
(mm)
Total
depth
(mm)
Metal
loss
(mm)
Total
depth
(mm)
Metal
loss
(mm)
Total
depth
(mm)
Metal
loss
(mm)
Total
depth
(mm)
Metal
loss
(mm)
Total
depth
(mm)
214
600
800H
310
S
C-276
0.010
0.025
0.079
0.033
0.010
0.033
0.081
0.046
0.018
0.020
0.023
0.036
0.145
0.066
0.061
0.086
0.046
0.053
0.150
0.071
0.018
0.038
0.031
0.031
0.224
0.163
0.066
0.132
0.097
0.061
0.257
0.175
0.023
0.127
0.043
0.086
0.315
0.300
0.150
0.252
0.191
0.152
0.353
0.320
0.013
0.330
0.203
0.191
0.419
0.419
0.051
0.386
0.424
0.246
0.472
0.450
99
100
101
Fig. 11 Laboratory test results for various alloys coated with a synthetic ash (37.5 mol%
Na 2SO4 , 37.5 K 2SO4 , 25 Fe 2O3 ) and exposed to a synthetic flue gas (80% N 2 15 CO2
4 O2 1 SO2 ) for 50 h at different temperatures. (From Ref. 97.)
102
Metal loss
(mm)
Maximum metal
affected (mm)
0.00
0.24
0.42
0.50
0.55
0.72
0.05
0.28
0.44
0.55
0.62
0.75
Fig. 12 Results of a field rack test in a NaClKClBaCl 2 salt bath at 840C for 1 month.
(From Ref. 111.)
Chemical feedstocks are commonly contaminated with S, Cl, K, and Na impurities, and salts are formed during the high-temperature calcining process. Recuperators used in industrial furnaces can also suffer sulfidation and/or chloride attack.
For example, severe corrosion attack can be experienced by stainless-steel recuperator tubes used in aluminum melting operations. The flue gas stream that exits
from an aluminum remelting furnace typically contains Cl, S, K, Na, and other
impurities as a result of the flux used in the aluminum melting process. These
impurities lead to sulfide and/or chloride salt deposits on the recuperator tubes
103
Table 15 Results of
Laboratory Tests in a NaCl Salt
Batha at 840C for 100 hours
Alloy
188
556
601
214
304
316
X
310
800H
625
RA330
617
230
S
600
Total depth
of attack b
(mm)
0.050
0.060
0.060
0.070
0.080
0.080
0.090
0.100
0.100
0.110
0.110
0.120
0.140
0.160
0.190
and rapid attack of the alloy. In one case, a Type 310 recuperator suffered approximately 3.9 mm of attack after 15 months of service at about 650C.
104
Given that molten salts are usually good fluxing agents which remove oxide
scale from a metal surface, the corrosion reaction proceeds by oxidation of the
alloy followed by dissolution of the oxide in the molten salt. For this reason, the
presence of oxygen and water vapor can accelerate the rate of molten-salt corrosion. This type of corrosion can also take place via mass transfer due to a thermal
gradient in the melt. This mode of attack involves the dissolution of an alloying
element at hot spots and deposition of that element at cooler spots, which can
subsequently result in fouling and plugging in a circulating system.
Chloride salts are commonly used in the heat-treating industry for annealing
and normalizing of steels at temperatures between 760C and 980C. Neutral salt
baths, as they are commonly called, typically consist of one or more of the following chlorides: barium, sodium, and potassium. Compositions of five of the more
common neutral salt baths are as follows (110):
50
50
20
25
21
NaCl50 KCl
KCl50 Na2CO3
NaCl25 KCl55 BaCl 2
NaCl75 BaCl 2
NaCl31 BaCl 2 48 CaCl 2
Jackson and LaChance (110) conducted an extensive study on molten-salt corrosion of cast FeNiCr alloys in a 20 NaCl25 KCl55 BaCl 2 salt bath. In this
study and as typically found in molten chloride salts, they found that the alloys
suffered intergranular attack significantly more than metal loss. Also, the results
showed that resistance to the molten salt increased with decreasing chromium
content and increased nickel content and that the intergranular attack generally
followed grain-boundary carbides. Thus, lowering the carbon content of a given
alloy could significantly improve its molten-salt corrosion resistance.
Lai et al. (111) conducted field testing on various iron-, nickel-, and cobaltbased wrought alloys in a NaClKClBaCl 2 salt bath at 840C for 1 month, with
the results shown in Fig. 12. In contrast to the above-discussed results for cast
alloys, two of the high-nickel alloys (Alloy 600 and Alloy 601) suffered more
corrosion attack than the stainless steels 304 and 310. Results from testing in a
NaCl salt bath at 840C are shown in Table 15 (111,112), and similar to the field
test results, CoNiCrW and FeNiCoCr alloys performed the best. At
lower temperatures, corrosion from molten salts typically decreases. Susskind et
al. (113) studied various alloys in a molten NaClKClMgCl 2 salt bath at temperature between 450C and 500C and found many alloys resistant to molten chlorides (see Table 16).
Nitrate or nitratenitrite salt baths are also used for heat-treating purposes,
with typical salt bath temperatures between 160C and 590C. Applications also
exist for use as a medium for heat transfer or energy storage. Slusser et al. (114)
evaluated the molten-salt corrosion of various alloys in a NaNO3 KNO3 salt bath
105
Maximum penetration
(mm/year)
0
0.08
0.01
0
0.01
0.12
0.03
0.05
0.01
0.05
0.05
0
0.07
at 675C for 336 h. In general, nickel-based alloys performed better than ironbased alloys. However, pure nickel exhibited a rapid rate of corrosion attack.
The corrosion rates of the various alloys plotted as a function of the nickel content
are shown in Fig. 13 (114). Longer-term testing (1920 h) showed corrosion rates
similar to the 336-h tests, except for Alloy 800, as shown by a comparison of
Fig. 13 and Table 17 (114). Also shown in Table 17 is the result of testing at
700C, where corrosion rates became much higher.
Exposure of metals to molten sodium hydroxide (NaOH) results in the formation of metal oxide, sodium oxide, and hydrogen (115). Nickel shows the best
resistance to molten NaOH (116119), particularly low-carbon nickel such as
the Ni 201 alloy (120). The corrosion rates of several nickel-based alloys exposed
to molten NaOH at temperatures between 400C and 680C were reported by
Gregory et al. (119) and are shown in Table 18. Based on these results, molybdenum and silicon were detrimental to the molten NaOH salt corrosion resistance.
Also, molybdenum and iron were found to be selectively removed from nickelbased alloys with less than 90% nickel, leading to the formation of internal voids
(121).
The molten-salt nuclear reactor uses a LiFBeF 2 base salt containing various amounts of UF 4 , ThF 4 , and ZrF 4 , as a fuel salt, and the reactor coolant is a
mixture of NaBF 4 NaF (122). Thus, the corrosion of alloys in molten fluoride
salts has been extensively studied for nuclear applications. The most corrosion-
106
675C
1920 h
700C
720 h
0.41
0.25
0.23
0.48
1.85
0.53
0.99
1.22
1.25
6.60
107
Table 18 Corrosion Rates of Selected Alloys Obtained from Static Tests in Molten
Sodium Hydroxide
Corrosion rate (mm/year)
Alloy
400C
500C
580C
680C
Ni-201
C
D
400
600
301SS
0.023
0.018
0.046
0.028
0.043
0.033
2.540
0.056
0.130
0.060
0.080
0.06
(a)
0.25
0.45
0.13
0.26
0.96
(a)
1.69
1.03
General a
Grain boundary b
15
90
90
45
155
185
130
165
270
15
140
30
120
105
108
crucibles with an argon cover gas and the results are shown in Table 19. Chromium appeared to be detrimental in the nickel-based alloys, but no such effect
was seen in the iron-based alloys.
Molten carbonates are generally less corrosive than molten chlorides or
hydroxides. Coyle et al. (125) evaluated a eutectic sodiumpotassiummagnesium chloride salt (33 NaCl21.5 KCl45.5 MgCl 2 ), a sodium hydroxide salt,
and a eutectic sodiumpotassium carbonate salt (58 Na 2CO3 42 K 2CO3 ) for a
possible heat-transfer and energy-storage medium capable of operating at 900C
for a solar power generation system. Both the NaClKClMgCl 2 and NaOH were
too corrosive for many commercial alloys; however, the Na 2CO3 K 2CO3 showed
promise because of its less aggressive nature. Results of testing in this molten
Alloy
Total depth
of attack a
(mm)
X
214
188
556
600 b
600 b
N
304
316
230
Nickel
800 b
800 b
S
0.12
0.19
0.22
0.26
0.34
0.44
0.51
0.54
0.63
0.77
0.30
0.25
0.8
1.43
109
carbonate salt at 900C for 504 h are shown in Table 20. The NiCrMo Alloy
S was severely corroded; the NiCrFeMo Alloy X performed the best. However, no systematic trend between alloying elements and performance was noted.
REFERENCES
1. GY Lai. High-Temperature Corrosion of Engineering Alloys. Metals Park, OH:
ASM International, 1990.
2. HE McGarrow, ed. The Making, Shaping and Treating of Steel. Pittsburgh, PA:
United States Steel Corporation, 1971, p. 1136.
3. HE Eiselstein, NE Skinner. ASTM STP No. 165. Philadelphia, PA: ASTM, 1954,
p. 162.
4. MF Rothman. Internal Technical Report No. 12034, Haynes International, 1985.
5. MA Harper, JE Barnes, GY Lai. CORROSION/97. Houston, TX: National Association of Corrosion Engineers, 1997, Paper 132.
6. B Gleeson, MA Harper. Oxid Met, Vol. 49, Nos. 3/4, 1998, p. 373.
7. JL Smialek, GM Meier. High-temperature oxidation. In CT Sims, NS Stoloff, WC
Hagel, eds. Superalloys II. New York: John Wiley & Sons, 1987, pp. 293323.
8. RJ Hendricks, KD Sheffler. Materials for advanced turbine enginesProject 3 design, fabrication and evaluation of an oxide dispersion strengthened sheet alloy
combustor liner. NASA CR-17491, February 1984.
9. JL Gonzalez-Carrasco, V Guttmann, H Fattori. Met Trans A 26A: 915, 1995.
10. GY Lai. Haynes International, unpublished research, 1991.
110
11. J Bailey. Volatile Cr contamination reduction in APCVD systems by alloy oxidation engineering. J Electrochem Soc, in press.
12. S Mrowec, K Przybylski. High Temp Mater Proc 6(1,2): 1, 1984.
13. S Mrowec. Oxid Met 44(1/2):177, 1995.
14. DJ Young. Rev High Temp Mater 4(4):299, 1980.
15. G Sorrell, WB Hoyt. Collection and Correlation of High Temperature Hydrogen
Sulfide Corrosion Data. Houston, TX: National Association of Corrosion Engineers, 1956.
16. G Sorrell. Compilation and Correlation of High Temperature Catalytic Reformer
Corrosion Data. Houston, TX: National Association of Corrosion Engineers, 1957.
17. EB Backensto. Corrosion in Catalytic Reforming and Associated Processes. 22nd
Midyear Meeting of APIs Division of Refining, Philadelphia, 1957.
18. EB Backensto, JW Sjoberg. Iso-Corrosion Rate Curves for High temperature Hydrogen-Hydrogen Sulfide. Houston, TX: National Association of Corrosion Engineers, 1958.
19. AO Schaefer. A program to discover materials suitable for service under hostile
conditions obtained in equipment for the gasification of coal and other solid fuels,
Metal Properties Council annual reports, 1976 through 1983.
20. MAH Howes. High temperature corrosion in coal gasification systems. Gas Research Institute Report GRI-8710152, 1987.
21. V Nagarajan, RG Miner, AV Levy. JECS 129(4):782, 1982.
22. JF Norton, M Maier, WT Bakker. Corrosion of candidate heat exchanger alloys
in complex simulated coal gasification atmospheres at 450C. CORROSION/97.
Houston, TX: National Association of Corrosion Engineers, 1997, Paper 144.
23. JF Norton, FG Hodge, GY Lai. High Temperature Materials for Power Engineering.
Netherlands: Kluwer Academic, 1990, p. 167.
24. GY Lai, JF Norton, FG Hodge. The corrosion behavior of a new sulfidation-resistant alloy in a sulfidizing/oxidizing/carburizing atmosphere. Proceedings of the
First International Conference on Heat-Resistant Materials, September 2226,
1991. Metals Park, OH: ASM International, 1991, p. 211.
25. MA Harper, JE Barnes. Haynes International, unpublished research, 1997.
26. RW Bradshaw, RE Scholts, DR Adolphson, Sandia National Laboratories Report
SAND 77-8277, 1977.
27. RL Santorelli, JF Norton, F Bregani. Werkst Korros 41:669, 1990.
28. GY Lai. In: MF Rothman, ed. High Temperature Corrosion in Energy Systems.
Warrendale PA: TMS, 1985, p. 227.
29. V Vasantasree, MG Hocking. Corros Sci. 16:261, 1976.
30. P Kofstad. High Temperature Corrosion. New York: Elsevier Applied Science,
1988.
31. JJ Barnes, GY Lai. CORROSION/90. Houston, TX: National Association of Corrosion Engineers, 1990, Paper 276.
32. R Viswanathan, CJ Spengler. Corrosion 26(1):29, 1970.
33. W Steinkusch. Werkst Korros 30:837, 1979.
34. LH Wolfe. Mater Perform 38, April 1978.
35. RH Kane. CORROSION/83. Houston, TX: National Association of Corrosion Engineers, 1983, Paper 266.
111
36. U Van den Bruck, CM Schillmoller. CORROSION/85. Houston, TX: National Association of Corrosion Engineers, 1985, Paper 23.
37. C Steel, W Engel. AFS Int Cast Metals J 28, September 1981.
38. O Demel, E Keil, P Kostecki. SGAW Report 2538. Studiengesellschaft fur Atoenergie, Osterreichische.
39. HJ Grabke, U Gravenhorst, W Steinkusch. Werkst Korros 27:291, 1976.
40. SK Bose, HJ Grabke. Z Metallk 69:8, 1978.
41. GY Lai. In: MF Rothman, ed. High Temperature Corrosion in Energy Systems.
Warrendale, PA: TMS, 1985, p. 551.
42. RH Kane, GM McColvin, TJ Kelly, JM Davison. CORROSION/84. Houston, TX:
National Association of Corrosion Engineers, 1984, Paper 12.
43. GY Lai, CR Patriarca. Metals Handbook, 9th ed. Metals Park, OH: ASM International, 1987, p. 1311, Vol. 13.
44. GM Tanner. Eng Fail Anal 1(4):289, 1994.
45. F Eberle, RD Wylie. Corrosion 15(12):622t, 1959.
46. WB Hoyt, RH Caughey. Corrosion 15(12):627t, 1959.
47. RA Perkins, WC Coons, FJ Radd. Properties of High Temperature Alloys. Pennington, NJ: Electrochemical Society, 1976.
48. RL Codwell, Corrosion in the Petrochemical Industry. Metals Park, OH: ASM International, 1994, p. 231.
49. ML Holland, HJ de Bruyn. Int J Pres Ves Piping 66:125, 1996.
50. GY Lai. J Metals 37(7):14, 1985.
51. HJ Grabke, R Krajak, JC Nava Paz. Corros Sci 35:1141, 1993.
52. HJ Grabke, CB Bracho-Troconis, EM Muller-Lorenz. Werkst Korros 45:215,
1994.
53. E Pippel, HJ Grabke, S Straub, J Woltersdorf. Steel Res 66:217, 1995.
54. A Katsman, L Klinger, LA Levin, T Werber. In: RY Lin, YA Chang, RG Reddy,
and CT Liu, eds. Design Fundamentals of High Temperature Composites, Intermetallics, and Metal-Ceramic Systems. Warrendale, PA: TMS, 1995, p. 413.
55. HJ Grabke, R Krajak, EM Muller-Lorenz, S Straub. Mater Corros 47:495, 1996.
56. J Klower, HJ Grabke, EM Muller-Lorenz, DC Agarwal. CORROSION/97. Houston, TX: National Association of Corrosion Engineers, 1997, Paper 139.
57. GL Swales, In: I Kirman, JB Marriott, M Merz, PR Sahm, and DP Whittle, eds.
Behavior of High Temperature Alloys in Aggressive Environments, Proceedings
of the Petten International Conference. London: Metals Society, 1980, pp. 45
82.
58. K Rorbo. Environmental Degradation of High Temperature Materials. London: Institution of Metallurgists, 1980, Series 3, No. 13, Vol. 2.
59. JJ Barnes, GY Lai. High temperature nitridation of Fe-, Ni-, and Co-base alloys.
In Proceedings of the TMS/AIME Symposium on Corrosion and Particle Erosion
at High Temperatures. Warrendale, PA: TMS, 1989, p. 617.
60. KM Verma, H Ghosh, JS Rai. Br Corros J 13(4):173, 1978.
61. JJ Moran, JR Mihalism, EN Skinner. Corrosion 17(4):191t, 1961.
62. DW McDowell Jr. Mater Protect 1(7):18, 1962.
63. GD Smith, PJ Bucklin. CORROSION/86. Houston, TX: National Association of
Corrosion Engineers, 1986, Paper 375.
112
64. VP Swaminathan, SJ Lukezich. Degradation of Transition Duct Alloys in gas Turbines. In: Proceedings of ASM 1993 Materials Congress Materials Week 93. Materials Park, OH: ASM International, 1993, pp. 99111.
65. GY Lai. Nitridation attack in a simulated gas turbine combustion environment.
In: D Coutsouradis et al. eds. Materials for Advanced Power Engineering, Part II.
Netherlands, Kluwer Academic, 1994, pp. 12631272.
66. WJ Kroll. Method of Manufacturing Ti and Alloys Therof, U.S. Patent 2205854,
January 1940.
67. WA Henderson. J Metals 16:155, 1964.
68. SM Shelton. In: B Lustman and F Kerze Jr, eds. The Metallurgy of Zirconium.
New York: McGraw-Hill, 1955, p. 59.
69. I Iwasaki, Y Takahasi, H Kahata. Trans SME AIME 243:308, 1966.
70. CL Mantell. Tin New York: Reinhold, 1949.
71. PL Daniel, RA Rapp. In: MG Fontana, and RW Staehle, eds. Advances in Corrosion
Science and Technology. New York: Plenum Press, 1970, Vol. 5, p. 55.
72. O Kubaschewski, E Evans. Metallurgical Thermochemistry, New York: Pergamon
Press, 1958.
73. RH Kane. In: BJ Moritz, WI Pollock, eds. Process Industries Corrosion. Houston,
TX: National Association of Corrosion Engineers, 1986, p. 45.
74. S Baranow, GY Lai, MF Rothman, JM Oh, MJ McNallan, MH Rhee.
CORROSION/84, Paper 16, Houston, TX: National Association of Corrosion Engineers, 1984.
75. JM Oh, MJ McNallan, GY Lai, MF Rothman. Met Trans A 17A:1087, 1986.
76. MH Rhee, MJ McNallan, MF Rothman. In: MF Rothman, ed. High Temperature
Corrosion in Energy Systems. Warrendale, PA: TMS, 1985, p. 483.
77. MJ McNallan, MH Rhee, S Thongtem, T Hensler. CORROSION/85. Houston, TX:
National Association of Corrosion Engineers, 1985, Paper 11.
78. S Thongtem, MJ McNallan, GY Lai. CORROSION/86. Houston, TX: National
Association of Corrosion Engineers, 1986, Paper 372.
79. MH Brown, WB DeLong, JR Auld. Ind Eng Chem 39(7):839, 1949.
80. MK Hossain, JE Rhoades-Brown, SRJ Saunders, K Ball. Proc. U.K. Corrosion/
83, p. 61.
81. J Stringer. Hot corrosion in gas turbines. Battelle Report MCIC-72-08, 1972.
82. J Stringer, RI Jaffee, TF Kearns, eds. High Temperature Corrosion of Aerospace
Alloys. London: Harford House, 1973.
83. JW Fairbanks, I Machlin, eds. Proceedings of the 1974 Gas Turbine Materials in the
Marine Environment Conference, Columbus, OH: Battelle Columbus Laboratories,
1974.
84. Hot Corrosion Problems Associated with Gas Turbines, Philadelphia, PA: ASTM,
1967.
85. AB Hart, AJB Cutler, eds. Deposition and Corrosion in Gas Turbines, London:
Applied Science Publishers, 1973.
86. JA Goebel, FS Pettitt, GW Goward. Met Trans 4:261, 1973.
87. RA Rapp. Corrosion 42(10):568, 1986.
88. RA Rapp, YS Zhang. J Metals 46(12): 4755, 1994.
89. PA Bergman, AM Beltran, CT Sims. Development of hot corrosion-resistant alloys
90.
91.
92.
93.
94.
95.
96.
97.
98.
99.
100.
101.
102.
103.
104.
105.
106.
107.
108.
109.
110.
111.
113
for marine gas turbine service. Summary Report to Marine Engineering Laboratory,
Navy Ship R&D Center, October 1, 1967.
MJ Zetlmeisl, DF Laurence, KJ McCarthy. Mater Perform 41, June 1984.
PA Bergman, CT Sims, AM Beltran. Hot Corrosion Problems Associated with Gas
Turbines. Philadelphia, PA: ASTM, 1967, p. 38.
GY Lai, JE Barnes. A burner rig investigation of the hot corrosion behavior of
several wrought superalloys and intermetallics. ASME Paper 91-GT-21, 1991.
C Cain Jr, W Nelson. J Eng Power Trans. ASME 468, October 1961.
RW Borio, AL Plumley, WR Sylvester. In: RW Bryers, ed. Ash Deposits and Corrosion Due to Impurities in Combustion Gases. New York: Hemisphere Publishing/
McGraw-Hill, 1978, p. 163.
T Flatley, CW Morris, UK Corrosion 83, Birmingham, UK: Institution of Corrosion Science & Technology, 1982, p. 71.
AH Rudd, JM Tanzosh. In: A Armor et al., eds. Proceedings of the 1st International
Conference on Improved Coal-Fired Power Plants, Palo Alto, CA: Electrical Power
Research Institute, 1986, pp. 2145.
J Stringer. Metals Handbook, 9th ed. Corrosion. Materials Park, OH: ASM International, 1987, p. 998, Vol. 13.
WT Reid. In: JM Beer, ed. External Corrosion and DepositsBoilers and Gas
Turbines. New York: American Elsevier, 1971.
DW McDowell Jr, JR Mihalisin. ASME Paper 60-WA-260, 1960.
BF Spafford. UK Corrosion 83. Birmingham, UK: Institution of Corrosion Science & Technology, 1982, p. 67.
GL Swales, DM Ward. CORROSION/79. Houston, TX: National Association of
Corrosion Engineers, 1979, Paper 126.
N Bolt. Proceedings of the 10th International Congress on Metallic Corrosion. New
Delhi: Oxford and IBH Publishing, 1988, p. 3593.
SK Srivastava, GY Lai, DE Fluck. CORROSION/91. Houston, TX: National Association of Corrosion Engineers, 1991, Paper 398.
JA Harris, WG Lipscomb, GD Smith. CORROSION/87. Houston, TX: National
Association of Corrosion Engineers, 1987, Paper 402.
GA Whitlow, PJ Gallagher, SY Lee. CORROSION/89. Houston, TX: National
Association of Corrosion Engineers, 1989, Paper 204.
GY Lai. CORROSION/91. Houston, TX: National Association of Corrosion Engineers, 1991, Paper 249.
JL Blough, GJ Stanko, MT Krawchuk. In Situ Materials Testing in a Waste-toEnergy Power Plant, 2nd International Conference on Corrosion in Advanced
Power Plants, Tampa, FL, 1997.
HH Krause, PW Cover, WE Berry. Proceedings of the 1st Conference on Advanced
Materials for Alternative Fuel Capable for Use in Directly Fired Heat Engines.
Castine, ME: Maine Maritime Academy, 1979.
HH Krause. CORROSION/91. Houston, TX: National Association of Corrosion
Engineers, 1991, Paper 242.
JH Jackson, MH LaChance. Trans ASM 46:157, 1954.
GY Lai, MF Rothman, DE Fluck. CORROSION/85. Houston, TX: National Association of Corrosion Engineers, 1985, Paper 14.
114
5
Corrosion of Copper and Its Alloys
Andrew James Brock
Metals Research Laboratories, Olin Corporation, New Haven,
Connecticut
I.
INTRODUCTION
Copper and its alloys are some of the earliest metals known to man. The use of
copper was recorded in northern Iraq as early as 8500 b.c. and in Egypt in 7000
b.c. Copper articles found in the Sinai Peninsula have been dated as being made
in 4000 b.c. The ease with which copper oxide ores can be reduced to copper,
together with the ability of the metal to alloy advantageously with other elements,
resulted in copper playing a critical role in the development of civilization.
The present wide use of copper results from a combination of good corrosion resistance in a multitude of environments, together with other desirable properties. These include its high electric and thermal conductivities, its mechanical
properties, and its ease of being formed. Not the least of its uses has been for
decorative purposes where the attractive color of the pure element and of alloys
such as the brasses, together with their ease of accepting a polish, have considerable aesthetic appeal. Copper and its alloys find other decorative uses such as
for the roofs or domes on churches and other institutional buildings. In this case,
slow corrosion of the metal at less than 7.5 mils in every 100 years is accompanied by the formation of characteristic green coatings or verdigris. Perhaps the
classic example of this is the verdigris on the Statue of Liberty. In other natural
environments, such as freshwater and saline waters and soil, copper and copper
alloys have high corrosion resistance. They are extensively used for underground
water lines, for plumbing tubes, and for condenser tubes in the power utility
industry.
Copper is a noble metal in that its electrochemical potential is above that
of hydrogen. Accordingly, it does not discharge hydrogen from nonoxidizing
115
116
Brock
117
Table 1 Types and Alloying Constituents in Selected Wrought and Cast Copper
Alloys
Alloy
Wrought
Coppers
High coppers
Brasses
Leaded brasses
Tin brasses
Phosphor bronzes
Aluminum bronzes
Silicon bronzes
Coppernickels
Cast
Coppers
High coppers
Brasses, leaded brasses, and
Manganese bronzes
Manganese and leaded
Silicon brasses and bronzes
Tin bronzes and leaded tin bronzes
Nickeltin bronzes
Aluminum bronzes
Coppernickels
Nickelsilvers
UNS No.
Alloying elements
C10100C15760
C16200C10600
C20500C28580
C31200C38590
C40400C49080
C50100C52400
C60600C64400
C64700C66100
C70000C79900
99.3% Cu
96% Cu
Zn
ZnPb
ZnSnPb
SnP
AlNiFeSiSn
SiSn
NiZn
C80100C81100
C87300C82800
C83300C85800
99% Cu
94% Cu
ZnSnPb
C86100C86800
C87300C87900
C90200C94500
C94700C94900
C90520C95800
C96200C96800
C97300C97800
ZnMnFePb
ZnSnSi
SnZnPb
NiSnZnPb
AlFeNi
NiFe
NiZnPbSn
sure in environments such as certain soils can result in similar etched surfaces
but only after several years.
B. Pitting
In certain environments, the corrosion of copper alloys can result in pitting. In
some cases, pitting occurs over the entire exposed surface of the alloy; in others,
pits are formed at discrete locations. In either case, the formation of pits is undesirable because it can lead to structural weakening of the alloy, or perforation in
the case of tubes or vessels.
Pitting is not the predominant type of attack seen with copper alloys. Environments which typically promote pitting are specific and these will be discussed
in later sections. Typically, these environments include sulfide-polluted seawater,
water under conditions of stagnation, and certain potable waters with very specific
chemistries. The pits often do not continue to grow with time, but attain a certain
depth beyond which further increases in depth do not occur. Alloys most resistant
118
Brock
to pitting include the high-copper alloys and tin bronzes. Those least resistant to
pitting are the low-zinc brasses and aluminum bronzes, whereas the copper
nickels and tin bronzes have intermediate resistance.
C. Crevice Corrosion
This is a form of attack which can occur where there is a crevice on the exposed
alloy surface. Such a crevice can be formed between two copper alloy surfaces
or between the alloy and a nonmetallic surface. The latter includes deposits on
the metal surface. In the case of copper-alloy condenser tubes, such deposits
could be fragments of shells or vegetation or localized deposits of silt. Cleaning
of the condenser tubes is one method of eliminating this form of attack.
Classically, two types of crevice corrosion are known. In one type, the area
inside the crevice is preferentially attacked because it can become depleted in
oxygen with respect to the environment outside. The crevice then serves to isolate
the anode reaction of alloy dissolution inside the crevice from the cathodic reduction of oxygen outside the crevice. Attack is then essentially due to differential
aeration of the metal inside the crevice and that outside it. In the second type of
crevice corrosion, initial corrosion of the alloy within the crevice results in an
increase in the concentration of metal ions in the solution. On the alloy surface
immediately outside the crevice, the metal ions can readily diffuse away or be
swept away by the flow of the environment. The overall results is that the areas
outside of the crevice suffer dissolution and the metal inside the crevice is ennobled because of the higher metal ion concentration. This form of attack is due
to the formation of a metal ion concentration cell. It is the type of crevice corrosion which is more typical of copper alloys. As with pitting, crevice corrosion
is a statistical phenomenon. In waters, it is promoted by high temperatures and
by the flow of the environment outside the crevice. The depth of attack is usually
less than that seen with pitting.
D. Dealloying
This is a form of localized attack which occurs in alloys where the alloying
constituents differ in activity. It has been observed in copper alloys containing
Ni, Al, Mn, or Zn. The attack is due to the selective dissolution of the more
active component, leaving behind a structurally weak deposit of the less active
component. Among copper alloys, dealloying of brasses is most common. It occurs in alloys with more than 15% zinc, the attack being termed dezincification.
In waters, this form of attack is favored by high temperature, water stagnation,
the presence of crevices where aeration is restricted, a high ratio of chloride ion
to bicarbonate ion, and a relatively high pH (1,2).
In single-phase brass alloys, the dealloying occurs over the entire alloy
119
Stress-Corrosion Cracking
Fig. 1 Cross sections of leaded brass alloys showing dezincification after exposure to
acidified copper chloride solution for 7 days: (a) shows fairly uniform layer-type attack
and (b) shows attack along phase in a cast alloy. Magnification 100.
121
The atmospheres which promote this form of attack are especially those
containing ammonia together with water or water vapor. The presence of oxygen
and carbon dioxide accelerate the rate of failure (12). Other corrodents which
have been claimed to promote stress corrosion in brasses are amines (13), citrate
and tartrate solutions (14), nitrites, carbonates, phosphates, and alkalis (15), sulfur
dioxide (16), and seawater (17).
Testing for susceptibility to stress-corrosion cracking is typically conducted
in Mattssons solution at pH 7.3. This is a solution of ammonium sulfate, copper
sulfate, and water with a copper ion content of 0.05 g ions/L and an ammonium
ion content of 1.0 g ions/L (18). In one form of the test, strip samples, 30 mils
thick and measuring 6 0.5 in. are bent around a 3/4-in. mandrel to provide a
permanent 90 set. The sample is then bent into a U and fixed in a jig so that
the legs are 3/4 in. apart. The stressed sample is then immersed in the Mattssons
solution. Withdrawal of the sample is practiced at intervals. The degree of
springback of the samples is compared against that of a control sample not immersed in the solution. Any decrease in springback over that of the control is
then due to stress corrosion. When the springback falls to below 80%, the sample
is said to have failed. A more rapid test is to expose similar U-bend samples
over 28% ammonia in a closed container.
Table 2 shows the response of various alloys to such testing (19). The
results indicate that many alloys are immune to cracking in the Mattssons solution. Those containing Zn are susceptible, the susceptibility increasing with increasing Zn content. Alloy C66900, containing Mn and Zn, is particularly susceptible. In moist ammonia, other alloys also show susceptibility to stress corrosion.
Of the alloys tested, only alloys C11000, C19400, and C65400 and the copper
nickel alloys are immune. Figure 2 illustrates the typical intergranular paths of
stress-corrosion cracks which formed in U-bends of alloy C26000 after only 4
h exposure over 28% ammonia. The many branches to the main cracks are characteristic of this corrosion phenomenon.
Another test used as an indicator of whether a component will suffer stress
corrosion consists of immersion in a mercurous nitrate solution (20). This test
is strictly only a means of indicating the presence of residual stress in an alloy,
the failure occurring by mercury embrittlement of grain boundaries. As such, it
also reveals the presence of residual stresses in alloys, such as the coppernickels,
which are not susceptible to stress-corrosion cracking. Therefore, this test should
be confined to testing parts fabricated from brass.
F.
Corrosion Fatigue
122
Brock
Table 2 Stress-Corrosion Data for Selected Copper Alloys in Moist Ammonia and
Mattssons Solution, pH 7.3
Time to failure (h)
Alloy
C11000
C19400
C22000
C23000
C24000
C26000
C26000
C35300
C35300
C42200
C42200
C42500
C42500
C44300
C44300
C51000
C63800
C65400
C66900
C68800
C70600
C71500
C72200
C75200
C75200
C76200
C77000
a
Condition
Hard
Hard
Extra hard
Cold rolled
Spring
Hard
Extra hard
Extra hard
Cold rolled
Cold rolled
Extra hard
Cold rolled
Spring
Cold rolled
Cold rolled
Extra hard
Extra hard
Cold rolled
Cold rolled
Cold rolled
Cold rolled
Cold rolled
Cold rolled
Extra hard
Cold rolled
Cold rolled
Extra hard
50%
50%
40%
40%
10%
40%
25%
25%
40%
50%
50%
10%
50%
50%
Composition (wt%)
99.9% Cu
2.4% Fe, 0.13 Zn, 0.04 P
10% Zn
15% Zn
20% Zn
30% Zn
30% Zn
35% Zn, 2% Pb
35% Zn, 2% Pb
12% Zn, 1% Sn, 0.2% P
12% Zn, 1% Sn, 0.2% P
9.3% Zn, 2% Sn, 0.2% P
9.3% Zn, 2% Sn, 0.2% P
28% Zn, 1% Sn, 0.04 As
28% Zn, 1% Sn, 0.04 As
5% Sn, 0.2% P
2.8% Al, 1.8% Si, 0.4% Co
3% Si, 1.6% Sn
15% Zn, 12% Mn
23% Zn, 3.4% Al, 0.4% Co
10% Ni, 1.4% Fe
30% Ni, 0.5% Fe
17% Ni, 0.75% Fe, 0.5% Cr
17% Zn, 18% Ni
17% Zn, 18% Ni
30% Zn, 12% Ni
27% Zn, 18% Ni
Mattssons
a
NF
NF
NF
34
15
4
4.7
2
2
NF
NF
NF
NF
42
13
NF
NF
NF
0.6
500
NF
NF
NF
880
530
29
60
Ammonia
NF
NF
11
2.4
1.2
1.2
3
17
10
37
29
1.2
16
393
28
NF
37
1
NF
NF
NF
422
16
NF no failures.
of stress compared to the stress level for the same number of cycles in air (21).
Failure under such conditions is termed corrosion fatigue. The process is characterized by cracks in the metal which are perpendicular to the tensile stress. Their
rate of propagation is usually faster than that of stress-corrosion cracks and they
are generally much straighter and with less crack branching. In contrast to stresscorrosion cracks, there is generally only one crack associated with failure by
corrosion fatigue. Crack initiation often occurs at the base of corrosion pits. Figure 3 shows a typical fatigue crack initiated on a pit in the outside of an Admiralty
alloy condenser tube. Cyclic stresses resulted from vibration of the tube within
123
the condenser, the corrosive environment being ammoniated condensate. The micrographs reveal that the crack is associated with a small pit on the alloy surface
and that the crack path is transgranular. Copper alloys most resistant to corrosion
fatigue are those with a high fatigue limit and a high resistance to corrosion in
the particular environment. Typical of such alloys are the berylliumcoppers,
phosphor and aluminum bronzes, and the coppernickels.
G.
Intergranular Corrosion
As its name implies, this is a form of corrosion in which attack penetrates along
grain boundaries, often to a depth of several grains. The more rapid attack of the
grain boundaries is generally the result of a difference in composition between
the metal in the grain boundary and that in the bulk. Such differences in composition can result from segregation of impurities at the grain boundaries. As intergranular corrosion proceeds in an alloy, the rate of metal loss can increase with
124
Brock
Fig. 3 Cross section through the wall of an alloy C44300 tube showing (a) fatigue crack
initiating from small pit at 50 and (b) the transgranular nature of crack at 500.
125
Fig. 4 Cross section of an alloy C44300 condenser tube revealing integranular corrosion
on the water side at 1000.
time. This is due to grains being isolated and removed from the alloy bulk. Copper
alloys most susceptible to this form of attack include Admiralty metal, aluminum
brasses, and silicon bronzes. Figure 4 shows an intergranular attack which occurred on the water side of an Admiralty alloy condenser tube which had been
in service for close to 30 years with lake water as coolant. Although the depth
of attack shown in the micrograph is only some three grains deep, intergranular
attack with removal of alloy grains by the flowing water had resulted in a loss
in wall thickness of the tube of up to 20 mils.
H. ErosionCorrosion
When copper alloys are exposed to an environment under conditions of flow,
such as in water tubes or condenser tubes, the flow rate can be significant in
determining the corrosion rate. This is because at high fluid velocities, the shear
stresses exerted by the fluid on protective films can lead to their removal. Syrett
and Lapel (22) defined a breakaway velocity beyond which the effects of erosion
cause the oxide to be partly removed. The rate of corrosion in the oxide-free
regions increases and is further accelerated by the galvanic action between the
oxide-free and oxide-covered areas. The nature of the attack often takes the form
126
Brock
Galvanic Corrosion
When two dissimilar metals are immersed in a solution and there is electrical
contact between the two, accelerated corrosion of the more electronegative metal
can occur while corrosion on the other is reduced. Such accelerated attack is
known as galvanic corrosion. Copper and its alloys are usually noble with respect
to other metals with which they come into contact. Thus, copper coupled to aluminum or steel will increase the corrosion of the aluminum and the steel while
being cathodically protected itself. In these instances, copper is the cathode in
the coppersolutionaluminum and coppersolutionsteel cells. The degree of
attack on the more active component is greatest at the point of contact with the
copper. Additionally, the attack is greater, the greater the ratio of the area of the
copper member to the more active member. An example of poor design with
respect to galvanic corrosion would be the use of steel rivets in copper plates or
steel nails in copper roof flashing. Very rapid deterioration of the steel would be
anticipated in such cases.
The electrochemical potential of a metal in solution is a function of the
solution concentration and type. Such potentials are not usually known. Table 3
(23) can be used as a general guideline for designing so that there is no problem
with galvanic corrosion. It lists the electrochemical potentials, with respect to a
127
Electrochemical
potential (V)
Magnesium
Zinc
Aluminum alloys
Mild steelcast iron
Low-alloy steel
Aluminum bronze
Naval brassyellow brass
Tin
Copper
Leadtin solder
Admiralty brassaluminum brass
Manganese bronze
Silicon bronze
Nickel silver
9010 copper nickel
Stainless steel 430
Lead
7030 coppernickel
Nickel 20
Silver
Stainless Steel, 302, 304, 321, 347
Stainless steel, 316, 317
Titanium
Platinum
Graphite
1.6 to 1.63
0.99 to 1.2
0.97 to 0.99
0.6 to 0.71
0.58 to 0.63
0.31 to 0.42
0.3 to 0.4
0.3 to 0.33
0.3 to 0.36
0.28 to 0.37
0.28 to 0.35
0.28 to 0.33
0.25 to 0.28
0.24 to 0.29
0.22 to 0.27
0.2 to 0.27
0.19 to 0.24
0.18 to 0.23
0.1 to 0.2
0.1 to 0.13
0.05 to 0.1
0 to 0.1
0.05 to 0.04
0.25 to 0.18
0.3 to 0.2
128
Brock
Uncoupled
C70600
C71500
C61400
Carbon steel
Titanium
Coupled
C70600
C61400
C70600
Carbon steel
C70600
Titanium
C71500
C61400
C71500
Carbon steel
C71500
Titanium
Corrosion
rate
(mils/year)
1.2
0.8
1.7
1.3
0.08
1
1.7
0.12
31
8.2
0.08
0.7
2.5
0.12
28
4.2
0.08
logarithmic law
1
t
k1 log
1
m
k2
m k 3 log
t
1
k4
129
where m is the weight of oxide formed, t is the time, and the other terms are
constants. At these temperatures, the thickness of the oxide film is usually of the
order of the wavelength of light. Because of this, oxides formed are often translucent and colored because of the property of light interference. The thickness of
the oxide films can be estimated from the color of the oxide as illustrated in
Table 5 (27).
At higher temperatures, the rate of oxidation is usually described in terms
of a parabolic rate equation (28) and sometimes in terms of a cubic rate equation
(29). The expressions are
m 2 k 5 t k 6 parabolic rate equation
m 3 k 7t k 8 cubic rate equation
where the terms have the same meaning as given previously. At temperatures in
the range 3501000C, thick oxide films are formed which spall off from the
metal when the samples are cooled.
Thickness
)
(A
380
420
450
480
500
830
880
970
980
1110
1200
1260
130
Brock
131
Fig. 5 Cross sections of copper silicon alloys oxidized for 16 h at 600C in nitrogen
4% hydrogen showing transition from internal oxidation to external film formation as
silicon content increases from (a) 0.2% through (b) 0.5% to (c) 2.5% at bottom. Magnification 800.
this coating takes the form of a green patina. On copper on roofs or statues, this
patina provides for an aesthetically pleasing appearance.
Vernon and Whitby (33) characterized the composition of the patina formed
on copper exposed in England for times ranging up to 300 years. Under most
conditions of exposure, they showed that the major constituent was brochantite,
a basic copper sulfate. A basic copper chloride, atacamite, was found in the corrosion products formed on copper near the sea. Lesser amounts of malachite, a
basic copper carbonate, were also detected.
It is evident from work described in the literature that the rate of corrosion
of copper in the atmosphere depends on the specific environment, be it urban,
marine, or rural, Other factors which play a role are the temperature and humidity,
the degree of pollution, and the relative time spent between wet and dry conditions. From measurements of the increase in resistance of copper and copper-
132
Brock
alloy wires exposed in an urban atmosphere, Hudson (34) showed that over a
time of close to 5 years, copper corroded at an average rate of 0.195 mils/year
Slightly lower corrosion rates were observed with silicon and tin bronzes but
significantly higher corrosion rates for 8020 and 7030 cupronickel alloys, 70
30 brass, and an aluminum bronze.
Thompson et al. (35) used weight-loss measurements to determine the rate
of corrosion of sheet and wire samples of 11 different copper alloys in various
atmospheres over a time of up to 2 years. Over this time, they determined average
corrosion rates of 3447, 2737, and 1217 in. per year in industrial, marine,
and rural atmospheres, respectively. Tracy (36) also summarized the corrosion
rates of copper and several copper alloys exposed for 20 years at eight different
sites. For most of the alloys, the corrosion rates were within the ranges 230,
1090, and 50120 in. per year. Exceptions to the behavior were two brasses,
for which the rates of attack were 180450 mil/year.
Holm and Mattsson (37) evaluated sheet and rod samples of a wide range
of copper alloys after exposure for times up to 16 years in rural, marine, and
urban atmospheres in Sweden. They observed that dark brown coatings consisting essentially of copper oxides together with some copper salts and alloying
constituents formed in the first year of exposure. In urban and marine environments, green coatings appeared on copper after 67 years and even earlier on
phosphor bronzes. However, green coatings were not formed on high-zinc alloys.
Such coatings were not formed on any of the alloys exposed at rural sites
even after 16 years exposure. Instead, black or brown films were formed. They
showed that the average rate of loss of metal by general corrosion after 7 years
was much lower than that after only 2 years of exposure but similar to that after
exposure for 16 years. The values were 1220 in./year in rural atmospheres,
2035 in./year in marine atmospheres, and 3552 in./year in urban atmospheres.
Penetration of the brasses by dezincification occurred at rates significantly
higher than those described. It was greatest for two-phase alloys and after 16
years had penetrated to 3.55.6 mils. Their results showed that two-phase alloys
containing additions of aluminum, tin, or arsenic improved the dezincification
resistance. Single-phase brasses containing arsenic additions were shown to have
good resistance to dezincification.
Costas (38) evaluated a range of copper alloys after exposure for up to 20
years in rural, marine and industrial locations. He observed that at industrial sites,
green coatings predominated on samples free of zinc or nickel. Blue or green
hues were seen on a few alloys exposed at both marine and rural sites but not
to the same degree as at the industrial sites. The average corrosion rates varied
from 9 to 90 in./year and were greatest at the industrial site.
Figure 6 shows weight losstime plots for alloy C26000 exposed at three
different sites and illustrates the dependency of the corrosion rate on the site
133
Fig. 6 Weight losstime plots for alloy C26000 exposed to the atmosphere at three
different sites.
conditions (19). The New Haven site is representative of an industrial atmosphere, Daytona Beach represents a marine environment with negligible industrial pollution, and East Alton represents a severe industrial atmosphere. The
corrosion rate of the brass is much less at the Daytona Beach site, reflecting the
absence of industrial pollutants. Figure 7 shows similar plots for three alloys
exposed at Daytona Beach. The corrosion rate of the high-copper alloy, C11000,
is significantly greater than those of brass, C26000, and of the coppernickel
alloy, C70600.
The plots also demonstrate that, with time, the corrosion rates generally
Fig. 7 Weight losstime plots for selected copper alloys exposed in New Haven.
134
Brock
Table 6
Sites
Alloy
C11000
C22000
C23000
C26000
C42500
C51000
C65400
C68800
C70600
C76200
Alloying
elements
(wt%)
Daytona
Beach
New
Haven
East
Alton
99% Cu
10% Zn
15% Zn
30% Zn
9.3% Zn, 1% Sn, 0.2% P
5% Sn, 0.2% P
3% Si, 1.6% Sn
2.8% Al, 1.8% Si, 0.4% Co
10% Ni, 1.4% Fe
3% Zn, 12% Ni
0.1
0.07
0.03
0.03
0.08
0.15
0.28
0.02
0.06
0.02
0.04
0.03
0.03
0.05
0.04
0.05
0.04
0.03
0.03
0.05
0.04
0.06
0.04
0.04
0.06
0.04
0.04
0.03
0.06
0.04
decrease because the corrosion product layers provide some resistance to further
attack. The rate of corrosion at any time is given by the slope of a particular plot
at that time. Table 6 shows the corrosion rates so determined for several alloys
after 2 years of exposure at the three sites.
Stress-corrosion cracking of certain copper alloys can also occur during
exposure to the atmosphere. Table 7 shows stress-corrosion data for copper-alloy
samples exposed as U-bends at New Haven, Daytona Beach, and Brooklyn, NY.
In contrast to the data shown in Table 3 for laboratory testing in moist ammonia
and the Mattssons solution, brasses with 15% or less Zn did not fail within the
10-year exposure time. However, failures were observed for alloy C66900 with
only 15% Zn but with 12% Mn. The results also demonstrate a sensitivity to the
site, with the marine site being least aggressive in promoting stress-corrosion
cracking.
Under certain conditions, much more rapid atmospheric corrosion of copper
alloys can occur. Typically, such conditions include the presence of sulfides such
as hydrogen sulfide, which leads to the formation of black, much less protective
films of copper sulfide.
V.
WATER STAINING
Very rapid staining of copper alloys can occur under specific conditions. Typically, this occurs on freshly cleaned strip or parts which have not been fully
135
Condition
Hard
Hard
Extra hard
Cold rolled
Hard
Extra hard
Cold rolled
Extra hard
Cold rolled
Extra hard
Cold rolled
Cold rolled
Extra hard
Extra hard
Cold rolled
Cold rolled
Cold rolled
Cold rolled
Cold rolled
Cold rolled
Extra hard
Cold rolled
Extra hard
Extra hard
50%
50%
40%
10%
40%
25%
20%
40%
50%
50%
10%
50%
Alloying elements
(wt%)
New
Haven
Daytona
Beach
East
Alton
99.5% Cu
2.4% Fe, 0.13% Zn, 0.04% P
10% Zn
15% Zn
30% Zn
35% Zn, 2% Pb
35% Zn, 2% Pb
12% Zn, 1% Sn, 0.2% P
9.3% Zn, 2% Sn 2% P
9.3% Zn, 2% Sn, 0.2% P
28% Zn, 1% Sn, 0.04% As
28% Zn, 1% Sn, 0.04% As
5% Sn, 0.2% P
2.8% Al, 1.8% Si, 0.4% Co
3% Si, 1.6% Sn
15% Zn, 12% Mn
23% Zn, 3.4% Al, 0.4% Co
10% Ni, 1.4% Fe
30% Zn, 0.5% Fe
17% Ni, 0.75% Fe, 0.5% Cr
17% Zn, 18% Ni
17% Zn, 18% Ni
30% Zn, 12% Ni
27% Zn, 18% Ni
NF a
NF
NF
NF
133
856
79
NF
NF
NF
NF
69
NF
NF
NF
11
2080
NF
NF
NF
NF
3100
667
396
NF
NF
NF
NF
140
NF
1256
NF
NF
NF
NF
61
NF
NF
NF
7.5
NF
NF
NF
NF
NF
NF
1105
NF
NF
NF
NF
NF
NF
NF
NF
NF
NF
NF
NF
NF
NF
NF
9.5
NF
NF
NF
NF
NF
NF
NF
146
dried. The staining occurs under water drops, which are the last to evaporate.
The situation contrasts with tarnishing in bulk water, which only occurs slowly
because the tarnishing process requires diffusion of oxygen to the metal surface.
Under a water drop, the diffusion distance of the oxygen from the air to the metal
surface is small. The rates of tarnishing are then much higher, and brown stains
of copper oxides can result in times of only a few minutes.
Under conditions of high humidity and temperatures, such rapid tarnishing
can also occur even if the metal is originally dry. On cleaned dry parts held in
a barrel, moisture can condense out when the temperature falls, as, for example,
during the night. This water is trapped at points where components are in contact
and held there by capillary action. Rapid tarnishing, because of the ease of diffu-
136
Brock
sion of oxygen, through the thin water layers results. Similar rapid tarnishing can
occur between two alloy strips when one is placed on top of the other. Moisture
condensing from the atmosphere is trapped in the space between the two strips
and drawn into the center of the strip by capillary action. During severe conditions
of humidity and temperature, tarnishing can be such as to result in the formation
overnight of violet films on the surfaces of the strips which are in contact, with
no tarnishing on surfaces not in contact.
The rapid tarnishing described occurs on all copper alloys and the rate of
tarnishing is worse if there are contaminants on the alloy surface, such as those
resulting from ineffective rinsing following an acid clean. The tarnishing can be
reduced by maintaining a low ambient humidity and temperature in the area
where parts are dried and stored. Significant decrease in the tarnishing rates can
also be obtained by final rinsing in water containing 0.10.5 wt% benzotriazole
(BTA). Such treatment results in the formation of a protective film of a BTA
thick. This serves to inhibit attack of the
copper complex only some 2040 A
alloy surface.
137
Fig. 8 Weight losstime plots for selected copper-alloy tubes resulting from New Haven
tap water flowing at 8 ft/s at ambient temperature.
Cohen and Brock (43) recently described the soil-side corrosion of copper
water service lines removed from various streets in Billings, Montana after service in the range 1070 years. Their findings attested to the excellent resistance
of copper in the alluvial soil characteristic of this location. In some instances,
tubes in use for 70 years had essentially no evidence of corrosion on the soil
side. The degree of corrosion varied from site to site, but in the worst case, pits
only 11 mils deep were found, and these were on a tube which had been in service
for 35 years.
138
Brock
Corrosion
rate
(mils/year)
0.02% P
2.4% Fe, 0.13% Zn, 0.04% P
15% Zn
20.5% Zn, 2% Al
10% Ni, 1.4% Fe
0.32
0.1
0.4
0.2
0.27
Alloy
C12200
C19400
C23000
C68700
C70600
and the highest for alloy C230 with 15% Zn. The corrosion rates are low and
all less than 0.5 mils/year.
Despite the outstanding corrosion resistance, there are isolated instances
where corrosion can lead to problems. Pitting of copper tubes, although infrequently found, can occur in certain aggressive waters and lead to early failure.
In water at temperatures above 140F, pitting is particularly unusual. Waters
which promote such pitting are soft, have a pH less than 7.4, and have a ratio
of bicarbonate to sulfate ion content of less than 1 (44). Failures have occurred
in Canada, Sweden, Europe, and the United Kingdom. The author has seen rapid
failure of copper tubes in hot, soft well water in Connecticut (U.S.A.). Pits which
form in hot water are deep and narrow and contain cuprous oxide. They are
generally capped by black or greenish-black mounds of copper oxide and basic
copper sulfate. The pits are often surrounded by a deposit high in alumina. In
some hot waters, the presence of manganese can promote pitting and form somewhat larger pits which are surrounded by black deposits of manganese dioxide.
Cold-water pitting has been widely described in the literature, the studies
mainly relating to failures in the United States, the United Kingdom, and Belgium. It requires an aggressive water. Such pitting typically arises in hard waters
from deep wells. The pits are usually hemispherical in nature and form under
green mounds of malachite, a basic copper carbonate. When such nodules are
removed, a brown film of cuprous oxide is revealed with a central hole (Fig. 9a).
Close examination of this reveals that it has an underlying crystalline nature (Fig.
9b). In cross section (Fig. 9c), this oxide is seen to form a membrane which
covers the underlying pit. The base of the pits are filled with copper chloride.
Lucey (45) proposed a mechanism by which such pits are formed. The copper
oxide membrane is critical to his theory. He proposed that the copper oxide membrane initially promotes the formation of pockets of copper chloride adjacent to
139
the metal surface. In continuing pit growth, the membrane then separates the
anodic process within the pit from the cathodic process of oxygen dissolution on
the water side of the membrane. His generally accepted mechanism clearly requires the presence of chloride and bicarbonate ions in the water, as well as
dissolved oxygen.
Cohen and Lyman (46) analyzed 65 waters in the United States where
pitting had occurred, and they showed that they typically contain over 5 ppm
carbon dioxide, have a pH value in the range 7.07.8, contain 1012 ppm oxygen,
and have a sulfate content generally three to four times that of the chloride ion
content. These findings are in accord with the work of OBrecht et al. (47), who
showed that aggressive waters contained high concentrations of dissolved carbon
dioxide and oxygen, a pH in the range 7.07.8, and a sulfate-to-chloride ratio
of from 34 to 1. Following analysis of many aggressive waters in the United
Kingdom, Lucey (48) produced a nomogram, the use of which permits termination of the pitting propensity of the water as determined from its sulfate, sodium,
chloride, and bicarbonate ion content, the concentration of dissolved oxygen, and
the pH.
The presence of carbon films can promote pitting in aggressive waters
(49,50). Such films can be present on a half-hard tube as a result of the decomposition of drawing oil during annealing operations. The films can promote pitting
by isolating pockets of copper chloride adjacent to the alloy surface and enhance
pitting by providing a large cathode-to-anode ratio.
In certain potable waters, corrosion of plumbing fixtures, which are usually
fabricated from leaded brass alloys, can occur. The factors which promote dezincification are high temperatures, water stagnation, the presence of crevices where
aeration is restricted, a high ratio of chloride ion to carbonate ion, and a relatively
high pH (2). In terms of plumbing fixtures, such an attack causes problems not
only because of metal degradation but also because of the formation of voluminous corrosion products which can cause blockages of valves and freezing of
valve stems.
B. Freshwater Cooling Systems
The excellent corrosion resistance of copper alloys in water results in their widescale utilization in power utility condensers and heat exchangers where the source
of water is from rivers or lakes. Alloys typically used for condenser applications
include C14200, C19400, C44300, C68700, C70600, and C71500. In systems
in which the cooling water is circulated through cooling towers, evaporation,
blowdown, and makeup of the water results in an increase in concentration of
the salts initially present in the water. The waters can then become more corrosive. Coppernickel alloys find more application in such systems.
140
Brock
141
C. Saltwater
Copper alloys have a high resistance to corrosion in seawater. Although copper
itself has good resistance, the coppernickel alloys are among the most resistant.
This is illustrated in the weight-loss results shown in Table 9 obtained for panels
of selected copper alloys exposed for various times on racks below the low-tide
level at Daytona Beach (19). These results are consistent with those obtained
from much longer immersion in tidal seawater (51). In these, the corrosion rates
of alloys C70600 and C71500 after immersion for 14 years were shown to be
only 1.1 and 0.8 m/year.
The good corrosion resistance of copper alloys in saltwater or seawater has
resulted in their wide-scale use in ships and in tidal power station condensers.
Alloys used in such applications include the inhibited Admiralty alloys C44300,
C44400, and C44500, and alloys C61300, C68700, C70600, C71500, and
C71640. Instantaneous corrosion rates and weight losses of selected copper alloys
after 1 year obtained from laboratory studies in 3.5% sodium chloride solution
142
Brock
Table 9 Weight Loss of Copper Alloys After Immersion Below Low-Tide Level for
Various Times at Daytona Beach
60 Days
156 Days
365 Days
99.5% Cu
2.4% Fe, 0.13% Zn, 0.04% P
15% Zn
30% Zn
10% Ni, 1.4% Fe
27% Zn, 18% Ni
10.6
9.6
8.9
17.2
3.6
9.7
14.6
11
102
15.6
3.4
14.2
19.7
16
16.1
21.9
5.1
20.9
Alloy
C11000
C19400
C23000
C26000
C70600
C77000
Weight loss
(mg/cm 2)
Alloying
elements
(wt%)
at 40C and flowing at 5 ft/s are shown in Table 10 (19). The various weight
losses reflect the more rapid rate of corrosion of the alloys which occurs during
the early stages of the tests.
The corrosion resistance of copper alloys in nonpolluted seawater results
from the formation of protective films over the surface of the alloy. These have
been shown to be cuprous oxide with outer layers enriched in iron and nickel
(52). Analysis of the protective films formed on a coppernickel condenser tube
removed from tidal power stations alloys confirms these findings (19). Highresolution scanning electron microscopy of cross sections of such films always
reveals an inner layer of cuprous oxide with layers high in nickel and then iron
Table 10 Corrosion Rates and Weight Losses of Copper Alloys After 1 Year
with 3.5% Sodium Chloride at 40C Flowing at 7 ft/s
Alloy
Alloying
elements
(wt%)
Corrosion
rate
(mils/year)
Weight
loss
(mg/cm 2)
C12200
C19400
C44300
C68700
C70600
C71500
0.02% P
2.4% Fe, 0.13% Zn, 0.04% P
28% Zn, 1% Sn, 0.04% As
20.5% Zn, 2% Al
10% Ni, 1.4% Fe
30% Ni, 0.5% Fe
0.6
0.1
0.45
0.1
0.1
0.1
38.6
16.0
10.3
4.0
4.7
3.8
143
oxides above this. Above these, are a layer of paratacamite, Cu2(OH)Cl, and,
invariably, a layer of material deposited from the water, which usually has a high
silica content often with significant concentrations of oxides of iron.
Examination of Alloy C687 condenser tubes from a coastal power station revealed that some 20% of the surface was covered with an orange layer
containing iron, copper, oxygen, and sodium (53). The remaining surface was
covered with a double layer, with the outer layer being a porous material
consisting mostly of iron oxide and an inner later consisting of hydrotacite,
Mg 6 Al 2 (OH) 16 CO3 4H2O, and lesser amounts of paratacamite. Further work on
tubes removed from many power stations revealed that there was always an outer
layer of -FeOOH with a thin underlying layer containing Mg, Al, Zn, and Cu
(54).
Selection of alloys for seawater condenser tubes is based on factors other
than their intrinsic corrosion resistance. An important consideration is the resistance of the alloy to erosioncorrosion resistance of the material. This determines
not only the maximum coolant velocity but also the resistance to increased turbulence round partial blockages in the tubes. Acceptable maximum velocities obtained from both laboratory and service performance are listed in Table 11 (55).
The high resistance of coppernickel alloys to erosioncorrosion by seawater
was demonstrated in the early pioneering work of Stewart and LaQue (56). They
showed that at velocities of 30 ft/s, the erosioncorrosion resistance of Cu10%
Ni alloys increased with increasing iron content and leveled off at the 1.4% iron
Maximum
velocity
(ft/s)
0.02% P
28% Zn, 1% Sn, 0.04% As
5% Al
7% Al, 0.3% Sn
20.5% Zn, 2% Al
1.5% Si
10% Ni, 1.4% Fe
30% Ni, 0.5% Fe
17% Ni, 0.75% Fe, 0.5% Cr
23
46
9
9
8
3
1012
15
30
Alloy
C12200
C44300
C60800
C61300
C68700
C65100
C70600
C71500
C72200
144
Brock
content. Some of this resistance was lost when the alloy was annealed. This precipitates the iron out in an ironnickelcopper phase. These findings were instrumental in setting the iron content of Alloy C70600 close to 1.4% with magnetic
permeability measurements being conducted to ensure that the iron is essentially
retained in solid solution. Similar increases in erosioncorrosion resistance of
copper30% nickel were observed when the iron content was increased from
0.05% to 0.46%. Increasing the iron content above these levels can result in a
increased tendency for pitting corrosion (57). The presence of 2% Mn and 2%
Fe in an alloy with 15% Ni, C71640, and the Cr addition in Alloy C72200 also
leads to high resistance to erosioncorrosion and in preventing attack in severe
conditions where there is entrained sand (57).
145
films. These render the films far less protective to subsequent corrosion. Pitting
of copper alloys is typical in sulfide-polluted conditions.
The sensitivity of the corrosion resistance to sulfide ions is remarkable, and
sulfides at the level of only 0.01 ppm can promote attack on Alloy C706 (62).
Other work has demonstrated that the adverse affect of sulfides increase with
increasing coolant velocity (63).
Polluted waters can lead to early failures of condensers. Elimination of the
source of the pollutant is an obvious step in preventing this type of attack. Decaying plant and animal life can be prevented from entering condensers by the
use of screens and filters. Aeration of the water is also beneficial in removing
sulfides. The intentional formation of hydrated iron oxides is also beneficial in
decreasing the rate of attack by polluted waters. Iron can be introduced into waters
either by an iron anode or by intermittently introducing ferrous ions into the
water by the addition of ferrous sulfate (64). The protective films are then deposited over the surface of the existing corrosion product films.
X.
STEAM
In the absence of carbon dioxide, ammonia, and oxygen, copper alloys are resistant to attack by steam. This property is important in the wide use of copper
alloys in condenser applications. At high steam velocities, such as may be seen
where steam enters a condenser, erosion of copper alloys can occur. Additionally,
if the steam is wet, then water droplets impinging on the metal can promote severe
attack in the form of a high frequency of narrow, deep pits. This is illustrated in
Fig. 10 for an Admiralty alloys tube which had been exposed to such an impingement attack (19). The attack can be averted by incorporation of appropriate baffles
to prevent the water droplets and steam from directly impinging on the tubes.
146
Brock
Fig. 10 Micrograph (at 100) of pits on condensate side of an alloy C706 condenser
tube resulting from impingement of water droplets in wet steam.
sate. This is supported by the data for various copper alloys under both laboratory
and field test conditions, as shown in Table 12 (65).
147
Plant A
Plant B
Plant C
Laboratory
C71500
C72200
C70600
C44300
0.0083
0.016
0.019
0.05
0.004
0.016
0.014
0.031
0.015
0.015
0.018
0.024
0.002
0.008
0.043
0.09
Note: Corrosion rates are after 2 years exposure with laboratory data extrapolated from 1000 h in 2
ppm ammonia at pH 9.4.
Source: Ref. 19.
of the metal. The rate of attack in acids is therefore proportional to the acid
concentration, the temperature, the degree of aeration, and the flow rate. Under
fairly mild conditions, copper alloys are successfully used for handling hydrofluoric, sulfuric, phosphoric, acetic, and other organic acids (6871). Alloys most
resistant are the tinbronzes, the aluminum bronzes, the silicon bronzes, and the
cupronickels.
B. Alkali Solutions
Copper alloys have good resistance to alkali solutions other than ammonium
hydroxide. Coppernickel alloys are most resistant and corrosion rates of less
than 0.2 mils/year are typical of Alloy C71500 in 1N to 2N sodium hydroxide
solutions. For the phosphorbronze alloys, the corrosion rates under the same
conditions are some 10 mils/year.
C. Neutral Solutions
Copper and its alloys are suitable for handling most neutral solutions of nonoxidizing salts. They are used for handling solutions of nitrates, sulfates, and chlorides. Solutions of oxidizing salts, such as those containing chromate, ferric, or
stannic ions, can promote rapid attack. Similarly, salts of metals more noble than
copper will promote attack while plating out on the metal surface.
D. Organic Compounds
Copper alloys are resistant to a wide range of organic compounds such as amines,
ester, ethers, ketones, alcohols, aldehydes, naphtha, and gasoline. In amines, the
148
Brock
REFERENCES
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
149
150
62.
63.
64.
65.
66.
67.
68.
69.
70.
71.
Brock
per mariners experience and economics. Society of Naval Architects and Marine
Engineers, 1976.
JP Gudas, HP Hack. Corrosion 78. Houston, TX: National Association of Corrosion
Engineers, 1978, Paper 23.
TS Lee, HP Hack, DG Tipton. Proceedings of the Fifth International; Congress on
Marine Corrosion and Fouling, Barcelona, 1980.
S Sato, K Nagata, S Yamauchi. Evaluation of various protective measures against
corrosion of copper alloy condenser tubes by sea water. Corrosion 81. Houston,
TX: National Association of Corrosion Engineers, Toronto, 1981, Paper 195.
GP Sheldon, NW Polan. Field testing of power utility condenser tube alloys. J Mater
Energy Syst 6(4), 1985.
Metals Handbook, Volume 13, Corrosion, 9th ed. Metals Park, OH: ASM International, 1987, pp. 627633.
H Leidheiser. The Corrosion of Copper, Tin, and Their Alloys. New York: John
Wiley & Sons, 1971, pp. 107119, 135143.
HH Uhlig. Corrosion Handbook, New York: John Wiley & Sons, 1948.
RJ McKay, R Worthington. Corrosion Resistance of Metals and Alloys. New York:
Reinhold, 1936.
FN Speller. Corrosion, Causes and Prevention. New York: McGraw-Hill, 1951.
JA Lee. Materials of Construction for Chemical Process Industries. New York:
McGraw-Hill, 1950.
6
Reactive and Refractory Alloys
Te-Lin Yau
Te-Lin Yau Consultancy, Albany, Oregon
I.
INTRODUCTION
This chapter covers three metals each of Group IVB [titanium (Ti), zirconium
(Zr), and hafnium (Hf )] and Group VB [vanadium (V), niobium (Nb), and tantalum (Ta)] in the periodic table. Because of their similarities, these metals are
referred to as the following:
1. Reactive metals, as they are highly active in the electromotive force
(EMF) series. Practically, the reactivity of these metals allows them
to spontaneously form protective oxide films in air and makes them
corrosion-resistant materials.
2. Refractory metals, as their melting points are above the range of iron,
cobalt, and nickel.
Although these metals are similar in many ways, they also show very significant differences. The attractive strength-to-density ratio makes titanium an
important structural material in the aerospace industry. The contrast between zirconium and hafnium in thermal neutron absorption cross section makes them
complementary in nuclear applications. Vanadium is a vital alloying element to
make steel and titanium strong and tough. Niobium has hot and cold applications
because of its strength at elevated temperatures and superconducting characteristics at low temperatures. Tantalum is widely used in electrolytic capacitors resulting from the high dielectric constant of its surface oxide film.
In addition to these standout applications, materials and corrosion professionals often specify Ti, Zr, Nb, Ta, and their alloys to handle highly corrosive environments. Moreover, current trends in the chemical process industries favor the increas151
152
Yau
ing usage of these metals. These metals are highly corrosion resistant over wide
ranges of media. When properly used, these metals can realize high returns-oninvestments due to low maintenance and replacement costs, improved process efficiency, added values of high-quality products, and compliance with safety and environmental protection requirements. However, there are some striking differences
among these metals, too. It is essential to fully understand the capabilities of these
metals. Otherwise, any misuses may result in costly mistakes. Consequently, this
chapter discusses environmental effects on reactive and refractory metals.
Property
Atomic number
Melting point (C)
Density (g/cm3)
Coefficient of thermal expansion (106 /C)
Thermal conductivity (W/mC)
Modulus of elasticity (GPa)
Tensile strength (MPa)
Yield strength (MPa)
Stress-relieving temperature (C)
Ti
Zr
Hf
Nb
Ta
22
1677
4.51
8.9
17
52
240655
170720
480
40
1857
6.51
5.89
22
99
165440
170310
540
72
2227
13.3
5.9
22.3
135
440
225
650
23
1917
6.11
9.4
31
120
200500
100200
900
41
2468
8.57
7.1
523
188
170500
75200
800
73
2996
16.69
6.6
544
185
170285
100170
900
Table 1
153
154
Yau
Table 2
ASTM Specifications
Products
Ti
Zr
Hf
Nb
Ta
Plate
Pipe
Tube
Wire
Ingots
Castings
Fittings
Forgings
B-265
B-337
B-338
B-348
B-367
B-363
B-381
B-551
B-658
B-523
B-550
B-495
B-752
B-653
B-493
B-776
B-737
B-393
B-394
B-392
B-391
B-708
B-521
B-365
B-364
Table 3
Condition
Ti
Zr
Nb
Ta
Hydrochloric acid
Sulfuric acid
Nitric acid
Oxidizing acids w/o Cl
Oxidizing acids with Cl
Acids with F
Caustics
Hydrogen peroxide
Dry chlorine
Wet chlorine
Ignition in oxygen
Abrasives
Excellent
Very good
Excellent
Excellent
Cautious
Poor
Excellent
Excellent
Very good
Cautious
Very good
Very good
Good
Good
Excellent
Excellent
Excellent
Fair
Poor
Fair
Excellent
Excellent
Excellent
Good
Very good
Excellent
Excellent
Excellent
Excellent
Poor
Poor
Fair
Excellent
Excellent
Excellent
Good
Poor
Poor
Very good
Very good
Excellent
Poor
Poor
Poor
Poor
Excellent
Good
Good
Ranking (excellent being the best and poor being the worst): Excellent, very good, good, fair, and
poor. Cautious means that other factors such as surface condition is important.
155
156
Yau
oxides, sulfides, carbonates, and chlorides, that do not affect Tis corrosion resistance. Also, chlorination treatments used to control sliming and biofouling have
no adverse effects on Ti.
157
3. Inorganic Acids
Titanium is regarded as a resistant metal in oxidizing acids, such as nitric and
chromic acids, over a wide range of concentrations and temperatures. It is fair
in mildly reducing acids, such as sulfurous acid, but is rather poor in strongly
reducing acids, such as sulfuric, hydrochloric, hydrobromic, and phosphoric
acids. Still, titanium is not as corrosion resistant as zirconium and tantalum to
nitric acid. Titanium resists nitric acid over a wide range of concentrations at
temperatures below boiling temperatures. At boiling and above, titaniums corrosion resistance is very sensitive to nitric acid purity. Generally, titanium is not
corrosion resistant in pure nitric acid. Similarly, titanium may be attacked in the
vapor of nitric acid where condensates may form. The higher the metallic ion
content of the acid, the better titanium will perform. In particular, the corrosion
of titanium to produce small amounts of titanium ions will result in the inhibitive
effect on titaniums corrosion in nitric acid. Therefore, titaniums corrosion may
decrease rapidly in nitric acid under a closed system. Unlike stainless steels,
titanium is uniquely suitable for handling recycled nitric acid. Consequently, titanium has many commercial applications.
When titanium experiences corrosion problems in nitric acid, the problems
cannot be solved by switching to normally more resistant titanium alloys, such
as Pd-containing alloys. These types of alloys are useful in improving Tis resistance in reducing acids. Because nitric acid is not reducing but oxidizing, the
switch will not offer any improvement. Moreover, titanium should not be considered for handling red fuming nitric acid because of the danger of pyrophoric
reactions.
Titanium has limited usefulness in strongly reducing acids (e.g., up to about
7% hydrochloric or sulfuric acid at room temperature). The resistance decreases
rapidly with increasing temperature. It improves when the acids contain small
amounts of oxidizing impurities, such as ferric ions or chlorine. For example,
the addition of 2 g/L ferric chloride will reduce the corrosion rate of Ti in 3%
HCl at boiling from 14 mm/year to less than 0.01 mm/year. Fortunately, it is
common to have this type of impurities in industrial acids. Titanium does have
numerous industrial applications involving reducing acids, such as in the mining
industry.
Titanium can have various corrosion problems in handling reducing acids,
such as crevice corrosion and vapor-phase attack or when the concentration gets
too strong. The addition of up to 0.25% Pd to Ti will significantly improve Tis
resistance in reducing acids. Traditionally, the Pd content is controlled in the
158
Yau
0.15% range. Still, this addition greatly increases the cost. Recently, new grades
of Ti alloys have been added by just having 0.05% Pd for improved corrosion
resistance with a minimum increase in cost.
4. Organic Solutions
Organic acids normally are mildly reducing. Titanium shows good corrosion resistance in most organic acids and other chemicals. It would be vulnerable to
corrosion in strong organic acids, such as formic, oxalic, sulfamic, and trichloroacetic acids. Generally, the presence of oxygen due to aeration and water presence
improves Tis resistance in organic media.
On the other hand, certain anhydrous organic media may attack titanium.
Without oxygen, it would be difficult for Ti to maintain its passivity. For example,
dry methyl alcohol can cause SCC in titanium, probably due to the breakdown
of passive film by halides. Once the passive film is broken down, there will not
be any repair if there is not any oxygen or water. Indeed, the combination of the
absence of water and the presence of halogens or halides is the major reason that
titanium experiences corrosion problems in organic solutions. The addition of
1.5% water can suppress titaniums susceptibility to SCC in methyl alcohol.
Another major problem for Ti in organic media is its susceptible to hydrogen embrittlement. Under reducing conditions, Ti will slowly absorb hydrogen
even when the corrosion rate is very low.
5. Alkalis
Titanium resists most alkalis except hot, strongly alkaline solutions. The major
problem is the excessive hydrogen uptake and eventual embrittlement of titanium
at temperatures above 80C when the pH is at or above 12. The presence of a
strong oxidizer, such as chlorine, makes Ti highly suitable for processing alkalis.
In fact, Ti is a useful structural material in the dual production of caustic soda
and chlorine by an electrochemical process.
6. Gases
Titanium has excellent resistance to air and oxygen at temperatures up to 370C.
Above this temperature but below 450C, titanium may form colored surface
films that thicken slowly with time. Above 650C, titanium will become brittle
due to poor oxidation resistance. Scales form rapidly at 930C. Because the oxidation is an exothermal reaction, titanium may ignite in pressurized oxygen under
a confined condition.
Nitrogen reacts much more slowly with titanium than oxygen. It reacts with
nitrogen to form a gold-colored film starting at 540C. Above 800C, excessive
159
diffusion of the nitride will occur and may cause metal embrittlement. Properly
formed nitride layer can enhance Tis resistance to abrasives.
The surface oxide film can protect Ti from absorbing hydrogen at temperatures below 80C. Absorption of several hundred parts per million (ppm) of hydrogen results in embrittlement and the possibility of cracking under conditions
of stress. The presence of moisture in hydrogen (e.g., as low as 2%) can effectively reduce hydrogen uptake.
Titanium is resistant to corrosion by sulfur dioxide and water-saturated
sulfur dioxide. Water-saturated sulfur dioxide may form the highly corrosive sulfurous acid that does not affect titanium. This capability allows Ti to be used in
various sulfur dioxide scrubber systems.
Titanium resists attack by wet and dry hydrogen sulfide. It is well known
that hydrogen can induce hydrogen embrittlement on many metals and alloys.
This possibility exits in Ti, too. Unlike most metals and alloys, Ti does not become brittle in wet hydrogen sulfide. However, in galvanic couples with certain
metals, such as steel, the presence of hydrogen sulfide in an aqueous solution
will promote hydriding in titanium.
Titanium is among the most resistant metal in wet chlorine and other chlorine chemicals because of their strongly oxidizing natures. This has been the
primary factor for using Ti in industrial service. Titanium equipment has been
relatively free of corrosion problems for decades. However, titanium is incompatible with dry chlorine that can cause a rapid attack of Ti and may even cause
ignition. As little as 1% water is sufficient for repassivation after mechanical
damage to Ti in chlorine gas under static conditions at room temperature.
7. Liquid Metals
Titanium has good resistance to many liquid metals at moderate temperatures.
It has been used in processing molten aluminum. Rapidly flowing molten aluminum, however, will erode titanium. Also, some liquid metals, such as mercury
and cadmium, can cause SCC in titanium.
B. Zirconium
Zirconium has the reputation of being one of the most corrosion-resistant metals.
It has a very strong affinity for oxygen. It is one of very few metals that even can
react with oxygen in water under highly reducing conditions to form an adherent,
protective oxide film on its surface. This film is self-healing and protects the base
metal from chemical and mechanical attack at temperatures to 350C.
Many engineering metals, such as iron, nickel, chromium, and titanium,
produce metal ions of a variable valency. Uniquely, zirconium is predominantly
quadrivalent in its oxides and many other compounds. It forms very few unstable
160
Yau
161
behavior after an initial period. Unlike the white, porous oxide films on unalloyed
zirconium, the oxide film on zircaloy-2 remains dark and adherent throughout
transition and in the posttransition region.
Zircaloy-4 differs in composition from zircaloy-2 in having a slightly
higher iron content but no nickel. Both variations are intended for reducing hydrogen pickup with little effect on corrosion resistance in reactor operation. For
example, in water at 360C, hydrogen pickup for zircaloy-4 is about 25% of
theoretical, or less than half that for zircaloy-2. In addition, hydrogen pickup for
zircaloy-4 is less sensitive to hydrogen overpressure than that for zircaloy-2. For
both alloys, hydrogen pickup is greatly reduced when dissolved oxygen is present
in the medium.
Zr2.5 Nb is considered to be somewhat less resistant to corrosion than
the zircaloys, with exception. Nevertheless, Zr2.5 Nb is suitable for many applications, such as pressure tubes in the primary loops of some reactors. Furthermore, the corrosion resistance of Zr2.5 Nb can be substantially improved by
heat treatments. Also, the Zr2.5% Nb alloy is superior to zircaloys in steam at
temperatures above 400C.
2. Saltwater
Zirconium has excellent corrosion resistance to seawater, brackish water, and
polluted water. Zirconiums advantages include its insensitivity to variation in
factors like chloride concentration, pH, temperature, velocity, crevice, and sulfurcontaining organism. Some of the results are summarized as follows.
Zirconium specimens with or without a crevice attachment were placed in
the Pacific Ocean at Newport, Oregon, for up to 129 days. All welded and nonwelded specimens exhibited nil corrosion rates. Marine biofouling was observed;
however, no attack was found beneath the marine organisms or within the crevices.
Laboratory tests were performed on Zr 702 (unalloyed Zr) and Zr 704 (nonnuclear grade of zircaloys) in boiling seawater for 275 days and in 200C seawater for 29 days. Both alloys were resistant to general corrosion, pitting, and crevice corrosion. Tests of U-bend specimens, with or without steel coupling, of Zr
702, nickel-containing Zr 704, and nickel-free Zr 704 were conducted in boiling
seawater for 365 days. No cracking was observed during the test period. Overstressing of the tested U-bends indicated that all specimens remained ductile except for the welded nickel-containing Zr 704 with steel coupling. Steel-coupled
nickel-containing Zr 704 showed much higher hydrogen and oxygen absorption
and formed hydrides. Chemical analyses and metallographic examinations on
other U-bends did not show evidence of hydrogen absorption and hydride formation.
162
Yau
3. Halogen Acids
Zirconium resists attack by all halogen acids, except hydrofluoric acid (HF). HF
vigorously attacks it at all concentrations. Acidic fluoride solutions are highly
corrosive to zirconium, too. Zirconiums corrosion resistance is not as poor in
fluoride salt solutions as in HF until fluorides become HF in pH 3 solutions.
This fact is taken advantage of in preparing zirconium surfaces using mixtures
of hydrofluoric and nitric acids for various fabrication steps and for improved
corrosion resistance in certain nuclear and chemical applications.
In recent years, it appears that the chance to have fluorides in the process
media has increased somewhat. One of the possibilities is the increased usage of
recycled chemicals. For example, recycled sulfuric acid may contain more than
100 ppm fluorides. When zirconium equipment faces fluoride-containing acids,
inhibitors that form strong fluoride complexes should be added for protecting
zirconium equipment. Effective inhibitors include zirconium nitrate, zirconium
sulfate, and phosphorus pentoxide.
The other halogen acids [i.e., hydrochloric (HCl), hydrobromic (HBr), and
hydriodic (HI) acids] do not attack zirconium. Yet, one of the most impressive
corrosion properties for zirconium is its excellent resistance in HCl at all concentrations and temperatures even above boiling. Because of its strong reducing
power, it is very difficult for most metals to form protective oxide films in HCl.
The presence of even a small amount of HCl in a medium may cause common
metals and alloys to suffer general corrosion, pitting, and SCC.
Zirconium is suitable for handling HCl at all concentrations. Moreover,
zirconium is not as susceptible to hydrogen embrittlement in HCl as tantalum is.
For example, tantalum became brittle when tested in 11M HCl and 11M HCl
7% GaCl3 for 1000 h at 70C. Under the same conditions, zirconium remained
unattacked and retained 100% of its ductility.
Zirconium is susceptible to localized corrosion, such as pitting, intergranular corrosion, and SCC when it is anodically polarized. Zirconium is susceptible
to pitting in 20% HCl, but to intergranular corrosion in 20% HCl. The same
types of corrosion problem may be developed in HCl when highly oxidizing ions,
such as ferric and cupric ions, are present. The presence of ferric ions may polarize the zirconium surface to a potential exceeding the breakdown potential. Thus,
a local breakdown of the passive surface at preferred sites occurs, and the condition favors the occurrence of localized corrosion. To eliminate preferred sites by
pickling zirconium in a mixture of hydrofluoric and nitric acids can suppress the
breakdown process of passive films. Alternatively, maintaining zirconium at a
potential in its passive region, which is arbitrarily set at 50100 mV below the
corrosion potential, can counteract the detrimental effects resulting from the presence of ferric ions.
163
4. Nitric Acid
Nitric acid (HNO3), because of its oxidizing power, is not considered to be a
difficult acid for passive metals to handle. Nevertheless, HNO3 becomes highly
corrosive when its temperature is high, its concentration is too high or not high
enough, or its purity is poor. The oxidizing power favors the formation of oxide
films, but it may also cause the passive films to break down.
Zirconium is considerably more suitable than most passive metals for handling HNO3, particularly when the acid is hot, impure, and/or variable in concentration. In certain conditions, zirconium is even more resistant than the noble
metals to the acid. Zirconiums temperature limit is somewhat higher than that
of noble metals. Traces of chloride may lead to rapid attack of noble metals, but
not of zirconium.
The excellent corrosion resistance of zirconium in HNO3 has been recognized for more than 30 years. Zirconium resists nitric acid up to the boiling point
and at 98% HNO3, and up to 250C and at 70% HNO3. Moreover, the corrosion
rates are nil in boiling 3070% HNO3 with up to 1% FeCl3, 1% NaCl, 1% seawater, 1% iron, or 1.45% type 304 stainless steel at 205C. These results indicate
that the presence of heavy metal ions and chlorides has little effect on the corrosion resistance of zirconium.
Zirconium is normally susceptible to pitting in oxidizing chloride solutions.
However, the NO3 ion is an effective inhibitor for the pitting of zirconium. The
minimum [NO3]/[Cl] molar ratio required to inhibit pitting of zirconium was
determined to be 15. Still, the presence of an appreciable amount of HCl should
be avoided because zirconium is not resistant to aqua regia.
The slow strain-rate technique can reveal zirconiums SCC susceptibility
in HNO3. The primary concern for using zirconium in HNO3 service would be
SCC in concentrated acid. Zirconium seems to be resistant to SCC in concentrated
acids when it is stressed below the yield point.
Other concerns include the accumulation of chlorine gas in the vapor phase
and the presence of fluoride ions. Chlorine gas can be generated by the oxidation
of chlorides in HNO3. Areas that can trap gases should be avoided when Cl is
present in HNO3; or, zirconium equipment should be pickled for much improved
resistance to pitting in wet-chlorine-containing vapors. As indicated previously,
the corrosion of zirconium in fluoride-containing acids can be controlled by adding an inhibitor, such as zirconium compounds, to convert fluoride ions into noncorrosive complex ions.
5. Sulfuric Acid
Sulfuric acid (H2SO4) is the most important acid for use in the manufacture of
many chemicals. For example, it is used as a dehydrating agent, an oxidizing
164
Yau
165
tem is imperfectly sealed. In a leaking system, the acid concentration can exceed
the concentration limit. Acid concentration can easily increase when the system
is under vacuum operation because water vapor is continuously taken away.
When the corrosion resistance limits of zirconium in H2SO4 are exceeded,
a pyrophoric surface layer may be formed on zirconium under certain conditions.
The pyrophoric surface layer on zirconium formed in 77.5% H2SO4 200 ppm
Fe3 at 80C consists of -hydride, ZrO2, Zr(SO4)2, and fine metallic particles.
The combination of hydride and fine metallic particles is suggested to be responsible for the pyrophoricity. Treating in hot air or steam can eliminate this tendency.
6. Phosphoric Acid
Phosphoric acid (H3PO4) is less corrosive than other mineral acids. Many stainless
alloys demonstrate useful resistance in the acid at low temperatures. As often
occurs, corrosion rates of common alloys in the acid increase with increasing
temperature, concentration, and impurities. Areas like the liquid-level line or the
condensing zones are particularly vulnerable to attack.
Zirconium resists attack in H3PO4 at concentrations up to 55% and temperatures exceeding the boiling point. Above 55% H3PO4, the corrosion rate could
increase substantially with increasing temperature. The most useful area for zirconium would be dilute acid at elevated temperatures. Zirconium outperforms common stainless alloys in this area.
If H3PO4 contains more than a trace of fluoride ions, attack on zirconium
may occur. Because fluoride compounds are often present in the ores for making
H3PO4, the use of zirconium has always been questioned. However, because P2O5
is an effective fluoride inhibitor and is usually present in large amounts in H3PO4
processes, tests should be conducted to determine zirconiums suitability in the
actual stream. Furthermore, zirconium compounds can be used to complex fluorides.
7. Other Acids
Zirconium has excellent corrosion resistance in up to 30% chromic acid at temperatures to 100C. It is not suitable for handling chrome-plating solutions that
contain fluoride catalysts.
Zirconium is also resistant to some mixed-acid systems. It can be in acid
mixtures of sulfuricnitric, sulfurichydrochloric, and phosphoricnitric. The
sulfuric acid concentration must be below 70%. Zirconium is aggressively attacked in 1:3 volume mixtures of nitric and hydrochloric acids (aqua regia). In
the 1:1 volume mixture, zirconium is attacked but much slower than in the 1:3
mixture. In mixtures greater than the 3:1 ratio, zirconium is resistant.
166
Yau
8. Alkalis
Zirconium resists attack in most alkalis, which include sodium hydroxide, potassium hydroxide, calcium hydroxide, and ammonium hydroxide. This makes zirconium distinctly different from other highly corrosion-resistant materials, such
as titanium, tantalum, graphite, glass, and polytetrafluoroethylene (PTFE).
Zirconium U-bend specimens have been tested in boiling 50% NaOH. During the test period, the concentration changed from 50% to about 85% and the
temperature increased from 150C to 300C. The PTFE washers and tubes used
to make the U-bends dissolved. However, the zirconium U-bends remained ductile and did not show any cracks after 20 days. It should be noted that stainless
steel is susceptible to SCC in alkaline solutions including NaOH solutions.
Zirconium coupons were tested in a white-liquor, paper-pulping solution,
which contained NaOH and sodium sulfide, at 120C, 175C, and 225C. All
coupons showed nil corrosion rates. In the same solution, graphite and glass both
corroded badly at 100C. Zirconium also exhibits excellent resistance to SCC in
simulated white liquors.
9. Salt Solutions
Zirconium is resistant to most salt solutions, which include halide, nitrate, carbonate, and sulfate salts. Corrosion rates are usually very low at temperatures at least
up to the boiling point. Solutions of strong oxidizing chloride salts, such as FeCl3
and CuCl2, are examples of the few exceptions. In strong oxidizing chloride solutions, zirconiums performance is very dependent on surface condition. Zirconium becomes quite resistant to pitting when it has a good surface finish, like
the pickled surface.
Although zirconium has good corrosion resistance in sodium fluoride and
potassium at low temperatures, resistance decreases rapidly with increasing temperature or decreasing pH. Consequently, zirconium is not ideal for handling
most fluoride-containing solutions unless fluoride ions are complexed.
Zirconium is considerably more resistant to chloride SCC than stainless
steels are. No failure was observed in U-bend tests conducted in boiling 42%
magnesium chloride (MgCl2). Another attractive property of zirconium is its high
crevice corrosion resistance. Zirconium is not subject to crevice corrosion even
in acidic chloride at elevated temperatures. No attack was observed on zirconium
in a salt spray environment.
Unlike many common metals, zirconium has very little affinity for sulfur.
Zirconiumsulfur compounds form only at temperatures above 500C. Furthermore, there is no instance of zirconiumsulfur bonds forming in aqueous systems.
Hence, hydrogen sulfide (H2S), which is highly corrosive to common metals and
alloys, is not expected to participate in the corrosion reactions of zirconium in
sulfide-containing solutions. Zirconium coupons and U-bends were tested in nu-
167
merous NaClH2S solutions at temperatures to 232C. No general corrosion, pitting, crevice corrosion, or SCC was observed.
168
Yau
sive to zirconium. They are not stable. They react violently with water. There is
no chance for water to be present in this type of halides, which are the most
undesirable organics for zirconium, and possibly other metals, to handle.
11. Gases
Zirconium forms a visible oxide film in air at about 200C. The oxidation rate
becomes high enough to produce a loose, white scale on zirconium at temperatures above 540C. At temperatures above 700C, zirconium can absorb oxygen
and become embrittled after prolonged exposure.
Zirconium reacts more slowly with nitrogen than with oxygen because it
has a higher affinity for oxygen than for nitrogen and it is normally protected by
a layer of oxide film. Once nitrogen penetrates through the oxide layer, it diffuses
into the metal faster than oxygen because of its smaller size. Clean zirconium
starts the nitride reaction in ultrapure nitrogen at about 900C. Temperatures of
1300C are needed to fully nitride the metal. The nitriding rate can be enhanced
by the presence of oxygen in the nitrogen or on the metal surface.
The oxide film on zirconium provides an effective barrier to hydrogen absorption up to 760C, provided that small amounts of oxygen are also present
in hydrogen for healing damaged spots in the oxide film. In an all-hydrogen
atmosphere, hydrogen absorption will begin at 310C.
Zirconium will ultimately become embrittled by forming zirconium hydrides when the limit of hydrogen solubility is exceeded. Hydrogen can be effectively removed from zirconium by prolonged vacuum annealing at temperatures
above 760C.
The corrosion resistance of zirconium and its alloys in steam is of special
interest to nuclear power applications. They can be exposed for prolonged period
without pronounced attack at temperatures up to 425C. In the 360C steam, up
to 350 ppm chloride and iodide ions, 100 ppm fluoride ions, and 10,000 ppm
sulfate ions are acceptable for zirconium in general applications but not in nuclear
power applications.
Zirconium is stable in NH3 up to about 1000C, in most gases (CO, CO2,
and SO2) up to about 300400C and in dry halogens up to about 200C. At
elevated temperatures, zirconium forms volatile halides. Depending on the surface condition, zirconium may or may not be resistant in wet chlorine. Zirconium
is susceptible to pitting in wet chlorine unless it has a properly cleaned surface.
169
Zirconium resists some types of molten metals, but the corrosion resistance
is affected by trace impurities, such as oxygen, hydrogen, or nitrogen. Zirconium
has a corrosion rate of less than 25 m/year in liquid lead to 600C, lithium to
800C, mercury to 100C, and sodium to 600C. The molten metals known to
attack zirconium are zinc, bismuth, and magnesium.
C. Hafnium
Hafnium and zirconium are chemically very similar. They occur naturally in ores.
However, hafnium has a high thermal neutron absorption cross section in contrast to the very low one of zirconium. Because of this dramatic difference, they
have to be separated from each other for nuclear applications but not corrosionresistant applications.
Similar to zirconium, hafnium has excellent corrosion resistance to many
media, including hydrochloric, nitric, and nonconcentrated sulfuric acids and alkalis. In fact, hafnium is superior to zirconium in corrosion resistance in water,
steam, molten alkali metals, and air.
Hafnium is soluble is hydrofluoric and concentrated sulfuric acids and aqua
regia. It begins to react slowly with air or oxygen at about 400C and nitrogen
at about 900C and rapidly with hydrogen at about 700C.
An interesting capability for hafnium is the combination of high neutron
absorption and excellent corrosion resistance to nitric acid. This capability makes
hafnium a uniquely reliable material for use in spent nuclear fuel reprocessing
plants. Hafnium is being used not just for its corrosion resistance but also for
preventing criticality.
D. Vanadium
Compared to other reactive and refractory metals, vanadium is inferior in aqueous
corrosion resistance. There is no known application for vanadium in the chemical
process industries. There is no established ASTM specification for vanadium,
either.
Vanadium resists attack by oxygen, nitrogen, and hydrogen at ambient temperatures. It oxidizes in air at different temperatures to form various oxides (i.e.,
trioxide, tetroxide, and pentoxide). It reacts with chlorine at temperatures greater
than 180C.
Vanadium is resistant to seawater, reducing acids, such as hydrochloric and
dilute sulfuric acids, and to alkaline solutions. It is poor in oxidizing acids, such
as nitric and concentrated sulfuric acids, and hydrofluoric acid.
Vanadium has one important corrosion property: its resistance in liquid
metals including lithium and sodium. This, combined with its neutron economy
170
Yau
and high-temperature strength, makes vanadium a leading candidate for first wall
and blanket structures of liquidmetal cooled-fusion energy systems.
E.
Niobium
171
To reduce the cost, a TiNb alloy can be used in ignition-sensitive areas. For
example, the Ti45% Nb alloy has found applications that employ dilute sulfuric
acid and pressurized oxygen, as in the gold mining industry.
Below 100C, niobium is inert in most common gases, including bromine,
chlorine, hydrogen, nitrogen, and sulfur dioxide (wet or dry). The resistance to
wet and dry chlorine is particularly interesting in corrosion-resistant applications.
Niobium reacts with nitrogen above 350C, with steam above 300C, with chlorine above 200C, and with carbon dioxide, carbon monoxide, and hydrogen
above 250C.
In the absence of nonmetallic impurities, such as gases, niobium has excellent corrosion resistance in liquid metals. The temperature limits for niobium in
liquid metals are 510C for bismuth, 400C for gallium, 850C for lead, 1000C
for lithium, sodium, potassium, and sodiumpotassium alloys, 600C for mercury, 850C for thoriummagnesium eutectic, 1400C for uranium, and 450C
for zinc.
The excellent resistance of niobium in sodium vapor leads to the use of
the Nb1 Zr alloy as the end caps in high-pressure sodium vapor lamps used for
highway lighting. Furthermore, liquid metals are excellent heat-transfer media;
they are ideal for use in compact thermal systems, such as fast breeder reactors,
reactors for space vehicles, and fusion reactors.
F.
Tantalum
172
Yau
7
Aluminum Alloys
N. J. Henry Holroyd
Luxfer Gas Cylinders, Riverside, California
I.
INTRODUCTION
The aim of this chapter is twofold: first, to briefly describe the various modes
of corrosion (time-dependent environment-induced degradation) suffered by aluminum and its alloys; second, to discuss and review the influence played by
surface films during these corrosion processes. This aspect has been chosen because although its importance is consistently highlighted in the published texts
on aluminum corrosion (14), little detailed discussion ensues. A good example
is the recently published book Corrosion of Aluminum Alloys (4) which reviews
many aspects of the corrosion of aluminum and its alloys. The detailed discussion
on the role of surface films during corrosion is limited to the statement, Aluminum owes its excellent corrosion resistance and its usage as one of the primary
metals of commerce to the barrier oxide film that is bonded strongly to its surface and to a page of text giving basic information on aluminum films. Hopefully, this chapter will contribute to addressing this position.
II. BACKGROUND
Aluminum and its alloys are currently used in a wide range of applications, including aerospace, automotive, building, electrical, marine, packaging, and transportation. A comprehensive summary of the specific alloys (wrought and cast alloys)
used for different applications has recently been published (4). Reasons why a particular alloy is selected are application dependent. For instance, for structural appli173
174
Holroyd
Alloy series
1000
2000
3000
4000
5000
6000
7000
8000
In most commercial applications the corrosion of aluminum and its alloys is not
an issue during service life because the thin surface oxide film formed during
heat-treatment and\or during product fabrication reduces corrosion to negligible
rates. In benign service environments this remains the case even if these preexisting surface films are mechanically disrupted because another protective film
rapidly forms locally preventing the initiation of corrosion. Situations do arise,
however, where this is not the case, and a surface that was protective, locally
loses this capability and a corrosion process initiates.
Aluminum Alloys
175
the initial surface film dissolving and the new film that develops being
nonprotective. The extent of the damage induced depends on the specific chemical composition of the electrolyte, its temperature, and its volume. At one extreme, the alloy can completely dissolve; at the other, the corrosion product consumes all the free water, resulting in a gel layer forming that can stifle further
attack and eventually may become a glasslike solid layer.
In highly acidic solutions (pH 2), general corrosion usually occurs via
a crystallographic mode, whereas in highly alkaline solutions (pH 11), attack
is usually associated with the formation of shallow cusps. General corrosion
is not usually experienced during a commercial products service life unless the
environmental conditions encountered stray well outside those anticipated for its
application.
B. Localized Corrosion
Localized corrosion is the most common form of corrosion experienced by aluminum alloys during service applications. It occurs when the surface oxide film,
for some reason, locally can no longer prevent the initiation of corrosion. The
most common forms of localized corrosion are pitting, intergranular corrosion,
crevice corrosion, galvanic corrosion, exfoliation, stress corrosion, and corrosion
fatigue. Less common forms, although in some cases commercially important
forms, include filiform corrosion, erosioncorrosion, poultice corrosion, biological corrosion, snowflake corrosion, and fingerprint corrosion.
C. Pitting Corrosion
Pitting corrosion is the most common form of localized corrosion suffered by
aluminum and its alloys, and although numerous pitting studies have been undertaken since the 1930s, the universal agreement on how pits initiate is still awaited.
An example of the pitting of an aluminum alloy is shown in Fig. 1 and a polished
metallographic cross section through a typical pit is shown in Fig. 2.
D. Intergranular Corrosion
Intergranular corrosion (IGC), second to pitting corrosion, is probably the next
most common form of corrosion suffered by aluminum alloys. A typical example
of intergranular corrosion is given in Fig. 3. It is potentially more damaging than
pitting because for a given loss of metal, the percentage reduction in a structures
load-bearing capability is significantly higher for IGC and the observed tendency
for self-stifling is less common than is found for pitting corrosion.
Numerous studies have been published over the years, and compared to
pitting corrosion, there is a greater consensus on the mechanisms involved; albeit,
details may be alloy system and temper dependent. In general, underaged tempers
176
Holroyd
Fig. 1 Pitting morphology before and after cleaning an aluminum alloy after long-term
exposure to a saline environment (reduction 3).
Fig. 2 A polished metallurgical cross section through a typical pit shown in Fig. 1 (magnification 200).
Aluminum Alloys
177
are more susceptible to intergranular corrosion that either peak or overaged tempers.
E.
Crevice Corrosion
An opportunity for crevice corrosion occurs when the local geometry provides
a region where a high surface area of the metal can be exposed to a relatively
small volume of an aqueous solution that has a restricted access to the bulk environment.
It has been argued that crevice corrosion is a variant of pitting corrosion,
where the initiation phase of pitting is eliminated by the crevice itself, providing
the restricted geometry (i.e., a very large pit) needed for the development of the
local environmental conditions for corrosion initiation. Although this is mechanistically incorrect, it is not an unreasonable visualization of why crevices should
be avoided in good designs and how tight crevices could lead to the initiation
of localized corrosion.
F.
Galvanic Corrosion
178
Holroyd
either pitting or crevice corrosion; however, severe galvanic stimulation can lead
to general corrosion.
G.
Exfoliation Corrosion
Aluminum Alloys
179
Fig. 5 Section of an agricultural vehicle showing severe exfoliation corrosion: magnification (a) 3/4; (b) 15.
180
Holroyd
regarded as corrosion occurring simultaneously along multiple parallel paths associated with a given microstructural region adjacent to the alloys external surface.
Exfoliation corrosion has been observed in both heat-treatable and nonheat-treatable alloys. For heat-treatable alloys, the corrosion paths follow grain
boundaries and/or regions adjacent to them, whereas for non-heat-treatable alloys, the regions followed involve those rich in transitional metals and the regions
immediately adjacent.
H. Stress-Corrosion Cracking
Stress-corrosion cracking (SCC) is a complex phenomena requiring the concurrent action of both local strain and corrosion. Generally, it is only of practical
concern for the higher-strength aluminum alloys, and then only when heat-treatable alloys [e.g., AlCuMg, AlZnMg(Cu), or AlLi] are in underaged or
peak-aged tempers. Non-heat-treatable AlMg alloys with magnesium concentrations above around 3% (wt) may be susceptible to SCC if they become sensitized
by a grain-boundary precipitation of an active phase during service life. (This
process tends to be accelerated in highly worked alloy microstructures.)
Medium-strength heat-treatable alloys such as AlMgSi (6xxx series)
alloys are rarely susceptible to SCC other than when highly alloyed and subjected
to unusual noncommercial laboratory heat treatments (3,5). Several extensive and
complete reviews of the SCC of aluminum alloys have been published (59) and
the reader is directed to these works, as this aspect of aluminum corrosion will
not be covered in this chapter.
I.
Corrosion Fatigue
Aluminum Alloys
181
182
Holroyd
peratures up to around 400C, as long as the alloy does not contain alloying
elements, such as magnesium or lithium, that react selectively to form their own
oxides. The principle difference for these higher-temperature films is that the
-thick compact
thickness of the outer -Al2O3 layer, forming over the 1015-A
inner layer, increases with temperature. Film thickness measurements quoted in
the literature vary significantly, reflecting that, as with corrosion, the surface preparation prior to oxidation often has a significant influence on the subsequent oxidation behavior (17). Notwithstanding this, it is reasonable to assume oxides will
thick [normally 100 A
(18)] for temperatures up to at 400C
be less than 200 A
and increase to around 400 A at 600C (19). Alloying elements such as copper
and manganese have been reported to enter and dope -Al2O3 at these higher
temperatures (20).
At temperatures above about 425450C, the amorphous outer layer of
the duplex film initially forming on aluminum alloys undergoes a discontinuous
structure change. This allows the rapid migration of oxygen to the oxidemetal
surface and the nucleation and growth of crystalline -Al2O3 as a new phase below
the amorphous layer (21,22).
The presence of alloying elements that react selectively, forming their own
oxides, slightly complicates the oxidation of aluminum alloys. For commercial
aluminum alloys, the principal elements doing this are magnesium and lithium.
For aluminum alloys containing more than 0.5 wt% magnesium, the oxides generated at temperatures up to around 300C are independent of the magnesium addition. However, for higher temperatures, the outermost surface of the
oxides formed is essentially a dense layer of MgO crystals (23). Electron microscopy (24) has shown that MgO crystals can also form at the metaloxide interface, thereby disrupting the inner compact alumina layer.
Zinc additions to aluminum alloys appear to have minor effects on an
alloys oxidation characteristics (25,26) with no evidence existing for zinc incorporation into the films formed on high-purity AlZnMg alloys (27) and only
limited evidence for commercial AlZnMgCu alloys (26,28). Experimental
evidence for these alloys shows that the alloy grain boundaries provide a shortcircuit diffusion path for magnesium to participate in the oxidation process (27).
The presence of lithium in aluminum alloys even at low concentrations has
a dominant effect on the oxidation behavior of aluminum alloys, overriding those
due to magnesium and other alloy additions to commercial alloys (25).
C. Oxide Films Formed on Aluminum Alloys in
Wet Environments
As stated earlier, the natural oxide formed on aluminum in dry air at room
-thick barrier layer next to the
temperature is duplex with a compact 1015-A
metal covered with a thicker less dense layer. The thickness of the outer layer,
although initially around 50 angstroms, is highly sensitive to the prevailing envi-
Aluminum Alloys
183
ronmental conditions (11,29). Although researchers have studied the filming characteristics of aluminum alloys in water for decades, accurate prediction of the
expected film thickness of aluminum alloy is not yet possible. This is due to the
multiplicity of variables involved which include: the nature of the pre-existing
surface oxide, details of the water chemistry composition, temperature, pH, oxygen content, etc. and the alloy composition.
An excellent review of the films forming in the aluminum-water system
has been provided by Alwitt (29), who states: It is an interesting fact that those
phases commonly present as surface films on aluminumpseudo-boehmite, bayerite, boehmite and -Al2O3 are all metastable phases of the Al2O3 water system and the film composition is often the most kinetically accessible product,
which often is only an intermediate product on the path to thermodynamic equilibrium.
Film forming on aluminum exposed to water at various temperatures can
be categorized by four temperature regimes: a) room temperature up to around
40C; b) above 40C to 100C; c) 100C to 150C and d) above 150C.
184
Holroyd
Fig. 6 Temperature dependence of the induction time for film growth on pure aluminium
exposed to distilled water in the temperature range 30100C showing that filming process
change occurs at temperatures around 45C. from Ref. 18; from Ref. 29; from
Ref. 89; and from Ref. 33.
containing excess water (18)] and an outer layer of bayerite crystals [a form of
aluminum hydroxide (32)].
Film growth for this temperature range proceeds in three stages: (1) an
induction period, during which no significant weight change is observed, (2) a
period during which pseudo-boehmite forms, and (3) bayerite crystallization onto
the pseudo-boehmite layer. Stages 2 and 3 can overlap, with bayerite crystal
nucleation occurring on the boehmite while the boehmite is still growing (29).
It has been suggested that the cessation of film growth is dictated by the completion of the bayerite layer and is independent on the formation of the underlying
pseudo-boehmite layer, which is considered to reach a limiting thickness rapidly (29).
The precise mechanism is not fully established. Scamans and Rehal (32)
have made a reasonable proposal based on published work, suggesting that the
pre-existing amorphous film thickens during the induction period [as proposed
Aluminum Alloys
185
by Hart (33)] with either proton, hydroxyl ions, and/or water penetrating the
film to establish a soluble aluminum species. A dissolutionprecipitation process
follows, leading to the rapid formation of a pseudo-boehmite layer on the aluminum surface that, once complete, can only thicken slowly via a process limited
by the diffusion of water into and through the film, as opposed to the outward
diffusion of Al3 ions. These authors suggest that the mechanistic details of the
bayerite formation process awaits clarification. However, it probably involves a
further dissolutionprecipitation process, in this case, occurring at the pseudoboehmitesolution interface with a typical bayerite layer thickness growing to
23 m at 70C.
Fig. 7 Schematic representations as given by Altenpohl (34) for cross sections through
boehmite films formed on superpure aluminum immersed in boiling distilled water for 1
-thick barrier layer at
h and 5 h. [Note: Subsequent studies indicate that a retained 10-A
the aluminumfilm interface is unlikely to be present (29).]
186
Holroyd
Fig. 8 Typical film thicknesstime data according to Altenpohl (34) for film growth on
AA1145 immersed in boiling distilled water.
thicknesstime data for their film growth on AA1145 exposed to distilled water
is shown in Fig. 8.
It is worth noting that Altenpohls filming model (34) envisions that the
thin, compact barrier oxide layer on the aluminum surface prior to any reaction
with water remains in situ throughout the boehmite filming process. Alwitt (29),
in his review of the aluminumwater system states that in his opinion, there
appears to little experimental evidence for such a film.
The benefits offered by boehmite films can be impressive. For instance,
data from Leidheisser and Kellermans study of straining aluminum alloy wires in
aqueous environments (37) shows that immersing commercially pure aluminum
(AA1100) in boiling water for 15 min prior to rapidly straining in a 0.1M NaCl
solution significantly improves the surface films ductility compared to that given
by an air-formed film or an anodized film exposed to boiling water (Fig. 9).
Another good example is the improved pitting-resistant boehmite films can
provide in potentially aggressive tap waters, as is shown in Table 1. Based on
Aluminum Alloys
187
Fig. 9 Effect of surface films on the electrochemical potential response after commercially pure aluminum (AA1100) wires are rapidly plastically strained 1.5% in a deaerated
0.1M NaCl solution, pH 5.5. (Data from Ref. 37.)
the data in Table 1 and other data, a significant interest was generated during the
1960s to attempt to utilize boehmite films commercially to render aluminum
cans suitable for packaging beer, milk, and other products (34).
Attractive as the boehmite films are, their commercial exploitation to
date has been restricted by issues associated with the conditions needed to successfully generate these films in either boiling water or in high-temperature steam.
For instance,
1. Protective boehmite films have only commercially been reproducibly formed on high-purity aluminum.
2. Trace additions of contaminants in boiling water or steam can either
stifle film growth or promote corrosion and/or the growth of a nonprotective film.
Re-examination of the boehmite film thickness data provided by Altenpohl
(34) for six superpure aluminum alloy compositions subjected to different surface
pretreatments prior to their immersion in boiling high-purity distilled water for
188
Holroyd
50C
70C
A
B
C
A 0.1N NH4OH
5
30
12
0
2
3
0
0
B 0.1N NH4OH
C 0.1N NH4OH
Air-formed film
0
0
18
0
0
3
Specific resistance of
distilled water (cm)
8 105
7 105
5 105
2.18 104 to 1.45 104
(start)
(finish)
Note: Boehmite films produced by immersion in boiling distilled waters for 4 hours, with and without
a 0.1N ammonium hydroxide addition.
Source: Data from Ref. 34; see reference for details on distilled-water compositions.
various times reveals that the film thickness generated is effectively controlled
by an alloys iron, copper, and silicon contents. This conclusion is based on the
following observation of high-purity aluminum (i.e., concentrations of iron, copper, and silicon will always be low):
1. Thicker films only occur if an alloys copper and iron contents are
below 0.0015 and 0.002 wt%, respectively (Fig. 10).
2. When copper and iron levels are sufficiently low, increasing an alloys
silicon weight percent favors thicker films (Fig. 10).
The latter observation is interesting in view of the well-known detrimental effects
trace additions of silicon to the alloy have on film formation at temperatures
above 150C or dissolved silicate ions additions to boiling distilled water have
on boehmite film formation (Fig. 11). Based on the latter, it is reasonable to
suggest that it is the free silicon in the alloy that is detrimental to boehmite
film formation, and when at appropriate low concentrations, iron and silicon can
form relatively benign precipitates, thereby allowing copper levels to control
boehmite film thickness (see Fig. 10).
Some published data on the influence of anions on the growth behavior of
boehmite films on aluminum immersed in boiling water are available (34,38,39).
Data presented by McCune et al. (38) for 99.99% aluminum (AA1100) in boiling
Aluminum Alloys
189
Fig. 10 Effect of silicon content on the boehmite film thickness developed on a range
of superpurity aluminum compositions immersed in boiling distilled water for either 2 or
8 days. (Data from Ref. 34.)
190
Holroyd
Aluminum Alloys
191
Fig. 12 Apparent pseudo-boehmite film thickness formed on 99.99% (AA1100) aluminum immersed in boiling distilled water containing various single-anion additions as sodium salts (1 g/L nitrate, 2 g/L sulfate, and 1 g/L silicate) for various up to 1 h. (Data
from Ref. 38.)
aluminum being exposed to water at temperatures up to around 350C. Aluminums attractions for this application were its low cost, ease of fabrication, and
low-adsorption cross section for neutrons. An intense research and development
(R&D) effort ensued in North America to overcome two issues:
1. To improve aluminums environmental performance in water at elevated temperatures and to avoid the situation where, above a critical
192
Holroyd
temperature, surface films lose their protective nature and allow intergranular corrosion and rapid loss of metal
2. To increase the strength of aluminum at temperatures around 300C
Aluminum alloy development was the R&D approach adopted to achieve the above.
This decision was based on the belief that appropriate alloying additions would
exist that could simultaneously increase an alloys strength while also increasing
the critical temperature above which aluminum corrodes rapidly, which was known
to increase as the alloys purity decreased (44). [Typical critical temperatures are
105C for 99.99% aluminum, 150C for 99.85% aluminum, 180C for 99.5% aluminum and a higher temperature for 2S aluminum (99.099.3% aluminum), which
normally performs well in water up to temperature around 200C (34).]
Work in the United States at the Argonne National Laboratories in 1955
established that a 1 wt% nickel addition to AA1100 prevented intergranular attack
at 350C (45), changing the mode to uniform corrosion (40). This led to the
development of the experimental alloy X-8001, which was superseded in 1957
by two further alloys based on AlNiFeTi (A203X and A198X), that both
offered a superior corrosion performance in water at 360C, which was attributed
to these alloys having a finer dispersion of second-phase precipitates.
Corrosion performances for the initial commercially cast versions of the
above AlNiFeTi alloys in high-temperature water were extremely variable,
and after detailed studies, it became evident that the corrosion performance was
extremely sensitive to the alloys purity levelin particular, the silicon weight
percent. A further alloy development program was undertaken by the General
Electric Company to generate an alloy that provided yet a better corrosion performance in water at 360C while also having a sufficient tolerance to the alloys
purity level that it could be readily cast by conventional casting practices. An
alloy based on Al1.8 Fe1.2 Ni resulted (46) that, as well as being sufficiently
tolerant to the alloys purity levels, was also tolerant of long-term thermal treatment at 550C and provided a superior corrosion performance.
In parallel with the United States effort, Krenz (42), working in Canada,
reported two AlNiFeSi alloys (Al2 Ni and Al0.5 Ni, both containing 0.5%
Fe and 0.2% Si) that performed reasonably well when exposed to water at 300C.
Initial attack of the alloys was rapid, but further attack slowed down to a linear
rate that was marginally lower for the higher-nickel-containing alloy.
MacLennan (47) exposed the above two AlNiFeSi alloys along with
2S aluminum to water at 300C for short periods of time (minutes) and then took
replicas of the inner and outer surfaces of the oxide films and evaluated them
using electron microscopy. Based on this study and further work involving longer
exposure times in water at temperatures in the range 150340C (48), MacLennan attributed the improved corrosion performance for the nickel-containing
alloys to their oxide films providing a local stress relief effect. This stress
relief effect was deemed to minimize any tendency for the films to crack, as is
Aluminum Alloys
193
Aluminum surfaces are normally covered by a protective thin oxide film. The
specific nature of this film depends on the conditions under which it was formed
194
Holroyd
and subsequently stored. Time-dependent processes are initiated when these surfaces are exposed to aqueous environments, the results of which are dependent
on several factors, including the aluminum alloy composition, temper and microstructure, the nature of the pre-existing surface oxide film itself, and the prevailing local electrochemical and environmental conditions (e.g., electrode potential, solution chemistry, temperature, pH, oxygen concentration, etc.).
When exposed to sufficiently aggressive aqueous environments, the protection provided by the pre-existing oxide film on aluminum alloys is negated
and general corrosion of the underlying metal is initiated. For the vast majority
of structural or engineering applications of aluminum alloys, these situations are
of no interest and these conditions need to be avoided at all costs. Exceptions
to this usually involve situations in which the alloy is intentionally being used
as (a) a battery anode [e.g., an aluminumair battery (56)] or (b) a sacrificial
anode (57) or aluminum-based spray coating (5860) to protect another material
Aluminum Alloys
195
such as a steel tension leg in an offshore oil platform structure (61). In these
cases, the general corrosion behavior itself has to be controlled to maximize
coulombic efficiencies (e.g., by alloy composition and microstructure manipulation). Another exception is the situation in which an aluminum alloy is subjected
to a short period of alkaline etching aimed at removing the pre-existing surface
film and the preferential removal of second-phase particles that otherwise would
act as a cathodic reaction sites during any subsequent corrosion processes (15).
Although it is true that aluminum alloys are only resistant to general corrosion when the solutions pHs are neither too high nor too low, it is now generally
accepted that an alloys corrosion performance is controlled by the nature of the
aluminum salt involved rather than the solutions pH produced by the hydrolysis
of the aluminum cation (62). Published information on the inhibition of aluminum
alloy corrosion in highly acidic (6365) and highly alkaline (6668) environments is available.
To simplify the following discussion we will initially concentrate on the
behavior of nonalloyed aluminum (i.e., high-purity and commercial purity grades
of aluminum alloys). Following this, we will examine the effects of single-alloy
additions (i.e., binary alloys) and, finally, consider the commercial alloy systems.
A. Unalloyed Aluminum Alloys
Although some controversy still exists in the literature on the details of corrosion
initiation, there is now a strong opinion that the initial process occurring when
an oxide-filmed aluminum surface is first introduced into an aqueous environment, although being sensitive to electrochemical potential, is predominantly a
chemical process. In a comprehensive review covering over 70 years of published literature Foley (69) came to the conclusion that the initiation of localized
corrosion of aluminum could be described by four steps involving the following:
1. Adsorption of a reactive species (usually an anion) on the oxidecovered aluminum alloy surface
2. A chemical reaction of the adsorbed species with the aluminum ion in
the aluminum oxide lattice or precipitated aluminum hydroxide
3. The effective thinning of the oxide film by dissolution or
anions penetrating the film
4. Direct attack of the exposed metal by the anion, possibly assisted by
anodic potentials (sometimes, Steps 3 and 4 can occur simultaneously)
Step 1: Adsorption A range of analytical tools, including autoradiography, secondary ion mass spectroscopy (SIMS), and X-ray photoelectron
microscopy (XPS), have provided substantial experimental evidence for
the competitive adsorption of anions on aluminum surfaces [e.g., chloride (7075), sulfate (38,70,72,75), perchlorate (72), nitrate (38,72),
196
Holroyd
chromate (72,76), and molybdate (38,77,78)] and the subsequent influence upon whether or not aluminum will subsequently undergo dissolution. This is particularly the case for the adsorption of chloride ions as
a preliminary step to aluminum alloys suffering pitting corrosion (70
75).
Observed adsorption behavior is dependent on the anion species involved.
For example, experimental evidence provided by Tomcsanyi et al. (70)
using a radio-tracer technique on 99.99% pure aluminum at its free-corrosion potential in various aqueous solutions has shown the following:
1. Adsorbed labeled sulfate ions can be displaced by further additions of
nonlabeled sulfate ions, whereas they are not displaced by the further
additions of chloride ions (Fig. 14).
2. Chromate ions displace adsorbed chloride ions and electrochemical potentials move to more positive values (Fig. 15), whereas further additions of nonlabeled chloride ions do not displace adsorbed labeled chloride ions.
Fig. 14 Accumulation of labeled sulfate ions on pure aluminum surfaces: (A) exchange
when additional nonlabeled sulfate ions are added and (B) effect of adding 5 103M
sodium chloride. (Data from Ref. 70.)
Aluminum Alloys
197
Fig. 15 Effect of chromate ion additions to the surface accumulation of chloride ions
on pure aluminum in a 1M sodium sulfate solution: (A) a 0.1M Na2CrO4 addition to a
solution containing 4 102M sodium chloride and (B) a 5 102M Na2CrO4 addition
to a solution containing 2 102M sodium chloride. (Data from Ref. 70.)
198
Holroyd
Fig. 16 Time dependency of the surface accumulation of chloride ions on pure aluminum exposed to various bulk solution chloride concentrations in a 1M sodium sulfate
supporting electrolyte. (Data from Ref. 70.)
adsorption of chloride ions], whereas the next two are consistent with the
adsorbed chloride ions chemically rather than electrochemically
reacting with the oxide film. Justification of the latter suggestion is
that the adsorption behavior for a given surface condition is insensitive
to electrochemical potential and that although the concentration of the
adsorbed chloride increases with increasing bulk chloride solution, the
absolute concentrations always start decreasing after 1 h. Rationalization of the final observation that the adsorption behavior was not reversible was that the adsorption is sensitive to the surface roughness of
the aluminum substrate, which will be modified by the electrode potential
being held at a cathodic potential.
Tomcsanyi et al.s (70) suggestion that chloride adsorption on oxide-covered aluminum is insensitive to electrode potential would seem inconsistent with Augustynskis (72) results showing that adsorbed chloride concentrations increased from around 3 at% at the free-corrosion potential to
around 1213 at% at potentials approaching the critical pitting potential.
Aluminum Alloys
199
Fig. 17 Evidence that the concentration of accumulated chloride ions on aluminum surfaces depend on both time and the current surface condition. Effects of cathodic polarization were not reversible. (Data from Ref. 70.)
Rationalization of these differences is probably associated with the timedependent changes occurring after the chloride ions are initially adsorbed;
this will be discussed below in Step 2 of the proposed process, which
covers the chemical reaction of the adsorbed species with the substrate.
Further evidence supporting the proposal that chloride ions adsorb on aluminum surfaces and then react with the oxide is provided by the simple but elegant experimental work published by Nguyen and Foley (73),
who measured the uptake of chloride ions and the production of Al3
ions as a function of chloride concentration when 10 g of either aluminum powder or aluminum oxide were mixed with 100 mL of an aqueous
sodium chloride solution for 16 h. The results generated showed that the
chloride ions adsorption is consistently higher on aluminum oxide than
on freshly generated aluminum powder, and for both substrates, the adsorption increased with increasing bulk solution chloride concentrations
up to a maximum at around 0.8M sodium chloride. The results for alumina are summarized in Fig. 18, along with data from similar experiments on aluminum powder conducted by Drazic et al. (74), who used
200
Holroyd
Fig. 18 Adsorption of chloride ions onto alumina from a range of sodium chloride solutions. from Ref. 73; from Ref. 74.
Aluminum Alloys
201
concentration increased linearly on a one-for-one basis with the increasing surface area of alumina.
An argument for anion adsorption being locally enhanced can be made
using the fact that metal surfaces are rarely homogenous and enhanced
localized adsorption and surface activity will tend to occur preferentially at imperfections, weak spots, and flaws in the surface oxide. In the
early 1970s, Richardson and Wood (81,82) proposed that oxide flaws
in aluminum oxide act as active centers for the initiation of localized
corrosion. They supported their proposal with experimental evidence
from transmission electron microscopy (TEM) studies for the existence
of large flaws in aluminum oxide layers electrochemically grown on
aluminum substrates. Further studies by Wood and later by Thompson
and their co-workers have led to numerous publications generating evidence demonstrating that the initiation of the pitting of aluminum could
be via an oxide flaw mechanism. Irrefutable evidence is awaited;
however, the recent evidence from ultramicrotomed sections of pure aluminum exposed to an aqueous environment (83) is very convincing.
Step 2: Chemical Reaction of the Adsorbed Species with the Surface
Oxide Evidence for the solubility of oxide films on aluminum substrates immersed in aqueous environments was provided by Lorking and
Mayne (84,85) in the early 1960s. These workers generated nonporous
anodized layers on aluminum and then exposed them to a wide range of
anions in aqueous environments which included chloride, sulfate, benzoate, acetate, citrate, phosphate, and chromate anions. In all cases, dissolved aluminum species were detected within 24 h and the concentrations detected were directly related to the initial corrosion rates for all
the environments evaluated other than chloride. On the basis of these
findings, the authors concluded that the initial anodic reaction in the chloride solutions was the formation of soluble aluminum chloride, whereas
in all the other aqueous solutions, the anions promoted the repair of the
anhydrous oxide.
Perhaps the clearest evidence that a chemical reaction occurs between aluminum oxide and an aqueous environment is provided by the experiments conducted by Nguyen and Foley (73). These experiments involved
the mixing of 10 g of aluminum oxide or 10 g of fresh aluminum powder
with 100 mL of various aqueous saline solutions for 16 h and the subsequent measurement of the chloride ions adsorbed and the aluminum species generated. The data obtained indicated that while the chloride ion
concentration adsorbed on the alumina increases with increasing initial
chloride concentration (Fig. 18), the concentration of the dissolved aluminum species found in the aqueous solution decreases, with the highest
concentration being recorded when the alumina is exposed to doubly-
202
Holroyd
distilled water (Fig. 19). Results for the aluminum powder were different
in that the dissolved aluminum concentrations were considerable lower
and independent of the solution chloride concentration, and although the
adsorbed chloride ions increased with increasing solution chloride concentration, the absolute concentrations were always considerable lower
than on alumina (73).
Based on the above results, it is reasonable to conclude that aluminum oxide
in the presence of water undergoes hydration and that when chloride ions
are available, the positive surface charge will lead to adsorption and the
generation of aluminum hydroxychloro species such as Al(OH)2Cl and
Al(OH)Cl2 or soluble complexes of the form Al(OH)2Cl2 (86).
Even more compelling evidence supporting the latter proposal was provided by Nguyen and Foley (73) when they mixed 10 g of alumina or
aluminum powder with 100 mL in a 1N sodium chloride solution acidified with 0.01N aluminum chloride. In the experiments involving alu-
Aluminum Alloys
203
mina, the solutions pH became alkaline along with the dissolved aluminum concentration falling dramatically from its initial value of 83 mg/
L to a final value of 3.29 mg/L and the formation of a heavy gelatinous
precipitate, presumably via a chemical reaction of the type
Al3 2OH Cl Al(OH)2Cl
For the aluminum powder case, the dissolved aluminum concentration
increased from 82.9 to 140.0 mg/L and the solution pH remained acidic.
Presumably, in this case, the dissolution of the thin layer of oxide
on the aluminum powder occurs without the solutions pH increasing
significantly and the surface layers of the aluminum powder can then
dissolve in the acidic solution.
The generation of a heavy gelatinous precipitate in the case when Nguyen
and Foley (73) mixed alumina with an acidified saline solution provides
clear evidence of the chemical reactivity of aluminum oxide in acidic
aqueous environments. Another good indicator of the chemical reactivity of oxide films on aluminum surfaces exposed to aqueous saline
environments is given by Tomcsanyi et al.s (70) observation that the
concentration of excess chloride adsorbed on aluminum surfaces decrease with time (Fig. 16). This phenomenon was deemed by Tomcsanyi
et al. (70) to be a multistep heterogeneous transformation of the original
air-formed oxide into a mixture of oxo-hydroxo- and chloro- complexes,
with the details of the process being influenced by the initial surface
oxide and the local solution chemistry at the oxidesolution interface.
It could be wrongly concluded that chemical reactivity between the initial
surface oxide and an aqueous environment necessitates the presence
of aggressive species. Clear evidence that aggressive species are not essential is provided by evidence published by Scamans and Rehal (32).
In their electron microscopy studies, small surface blisters were observed
to form in the oxide when a wide range of aluminum-oxide-covered aluminum alloys (pure Al, AlMg, AlMgSi and AlZnMgCu) were
exposed to water-vapor-saturated air at 70C. Further evidence is provided by the experimental observation for 99.95% purity aluminum artificial crevices exposed to high-purity water, where the local solutions
pH conditions within the crevices become acidic, following a similar
time dependence as observed for most aluminum alloys exposed to aqueous saline solutions (87).
On the basis of the above discussion, it is reasonable to conclude that the
initial surface oxides chemically react with aqueous environments to
form modified surface films that, in the absence of aggressive species,
will exist in a state of equilibrium with the aqueous environment.
Although not yet proven, it would seem reasonable that inhibitor anions
204
Holroyd
such as chromate initially function by competitive adsorption on the initial oxide surface followed by the adsorbed species somehow modifying
the subsequent film-formation process and the film properties. Augustynskis (72) results from XPS experiments supports this view and
he concluded that after the chromate anions adsorb on aluminum oxide
surfaces, chemical reactions lead to hydrated chromium III and aluminum III oxides as well as chromium IV species being present in the
protective film.
Step 3: Effective Thinning of the Oxide Film by Dissolution or
Anion Penetrating the Film The traditional view of the oxide film
on aluminum participating in a corrosion reaction in an aqueous environment has been that it acts essentially as an anhydrous inert barrier
(88,89). Despite all the evidence to the contrary, it is still common for
the oxide to be considered as chemically inert in aqueous solutions
with pHs in the range 49. This is not an accurate view of the oxide
films.
A more realistic view of surface films participating in corrosion reactions
should be of hydrated aluminum oxide layers, probably in a colloidal
state (69) or as a semipermeable membrane (90). Unfortunately, these
ideas of films existing in a colloidal state or as semipermeable membranes have not received the attention they deserve. An exception to this
is the work conducted by Liepina and co-workers during the 1950s and
1960s (9195), in which the corrosion process was envisaged in terms
of colloidalchemical effects occurring on the metal surface. For aluminum in an aqueous potassium chloride solution, these authors suggest the
potential reaction sequence AlAlCl3 polyoxychloride intermediates
amorphous gelsboehmitebayeritehydrargillite, with the proviso that
if the aluminum oxide is in a colloidal state, the chloride ion would
peptize it and render it dispersible. [Interestingly, in 1967, Hoar (96)
suggested that the strong negative charge caused by adsorbed anions
could peptize oxide films. However, to the authors knowledge, this idea
was not pursued.]
As will be discussed further in Step 4 of the proposed process, the modified
oxide that is generated on aluminum surfaces exposed to aqueous
environments must have a significantly higher ductility than those reported for anhydrous alumina films either stripped from or while adhered
to an aluminum substrate (97).
In the very early investigations of the pitting of aluminum in chloride solutions, it was believed that the chloride ion had the ability to penetrate
the surface oxide (88,89), with the anion directly diffusing through the
aluminum oxide lattice. This mechanism has now been discounted and
Aluminum Alloys
205
the concept of penetration has been replaced with ones involving the
following:
1.
2.
206
Holroyd
Fig. 20 Blister growth potential given by Bargeron and Givens (105) compared with
the pitting potentials for pure aluminum in various aqueous chloride solutions quoted by
various authors. from Ref. 120; from Ref. 144.
2. Blister growth proceeds along the metaloxide interface with the oxide
exhibiting plastic behavior.
3. Blisters generally are circular, but the shape can be locally distorted
by grain boundaries and other microstructural features.
4. Corrosion initiates locally in a peripheral zone toward the blisters perimeters, prior to the blister bursting due to excessive internal gas pressure.
5. An oxide skin of some description remains in place over the ruptured blister which is capable of maintaining an occluded cell and
a localized solution chemistry.
6. Gas compositions generated from the blister/pits are specific to the
anions in the aqueous environment; for instance, Bargeron and Benson
(108) detected hydrogen exclusively for chloride solutions and de-
Aluminum Alloys
207
2.
208
Holroyd
Fig. 21 Schematic of a typical anodic polarization curve for aluminum and its alloys
in a saline environment.
Aluminum Alloys
2.
3.
209
Pits
Pits
Pits
Pits
4.
210
Holroyd
Fig. 22 Typical potentiostatic polarization curves for pre-anodized 99.99% pure aluminum in 3M KCl. [Note the 8-min induction time for pitting at E 0.81 V (SCE).] (Data
from Ref. 120.)
Aluminum Alloys
211
tion studies being more noble than values obtained from potentiostatic methods.
The reason for this is that by the time the induction time has elapsed, the electrochemical potential in a dynamic experiment has moved on from the potentiostatic Ep to a more positive value. Typical examples of this effect are provided
by the Ep data published by Foroulis and Thubrikar (119,120) from potentiostatic
and potentiodynamic polarization experiments on pure aluminum (99.99%) in
aqueous potassium chloride solutions at various chloride concentrations for a
given temperature and at various temperatures for a given solution chloride concentration. These data are reproduced in Figs. 23 and 24, along with comparable
data from other studies.
Induction times are a common feature of pit initiation under potentiostatic test conditions. Published data indicate that it is a statistical variable
(121,122) with induction times, :
Decreasing with increasing chloride concentration (98,122), potential
(112,122), or temperature (98,122),
212
Holroyd
Fig. 24 Effect of temperature on the pitting potential for pure aluminum in a 3M KCl
solution given by potentiostatic and potentiodynamic polarization methods. , from
Ref. 120; from Ref. 123.
Aluminum Alloys
213
Fig. 25 Experimental pitting potentials determined obtained using various anodic potential sweep rates for pure aluminum in de-aerated 3.5% NaCl and plotted on normal probability paper. (Data from Ref. 121.)
214
Holroyd
Fig. 26 Median values of the pitting potentials for pure aluminum in a de-aerated 3.5%
NaCl solution at various temperatures, plotted as a function of the square root of the anodic
polarization potential sweep rate. (Data from Ref. 121.)
Aluminum Alloys
215
216
Holroyd
concentrations, they are present in the form of a solid solution, whereas at higher
concentrations, they are also present as a second phase. For heat-treatable aluminum alloys, the situation is further complicated by the fact that heat treatment
may modify the microstructural form adopted in the alloy, and for non-heattreatable alloys, the same may be promoted by cold or warm working processes.
In view of the above situation, it is not too surprising that the precipitation
of electrochemically distinguishable second phases (e.g., CuAl2 in AlCu alloys
and Mg3Al2 in AlMg alloys) at preferential microstructural sites means that
alloys can suffer additional forms of localized corrosion to the pitting suffered
by unalloyed aluminum. These various forms of localized corrosion will be covered in the following discussion below, which will initially focus on binary aluminum alloys, move on to ternary alloys, and, finally, deal with the more complicated commercial aluminum alloys that contain multiple deliberate alloying
additions along with others present as impurities.
C. Binary Alloys
Many studies have been published on the corrosion of binary aluminum alloys
in aqueous environments. Initial studies concentrated on binary systems involving
the elements used as major alloying additions (i.e., AlCu, AlMg, AlZn, Al
Mn, and AlLi). In more recent times, in response to the goal of generating a
stainless aluminum alloy, corrosion studies have involved binary aluminum
alloys bases containing elements known to enhance the passivity of stainless
steels (e.g., Cr, Mo, Ta, Zr, Nb, and W). Because these elements all exhibit extremely low solid solubilities in aluminum (typically less than 1 at%), these researchers have had to employ novel techniques such as rapid solidification processes (e.g., splutter deposition or melt spinning) or ion implantation to generate
the alloys.
Detailed studies have also been conducted on various binary alloys containing alloying elements that activate aluminum (e.g., Hg, In, Sn, Zn) (57,131
133). This aspect of aluminum alloy corrosion will not be discussed in detail
here, as our focus is on aluminum alloys used in structural and engineering applications rather than those employed as sacrificial anodes or battery anodes.
The earliest corrosion studies on aluminum binary alloys not surprisingly
were conducted on the AlCu system, which, after being discovered in Germany
in 1906, resulted in the alloy duralumin and the first commercial exploitation of
an age-hardenable aluminum alloy in 1908 (134). It has been appreciated since
the early 1940s that the copper in a Al4% Cu (wt%) binary alloy may be retained
in solid solution if the alloy is solution heat treated and sufficiently rapidly
quenched. In this metallurgical state, the dissolved copper is thermodynamically
unstable and will preferentially precipitate as CuAl2 at grain boundaries if the
alloy is exposed to temperatures above around 120C or will have already precipi-
Aluminum Alloys
217
tated to some extent if the quench rate employed after solution heat treatment
was insufficient. This miscrostructural condition promotes a susceptibility to intergranular corrosion and researchers quickly realized that the localized corrosion
was related to the grain-boundary precipitation of CuAl2 (135,136). Dix and coworkers developed an electrochemical based model for the intergranular corrosion promoted in these situations. In this model (137), they suggest that the intergranular corrosion occurs along a narrow zone adjacent to the grain boundary
that is relatively depleted in copper and anodic to both the main body of the
grain and the CuAl2 precipitates on the grain boundary. (A schematic of the local
microstructure is given in Fig. 27.)
The weakness of the Dix et al. model for intergranular corrosion as it was
initially presented is that it relies solely upon whether a given phase is anodic
or cathodic with respect to another phase when exposed to a saline solution doped
with hydrogen peroxide. No consideration is given to the magnitude of the current
that would flow in a local cell or whether the poly-phase systems free-corrosion
potential would be greater or smaller than that needed to promote local corrosion.
Ideas presented by Galvele and his co-workers (138140) show that the Dix et
al. model for the intergranular corrosion of AlCu alloys can be justified if it
includes a consideration of the breakdown potentials of the phases present in the
system (141).
218
Holroyd
Fig. 28 Pitting potentials for various aluminum binary alloys in a de-aerated 1M NaCl
solution at 25C. from Ref. 140; from Ref. 140; from Ref. 142.
Aluminum Alloys
219
Fig. 29 Corrosion mode, pitting potential, and alloy hardness for an A13.3% Cu (wt%)
as a function of aging time at 240C. (Data from Ref. 142.)
220
Holroyd
Based on the above, it would seem reasonable that suitably heat-treated AlCu
binary alloys should display two pitting potentials during anodic polarization
studies: one for the grain-boundary region and a second at a higher electrode
potential for the grain interiors. Such results have been reported by Urushino and
Sugimoto (143) for an Al4 Cu alloy aged at 170C and polarized in a de-aerated
1M NaCl solution with its pH adjusted to 10. These results are reproduced in
Fig. 30 and the good fit with the Muller and Galvele data is obvious.
Muller and Galveles corrosion studies also included work on the AlMg
and AlZn binary-alloy solid solutions (140). Significant differences were readily
apparent between the various alloy systems. Alloying additions up to around 5%
(wt) led to pitting potentials in an aqueous 1M NaCl solution displaying positive
shifts for AlCu alloys, small effects for AlMg alloys, and shifts in the negative
direction for AlZn alloys (Fig. 28). A further significant difference noted was
the electrochemical potentials for the intermetallic phases precipitating from supersaturated solid solutions and those for the denuded zones adjacent to these
precipitates. In the case of AlCu alloys, the intergranular precipitation of Al2Cu-
Fig. 30 Effect of aging time at 170C on the grain boundary and matrix pitting potentials
for an A14 Cu (wt%) in a deaerated 1M NaCl solution with its pH adjusted to 10. (Data
from Ref. 143.)
Aluminum Alloys
221
generated adjacent regions that were relatively anodic, whereas for the Mg3Al2
precipitation in AlMg alloys, the precipitate itself was relatively anodic; see
data given in Table 2.
Pitting attack morphologies for the AlMg alloys were crystallographic,
closely resembling that previously reported by Galvele and de Micheli (144) for
pure aluminum, and there was no evidence of any magnesium buildup within
the pits. The pitting attack for the AlZn alloys was dependent on an alloys zinc
concentration, and microprobe studies indicated that zinc accumulation occurred
within pits. For low zinc levels, the pitting was crystallographic; however, as the
zinc level increased, the pits became more irregular and more tunnellike, probably
with some subsurface propagation (140).
Based on results from their experiments involving scratching oxide surfaces
during anodic polarization experiments, Muller and Galvele (140) concluded that
the oxides forming on various aluminum alloys had different properties. The
presence of oxide films on pure aluminum, AlCu, and AlZn did not interfere
with or accelerate the initiation of pitting, whereas films forming on the AlMg
alloys acted as a barrier to the pitting process, with pits for these alloys, unlike
the others, nucleating preferentially on the scratch lines.
Muller and Galvele (140) noted during their polarization studies on the
higher magnesium binary alloys (Al3 Mg and Al5 Mg) that the measured
currents became unsteady at potentials just below the pitting potentials. The ob-
Table 2 Published Pitting Potentials, Ep, for Various Aluminum Binary Alloys and
Aluminum-Based Precipitates in De-aerated 1M NaCl at 25C and 53 g/L NaCl 3
g/L H2O2
Alloy
Pitting potential of
de-aerated 1M NaCl
[V (SCE)] (Ref.)
Al2 Cu
Al4 Cu
0.774
0.777
0.654
0.594
Al1 Zn
Al3 Zn
Al3 Mg
Al5 Mg
Al2Cu
Al3Mg2
0.864 (140)
0.994 (140)
0.814 (140)
0.824 (140)
0.654 (144)
0.994a
Pure aluminum
(144)
(145)
(142)
(142)
Pitting potential
53 g/L NaCl 3 g/L H2O2
[V (SCE)] (Ref.)
0.764 (144)
0.644 (141)
0.598 (135)
0.604 (141)
0.874 (141)
0.994 (141)
0.784 (141)
0.794 (141)
0.644 (6, 144)
0.978 (135)
222
Holroyd
served transient peak currents were two to three times those needed to maintain
the passive film and occurred at increasing frequencies as the potential approached the alloys pitting potential. These authors did not comment on whether
these effects also occurred during their experiments on AlZn binary alloys; however, as will be discussed later, these effects are now known to occur for Al
Zn binary alloys in aqueous saline solutions (124,145).
Perhaps the most significant outcome from the above studies by Galvele
and his co-workers is that the results are consistent with his previously proposal
pitting model, where the pitting potential is the minimum potential for sufficient
localized acidification to be maintained at the metalsolution interface within a
pit (138). According to Galvele (138), the pitting potential is given by
Ep E p* Einh
(1)
where Ep is the pitting potential, E p* is the pitting potential of the alloy in a pitlike
environment, is the overpotential necessary to draw a net anodic current
through the pit, is the potential gradient through the pit, and Einh is the contribution due to inhibitors present in the solution.
Substitution of appropriate values for 99.99% purity aluminum, Al3 Cu,
Al3 Mg, and Al3 Zn into Eq. (1) yields the results given in Table 3. The
predicted pitting potentials are in excellent agreement with those determined experimentally. [Values for E p* are based the alloys corrosion potential measured
in a saturated AlCl3 solution after 2 h exposure (140). Values for were taken
as the polarization above the corrosion potential necessary to maintain an alloys
anodic current density at 1 mA/cm2 in a saturated AlCl3 solution; see Table 3.
for a 1M NaCl solution is approximately 0.050 V (138) and because there are
no inhibitors Einh 0.]
Sato and Newman (145) have recently published elegant experimental studies on the role zinc additions up to 5% (wt) play during the activation of pitting
corrosion for high-purity binary AlZn alloys in a de-aerated aqueous 0.5M NaCl
solution. Their results are consistent with Zn effects on the pitting potential being
the following:
E *p
ETheory
Ep
1.00
0.76
0.10
1.06
0.17
0.04
0.15
0.01
0.05
0.05
0.05
0.05
0.78
0.67
0.80
1.0
0.77
0.64
0.81 to 0.77
0.99
Aluminum Alloys
223
224
Holroyd
Fig. 31 Normal probability plot of the pitting potential for superpure (99.999%) aluminum and A10.13 Zn (wt%) in a de-aerated aqueous 0.5M NaCl solution. (Data from
Ref. 124.)
Aluminum Alloys
225
Corrosion studies on AlLi-based alloys increased when the aerospace industry became interested in the alloy systems potential promise of a 10% increases in stiffness combined with a 10% density reduction over the conventional
high-strength aluminum alloys (146). Published work on AlLi binary alloys
indicates that the corrosion resistance in saline solutions is no worse than that
observed for pure aluminum. Corrosion occurs as crystallographic pitting and is
insensitive to thermal aging, irrespective of whether the AlLi binary alloy is
single phase, with the lithium totally in solid solution, or the lithium also present
in the alloy as the normal age-hardening precipitate, , Al3Li (147,148). The
only exception is when AlLi binary alloys are subjected to extended overaging
heat treatments and enhanced corrosion rates are observed with pits initiating
and spreading from grain boundaries (147). The latter has been attributed to the
extended overaging promoting grain-boundary precipitation of , AlLi, a highly
anodic phase (149,150).
The reason why, unlike AlCu and AlMg binary alloys, the corrosion
behavior of AlLi binary alloys are insensitive to the alloy addition being either
in solid solution or precipitated as a hardening phase is twofold. First, the Al3Li
phase promoting age hardening is a coherent phase and the electrochemical potential differences between it, the adjacent solute-depleted zones, and the grain interiors are believed to be minimal and, second, lithium dissolution can lead to
beneficial modifications to the surface films generated in aqueous environments.
Over the last 20 years, attempts have been made to develop stainless
aluminum alloys that will offer high resistances to localized corrosion in chloride
environments. The approach adopted has been to use nonequilibrium methods
that will simultaneously do the following:
1. Eliminate the second phases, usually present in conventionally cast
alloys, that are known to be detrimental to corrosion resistance
2. Produce supersaturated alloys containing elements thought to enhance
passivity
Nonequilibrium methods employed to introduce a wide range of elements in aluminum include ion implantation (109,151153) and rapid solidification methods
such as melt-spinning (154) and sputter deposition (155161).
One of the earliest studies involved using ion implantation to introduce
Mo ions or Ar ions into 99.99% pure aluminum and a high-strength AlZn
MgCu alloy, 7075-T6 (151). Implantation of Mo ions into the pure aluminum
promoted a measurable improvement in the pitting corrosion resistance in an
aqueous saline environment with the pitting potential, Ep, increasing by around
200 mV, whereas for 7075-T6, the effect was less pronounced with the Ep only
increasing by around 100 mV. The authors suggested that these beneficial effects
were due to either the incorporation of Mo into the passive film or the reprecipitation of some Mo-containing species on the passive film. Effects due to Ar ion
226
Holroyd
implantation into either pure aluminum or the 7075-T6 were limited to small
increases in the measured Ep values which were attributed to a slight thickening
of the air-formed oxide films induced by the ion implantation.
Corrosion data in de-aerated 0.5N NaCl has been provided by Yoshioka et
al. (154) for a wide matrix of rapidly solidified aluminum binary alloys containing
2, 4, and 6 (at%) Mg, Ti, Mn, Cr, Fe, Ni, Cu, Zn, Zr, and Si produced by a meltspinning process. Pitting potentials for all the alloys were significantly ennobled
(and for pure aluminum) except for the alloys containing Mg, Fe, and Zn. Corrosion rates at the free-corrosion potential showed similar trends that reflected the
observed decreases in both the anodic and cathodic current densities. These beneficial effects were attributed to both the supersaturation of the aluminum solidsolution phase with solute atoms and to the elimination or decrease in the number
of intermetallic phases available to initiate pitting.
Extensive work studying the evolution of passive film chemistry on a range
of sputter-deposited AlX binary alloys, where X included Cr, Ta, Mo, and W,
was conducted at the Martin Marietta Research Laboratories by Moshier and
coworkers during the mid-1980s and the early 1990s (155159). Beneficial effects on pitting resistance promoted by the various additions should be obvious
from Fig. 32, showing a compilation of anodic polarization curves from these
studies for pure aluminum and a range of the supersaturated aluminum binary
alloys obtained in 0.1M NaCl. X-ray photospectroscopy (XPS) studies on the
passive films formed on AlMo, AlCr, and AlTa alloys after exposure to a
0.1M NaCl solution revealed that the films contained 510% of the oxidizing
solute as MoO4, CrOOH, and Ta2O5, respectively, with the concentrations increasing fivefold as the overpotential increased from the open-circuit potential
to the pitting potential (156,157). On the basis of these results, the authors proposed that the enhanced pitting corrosion resistance resulted from the above species somehow impeding chloride ingress through the passive film. A different
explanation was invoked for AlW alloys because although W was very effective
in improving the pitting resistance (Fig. 32), the XPS studies failed to detected
any W in the passive films. Here, it was proposed that an undefined synergistic
effect between W and Al2O3 somehow led to a more stable oxide layer at the
metaloxide interface (158).
Natisham et al. (152) offer an alternative hypothesis to account for the
beneficial effects on the pitting corrosion resistance found for a range of aluminum binary alloys produced by an ion-implantation method. Here, the authors
assume that the rate-determining step (RDS) for pitting is the adsorption of chloride ions on to an alloys filmed surface and, hence, it is critical that the oxide
film covering the alloy must be positively charged so that chloride ions will be
adsorbed. As evidence supporting this proposal, they presented a correlation between the pitting potentials for Ta, Ti, Cr, Cu, Fe, Al, Zn, and Mg and the pHZCH
Aluminum Alloys
227
(pH of zero charge) for the metals hydrated oxide in an aqueous 1N NaCl solution
(152). More recent studies have questioned the validity of Natisham et al.s mechanistic interpretation and the relevance of the correlation between the pitting potential and pHZCH. Perhaps the most convincing of these arguments is provided
by the work of Vijh (162), who suggests that the correlation between Ep and pHZCH
exists because a linear relationship also exists between pHZCH and the metalmetal
bond energy b(MM) (163). Vijh suggests that the critical relationship is the one
between Ep and b(MM) (Fig. 33) and not the one cited by Natisham et al. The
acceptance of b(MM) as the dominant parameter rather than pHZCH is relatively
straightforward if the RDS for pitting is attributed to a process such as the devel-
228
Holroyd
Fig. 33 A plot of pitting potentials of various aluminum alloys against the metalmetal
bond energy b(MM) values of the alloying element in the binary alloy. (Data from Ref.
162.)
opment of appropriate local solution conditions within a pit rather than one involving an aggressive species surface adsorption on to or its transport through
a passive film.
Szklarska-Smialowska (164) has proposed yet another mechanistic interpretation. The RDS for stable pit growth in her model is deemed to be the establishment of an appropriate solution chemistry within a pre-existing defect in the
film that can support active dissolution of the metal at the metaloxide interface.
This is a variant of the Galvele model for pitting (138), with the main difference
being an extra caveat in that the envisioned role for the alloying addition is to
establish the level of acidification needed to dissolve the alloying additions oxide
Aluminum Alloys
229
in the pit environment. For instance, in this model, an alloying addition can either
cause Ep to increase by providing an oxide that is less soluble than alumina in
an acidified pit environment (e.g., Cr and W oxides) or to decrease Ep by providing an oxide that is more soluble. Justification for the passive films themselves
not being involved in the RDS is based on the following premise:
1. The observation that the passive current densities for pure aluminum
and AlX alloys in a given saline environment are effectively identical
(see data in Fig. 32) and, hence, so are the physical and/or electrical
properties of the passive films.
2. The passive films forming on pure aluminum in aqueous saline environments are not effective obstacles to the penetration of chloride ions
and water to the metal surface (165)
3. Support for item 2 is provided by the immediate generation of electrochemical current noise when aluminum is immersed in chloride solutions as reported by Uruchurtu and Dawson (166)
Inturi and Szklarska-Smialowska (165) used the model to explain the beneficial
effects on the pitting characteristics of sputter-deposited Al4 Cr, Al9 Cr, Al
10 Ta, and Al17 Ta (at%) alloys exposed to a de-aerated 0.1M NaCl solution.
The observed open-circuit potentials (OCP) and the passive current densities for
these alloys were similar to pure aluminums, other than the OCP for the Al
17 Ta alloy, which was more positive. This was attributed to the Al17 Ta alloys
air-formed oxide containing Ta5 species as well as Al3 species, whereas Auger
electron spectroscopy only detected Al3 cationic species in the air-formed oxide
films on the other alloys.
Vijh (162) has suggested that the correlative trend between the aluminum
binary-alloy pitting potentials and the solubility of the alloying element oxide
quoted by Szklarska-Smialowska (164) is heavily weighted by the presence of
W and Ta among the nine elements studies. He presents an alternative interpretation of the data presented by Inturi and Szklarska-Smialowska (165) based on a
Galvele-type model with active dissolution of bare metal initiating in Richardson
and Wood (81,82) type of microfissures and defects in the oxide films. In support
of his interpretation, Vijh demonstrates that the pitting potentials of the binary
aluminum alloys quoted by Inturi and Szklarska-Smialowska (165) show the expected relationship with the solid-state cohesion metal-to-metal bond energies
b(MM) of the alloying element (163); see Fig. 33.
Some of the controversies outlined above may be reconciled using data
and ideas from a study on the pitting resistance of sputter-deposited aluminum
alloys presented by Frankel et al. (161). These authors used thin-film sputter thick) and measured pitting potentials, Ep, and
deposited alloys (10002000 A
the repassivation potentials, ER, in an aerated aqueous 0.1M NaCl solution. The
230
Holroyd
alloys evaluated included AlNb, AlMo, AlCr, and AlW with solute concentrations up to around 35 (at%).
A major advantage for the Frankel et al. study was that the thin films used
in the corrosion studies led to two-dimensional pits that accurately simulate the
behavior of small three-dimensional pits (167). This overcomes the usual problems encountered with pit depth and ohmic potential drop and/or diffusional path
issues and facilitates the generation of highly reproducible repassivation potentials, ER (167). Important findings from this study include the following:
1. For a given alloy system, the anodic pit current densities just prior to
repassivation do not increase with the pitting resistance or an alloys
solute content.
2. Although the pitting potentials of many of the sputter-deposited alloys
were significantly higher than that for pure aluminum, the Ep for a
given alloy was only slightly higher than its repassivation potential.
Based on these findings, the authors concluded that although alloy solute enrichment in the alloys influenced the local environmental conditions needed for pit
growth, any effects on the passive films themselves were relatively minor, even
when the beneficial effects due to film modification from long-term aging
(laboratory air exposure for 4 years), displayed by some of the alloying systems,
was taken into account.
In essence, the work by Frankel et al. provides a link between the nonequilibrium aluminum binary alloys and the earlier pioneering work of Galvele and
his co-workers on conventional cast-aluminum binary alloys.
D. Ternary and Commercial Alloys
As discussed earlier, the corrosion behavior of AlLi binary alloys in saline solutions is independent of alloy heat treatment other than when alloys are grossly
overaged and ,AlLi, a highly anodic phase, forms at grain boundaries
(149,150). Copper additions to AlLi alloys significantly modifies this situation
with thermal aging at temperatures around 170C promoting copper-containing
precipitation at subgrain boundaries, T1(Al2CuLi) followed by T2(Al6CuLi3) at
high-angle grain boundaries when tempers approach peak strengths (168). The
intergranular corrosion susceptibility for both AlLiCu alloys and the commercial alloy 2090 have been explained by the selective dissolution of a copperdepleted zone (147), as previously proposed for AlCu alloys (137,142). The
situation for AlLiMgCu alloys (e.g., 8090) is slightly different, with the presence of magnesium leading to different precipitates being developed during
aging. The influence of alloy temper on the mode of localized attack and pitting
potentials for these alloys, including 2091 (169), however, closely resembles that
Aluminum Alloys
231
Fig. 34 Effect of aging time at 170C on the grain boundary and matrix pitting potentials
for 2024 in a de-aerated 1M NaCl solution adjusted to 10. (Data from Ref. 143.)
observed for the AlCu and the AlMgCu alloy systems, (compare Figs. 30,
34, and 35 for Al3.3 Cu, 2024, and 8090, respectively).
Similar trends have been reported for AlZnMgCu alloys by Maitra and
English (170). They report that the temper dependence of the pitting potentials
for the AlZnMgCu alloy 7075 in an aqueous saline solution is remarkably
similar to that reported by Muller and Galvele (142) for AlCu alloys. Their
anodic polarization studies on 7075 in a de-aerated 3.5% NaCl solution show the
following:
1. A single pitting potential was detected for the solution heat-treated and
rapidly quenched condition.
2. Two pitting potentials existed for the peak aged temper, the less noble
one being associated with the local breakdown in the grain-boundary
region and second corresponding to pitting of the matrix.
3. Sufficiently overaged alloy microstructures displayed a single pitting
potential that corresponded to pitting of the copper-depleted solidsolution matrix.
232
Holroyd
Fig. 35 Efect of aging time at 170C on the grain boundary and matrix pitting potentials
for 8090 in a de-aerated 3.5 (wt%) NaCl solution adjusted to 6.5. (Data from Ref. 148.)
Although the microstructural details of the precipitates formed differ for the various age-hardening aluminum alloy systems, the influence of alloy temper on the
modes of localized corrosion suffered are self-consistent for the AlCu(Mg)-,
AlZnMgCu-, and AlLi-based systems (i.e., 2xxx, 7xxx, and 8xxx series
alloys).
Aluminum Alloys
233
234
Holroyd
in 1M NaCl conducted by Baumgartner and Kaecshe (175). These researchers showed the following:
(a) Crevice corrosion can be triggered by unstable micropitting at potentials as low as around 1.040 V (SCE), some 0.3 V below the
pitting potential Ep and at least 0.1 V (SCE) below the repassivation potential ER for the 1M NaCl bulk environment.
(b) Unstable micropitting occurred during the incubation period for
crevice corrosion, which also occurs when aluminum surfaces in
noncrevice situations are exposed to aluminum chloride or acidified NaCl solutions.
(c) When a critical solution chemistry is established inside a crevice,
the mode of attack switches to a more general type of attack, operating at a low current density of around 10 A/cm2.
Based on these findings, Baumgartner and Kaecshe (175) suggest that
the buildup of a critical acidic electrolyte is a sufficient requirement for
the initiation of crevice corrosion. Other researchers have disagreed,
suggesting that the requirement is either a critical concentration of aluminum ions that is the chloride ion concentration dependent (174) or
the presence of basic aluminum chloride (171).
4. Aggressive ionic species in the aqueous environment modify the oxide hydration process and gel-like anion-selective surface membranes form (90), allowing water entry and an inward migration of
specific anions. In this scenario, dissolved aluminum species (mainly
Al3) and aggressive anions (e.g., Cl) build up at the metalfilm interface beneath the gel-like films and lead to the initiation of localized
corrosion at active microstructural sites (Fig. 36). (It may be argued
that situation 3 is a special case of situation 4) for which the local
bulk environment inside a restricted geometry (e.g., a crevice or
crack) changes and then process 4 occurs.)
Recent time-lapse video studies (103) have shown that continuous gel-like
corrosion films form on aluminum alloy surfaces exposed to saline environments.
Gas bubble generation occurs at sites beneath these films and, in some cases,
leads to local eruptions and the exposure of sites where localized corrosion has
become established. Although it is clear that the mechanism of localized corrosion initiation needs further detailed study, it is reasonable to conclude that the
initiation of localized corrosion for aluminum and its alloys in aqueous environments normally occurs by either situation 3 or 4, irrespective of whether it occurs
under total immersion, thin-film, or wetdry environmental conditions.
Surface films may have several roles during the propagation phase of localized corrosion:
Aluminum Alloys
235
236
Holroyd
Fig. 37 Typical surface films and the solution is pHs developed as a function of depth
within restricted geometries for aluminum and its alloys exposed to saline solutions.
Saturation
concentration
(M)
Solution pH
(90% sat.)
3.02
3.12
2.34
3.11
2.85
2.94
3.67
0.15
Aluminum Alloys
237
Further support for these ideas is provided by Wong and Alkire (179), who have
characterized solution chemistries developing within naturally occurring 100m-deep pits in 99.999% purity aluminum in 1M NaCl. Their results indicate
the following:
1. The solutions pHs in pits are between 3 and 4 and the acidity can
be explained by the hydrolysis of Al3 ions, as has been suggested by
others for solutions within pits (181), crevices (87,182,183), and cracks
(87,183188).
2. Eighty-four percent of the measured dissolve aluminum in the pits is
present as monomeric species and the remainder is present as polymeric species with no evidence of solid gels.
3. The hydrolyzed aluminum species react within the pit environment to
form basic salts that have nuclear magnetic resonance spectra very similar to those given by synthetic solutions of Al(OH)Cl2 or Al(OH)2Cl.
4. Current densities from single pits decreased roughly with the square
root of time and are consistent with the corrosion rate being controlled
by the rate of dissolution of an aluminum oxychloride salt film formed
by the precipitation of dissolution products.
Several authors (181,185,186,189,190) in the past have suggested that the
solutions pHs developed within restricted geometries (e.g., pits, crevices, and
cracks) for aluminum alloys exposed to aqueous saline environments are controlled by acid hydrolysis involving Al(OH)3. For instance, Pryor (181) suggested
that in practical situations, the solutions pH in aluminum pits would not fall
below 3.54.0 and he calculated a minimum possible pH of 2.8 using thermodynamic data for the chemical equilibrium process:
Al3 3OH Al(OH)3
(2)
A major flaw in the above argument is that the species Al(OH)3 is unstable in
solutions with pHs below 5 (87,177,191,192) and, therefore, it is unreasonable
to assume that pHs are being controlled by Eq. (2). Recent theoretical work by
Galvele (192) indicates that no solid-reaction products are expected to form in
aluminum crevice solutions with pHs of 3 and below, irrespective of bulk solution pHs, and only 10% of the products would be solid in a crevice solution
with a pH of 4.
Several researchers have suggested that the solutions pHs within restricted geometries are well explained by the hydrolysis of Al3 ions to form
AlOH2 and H ions via the equilibrium process (62,87,177,178):
Al3 H2O AlOH2 H
(3)
The more recent work of Wong and Alkire (179) is consistent with Eq. (3) but
indicates the difficulty in differentiating between Al(OH)2Cl and Al(OH)Cl2 and
238
Holroyd
AlOH2 H
Cl
Cl
AlCl2
Al(OH)Cl
Al(OH)2Cl2
(1)
along with published pitting potentials and taking Einh 0, we can calculate
values for where is the overpotential needed to draw a net anodic current
through a restricted geometry and is the potential gradient within the restricted
geometry. Experimental work is needed to validate the ( )calc values given
Table 6.
Vermilyea and Vedder (39) studied the filming behavior of pure aluminum
Aluminum Alloys
Table 5
Environmental Conditions Developed Within Restricted Geometries for Various Aluminum Alloys
Restricted geometry
Alloy
Solution
pH
Internal
potential
External
potential
Al3 conc.
(M)
Cl conc.
(M)
Bulk NaCl
conc. (M)
Ref.
Crevices
1 mm diameter
10 mm deep
80 mm deep, 90 m wide
15 mm diameter 30 mm
Cracks
6 mm long
15 mm long
20 mm long
Pit
100 m deep
1199
2017
3003
6063
7475-T6
7475-T6
LC4-T6
7075-T6
4
4
3.5
3.5
4
7475-T6
LC4-T6
3.4
33.2
2.7
3.2
S.P. Al
34
0.830
0.825
0.844
0.916
0.940
0.870
0.920
0.925
0.780
0.773
0.890
0.780
0.790
0.830
0.850
0.832
0.830
0.003
0.25
0.0250.37
0.1
0.75
23
1.82
1.0
0.5
0.5
0.5
0.6
0.51
0.6
1.0
1.0
0.51
0.6
187
182
182
182
194
87
184
187
177
87
184
179
239
240
Holroyd
Table 6 Calculated Values for ( ) from Eq. (1) Using Published Data
Alloy
Pure Al
(in 1M NaCl)
3003
(in 0.5M NaCl)
7475-T6
(in 0.6M NaCl)
Ep (Ref.)
E*p (Ref.)
( )calc.
0.740 (222)
0.830 (187)
crevice
0.844 (182)
crevice
0.870 (194)
crevice
0.090
0.713 (222)
0.760 (170)
0.131
0.110
in boiling water containing 103M of various anions. Takahashi et al. (176) have
classified these results into four groups as a function of their ability to inhibit
hydration:
1. No inhibition: borate, MnO4, ClO4, NO3, NO2, Cl, and CO32
2. Moderate inhibition: IO3, SO42, SeO42, GeO42, CrO42, citrate, oxalate, and MNO42
3. Strong inhibition: Teo42, SiO42, WO42, AsO42, and IO42
4. Extremely strong inhibition: PO43
Generalization of these findings needs care because the behavior of some anions
are sensitive to their concentration and/or pH and temperature. Phosphate ions
is a good example. At low concentrations (103M102M), oxide hydration is
completely suppressed (as discussed earlier in this chapter); at intermediate concentrations (101M), oxide films initially suffer both dissolution and hydration
with a dissolutiondeposition process generating a surface film; at higher
phosphate concentrations, the dissolution of the oxide is accelerated by the formation of anion complexes that themselves hydrolyze to form Al-hydroxo-phosphate
complexes that then undergo polymerization to deposit basic phosphate or
hydrated oxides containing phosphate layers (176).
The presence of several anions in an aqueous solution may promote unexpected results with regard to the corrosion performance of aluminum alloys in
aqueous environments. Three examples follow.
Becerra and Darby (195) found that the corrosion rates for AA1100-H14,
AA5052-H32, and AA6063-T4 sheet materials all increased with additions of
sodium bicarbonate (0225 ppm) or cupric sulfate (02 ppm) added as single
additions to a 3.5 wt% NaCl solution. The increased corrosion rates were greater
than simply additive when the additions were added simultaneously. The synergistic effect was attributed to copper plating out on the aluminum alloy surface
and acting as a cathode [as suggested by Davies (196) and Fraker and Ruff (197)]
and the bicarbonates buffering capability limiting the local solutions pH
Aluminum Alloys
241
changes at the cathodes and reducing the tendency for local protective film formation.
Overaged AlZnMgCu alloys usually provide a good resistance to intergranular corrosion in aqueous saline solutions (170). Maitra and English (198)
have shown that small additions of nitrate and sulfate ions (particularly nitrate
ions) can cause 7075-T7351 to suffer intergranular corrosion during laboratory
testing. They also found that the relative susceptibilities promoted by the different
solution chemistries were reflected by the difference between the grain-boundary
and matrix pitting potentials. These results suggested the possibility that overaged
high-strength AlZnMgCu alloys, although being highly resistant in a saline
environments containing just chloride ions, may be prone to either intergranular
corrosion or stress-corrosion cracking (SCC) in certain industrial environments
(198). To date, there is no evidence of service performance issues with localized
corrosion occurring in overaged high-strength AlZnMgCu alloys. However,
it is interesting to note that the stress-corrosion performance of 7075-T7551 and
7075-T7351 during a extensive ongoing test program on various aluminum alloys
in rural, industrial, and seacoast environments (199) has identified that these alloy
tempers are significantly more susceptible to SCC in an industrial environment.
Yet another example of synergetic effects between anions influencing the
localized corrosion behavior of aluminum alloys is that reported for the AlLi
MgCu alloy (8090) and the AlLiCu alloy (2091) where small additions of
sodium sulfate to aqueous sodium chloride solutions have had marked effects
on the alloys SCC susceptibility (200,201). For 8090-T651, Craig et al. (200)
demonstrated that whereas 8090-T6 was resistant to SCC when under totalimmersion conditions in an aqueous 0.6M sodium chloride solution, it became
susceptible with small additions of sulfate ions but highly resistant with higher
levels of sulfate additions (Fig. 38). Similar findings have been reported by Marsac et al. (201) for 2091, albeit this alloy simultaneously suffers a susceptibility
to intergranular corrosion that seems to be independent of the solutions sulfate
concentration. A likely explanation of the sulfate effect at low concentrations is
that the surface filming process is modified by the introduction of a thermodynamically more attractive reaction path allowing the formation of aluminum oxysulfate species which will form in competition with the aluminum oxychloride
species, as has been described by Foley and Nguyan (202). For the high concentrations of sulfate ions, it could be that basic aluminum sulfate films form.
Based on their findings, Craig et al. (200) predicted that 8090-T651 would
suffer SCC when under total-immersion conditions in artificial or natural seawater. This was observed, and the susceptibility being lower than predicted on the
basis of the solutions sulfate-to-chloride ratio (Fig. 38) was attributed to filming
effects promoted by the additional soluble additives in seawater (e.g., magnesium
salts).
242
Holroyd
Fig. 38 Time to failure for peak aged 8090 stress-corrosion samples tested under totalimmersion conditions in a range of aqueous environments containing various concentrations of sodium sulfate and sodium chloride at 20C. (Data from Ref. 200.)
Aluminum Alloys
243
244
Holroyd
flowing conditions, despite the combination of localized corrosion, general corrosion, and bulk environmental flow, creating a complex set of differential equations that are difficult to solve (217).
Most published studies on the effect of flowing conditions on the corrosion
of aluminum alloys have involved marine environments (212216,218220) and
are associated with applications such as marine vessels, structural and nonstructural uses on offshore oil platforms, ocean thermal energy conversion (OTEC) heat
exchangers, and desalination plants. A notable exception being Witts (221) studies on the effects of flowing conditions have on aluminum alloy (AA6063 type)
domestic radiator systems.
Godard and Booth (218) reported that the pitting corrosion rates for seawater flowing through 1S-H18 aluminum (0.02 Cu, 0.21 Fe, 0.11 Si, 0.016 Ti) pipes
initially decreased with increasing flow rates and then increased with further increases in flow rate. Unfortunately, only relative flow rates are available. Beneficial effects of low flow rates on localized corrosion rates have also been reported
for various alloys used in OTEC heat exchangers (214,215) and in domestic hotwater radiator systems (221).
Larson-Basse and co-workers (214216,219) have evaluated the corrosion
performance of a range of aluminum alloys [AA3003 and 5052 and Alcald (7072)
3003 and 3004] in flowing seawater in OTEC heat exchangers using flow rates
in the range 1.42.4 m/s. The observed localized corrosion behavior in the tropical surface water differs significantly from that promoted in the cold, deep ocean
water (214,216). In the warm surface water, a surface film forms after an initial
short period (510 days) of relatively rapid corrosion which is comprised of both
scale minerals precipitated from the seawater and aluminum corrosion products
that are alloy dependent (214). This film is protective the corrosion rates fall to
around 3 m/year for all the alloys tested. Unfortunately, the film thickness increases in a parabolic manner and is sufficiently detrimental to the systems heattransfer characteristics that it has to be regularly removed. This introduces challenging problems with both material selection and choice of method to remove
scales (216).
Exposure of aluminum alloys to the cold, deep ocean water under flowing
conditions as used in OTEC heat exchangers introduces little or no tendency for
biofouling; however, the alloys usually suffer localized pitting, with the incubation times being sensitive to small variations in solution chemistry (214) and
pitting tendencies increasing significantly with decreasing flow rates (214).
Witt (221) has reported that residual water left in a hot-water radiator system for a few hours after a successfully operating system has been emptied can
locally disrupt the protective film that had previously provided protection, and
intergranular corrosion can result during the subsequent operation of the radiator
system. Once again, this provides clear evidence that the corrosion performance
of an aluminum alloy in a given aqueous environment is not always predictable
Aluminum Alloys
245
and can be dictated by the surface film already present or one forming upon
immersion in an aqueous environment.
REFERENCES
1. RB Mears. Aluminum and Aluminum Alloys. In: HH Uhlig, ed. Corrosion Handbook. London: J Wiley, 1948, p. 48.
2. WW Binger. Aluminum and its alloys. In: FL LaQue, HR Copson, eds. Corrosion
Resistance of Metals and Alloys, 2nd ed. New York: Reinhold Publishing Corp.,
1963, p. 181.
3. HP Godard. Aluminum. In: HP Godard, WB Jepson, MR Bothwell, RL Kane, eds.
The Corrosion of Light Metals. New York: J Wiley, 1967, p. 3.
4. JR Davis, ed. Corrosion of Aluminum Alloys. Materials Park, OH: ASM International, 1999.
5. DO Sprowls, RH Brown. Stress corrosion mechanisms for aluminium alloys. In:
RW Stechle, AJ Forty, D van Rooyen, eds. Fundamental Aspects of Stress Corrosion Cracking. Houston, TX: National Association of Corrosion Engineers, 1966,
p. 466.
6. RH Brown, DO Sprowls, MB Shumaker. The resistance of wrought high strength
aluminium alloys to stress corrosion cracking. In: Stress Corrosion of MetalsA
State of the Art. Philadelphia: American Society for Testing and Materials, STP
518, 1972, p. 87.
7. MO Speidel. Metall Trans 6A:631, 1975.
8. NJH Holroyd. Environment induced cracking of high strength aluminium alloys.
In: RP Gangloff, MB Ives, eds. Environment-Induced Cracking of Metals. Houston,
TX: National Association of Corrosion Engineers, 1989, p. 311.
9. TD Burleigh. Corrosion 47:89, 1991.
10. RP Gangloff. Corrosion fatigue crack propagation in metals. In: RP Gangloff, MB
Ives, eds. Environment-Induced Cracking of Metals. Houston, TX: National Association of Corrosion Engineers, 1989, p. 55.
11. HP Godard. Aluminum. In: HP Godard, WB Jepson, MR Bothwell, RL Kane, eds.
New York: Wiley, 1967, p. 11.
12. H Dolling, FE Faller, W Gruhl, E Cossack, G Scharf, T Singe. Erzmetallurgica,
32:161, 1979.
13. M Textor, R Grauer. Corros Sci 23:41, 1983.
14. A Csanady, T Turmezey, I Ime-Baan, A Griger, D Marton, L Fodar, L Vitolis.
Corros Sci 24:237, 1984.
15. O Lunder, K Nisancioglu. Corrosion 44:414, 1968.
16. K Wafers, C Misra. Alcoa Technical Paper No. 19 (Revised), Alcoa Laboratories,
Pittsburgh, PA, 1987.
17. PE Blackburn, EA Gilbransen. J Electrochem Soc. 107:944, 1960.
18. W Vedder, DA Vermilyea. Trans Faraday Soc. 65:561, 1969.
19. CN Coehran, WC Sleppy. J Electrochem Soc. 108:322, 1961.
20. AJ Brock, MJ Pryor. Corros Sci 13:199, 1973.
246
Holroyd
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
54.
55.
56.
57.
58.
Aluminum Alloys
247
248
Holroyd
93. L Liepina, A Vaivade. Latvijas PSR Zinatnu Akad Vestis Kim Ser No. 3:297, 1963;
No. 6:683, 1963; 441, 1964; No. 4:527, 1964.
94. L Liepina, V Kadek. Corros Sci 6:177, 1966.
95. L Liepina. Latvijas PSR Zinatnu Akad Vestis No. 3:55, 1967.
96. TP Hoar. Corros Sci 7:341, 1967.
97. JC Grossman. J Electrochem Soc 116:1232, 1969.
98. ZA Foroulis, MJ Thubrikar. J Electrochem Soc 122:1296, 1975.
99. RT Foley. J Electrochem Soc 122:1493, 1975.
100. GT Burstein, RJ Cinderey. Corros Sci 33:475, 1992.
101. GT Burstein, RJ Cinderey. Corros Sci 3:1195, 1991.
102. PMA Sherwood, JA Rotole. U.S. Patent 6,066,403, May 23, 2000.
103. CJ Newton, NJH Holroyd. Time lapse-video techniques in corrosion testing aluminum alloys. In: VS Agarwala, GM Ugiansky, eds. New Methods for Corrosion
Testing of Aluminum Alloys. Philadelphia: ASTM, STP 1134, 1992, p. 153.
104. CB Bargeron, RB Givens. J Electrochem Soc 124:1845, 1977.
105. CB Bargeron, RB Givens. Corrosion 36:618, 1980.
106. CB Bargeron, RB Givens. J Electrochem Soc 124:1230, 1977.
107. CB Bargeron, RB Givens. J Electrochem Soc 129:340, 1982.
108. CB Bargeron, RC Benson. J Electrochem Soc 127:2528, 1980.
109. PM Natishan, E McCafferty. J Electrochem Soc. 136:53, 1989.
110. M Pourbaix. Corrosion 26:431, 1970.
111. A Broli, H Holtan. Corros Sci 13:237, 1973.
112. A Broli, H Holtan. Werkstaffe Korros 26:629, 1975.
113. A Broli, H Holtan. Corros Sci 17:59, 1978.
114. K Nisancioglu, H Holtan. Corros Sci 18:835, 1978.
115. K Nisancioglu, H Holtan. Corros Sci 18:1011, 1978.
116. DA Vermilyea. J Electrochem Soc 118:529, 1971.
117. N Nilsen, E Bardal. Corros Sci 17:635, 1977.
118. DM Aylor, PJ Moran. J Electrochem Soc 133:868, 1986.
119. ZA Foroulis, MJ Thubrikar. Electrochim Acta 21:225, 1976.
120. ZA Foroulis, MJ Thubrikar. Werkstaffe Korros 26:350, 1975.
121. T Shibata, M Sudo. Denki Kagaki 58:227, 1990.
122. TH Nguyen, RT Foley. J Electrochem Soc 126:1855, 1979.
123. H Bohni, HH Uhlig. J Electrochem Soc. 116:906, 1969.
124. F Sato, RC Newman. Corrosion 54:955, 1998.
125. BN Stirrup, NA Hampson, IS Midgley. J Appl Chem 5:229, 1975.
126. H Kaesche. Z Phys Chem 34:87, 1962.
127. A Bond, G Rolling, H Domian. J Electrochem Soc. 113:773, 1966.
128. N Lampas, PG Koutoukos. Corros Sci 36:1011, 1994.
129. RA Bonewitz. Corrosion 29:215, 1973.
130. RA Bonewitz. Corrosion 30:53, 1974.
131. JT Reding, JJ Newport. Mater Protect 5:15, 1966.
132. CDS Tuck, JA Hunter, GM Scamans. J Electrochem Soc 134:2070, 1987.
133. Q Song, RC Newman, RA Cottis, K Sieradzki. Corros Sci 31:621, 1990.
134. JT Staley. History of wrought aluminum alloy development. In: AK Vasudevan,
RD Doherty, eds. Aluminum AlloysContemporary Research and Applications.
New York: Academic Press, 1989, p. 3.
Aluminum Alloys
249
250
Holroyd
172.
173.
174.
175.
176.
177.
178.
179.
180.
181.
182.
183.
184.
185.
186.
187.
188.
189.
190.
191.
192.
193.
194.
195.
196.
197.
198.
Aluminum Alloys
199.
200.
201.
202.
203.
204.
205.
206.
207.
208.
209.
210.
211.
212.
213.
214.
215.
216.
217.
218.
219.
220.
221.
251
8
Magnesium Alloys
Mike J. Danielson
Pacific Northwest National Laboratory, Richland, Washington
I.
INTRODUCTION
Aluminum and zinc are the two major alloying elements added to magnesium
to increase its strength, ductility, and resistance to general corrosion attack. Magnesium alloys are of engineering interest because of their low density and high
strength-to-weight ratio. Magnesium alloys are about two-thirds the density of
aluminum alloys and about 25% lighter than aluminum alloys with similar stiffness. Other advantages are good machinability and castability into complex configurations. Magnesium alloys are used in applications where strength and
lightness are at a premium, such as aircraft, space applications, automotive applications, and certain consumer uses (i.e., luggage, computer frames, audio equipment, lawnmowers). In 1990, most alloy use was in die castings involving one
composition, AZ91D (1). Under dry or low-humidity applications, these alloys
have excellent materials properties, but when immersed or at high relative humidity, they can suffer severe localized corrosion attack, stress-corrosion cracking,
and corrosion fatigue.
There have been a number of reviews (26) on the materials performance
of magnesium alloys. This review will focus on the effect of the environment
on the materials performance, including general and localized corrosion attack,
stress-corrosion cracking, and corrosion fatigue under a variety of relevant engineering environments. This review will (a) summarize the early work and (b)
cover the more recent work since 1993, the most recent review. The intent will
be (a) to describe the underlaying principles that control the behavior of these
alloys, (b) describe the failure modes and their character, (c) describe methods
253
254
Danielson
that mitigate the failure processes, (d) direct the engineer in searching and evaluating the literature, and (e) assist the engineer in carrying out new testing in
relevant environments. The literature is quite complete on the subjects of generalized and localized attack but very sparse on stress-corrosion cracking and corrosion fatigue. On these latter two topics, benchmark papers are pointed out that
will be used to reveal the general character of magnesium alloys. After reading
this chapter, the engineer will have a good grounding in using these alloys in
practical applications and avoiding corrosion-related failures.
II. NOMENCLATURE
Successful use of these alloys requires an understanding of how the alloys are
identified. The American Society for Testing and Materials (ASTM) created a
standard system for naming magnesium alloys and tempers which has been in
use since 1952 (7). This system is clearly described by Hough et al. (8), and an
abbreviated version of their original table is shown in Table 1. The system is
most easily learned by following an example while referring to Table 1, using
AZ91E-T6 as the magnesium alloy. The letters AZ refer to the two alloying
elments in highest concentration, aluminum and zinc (highest concentration
shown first). The next two numbers, 91, refer to the approximate amount of each
alloying element, namely 9 wt% Al and 1 wt% Zn, rounded up to the largest
whole number. The third part of the description, E, is an additional designator
of the alloy and may have importance in specifying additional compositional
details. The fourth part, T6, refers to the temper, which in this case is a solution
anneal followed by some artificial aging. Table 2 shows the composition and
yield strength of some commonly used cast magnesium alloys.
Magnesium Alloys
255
aluminum
bismuth
copper
cadmium
rare earth
iron
magnesium
thorium
zirconium
lithium
manganese
nickel
lead
silver
chromium
silicon
tin
yttrium
antimony
zinc
Second part
Third part
Fourth part
Temper designation:
F as fabricated
O annealed
H10, H11
slightly strain
hardened
H23, H24, H26
strain hardened and partially annealed
T4 solution
heat treated
T5 artificially
aged, only
T6 solution
heat treated
and artificially
aged
T8 solution
heat treated,
cold worked,
and artificially
aged
sium is 6.2 wt% at 340C, and exceeding this concentration will result in the
formation of a variety of complex intermetallics, such as MgZn2. Because the
solubility of zinc also decreases with temperature (1.7 wt% at 150C), zincrich intermetallics may also be present. Mn and, more recently, Zr are added to
the alloy to act as grain refiners to improve the strength. Due to their fairly low
solubility, they will also be present as intermetallics in the matrix. Fe, Ni, and
Cu are very insoluble impurities, which above a certain threshold concentration
are well known to have extremely detrimental effects on the localized corrosion
behavior. They are present as intermetallics which act as cathodic depolarizers
in the creation of a localized microgalvanic cell. Magnesium alloys make use of
a number of other minor metal additions of low solubility such as the rare earths,
256
Table 2
Alloy
Mn
Zn
10.1
6.0
7.6
8.7
8.7
9.0
0.1
0.15
0.13
0.13
0.13
0.10
3.0
0.7
0.7
0.7
2.0
2.7
3.3
3.3
1.0
0.250.75
6.0
4.2
5.8
5.7
4.6
6.0
Other
Th
0.7
0.6
0.7
0.7
0.7
0.7
0.7
0.7
0.7
2.7
0.7
0.7
Zr,
Zr,
Zr
Zr
Zr
Zr,
Zr,
Zr,
Zr,
Cu
Zr,
Zr,
1.8
0.7 Zr
0.7 Zr
Note: In most cases, Cu (0.08%), Ni (0.01%), and Fe (0.005%) are kept at very low values.
Source: Refs. 7 and 8.
2.5
2.5
4.0
5.2
Ag, 2.1 Di
Ag, 1.0 Di
Y, 3.4 RE
Y, 3.00 RE
1.2 RE
2.6 RE
Danielson
AM100A-T61
AZ63A-T6
AZ81A-T4
AZ91C
AZ91E-T6
AZ92A-T6
EQ21A-T6
EZ33A-T5
HK31A-T6
HZ32A-T5
K1A
QE22A-T6
QH21A-T6
WE43A-T6
WE54A-T6
ZC63A-T6
ZE41A-T5
ZE63A-T6
ZH62A-T5
ZK51A-T5
ZK61A-T6
Al
Magnesium Alloys
257
which improve certain mechanical properties but also may form precipitates (contribute to the strengthening by precipitation hardening) within the matrix.
An article by Beldjoudi et al. (10) examined the microstructure of Mg9
Al, AZ91, and Mg3 Al alloys. In the T4 (solution annealed) condition, the alloys
were all single phase (dendritic substructure) with the Al uniformly distributed
in the Mg matrix, there being no evidence of segregation. Precipitates of the type
AlMnFe (globular) and Mg2Si (polygonal) were distributed uniformly throughout
the matrix. Aging the Mg9 Al alloy to the T6 condition resulted in the decrease
of solid-solution Al from 9% to 3% and the formation of a lamellar -Mg17Al12
at the grain boundaries. The T6 condition did not affect the size and distribution
of the AlMnFe-containing precipitates, but the Mg2Si were now present primarily
along grain boundaries. Zn did not affect the microstructure. The observations
of Lunder (11,12) on AZ91 in the F, T4, and T6 tempers reinforce those of
Beldjoudi, with the additional observation that the Zn is always in solid solution.
The as-cast condition (F) contained a great deal of -Mg17Al12 at the grain boundaries as well as being distributed throughout the matrix, and the solid-solution
Al was distributed nonuniformly. The intermetallics, AlMnFe and Mg2Si, were
present throughout the matrix and did not appear to be altered by heat treatment.
258
Fig. 2
Danielson
Magnesium Alloys
259
is more than 2 V below the hydrogen evolution a line, indicating that there
is a large thermodynamic driving force for the spontaneous reaction (corrosion)
of the Mg with water with the simultaneous evolution of hydrogen gas. Pourbaix
diagrams only show the thermodynamic propensity for reaction; the actual kinetic
rate must be determined by a direct physical measurement. Figure 4 shows the
measured open-circuit corrosion potential of various Almagnesium alloys in a
common testing environment, 3.5 wt% NaCl. The significant feature of this plot
is that the potential (written as a reduction potential) becomes more negative as
the magnesium content increases, indicating that the driving force for the corrosion reaction increases with Mg content. Because many of the alloying agents
260
Danielson
Fig. 4 Effect of the magnesium composition in AlMg alloys on the open-circuit corrosion potential exposed to 3.5% NaCl.
Magnesium Alloys
261
reactive than the matrix, an unusual type of localized attack occurs: the precipitates are protected and attack starts adjacent to them in the matrix. Careful microstructural evaluation is needed to reveal this type of morphology.
Magnesium corrodes in neutral and basic water to form a magnesium hydroxide film that can be very protective under certain specific conditions:
Mg 0 2H 2O Mg(OH) 2 H 2 (g)
The pH and certain anions, such as chloride, can damage the protective film,
leading to large increases in the corrosion rate. The composition of the film is a
function of the alloy composition and the impurities in the environment (16,17).
In summary, magnesium alloys are vulnerable to corrosive attack because
of the extreme reactivity of the magnesium within the alloy and the intimate
mixture of intermetallics that can act as localized cathodes, creating microgalvanic cells. The composition of the intermetallics can strongly affect the corrosion
performance of the magnesium alloy.
C. Localized and General Corrosive Attack
(Aqueous Solutions)
Magnesium alloys are prone to attack in aqueous solutions, particularly saltwater.
The presence of certain metallic impurity phases that makes the alloys vulnerable
has been recognized for over 50 years. Hannawalt et al. (14) carried out a pioneering study in 1942 that defined the tolerance limits for a number of the common
impurities. Their effects on the corrosion rate in saltwater are shown in Fig. 5.
Basically, Na, Si, Pb, Sn, Mn, and Al have negligible effects up to 5 wt% and
Fe, Ni, Co, and Cu have extremely detrimental effects at very low concentrations.
These tolerance limits are approximately 0.017, 0.1, and 0.004 wt% for Fe, Cu,
and Ni, respectively, and these elements must be maintained at very low concentrations in order for magnesium alloys to have adequate corrosion resistance.
Tolerance limits (3) are shown for some modern cast alloys in Table 3. Manganese is commonly added to commercial magnesium alloys at the nominal 0.2%
level to act as a grain refiner and to increase the tolerance for Fe. Recent work
indicates that the corrosion resistance is better correlated to the Fe/Mn ratio than
to the Fe content (18). Manganese is believed to act by (a) forming AlMnFe
intermetallic particles that precipitate out and fall to the bottom of the crucible,
thereby reducing the iron content, and (b) the fact that the AlMnFe intermetallics
remaining within the matrix are less efficient as cathodes as compared to metallic
Fe, reducing the driving force for the galvanic cell. Table 4 (19) is included in
order to benchmark the corrosion rate of a common magnesium alloy, AZ91, as
a function of purity (Fe content) and heat treatment under ASTM B-117 salt
spray conditions. Clearly, the corrosion rate of the commercial purity alloy (C
suffix) is unacceptable in saltwater conditions in any heat-treatment condition,
262
Fig. 5
Danielson
AZ91
AM60
AS41
AE42
64
50
400
42
30
100
20
40
200
40
200
1000
Magnesium Alloys
Table 4
Alloy
AZ91C
AZ91C
AZ91E
AZ91E
AZ91E
AZ91E
Mn%
Ratio
Fe/Mn
T4
T6
T5
0.18
0.16
0.23
0.26
0.33
0.35
0.087
0.099
0.008
0.008
0.004
0.004
18
17
0.64
2.2
0.35
0.72
15
18
4
1.7
3
0.82
15
15
0.15
0.12
0.22
0.1
0.12
0.12
0.12
0.1
263
264
Danielson
the -Mg 17Al 12 phase at the grain boundaries which are cathodic to the matrix,
and yet this heat treatment results in the most beneficial condition for decreasing
localized attack. The solution anneal case (T4) shows an anomolously high corrosion rate relative to the as-cast (F) and aged conditions (T6) (12,19) that is not
satisfactorily explained (the solution annealed alloy should be more homogeneous
that the cast or aged material). Beldjoudi et al. (10) reproduced these effects
of heat treatment and clearly showed by electrochemical measurements that Mg 17Al 12 is less reactive than the matrix. Lunder et al. (11) examined the attack
morphology on AZ91 in saltwater and found the pitting attack initiated adjacent
to the intermetallic particles consisting of type AlMnFe, -Mg 17Al 12 , or Mg 2Si.
Attack followed the dendritic arms in the as-cast material, whereas filiform corrosion was observed for the T4 condition. Other alloying agents (Zr, Y, and rare
earths Nd, Gd, Tb, Dy) are used to increase the strength of magnesium alloys,
and they appear to have a small effect on the corrosion resistance as reported by
King et al. (20) and Kamado et al. (21) in salt spray tests. King et al. (20) report
corrosion rates for WE54 and WE45 of about 0.7 mg/cm 2 /day (1.4 mm/year),
which are still substantial. Lunder et al. (12) determined that rare earths and
silicon additions have a beneficial effect on the corrosion resistance, whereas
zinc has a negligible effect.
Magnesium alloys are attacked by acidic solutions but are quite resistant
to alkaline environments and many hydrocarbon environments. Two exceptions
to the acidic generalization are HF and chromic acid solutions, in which the alloy
can be quite resistant (2,3). The alloys are very resistant to deionized water but
will pit in potable water, but the threshold halide concentration for pitting is not
defined in the literature. Except for fluoride, any of the halides, sulfates, and
nitrates are damaging. Solutions containing dissolved Cu, Ni, or Fe ions are especially damaging because the ions will be reduced onto the alloy surface and act
analogously to what happens when they are present within the alloy as impurities.
Froats et al. (22) have a useful chart of organic and inorganic environments with
the corrosion response of the magnesium alloys. Many of the results are prefaced
with the excellent advice that the material should also be tested in the actual
engineering environment to ensure a satisfactory result.
Hallopeau et al. (23) examined the behavior of certain traditional inhibitor
anions (SiO 32 , PO 43 , CrO 42 , MoO 42 ) on a Mg9 Al alloy in a 0.5M sodium
sulfate solution. The corrosion behavior was determined by using potentiodynamic and polarization resistance methods, which only evaluate the short-term
performance. As expected, the Mg was selectively dissolved from the alloy, leading to a surface film enriched in Al. Some inhibition was observed in alkaline
solutions, with SiO 32 and CrO 42 having an inhibiting effect in neutral solutions.
No effect was observed for MoO 42 and PO 43.
The effect of increasing the temperature is to increase the severity of attack.
As an example, the corrosion rate, which is negligible in deionized water at ambi-
Magnesium Alloys
265
ent temperature, increases to about 0.5 mm/year for the AZ alloys at 100C (16).
Clearly, magnesium alloys do not have adequate corrosion resistance for applications above ambient temperature. The literature does not give any guidance as
to the existence of a threshold temperature, above which the material will corrode
excessively.
In summary, magnesium alloys have their poorest corrosion behavior in
immersed, aqueous environments. Generally, the corrosion rates are reported in
terms of weight loss rather than penetration depth, and this can be misleading
because this material is prone to pitting rather than uniform attack. As a caveat,
the corrosion behavior of magnesium alloys under any immersion condition
should be regarded with suspicion until proven safe.
D. Localized and General Corrosive Attack (Gas Phase
and Partial Immersion)
Under most conditions, a partial immersion (part gas phase and part liquid phase)
is at least as damaging as a full immersion, so that all the observations applicable
to full immersion are relevant to a partial immersion. There is the additional
complication that the current lines are focused at the liquid side of the gasliquid
phase, resulting in a knifeline attack that may increase the penetration rate. The
presence or absence of oxygen does not appear to affect the process.
Magnesium alloys have insignificant corrosion rates in clean air if the relative humidity is below about 65% (2); if the surface remains clean of salt deposits,
the alloys can remain corrosion-free up to relative humidities of 90%. When
the surface becomes contaminated with dirt (which contains salts), the corrosion
rates increase and the damage is revealed as shallow pits.
Benchmark results for a dirty industrial environment are shown in Table
5. These results can be quite variable, with a rural environment having lowest
corrosion rates and a marine environment might be somewhat greater. These
results indicate that air exposures are relatively benign for magnesium alloys
compared to total immersion.
Dry gases such as chlorine, iodine, bromine, and fluorine result in little
corrosion of magnesium alloys under ambient temperature conditions. However,
the addition of moisture can greatly increase the corrosion rate.
Magnesium alloys have linear reaction kinetics in oxygen gas at elevated
temperatures and these rates can be very high. For instance, as a benchmark rate,
Leontis and Rhines (24) have measured a penetration rate of about 1.4 mm/year
for a 9% Almagnesium alloy at 400C. The rate would be lower in air and at
lower temperatures, but it is not clear what would constitute safe operating conditions; this would have to be determined on a case-by-case situation. Magnesium alloys can be used successfully under inert or low-reactivity environments.
There is a successful application (350C) in the gas-cooled nuclear industry for
266
Danielson
Penetration rate
(mm/year)
0.049
0.047
0.069
0.074
0.084
0.058
Source: Ref. 2.
Magnesium Alloys
267
Halide ions are particularly aggressive toward SCC, but almost any dissolved
salt increases the rate over that observed in deionized water. The environments
that do not induce SCC are those in which the magnesium alloys are perfectly
passive, such as dilute alkalies, concentrated HF, and chromic acid. Nitrates and
carbonates also have an inhibiting effect. The threshold concentrations at which
dissolved ions can cause SCC are not well defined in the literature, but dilute
solutions are capable of accelerating SCC. A popular laboratory test environment
for ranking alloys is one containing NaCl K 2Cr 2O 4 because it results in rapid
SCC. Unfortunately, the results obtained in this environment have not correlated
well with field failure experience. The literature is filled with conflicting observations about the effects of wrought versus cast alloys, the effects of cold work,
and the effects of heat treatment on susceptibility to SCC. Curiously, the cracking
process is slowed or even stopped by cathodic polarization while it is increased
with anodic polarization (relative to the open-circuit electrochemical potential);
yet, the cracking process is hypothesized to be driven by hydrogen embrittlement.
Fracture-mechanics-based SCC crack growth rate data are uncommon in
the literature, but some benchmark fracture mechanics data for this process are
shown in Fig. 6 in a work by Speidel et al. (27) with a somewhat SCC-resistant
alloy, ZK-60A [Al-free alloy containing zirconium, yield strength 296 MPa
(43 ksi)]. The cracking rate in the two salt solutions was greater than 3 10 4
cm/s, a very high rate, with the K 1scc 7.7 MPa m 1/2 (7 ksi in.1/2 ). The effects
of the dissolved salt are to increase the cracking rate and lower the K 1scc relative
to the distilled-water environment. The authors make the important comment
that in their experience, all high-strength magnesium alloys behave similarly to
environment-enhanced subcritical crack growth and that this behavior does not
vary significantly with alloy composition and heat treatment. Both SCC and corrosion fatigue cracks propagated in a mixed transgranular and intergranular mode
in this testing.
The same environments that cause SCC also reduce the corrosion fatigue
(CF) performance. Magnesium alloys have their best CF performance in vacuum,
followed by air, deionized water, and salt solutions. The CF strengths can be as
low as 10% of those in air. As with SCC, the aluminum- and zinc-containing
alloys are especially susceptible, even in mildly corrosive environments (28). A
significant amount of SN data exist in the literature, but most of it is old. The
majority of the modern fracture-mechanics-based fatigue crack growth data
comes from Russia. The effects of wrought versus cast alloys, the effects of cold
work, and the effects of heat treatment are largely unexplored with CF. Some
fracture-mechanics-based (da/dNK) benchmark data for CF are shown in Fig.
7 (27). This work clearly demonstrates that salt solutions significantly increase
the CF rate and that distilled water is more aggressive than dry argon (particularly
in stage III crack growth). In a more recent study, Stephens et al. (29) examined
the CF response of a modern, high-purity sand-cast alloy, AZ91E-T6, in a 3.5
268
Danielson
Fig. 6 The effect of stress intensity and environment on the velocity of SCC cracks for
a high-strength magnesium alloy.
wt% NaCl solution and air (R 0.05 and 0.5, respectively). Their CF testing
was limited to the stage II regime, but the results were very similar to the results
of Spiedel et al., implying that their earlier comment about the similarity of all
magnesium alloys with regard to SCC behavior might also be extended to their
CF behavior. The threshold K for CF is 2 MPa m 1/2 which is even lower than
the K1scc for SCC. These are very low values.
Increasing the temperature generally acts to increase the propensity for SCC
and CF, but the literature data are extremely sparse on this variable (6).
In summary, the SCC and CF behavior of magnesium alloys is seriously
Magnesium Alloys
269
Fig. 7 The effect of cyclic stress intensity range and environment on the velocity of CF
cracks for a high-strength magnesium alloy.
270
Danielson
increases. Bothwell (16) reports that an AZ31 alloy lost CF strength once the
relative humidity exceeded 50%, and at 93% relative humidity, it had a fatigue
strength of about 75% that of dry air.
G.
Two rapid-cooling methods have been utilized to improve the corrosion resistance of magnesium alloys: rapid solidification (RS) and laser annealing (LA).
Both methods are based on rapid cooling of the melt such that metastable, supersaturated solid solutions are obtained. The basic idea is to minimize the amount of
the various intermetallic compounds that could act as initiation sites for localized
corrosion processes. The RS methods result in the formation of fine powders or
foils that must be later extruded into structural forms. LA takes place after the
part is nearly machined to the final dimensions, whereupon a high-energy laser
beam is focused on the surface and melted to a depth of a few micrometers, and
then the part is machined to the final dimensions. The work of Chang and coworkers (30,31) demonstrated the improved general and localized corrosion resistance of the RS alloys as compared to conventional alloys in 3% NaCl. These
studies were carried out on RS-formed MgAlZn alloys with Mn, Si, and rare
earth additions. The materials were extruded into a bar before corrosion testing.
The measured corrosion rate for the MgAlZnY alloy was about 400 m/
year (10 mpy), one of the lowest rates ever measured for a magnesium alloy in
a salt solution. Makar et al. (32) studied the SCC of RS MgAl alloys in NaCl
K 2 CrO4 solution using the slow-strain-rate method. The materials showed the
classic transgranular SCC morphology, but no crack growth rate data were reported, so that a comparison could be made with conventional materials. The RS
alloy showed an improved general corrosion behavior and very rapid repassivation kinetics when the electrode was scratched.
The LA technique has been applied to the commercial alloy AZ91C by
first depositing a thin layer of Al, Cr, Zn, and so forth on the surface, followed
by laser annealing to melt the surface in order to incorporate the coating into the
surface alloy layer (33,34). Both Zn and Al deposits resulted in an improved
general corrosion rate behavior, but this was determined by potentiodynamic
scans which measures only short-term behavior. Once the thin modified layer
is perforated by corrosive attack, the material will probably revert to its bulk
behavior.
Another new surface modification technique is ion implantation (5). With
this technique, the part must be placed in a vacuum chamber and ions (to be
implanted) are accelerated and slammed into the surface to a penetration depth
of 50500 nm. The method is only mentioned here for information because the
penetration depth is extremely shallow and the results have had little effect on
the corrosion rate.
Magnesium Alloys
271
272
I.
Danielson
IV. CONCLUSIONS
Magnesium alloys have very desirable structural properties from the standpoint of weight and strength, but these properties are offset by their unusual propensity for corrosive attack by general, localized, stress-corrosion cracking, and
corrosion fatigue under aqueous environments. The first line of defense is using
high-purity alloys containing low amounts of Fe, Cu, and Ni. There appears to
be no inhibitor for these degradation processes in most engineering environments. Coatings can improve the cosmetic appearance and extend the useful operation life in marginal environments. Accelerated testing in the laboratory
has a history of not predicting field experience, and the technical literature
usually reports test results in these accelerated environments. Clearly, laboratory
testing must realistically replicate the application environment, particularly with
respect to SCC and CF processes, if these alloys are to be economically and
safely utilized.
Magnesium Alloys
273
REFERENCES
1. MO Pekguleryuz, MM Avedesian. Magnesium alloyingSome metallurgical aspects. In: BL Mordike, F Hehmann, eds. Magnesium Alloys and Their Applications.
Oberursel: Informationgesellschaft-Verlag, 1992, pp. 213220.
2. KG Adamson, DS Tawil. Magnesium and magnesium alloys. In: LL Shreir, RA
Jarman, GT Burstein, eds. Corrosion, 3rd ed. Boston: Butterworth/Heinemann,
1994, pp. 4:984:115.
3. JE Hillis. Magnesium. In: R Baboian, ed. Corrosion Tests and Standards: Application and Interpretation. Philadelphia: ASTM, 1995, pp. 438446.
4. A Luo, MO Pekguleryuz. Cast magnesium alloys for elevated temperature applications. J Mater Sci 29:52595271, 1994.
5. GL Makar, J Kruger. Corrosion of magnesium. Int Mater Rev 38:138153, 1993.
6. WK Miller. Stress-corrosion cracking of magnesium alloys. In: RH Jones, ed. StressCorrosion Cracking. Materials Park, OH: ASM, 1992, pp. 251263.
7. Standard Practice for Codification of Certain Nonferrous Metals and Alloys, Cast
and Wrought. B 275-95a. In: ASTM, Volume 2.02, Aluminum and Magnesium
Alloys. West Conshohochon, PA: ASTM, 1996.
8. S Housh, B Mikucki, A Stevenson. Selection and application of magnesium and
magnesium alloys. In: Metals Handbook, Properties and Selection: Nonferrous
Alloys and Special-Purpose Materials, 10th ed. Materials Park, OH: ASM, 1990.
9. TB Massalski (Editor in Chief). Binary Alloy Phase Diagrams. Materials Park, OH:
ASM, 1986.
10. T Beldjoudi, C Fiaud, L Robbiola. Influence of homogenization and artificial aging
heat treatments on the corrosion behavior of MgAl alloys. Corrosion 49:738745,
1993.
11. O Lunder, JE Lein, SM Hesjevik, TK Aune, K Nisancioglu. Corrosion morphologies
on magnesium alloy AZ91. Werkstoffe Korros 45:331340, 1994.
12. O Lunder, M Videm, K Nisancioglu. Corrosion resistant magnesium alloys. Soc
Automotive Eng, 1995, Paper 950428.
13. M Pourbaix. Atlas of Electrochemical Equilibria in Aqueous Solutions. Houston,
TX: National Association of Corrosion Engineers, 1974.
14. JD Hannawalt, CE Nelson, JA Peloubet. Corrosion studies of magnesium and its
alloys. Trans AIME 147:273299, 1942.
15. O Lunder, K Nisancioglu, RS Hansen. Corrosion of die cast magnesiumaluminum
alloys. Soc Automotive Eng, 1993, Paper 930755.
16. MR Bothwell. In: H Godard, WB Jepson, MR Bothwell, RL Kane, eds, The Corrosion of Light Metals. 1967, pp. 259311.
17. GR Bengough, L Whitby. Trans Inst Chem Eng 11:176189, 19318.
18. RS Hanson. Review of corrosion behavior of Mg-alloys. In: BL Mordike, F Hehmann, ed. Magnesium Alloys and Their Applications. Oberursel: Informationgesellschaft-Verlag, 1992, pp. 327334.
19. KN Reichek, KJ Clark, JE Hillis. Controlling the Salt Water Corrosion Performance
of Magnesium AZ91 Alloy. Soc Automotive Eng, 1985, Paper 850417.
20. J King, S Thistlethwaite. New corrosion resistant wrought magnesium alloys. In:
274
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
Danielson
BL Mordike, F Hehmann, ed. Magnesium Alloys and Their Applications. Oberursel:
Informationgesellschaft-Verlag, 1992, pp. 327334.
S Kamado, Y Kojima, Y Negishi, S Iwasawa. Corrosion resistance of Mgheavy
rare earth metal alloys. In: Light Metals Processing and Applications, Proceedings
of the International Symposium on Light Metals Processings and Applications, 1993,
pp. 849858.
A Froats, TK Aune, D Hawke, W Unsworthm, J Hillis. Corrosion of magnesium
and magnesium alloys. In: Metals Handbook, Volume 13, Corrosion, 9th ed. Metals
Park, OH: ASM 1987, pp. 740754.
X Hallopeau, T Beldjoudi, L Robbiola, C Fiaud. Electrochemical behavior of Mg
9 Al alloy in aqueous electrolyte solutions containing XO yn inhibiting anions. 8th
European Symposium on Corrosion Inhibitors, 1995, Vol. 2, 913927.
TE Leontis, FN Rhines. Rates of high temperature oxidation of magnesium and
magnesium alloys. Trans AIME 166:265294, 1946.
O Kubaschewski, BE Hoptins. Oxidation of Metal and Alloys. London: Butterworths, 1962, pp. 210277.
PL Miller, BA Shaw, RG Wendt, WC Moshier. Improving corrosion resistance of
magnesium by nonequilibrium alloying with yttrium. Corrosion 49:947950, 1993.
SMO Speidel, MJ Blackburn, TR Beck, JA Feeney. Corrosion fatigue and stress
corrosion crack growth in high strength aluminum alloys, magnesium alloys, and
titanium alloys exposed to aqueous solutions. In: AJ McEvily, RW Staehle, ed. Corrosion Fatigue: Chemistry, Mechanics and Microstructure. Houston, TX: National
Association of Corrosion Engineers, 1972, pp. 324345.
VV Ogarevic, RI Stephens. Fatigue of magnesium alloys. Annu Rev Mater Sci 20:
141177, 1990.
RI Stephens, CD Schrader, KB Lease. Corrosion fatigue of AZ91E-T6 cast magnesium alloy in a 3.5% NaCl aqueous environment. J Eng Mater Tech 117:293298,
1995.
CF Chang, SK Das, D Raybould, A Brown. Met Powder Rep 41:302308, 1986.
CF Chang, SK Das, D Raybould. In: Rapidly Solidified Materials. Metals Park, OH:
ASM, 1986, pp. 129135.
GL Makar, J Kruger, K Sieradzki. SCC of rapidly solidified MgAl alloys. Corros
Sci 34:13111342, 1993.
S Akavipat, EB Hale, CE Haberman, PL Hagans. Mater Sci Eng 69:311316, 1984.
P Hagnas, C Langhoff, D Moll, D Perettie, R Yates. In: Laser Assisted Deposition,
Etching, and Doping, Proceedings, Volume 459. Society of Photo-Optical Instrumentation Engineers, 1984, pp. 103107.
AL Olsen. Designing galvanic corrosion out of magnesium driveline components.
Soc Automotive Eng, 1987, Paper 870364.
9
Environmental Embrittlement of
Nickel-Based and Iron-Based
Intermetallics
Norman S. Stoloff
Rensselaer Polytechnic Institute, Troy, New York
I.
INTRODUCTION
Many investigations have shown that low ductility and brittle fracture in intermetallics are caused not only by intrinsic factors (such as lack of sufficient deformation modes, poor cleavage strength, weak grain boundaries, etc.) but also by extrinsic factors. Environmental degradation, an extrinsic factor, is found to be a
major cause of brittle fracture in many ordered intermetallics, particularly those
with high crystal symmetries (i.e., cubic L12 and B2 and hexagonal DO19), as
outlined in several recent reviews (13).
There are two types of environmental embrittlement observed in intermetallics. One is hydrogen-induced embrittlement occurring at ambient temperatures.
Many intermetallic alloys show a substantial decrease in room-temperature tensile ductility due to moisture-induced hydrogen embrittlement in air, or as a result
of direct exposure to hydrogen by precharging or testing in hydrogen gas. The
other is oxygen-induced embrittlement in air at elevated temperatures. In most
cases, the embrittlement is due to a dynamic effect involving generation and
penetration of an embrittling agent (i.e., hydrogen or oxygen) during testing.
In this chapter, the two types of environment-sensitive degradation are treated
separately. The concluding section discusses metallurgical means of alleviating
environmental degradation.
275
276
Stoloff
Table 1
Alloy
Ni3(Al,Ti)
(single crystal)
Ni3AlB
Ni3AlBe
Ni3(Al,Mn)
Ni3Si
Ni3(Si,Ti)
Ni3(Si,Ti)B
(Co,Fe)3V
Ni3Fe
FeAl
Fe3Al
Crystal
structure
Environmental
embrittlementa
Ref.
L12
L12
L12
L12
L12
L12
L12
L12
L12
B2
DO3
5
6
7
8
6
9
10
11
12
: Environmental embrittlement observed in moist environments; : not observed in moist environments, but embrittled by hydrogen.
Source: Ref. 2.
a
277
1. Ni3Fe
Ni3Fe is a model intermetallic for the study of hydrogen embrittlement in alloys
displaying a critical ordering temperature, Tc. This has permitted a direct comparison to be made between the ordered and disordered conditions of Ni3Fe, as shown
in Fig. 1 (13). In each case, hydrogen was cathodically charged for 1 h prior to
tensile test at room temperature. Ni3Fe is slightly embrittled by precharging in
both the ordered and disordered conditions; simultaneous charging and testing
causes greater loss of ductility than precharging, with an especially severe effect
for the ordered condition, perhaps due to the transport of hydrogen by moving
dislocations. Fractographic studies showed a change in fracture mode from transgranular microvoid coalescence to intergranular fracture in hydrogen accompanying the reduction in elongation. However, the intergranular embrittlement zone
was only about one-third as deep in precharged samples of ordered Ni3Fe as in
disordered samples. Chia and Chung (14) report that Ni3Fe is not embrittled by
moisture in air or by water, in marked contrast to the effects of hydrogen.
2. Ni3 Al
a. Effects of Composition Although the embrittling effects of hydrogen
on fracture behavior of Ni3Al alloys have been well documented (6,15), only
recently has evidence been provided for moisture-induced cracking in these
alloys.
Liu (16) has reported that the tensile ductility of Ni-24 at% Al is increased
from 2.6% to 7.2% by testing in oxygen rather than in air. A similar result was
noted with Ni23.5 at% Al (2.5% versus 8.2% elongation in air and oxygen,
respectively). Higher ductilities are observed in vacuum than in air for boronfree Ni23.4 at% Al, but the difference is strain rate sensitive; see Fig. 2 (17).
Also, the vacuum level plays a role, with the ductility of polycrystalline Ni23.4
at% Al increasing from 7.9% to 23.4% as the vacuum level changes from 101
to 3.6 108 Pa (17).
Additional striking effects of the environment have been reported for a Ni
22.65 Al0.26 Zr alloy, recrystallized from a 110-oriented single crystal and
tested at room temperature in two orientations; see Table 2 (18). This boron-free
alloy displayed 8.7% elongation in water, 13.2% in air, and 50.6% in oxygen in
the 0 orientation. Fracture was predominantly intergranular in all environments.
Although the results could have been influenced by texture (unknown in these
alloys) or by a preponderance of fracture-resistant grain boundaries in this material, the authors suggested that the principal role of boron in ductilizing Ni3Al
278
Fig. 1
Stoloff
279
Fig. 2 Effect of strain rate on room-temperature tensile ductility of polycrystalline Bfree Ni3Al (23.4% Al) in air and vacuum. (From Ref. 17.)
280
Table 2 Effects of Environment and Specimen Orientation on Tensile Properties of Ni22.65% Al0.26% Zr, 25C,
5.3 103 s1
Specimen
orientation
0
45
Test
environment
Elongation to
fracture (%)
Yield strength
(MPa)
Ultimate tensile
strength (MPa)
10 5a
(MPa)
15 5a
(MPa)
Water
Air
Oxygen
Water
Air
Oxygen
8.7
13.2
50.6
6.3
10.7
47.8
322
324
326
331
341
327
528
661
1451
473
603
1438
133
127
127
143
250
271
a
Difference in flow stress at 10% and 5% (or 15% and 5%) strain.
Source: Ref. 18.
Stoloff
Effect of Test Environment on the Room-Temperature Tensile Properties of B-Free and B-Doped Ni3Al
Alloy chemistry
(at.%)
Ni23.4 Al
Ni24 Al0.2 B
Test
environment
Elongation to
fracture (%)
Yield strength
(MPa)
Ultimate strength
(MPa)
Air
Oxygen
UHV
Water
Air
Oxygen
3.1
15.8
23.4
36.8
39.3
42.8
308
336
a
288
280
289
392
681
1199
1241
1315
Table 3
a
Not measured.
Source: Ref. 17.
281
282
Stoloff
Fig. 3 Elongation to fracture of three B-doped Ni3Al (24% Al) alloys as a function of
hydrogen pressure. Open and closed symbols refer to the ion gauge on and off, respectively. (From Refs. 22 and 23.)
283
pressures exceeding about 1 Pa. Because atomic hydrogen is the cause of embrittlement, this surprising result suggests that boron promotes the dissociation of
molecular to atomic hydrogen. This study also showed that when all sources of
hydrogen were minimized, the ductility of polycrystalline Ni3Al exceeded 40%,
with a predominantly transgranular fracture path.
Zirconium, as cited previously, is another important dopant that can reduce
environmental embrittlement in Ni3Al. George et al. (19,24) and Lin et al. (25)
suggested that zirconium enhances grain-boundary cohesion. Chiba et al. (26), on
the other hand, proposed that zirconium reduces the extent of sulfur segregation to
grain boundaries, thereby inhibiting sulfur-induced loss of cohesion. Itoh et al.
(27) recently attempted to resolve these conflicting views by studying the ductility
of sulfur-doped Ni3Al containing different levels of zirconium. Although they
confirmed that zirconium alleviated environmental embrittlement and found that
ZrS particles formed when sulfur levels were high, they could not definitively
identify the mechanism of the improved behavior. Zirconium has a similar beneficial effect on ductility and fatigue crack growth rates in Fe3Al alloys [see Sec.
II.C.2, (28)], but it is difficult to see how environmental effects in the two alloy
systems are related because moisture induces intergranular fracture in Ni3Al but
does not change the characteristic transgranular crack path in Fe3Al. Clearly,
much more work has to be done to identify the cause of the benefits bestowed
by small additions of zirconium.
Apart from boron and zirconium, several other solutes can increase the
ductility and change the crack path of polycrystalline Ni3Al in air; see Table 4
(29). Transition metals such as chromium, iron, and manganese improve ductility
Alloy composition
(at.%)
Tensile ductile
(%)
Fracture
mode
13
3554
50
16
8
11
5
4
6
13
Intergranular
Transgranular
Transgranular
Transgranular
Mixed
Intergranular
Intergranular
Intergranular
Intergranular
Intergranular
284
Stoloff
in air, but the effects of these elements may be due to the introduction of the
disordered -phase in the microstructure. Recently, palladium has been reported
to increase ductility in air of single-phase Ni3Al (30). Palladium in the range 1
3 wt% also increases ductility in the presence of hydrogen, although considerable
scatter in the data was noted (31). Further, Pd-free Ni3Al containing 5 wppm
hydrogen exhibited cleavagelike facets, whereas an alloy containing 0.5% Pd
exhibited microvoids on the fracture surface. A similar beneficial effect of Pd
has been noted in steels, for which the fracture toughness and threshold stress
in constant-load tests were significantly improved by Pd additions (32).
The partial replacement of aluminum by titanium in Ni3Al single crystals
does not change the pattern of embrittlement by moisture, water vapor, and hydrogen; see Fig. 4 (33). A strain rates approaching 103, embrittlement disappeared
in all environments. This work also demonstrated that when grain boundaries are
absent, embrittlement still can occur, along low-index cleavage planes.
b. Effects of Predeformation Moisture-induced embrittlement of Ni3Al
(34) and Ni3Si,Ti (35) can be reduced by predeformation (e.g., by shot peening
or by tensile prestrain), as shown in Fig. 5. The effect varies with temperature
of the predeformation; elongation increases with decreasing predeformation temperature. Corresponding to increased tensile ductility, the fracture mode changes
from intergranular to transgranular. The effect of prior deformation was attributed
Fig. 4 Effects of environment and strain rate on the ductility of 100-oriented Ni3Al,
Ti single crystals. (From Ref. 33.)
(a)
285
(b)
to trapping of hydrogen atoms at vacancies or dislocations or an increase of soluble hydrogen in the lattice.
c. Stress-Corrosion Cracking Ricker et al. (36) have shown that stresscorrosion cracking of Ni3Al alloyws such as IC-50 (Ni22 at% Al, 0.01 Zr, 0.08
B) and IC-218 (Ni17 at% Al, 8.18 Cr, 0.127 Zr, 0.098 B) in acidic solutions
(HNO3, H2SO4) arises from the liberation of hydrogen. Slow strain-rate tests in
low-pH solutions show a large reduction in ductility and a change in fracture
mode from transgranular to intergranular compared to tests in neutral or alkaline
solutions. Preexposure to sulfuric acid also reduces ductility in a subsequent slowstrain-rate test conducted in air, again due to release of hydrogen at the specimen
surface (36,37).
3. Ni3Si
Intergranular fracture and environmental embrittlement have been studied in
alloys based on Ni3Si (7,38). Ni3Si showed no appreciable plastic deformation
when tested in moist air but displayed an elongation of 7.5% when tested in dry
oxygen (Table 5), demonstrating that Ni3Si is prone to environmental embrittlement (39). Because the elimination of the environmental effect by testing in dry
oxygen does not lead to extensive ductility (e.g., 30% or more) and complete
suppression of intergranular fracture in Ni3Si, moisture-induced hydrogen embrittlement does not appear to be the sole source of grain-boundary brittleness in
the silicide. Boron additions segregate strongly to grain boundaries in Ni3Si and
suppress environmental embrittlement, as is clearly shown in Table 5 (39). Note
286
Stoloff
Test environment
Tensile ductility
(%)
Yield strength
[MPa (ksi)]
Ultimate
tensile strength
[MPa (ksi)]
0 ppm Ba
Air
Vacuum
Oxygen
0
4.7
7.5
677 (98.3)
685 (99.4)
627 (91.0)
853 (124)
1040 (151)
774 (112)
813 (118)
1151 (167)
1268 (184)
655 (95.0)
659 (95.6)
703 (102)
819 (119)
823 (120)
837 (136)
610 (88.6)
590 (85.6)
875 (127)
854 (124)
50 ppm Bc
Air
Vacuum
8.1
8.9
100 ppm Ba
Air
Vacuum
Oxygen
5.0
5.0
5.9
150 ppm Bc
Air
Oxygen
7.0
6.6
that as little as 50 ppm boron provides essentially the same ductility in air and
in vacuum, whereas the undoped alloys display no ductility in air and 4.7% in
vacuum. Other elements that reduce the effects of moisture in air or of distilled
water on the ductility of Ni3(Si,Ti) alloys are chromium, manganese, and iron,
in that order of effectiveness (40). Accompanying improved ductility is a reduction in the extent of intergranular fracture. Reduced strain rates are detrimental
to embrittled alloys, as is common with other intermetallics. However, the mechanism of improvement in ductility with transition metal solutes is not clear, with
several hypotheses being offered. Furthermore, the addition of these transition
elements was less effective in reducing embrittlement induced by hydrogen gas.
This implies that decomposition of H2 into H at the surface is not markedly affected by these solutes.
B. NiAl
Little is known about possible environmental embrittlement of NiAl. Lahrman
et al. (41) reported that the tensile properties of NiAl single crystals are about
287
the same in air, in vacuum, and in oxygen at room temperature. However, Bergmann and Vehoff (42) have reported that gaseous hydrogen, charged into single
crystals at 1000C, promotes stable crack growth in 110-oriented single crystals
at temperatures above 475K. Extensive scatter in the data precluded any conclusions as to the possible influence of hydrogen on the toughness of polycrystals.
C. Iron Aluminides
The two iron aluminides, FeAl and Fe3Al, form back-centered-cubic (bcc)-ordered crystal structures of B2 and DO3 type, respectively. Despite the difference
in crystal structure, these aluminides show somewhat similar patterns of embrittlement in hydrogen-containing atmospheres at room temperature. Both aluminides exhibit only a few percent ductility (15%) when tested at ambient temperature in air. This led many researchers to conclude that these alloys are inherently
brittle. However, it has been demonstrated conclusively that when water vapor
and hydrogen are eliminated from the external environment, both alloys exhibit
considerable ductility.
1. FeAl
Tensile elongations up to 19% have been achieved in FeAl alloys tested in dry
environments such as oxygen (11). An increase in ductility from 2% in water to
18% in oxygen was accompanied by a change in fracture mode from transgranular
cleavage in air to mainly grain-boundary separation in dry oxygen. These observations suggest that cleavage planes are more susceptible to embrittlement than
are grain boundaries. The maximum degree of moisture-induced embrittlement
occurs on either side of room temperature; see Fig. 6 (1). At higher temperatures,
in situ protective oxide films can form readily on specimen surfaces, whereas at
lower temperatures, the aluminummoisture reaction is slowed and the equilibrium moisture content in air also is lowered.
a. Effects of Aluminum Content The ductility of FeAl decreases with increasing aluminum content in the range 3848 at%, as shown in Fig. 7 (43).
Environmental embrittlement of FeAl, defined as the difference in ductility between tests in air and in oxygen, is considerably reduced as the aluminum content
increases. For Fe40% Al, ductility is about 4% in air, in 2 107 torr vacuum,
or after hydrogen charging (44). Gaydosh and Nathal (45) reported that the ductility of Fe40% Al is sensitive to microstructure (annealed and furnace-cooled
material displayed 9% elongation in vacuum compared to 5% for an as-extruded
sample). However, boron increased ductility of as-extruded Fe40% Al to 9%
in vacuum. Fe50% Al, on the other hand, was brittle in vacuum for both the
as-extruded and furnace-cooled conditions. For Fe43 at% Al, the ductility is
nil in air as well as in dry oxygen; all specimens fail intergranularly. This differ-
288
Stoloff
Fig. 6 Effect of test temperature on tensile elongation of FeAl (36.5% Al) in air. (From
Ref. 1.)
Fig. 7 Effect of aluminum concentration and test environment on the room temperature
tensile ductility of B-doped and B-free FeAl. (From Ref. 43.)
289
ence in behavior with aluminum content suggests that grain boundaries in FeAl
alloys with Al 38% are intrinsically weak. Therefore, environmental embrittlement and intrinsic effects must be distinguished in order to establish strategies for
reducing brittleness. It has been demonstrated that the intrinsic grain-boundary
brittleness in FeAl and other intermetallics can be alleviated by microalloying
with boron (45,46).
Figure 7 shows that environmental embrittlement does not occur in borondoped Fe48% Al, but ductility is low in both inert and aggressive environments
(42). As aluminum content decreases, embrittlement is increasingly severe in
boron-doped alloys, as measured by the difference in ductility between oxygen
and air. It is clear from Fig. 7 that the intrinsic ductility of boron-doped FeAl
drops sharply with increasing aluminum content, paralleling observations on boron-free material cited earlier.
b. Effects of Annealing Temperature It has been widely reported that furnace cooling from elevated temperatures or the imposition of low-temperature
anneals markedly increases the ductility of FeAl alloys (4751). This effect has
been linked to the elimination of excess vacancies, which cause hardening, as a
result of these treatments. The influence of annealing temperature on environmental embrittlement of Fe36% Al has been studied by Lynch and Heldt (49,50).
Ductility in air is much lower than in dry oxygen at all annealing temperatures,
but there is a detrimental effect of raising annealing temperature in either environment. It was concluded that excess vacancies resulting from cooling affect the
intrinsic ductility of iron aluminides and not their susceptibility to moisture.
c. Effects of Grain Boundaries There have been no systematic studies of
the influence of grain size on the susceptibility of FeAl alloys to environmental
embrittlement. However, smaller grain sizes tend to favor increased ductilities
for similar heat treatments, as noted by both Klein and Baker (51) and Gaydosh
et al. (52).
Lynch et al. (53) have shown that the tensile elongation of single crystals
of Fe35% Al is extremely sensitive to the environment; see Fig. 8. Note that
the ductility in air is about 4%, independent of orientation, whereas ductilities
in oxygen are 18.5% for crystal No. 1, oriented with the tensile axis 40 from
the (100) pole and 22.7% for crystal No. 2, oriented with the tensile axis 40
from the (100) pole. These differences in ductility are similar to those observed
for polycrystals, suggesting that grain boundaries are not necessary for moistureinduced embrittlement of FeAl alloys. (The same is true for Fe3Al alloys, in which
transgranular cleavage is usually observed in air as well as in oxygen or vacuum.)
The similarities between water-vapor-induced embrittlement and hydrogen
embrittlement are now well established for both FeAl and Fe3Al alloys. For example, lower strain rates exacerbate embrittlement in air, exactly as would be expected for a hydrogen-related phenomenon (54,55). Recently, Kasul and Heldt
290
Fig. 8
Stoloff
Elongation of Fe35 Al single crystals in oxygen and air. (From Ref. 53.)
(56) have provided additional evidence to link both types of embrittlement. Subcritical crack velocities were measured in Fe35 at% Al under constant-load
conditions; these were compared with strain-rate effects on ductility in air, as
well as with results of cathodic charging experiments. It was shown that the strain
rate has essentially no effect on ductility in 103 Pa vacuum, whereas ductility
was sharply reduced with decreasing strain rate in air (56). Similar results in air
have been reported by Shan and Lin (55), but they are at variance with a report
by Nagpal and Baker (54) of a sharp drop in ductility at a particular strain rate.
At a strain rate of about 0.3 s1 the ductilities in air and vacuum were equal; this
strain rate corresponded reasonably well with the established critical strain rate
above which enhanced hydrogen transport on moving dislocations is no longer
likely (56). The role of moving dislocations in transporting hydrogen to potential
crack initiation sites has been discussed by several research groups (13,56), but
there is, as yet, no consensus that this mechanism is applicable to environmental
embrittlement phenomena.
d. Cathodic Charging Experiments Cathodic charging experiments on
Fe35 at% Al have shown a clear correlation among hydrogen content, charging
time, and ductility (56). For example, after 24 h of charging in 1N H2 /SO4, the
hydrogen content is about 1.3 ppm and the ductility in subsequent tests in vacuum
has dropped from 8% to 1.3% (56). Baking treatments of sufficient duration at
high temperature (800C) had previously been shown by Kasul and Heldt (57)
291
to completely restore the ductility of cathodically charged Fe3Al. For Fe35% Al,
baking at 400C for more than 1 h was sufficient to completely restore ductility. A
valuable result of this study was an estimate of the diffusivity of hydrogen in
Fe35% Al at room temperature: about 4 1012 cm2 /s. Because this diffusivity
is too low to allow significant hydrogen penetration ahead of a crack moving
under constant stress in stage 2, it was suggested that hydrogen transport by
moving dislocations may occur.
e. Fracture Toughness Relatively few studies of fracture toughness in the
iron aluminides have been reported. Klein et al. (58) have shown that the fracture
toughness of FeAl increases from 23 to 30 MPa m1/2 for blunt notched specimens
as the strain rate increased. A similar trend was noted for precracked samples.
Ko et al. (59) used the standard multispecimen JIc procedure on side-grooved
specimens to estimate fracture toughness of Fe35 at% Al as a function of test
environment. It was shown that toughness increased from air to vacuum to oxygen environments; see Fig. 9. Also shown are toughness data for an Fe3Al alloy,
Fe28 at% Al, tested in both the B2 and DO3 conditions. For both conditions,
toughness in the vacuum and oxygen environments was lower than for the FeAl
Fig. 9 Fracture toughness of iron aluminides in different environments. (From Ref. 59.)
292
Stoloff
alloy, whereas tests in air resulted in the lowest toughness in FeAl. These results
were very similar to tensile data reported for the two alloys by the same group.
Under both tensile and fracture toughness loading conditions transgranular cleavage was the dominant fracture mode seen in air. Some intergranular fracture
regions were seen in B2-ordered Fe28 Al and Fe35 A1 specimens tested in
oxygen. This study indicated that Fe35 Al is more susceptible to environmental
embrittlement than is Fe28 Al and that grain boundaries are intrinsically weaker
in the FeAl alloy. However, complicating the interpretation of the data is the fact
that the B2 condition of Fe3Al alloys is more ductile and displays higher toughness than the DO3 condition. This suggests that it is the higher Al content in the
Fe35 Al alloy, rather than the crystal structure, that is responsible for the greater
effect of the environment.
f. Fatigue Crack Growth Recent work in our laboratory has shown that
hydrogen severely embrittles FeAl under cyclic loading conditions; see Fig. 10
(60). It appears that moisture is embrittling only at high cyclic stress intensities
for this alloy.
2. Fe3 Al
Fe3Al alloys form in the range of about 1832 at% Al. When near-stoichiometric
alloys are quenched from above the critical ordering temperature, Tc, they display
a partially ordered B2 structure. Slow cooling through Tc or extended annealing
Fig. 10 Effects of environment on fatigue crack growth of Fe35 at% Al at 25C. (From
Ref. 60.)
293
just below Tc results in a highly ordered DO3 structure at room temperature. There
is considerable evidence that environmental embrittlement, resulting from moisture in air, is severe in binary Fe3Al alloys. For example, tensile elongation in
air is 4% but 19% in vacuum, whereas less than 1% strain is noted after precharging with hydrogen, as shown in Fig. 11 (44). Ductility of Fe28% Al alloys in
air is significantly increased by the addition of chromium (61). Although the
mechanism of improvement in ductility is unknown, it may result from modification of the Al2O3 coating that naturally forms on Fe3Al in such a way that hydrogen liberation from water vapor is reduced. It is unlikely that Cr2O3 replaces
Al2O3 as the protective oxide at these chromium levels. Unfortunately, chromium
additions do not suppress embrittlement by gaseous hydrogen or by hydrogen
introduced by electrolytic charging (44).
a. Effects of Strain Rate Reducing strain rate sharply reduces the ductility
of Fe24 at% Al tested in air at room temperature, as was discussed previously
for Ni3Al (Fig. 2) and FeAl alloys (60). Such behavior further supports the concept of environmentally induced embrittlement, because lower strain rates provide more time for release of hydrogen from the water vapor in air, and for
penetration into the alloy. The influence of strain rate on ductility of FeAlCr
alloys in air has been confirmed by Shan and Lin (55). Further, Shea et al. (44)
have shown the same trend of increasing embrittlement with lower strain rate in
hydrogen-charged Fe23.5 at% Al.
294
Stoloff
Fig. 12 Effects of environment on fatigue crack growth of Fe3Al alloy FA-129 at room
temperature. (From Ref. 65.)
295
cooled DO3 condition. Embrittlement does not seem to be influenced by the microstructure, as both fully recrystallized and partly recrystallized samples showed
similar crack growth rates. The differing behavior between monotonic and cyclic
loading may be due to the repeated rupturing of oxide films that occurs under
cyclic loading. Therefore, any beneficial effect that chromium confers on the
protectiveness of Al2O3 under monotonic conditions is lost under cyclic loading.
Other solutes also have an effect. For example, the addition of 0.5% Zr to
Fe28 Al5 Cr results in a decrease in the crack growth rate as compared with
other alloys; see Fig. 13 (28). Carbon additions have been found to increase the
critical stress intensity with little effect on the crack growth rate. There exists a
limit to the beneficial effects of alloying with Zr, as a 1% Zr alloy has a higher
crack growth rate than either of the 0.5% Zr alloys.
The fatigue crack growth behavior of the iron aluminides demonstrates
typical corrosion fatigue characteristics. A more inert environment results in a
lower crack growth rate. In the alloys studied, the environment was found to
have an effect on the threshold and critical stress intensities, except in the Zrcontaining alloys. In those alloys, the threshold stress intensity was found to be
Fig. 13 Fatigue crack growth of iron aluminides in air at 25C. Note low growth rates
in alloys containing Zr. (From Ref. 28.)
296
Stoloff
297
Embrittlement Mechanisms
As mentioned in the previous sections, many ordered intermetallics display environmental embrittlement in hydrogen (charged or uncharged) environments or
in moist air at ambient temperatures. It has been shown that some intermetallics
containing reactive elements, such as FeAl and Fe3Al, exhibit more severe embrittlement in moist air than in dry hydrogen. The proposed chemical reaction for
the moisture-induced embrittlement of aluminides is (1)
2 M 3 H2O Al2O3 6 H
(1)
An alternative reaction between H2O and Ni3Al,Ti, as revealed by x-ray photoelectron microscopy (XPS) is (72)
Al x H2O Al(OH)x x H
(2)
298
Stoloff
Environmental and hydrogen embrittlement in the bcc-ordered iron aluminides and in FeCoV occurs along cleavage planes rather than grain boundaries,
suggesting that cleavage strength is reduced by absorbed or adsorbed hydrogen.
Experimental results are supported by the previously cited first-principles quantum-mechanical calculations, which indicate that absorbed hydrogen significantly
reduces the cleavage strength and energy of FeAl (by as much as 2070%, depending on the hydrogen concentration) (71). Superdislocations have been suggested to be the carriers for enhanced diffusion of hydrogen at crack tips (21). For
face-centered-cubic (fcc)-ordered intermetallics, hydrogen embrittlement usually
causes intergranular fracture, suggesting that reduction of cohesive strength along
the boundaries is responsible. However, single crystals of various L12 alloys also
are embrittled (8) so that grain boundaries clearly are not necessary to observe
reduced ductility. Bond et al. (75) have shown by in situ transmission electron
spectroscopy (TEM) that hydrogen enhances dislocation mobility and crack
growth in Ni3Al, but these authors also suggested that decohesion is the failure
mechanism. More work is needed to determine whether enhanced dislocation
mobility in the presence of hydrogen occurs in other intermetallic systems.
299
Alloy
Crystal
structure
Environmental
embrittlementa
Ref.
Ni3Al Hf B
Ni3Al Cr
(Ni,Co)3Al
Ni3Si
Ni3Si Cr
Ni3(Si,Ti) B
(Fe22Co78)3V
Ni3(Si,Ti) B
FeAl
Fe3Al
L12
L12
L12
L12
L12
L12
L12
L12
B2
DO3
76
76
93
80
80
38
81
38
95
95
A. L12 Alloys
1. Ni3Al
Tensile properties of Ni3Al are sensitive to test temperature and environment.
Figure 14 compares the tensile elongation of a Ni3Al alloy (Ni21.5 Al0.5 Hf
0.1 B) (IC-145) tested in air and vacuum (103 Pa) as a function of test temperature (76). The alloy tested in air showed appreciably lower ductility than that
tested in vacuum at temperatures above 300C, and the severest embrittlement
occurred near 750C, despite the fact that Ni3Al alloys exhibited good oxidation
resistance in air. The loss in ductility generally is accompanied by a change in
fracture mode from ductile transgranular to brittle intergranular. Similar embrittlement has been observed in other Ni3Al alloys, such as B-doped Ni3Al containing iron or hafnium (77). In these cases, oxygen has been identified as the
embrittling agent. The similarity in the shape of the two curves in Fig. 14 further
indicates that embrittlement cannot be completely suppressed by a conventional
vacuum of 103 Pa. The role of air pressure in embrittlement of Ni3Al alloys has
been shown clearly for IC-136 (Ni23 at% Al0.5% Hf0.07% B) tested at
760C. There is a rapid increase in elongation as air pressure is reduced from 1
to 103 torr, and continued to increase in ductility to a pressure of 107 torr (78).
300
Stoloff
Fig. 14 Tensile ductility of IC-145 (Ni21.3 at% A10.5 Hf0.1 B) in air and in vacuum. (From Ref. 76.)
2. Ni3Si Alloys
As in the case of Ni3Al, Ni3Si alloys exhibit severe environmental embrittlement
in oxidizing environments at elevated temperatures. For Ni3Si and Ni3(Si,Ti)
alloys doped with and without boron, tensile ductility decreases sharply at temperatures above 300C in moist air. The ductility of low-Si alloys exhibits a
minimum at 600C, as seen in Fig. 16 (79); above that temperature, the ductility
increases sharply. Ductility in vacuum at 600C increases by a factor of 20. As
in the case of Ni3Al, Cr additions in the range 26 at% effectively reduce embrittlemen (81). For high-(SiTi) alloys (e.g., 21%) ductility decreases continuously
with increasing temperature and approaches zero at temperatures above 600C.
No improvement in ductility results in a conventional vacuum at these temperatures.
301
B. Other Intermetallics
(CoFe)3V alloys, which are susceptible to embrittlement by water or moist air (9),
also display a sharp loss in ductility at temperatures above 500C. The minimum
ductility in both air and vacuum (5 104 Pa) is noted near Tc (910C) (81).
Ductility is lower in air than in vacuum between 500C and 910C. Embrittlement
below Tc was attributed to oxygen-induced penetration of grain boundaries. These
alloys are not embrittled at temperatures above Tc, probably due simply to the
lack of long-range order.
Iron aluminide alloys based on Fe3Al or FeAl do not appear to be susceptible to elevated-temperature embrittlement in oxidizing environments, in spite of
their extreme ambient-temperature embrittlement in moist air. The absence of
elevated-temperature embrittlement in moist air. The absence of elevated-temperature embrittlement in these aluminides is possibly related to the lack of a substantial yield anomaly, together with rapid formation of protective oxide films
302
Stoloff
Fig. 16 Yield strength and elongation of Ni18.9 at% Si, tested in air as a function of
temperature. (From Ref. 79.)
due to rapid diffusion. Further experimental studies are required to clarify these
points.
C. Embrittling Mechanisms
Embrittlement (other than via the pest reaction) has been suggested to be caused
by a dynamic effect simultaneously involving high localized-stress concentration,
elevated temperature, and gaseous oxygen (2). Such a dynamic effect involves
repeated weakening and cracking of grain boundaries as a result of oxygen absorption and penetration at crack tips. Based on a detailed study of crack growth
in Ni3Al alloys tested in oxidizing environments, a fracture mechanism of stressassisted grain-boundary oxygen penetration has been suggested to explain the
elevated-temperature embrittlement (82). This model consists of four sequential
steps: (1) occurrence of gaseous oxygen to the crack tips where a high localized
stress field is involved, (3) oxygen penetration on its atomistic form to the stress
field ahead of tips, and (4) inward development of surface cracks preferentially
along the grain boundaries, leaving some secondary cracks (83). Steps (2) and (4)
proceed continuously and repeatedly during deformation, leading to premature
303
304
Stoloff
tensile ductility of FeAl and Fe3Al alloys in air at ambient temperatures and the
ductility of B-doped Ni3Al alloys at elevated temperatures (76). Unfortunately,
the oxide films crack after stretching a few percent and their protective effect
disappears. Surface coatings also should be useful in protecting underlying alloys
from hydrogen or oxygen penetration along grain boundaries or through the lattice; however, this effect has not yet been well demonstrated.
B. Grain Size and Shape
Columnar-grained structures have proven to be effective in reducing environmental embrittlement in Ni3Al alloys of varying aluminum contents, tested in moist
air at room and elevated temperatures (8890). For example, formation of a columnar-grained structure in boron-doped Ni3Al produced by directional levitation-zone remelting increases the ductility in air from 0.2% to 33% at temperatures in the range 600700C (88). The loss of ductility in air is accompanied
by a change in fracture mode from microvoid coalescence to intergranular in
equiaxed material. However, the columnar structure displayed mainly transgranular failure in both air and vacuum. The beneficial effect of the columnar-grained
structure with grain boundaries oriented parallel to the stress axis is attributed
to minimizing the normal stress across grain boundaries, thereby suppressing
nucleation and propagation of cracks along boundaries even when those boundaries are weakened by oxygen penetration. Hirano (89) has reported high ductility
in stoichiometric Ni3Al tested both parallel and perpendicular to the growth direction. More recent work shows that elongation increases with decreasing solidification rate; see Fig. 17 (90). The technique also is effective in Al-rich Ni3Al,
leading the authors to suggest that unidirectional solidification may be effective
in improving ductility of two-phase Ni3AlNiAl alloys. A detailed analysis of
grain-boundary chemistry and misorientation is required to fully understand these
results. However, it appears that the mostly low-angle and 3-type boundaries
resulting from unidirectional solidification are less affected by moisture than are
general grain boundaries in conventionally processed material (91). Studies of
unidirectional solidification should be extended to other intermetallic systems.
The role of grain size seems to be the same for intergranular and transgranular fracture in that refining grain size tends to reduce susceptibility to embrittlement both at room and elevated temperatures. Takeyama and Liu (92) have shown
that heat treatment of Ni3Al alloys in oxidizing atmospheres can cause embrittlement in subsequent tests at both ambient and elevated temperatures. The degree
of embrittlement is essentially zero for fine-grained (20 m) alloys, but it is very
pronounced at 200 m, as shown in Fig. 18 (92). Apparently, a thin protective
Al-rich oxide film is formed on fine-grained material, whereas a less protective,
predominantly Ni-rich film forms on coarse-grained samples. The aluminum-rich
oxide on fine-grained material may form as a result of short-circuit diffusion of
aluminum atoms from the interior to the surface.
305
Fig. 17 Effect of growth rates (mm/h) on ductility of Ni26 at% Al. (From Ref. 90.)
Fig. 18 Effect of grain size on ductility of bare and preoxidized Ni23 at% Al0.5 Hf
0.5 B at 600C and 760C in vacuum. (From Ref. 92.)
306
Stoloff
C. Alloy Additions
Several striking examples of alleviation of embrittlement by alloying may be
cited. The ductilizing effect of boron on low-temperature fracture of Ni3Al and
Ni3Si in air is clearly associated with the inhibition of moisture-induced embrittlement (22,23). Similarly, chromium enhances high-temperature ductility of Ni3Al,
presumably by altering the kinetics of oxide film formation (76). Chromium also
has a beneficial effect on the low-temperature ductility of Fe3Al alloys in air (61).
However, chromium does not prevent environmental embrittlement under cyclic
loading conditions (44,60). Finally, zirconium additions have been shown to improve both ductility and fatigue crack growth behavior in air of Fe28 Al5 Cr
alloys, although the mechanism for such improvements remains obscure (28).
D. Processing Techniques
Boron-doped Ni3Al alloys prepared by rapid solidification are more prone to environmental embrittlement than alloys produced by conventional casting and thermomechanical processing (93). The latter exhibit much higher ductility at 750C
in vacuum than does the rapidly solidified material, although at 600C, there is
little effect of microstructure. The same workers reported that spray-formed Ni
10 at% Co24% Al with boron exhibited zero ductility in vacuum at 760C,
but ductility could be substantially improved by subsequent thermomechanical
processing. A number of factors could be influencing the results obtained with
differing processing techniques, including grain size and shape, grain-boundary
energy, oxygen and other impurities, mobile dislocation density, thermal history,
and texture. Of these, the only factor definitely linked with low ductility in many
alloys is impurity content.
E.
Prestrain
There is now considerable evidence that prestrain can reduce the severity of environmental embrittlement in Ni3Al and Ni3Si alloys (35,40,94). The magnitude of
the effect depends on the degree of prestrain as well as the temperature at which
it is performed. It remains to be seen whether this effect also can be found in
other intermetallics, especially the iron aluminides. There are several possible
mechanisms for the prestrain effect, such as dislocations acting as traps for hydrogen or the creation of compressive residual stresses in the case of prestrain by
shot peening. It has been argued that the beneficial effects of prestrain must mean
that dislocation-assisted transport of hydrogen does not occur. However, it must
be pointed out that the dislocation transport mechanism depends on the simultaneous application of strain and exposure to hydrogen, a condition clearly unmet by
the prestrain experiments.
V.
307
SUMMARY
Many nickel- and iron-based intermetallics are embrittled by moisture and other
hydrogen-containing environments at low temperatures and by oxygen at elevated
temperatures. Crystal structure, composition, metallurgical variables, and test
conditions all play a role in determining the degree of embrittlement. Dopants
such as boron and zirconium have proven to be effective in inhibiting embrittlement under some conditions, but the effects are complex. The emphasis in the
past has been on empirical observations of behavior; relatively little theoretical
modeling has been conducted. Nevertheless, several methods of alleviating embrittlement have been noted, including compositional variations, control of grain
structure, prestrain, and the use of coatings. Such efforts must be pursued vigorously in order for these intermetallics to reach their potential as structural materials.
ACKNOWLEDGMENTS
The author is grateful to the Department of Energy, Fossil Energy AR&TD Materials Program for financial support under Subcontract No. 19X-SF521C with Martin Marietta Energy Systems.
REFERENCES
1. CT Liu. In: O Izumi, ed. Intermetallic CompoundsStructure and Mechanical Properties. Sendai: Japan Institute of Metals, 1991, pp. 703711.
2. M Takeyama, CT Liu. Mater Sci Eng A153:538547, 1992.
3. NS Stoloff. In NS Stoloff, VK Sikka, eds. Intermetallic Compounds, Physical Metallurgy and Processing. New York, Chapman & Hall, 1996, pp. 479516.
4. T Takasugi. Acta Metall 39:21572167, 1991.
5. T Takasugi, N Masahashi, O Izumi. Scripta Metall 20:13171321, 1986.
6. N Masahashi, T Takasugi, O Izumi. Metall Trans 19A:353358, 1988.
7. CT Liu, WC Oliver. Scripta Metall 25:19221937, 1991.
8. T Takasugi, O Izumi. Acta Metall 34:607618, 1986.
9. C Nishimura, CT Liu. Scripta Metall 25:791794, 1991.
10. CT Liu, M Takeyama. Scripta Metall 24:15831586, 1990.
11. CT Liu, EH Lee, CG McKamey. Scripta Metall 23:875880, 1989.
12. CT Liu, CG McKamey, EH Lee. Scripta Metall 24:385390, 1990.
13. GM Camus, N Stoloff, DJ Duquette. Acta Metall 37:14971501, 1989.
14. WJ Chia, YW Chung. Intermetallics 4:283288, 1996.
15. AK Kuruvilla, NS Stoloff. Scripta Metall 19:8387, 1985.
16. CT Liu. Scripta Metall 27:2528, 1992.
308
Stoloff
309
310
72.
73.
74.
75.
76.
77.
78.
79.
80.
81.
82.
83.
84.
85.
86.
87.
88.
89.
90.
91.
92.
93.
94.
95.
Stoloff
WJ Chia, YW Chung. J Vac Sci Technol A13:1687, 1995.
WJ Chia, YW Chung. Intermetallics 3:505509, 1995.
YF Zhu, CT Liu, CH Chen. Scripta Metall 35:14351439, 1996.
GM Bond, IM Robertson, HK Birnbaum. Acta Metall 37:14071413, 1989.
CT Liu, VK Sikka. J Metals 38(5):1921, 1986.
CT Liu. In: NS Stoloff, CC Koch, CT Liu, O Izumi, eds. High Temperature Ordered
Intermetallic Alloys II. Pittsburgh, PA: Materials Research Society, 1987, Vol. 81,
pp. 355367.
DJ Alexander. Technology Transfer Conference and Workshop on Nickel and Aluminides, Oak Ridge, TN, 1991.
WC Oliver. In: CT Liu, A Taub, NS Stoloff, CC Koch, eds. High Temperature
Ordered Intermetallics III. Pittsburgh, PA: Materials Research Society, 1989, Vol.
133, pp. 397402.
NS Stoloff, GE Fuchs, AK Kuruvilla, SJ Choe. In: NS Stoloff, CC Koch, CT Liu,
O Izumi, eds. High Temperature Ordered Intermetallic Alloys II. Pittsburgh, PA:
Materials Research Society, 1987, Vol. 81, pp. 247261.
S Miura, CT Liu. Scripta Metall 26:17531758, 1992.
CA Hippsley, JH Devan. Acta Metall 37:14851496, 1989.
W Matuszyk, G Camus, DJ Duquette, NS Stoloff. Metall Trans 21A:29672976,
1990.
K Aoki, O Izumi. J Mater Sci 14:18001806, 1979.
Z Jishan, Z Zhiya, T Yajun, Z Jinghua, Y Yang, L Yengao, D Dongsheng, H Zhiangqi. In: CT Liu, A Taub, NS Stoloff, CC Koch, eds. High Temperature Ordered
Intermetallic Alloys III. Pittsburgh, PA: Materials Research Society, 1989, Vol. 133,
pp. 549554.
M Nazmy. ABB Power Generation Ltd., Baden, Switzerland, unpublished work.
Z Yizhang, Z Tianxiang, T Yingjie, Z Bingda, Z Yaoxiao, H Zhuangqi. In: CT Liu,
A Taub, NS Stoloff, CC Koch, eds. High Temperature Alloys III. Pittsburgh, PA:
Materials Research Society, 1989, Vol. 133, pp. 555559.
CT Liu, WC Oliver. J Mater Res 4:294299, 1989.
T Hirano. Scripta Metall 25:17471752, 1991.
T Hirano, T Mawari. In: NS Stoloff, RH Jones, eds. Processing and Design Issues
in High Temperature Materials. Warrendale, PA: TMS, 1996, pp. 425432.
EP George, M Imai, T Hirano. Intermetallics 5:425432, 1997.
M Takeyama, CT Liu. Acta Metall 37:26812688, 1989.
A Taub, KM Chang, CT Liu. Scripta Metall 20:16131618, 1986.
CL Ma, T Takasugi, S Hanada. Scripta Metall 34:16331639, 1996.
I Baker, DJ Gaydosh. In: NS Stoloff, CC Koch, CT Liu, D Izumi, eds. High Temperature Ordered Intermetallic Alloys II. Pittsburgh, PA: Materials Research Society,
1987, Vol. 81, pp. 315320.
10
Nonoxide Ceramics
Nathan S. Jacobson
NASA Glenn Research Center, Cleveland, Ohio
Elizabeth J. Opila
Cleveland State University, Cleveland, Ohio
I.
INTRODUCTION
Nonoxide ceramics such as silicon carbide (SiC) and silicon nitride (Si3N4) are
promising materials for a wide range of high-temperature applications. These
include such diverse applications as components for heat engines, high-temperature electronics, and reentry shields for space vehicles. Table 1 lists a number
of selected applications. Most of the emphasis here will be on SiC and Si3N4.
Where appropriate, other nonoxide materials such as aluminum nitride (AlN) and
boron nitride (BN) will be discussed. Proposed materials include both monolithic
ceramics and composites. Composites are treated in more detail elsewhere in this
volume; however, many of the oxidation/corrosion reactions discussed here can
be extended to composites. In application, these materials will be exposed to a
wide variety of environments. Table 1 also lists reactive components of these
environments.
It is well known that SiC and Si3N4 retain their strength to high temperatures. Thus, these materials have been proposed for a variety of hot-gas-path
components in combustion applications. These include heat-exchanger tubes,
combustor liners, and porous filters for coal combustion products. All combustion
gases contain CO2, CO, H2, H2O, O2, and N2. The exact gas composition is dependent on the fuel-to-air ratio or equivalence ratio. The equivalence ratio (EQ) is
311
Nonoxide ceramic
Application
Turbine engine components
Combuster liners
Blades and vanes
Piston engine components
Pistons
Valves
Industrial furnaces
Heat exchangers
Coal combustion
Particulate filters
Chemical process vessels, coal gasifiers, waste incinerators
Reentry shields
312
Table 1
Approximate use
temperatures
(C)
SiC, Si3N4
Composites
9001400
SiC, Si3N4
Composites
9001400
SiC
Composites
SiC
Composites
SiC, Si3N4
9001400
7001000
9001400
Electronic substrates
AlN
BN
Crucibles, insulators
Liquid metal containers
Processing, heat transfer
Pump bearings, cooling lines for
nuclear factors
BN
Various carbides and nitrides
9001400
6001400
SiC, Si3N4
300600
10001500
Use: 600
Processing: 1200
Use: 200
Processing: 1000
9001400
Ref.
4
5
6, 7
8
9
10
11
12
13
High-temperature semiconductors
SiC
Composites
SiC
Environment
Nonoxide Ceramics
313
Fig. 1 Equilibrium gas composition versus equivalence ratio. (From Ref. 24.)
314
(1)
(2)
(3)
(4)
4. Carbides
MC O2(g) MO CO(g)
(5)
5. Nitrides
MN O2(g) MO NO(g)
(6)
or
2MN O2(g) 2MO N2(g)
(7)
Nonoxide Ceramics
315
Fig. 2 Recession due to oxidation for selected materials, with protective oxide scale
indicated. (From Ref. 14.)
Most of these materials oxidize according to a parabolic rate law, but some oxidize according to linear kinetics. Hence, Fig. 2 gives oxidation kinetics in terms
of recession to account for both kinetic laws. The reader is referred to Ref. 14 and
references contained therein for more details on the kinetics of these reactions.
The important point from Fig. 2 is that only materials which form protective
silica or alumina films are useful for long times in oxidizing environments. Most
of the borides, carbides, and nitrides do not form protective metal oxide scales.
Hence, the focus of this chapter will be on silica-forming ceramics, SiC and Si3N4,
and the alumina-forming ceramic, AlN. We shall also include a brief discussion
of BN, which is an important material in many areas of technology.
316
Nonoxide Ceramics
317
Fig. 3 Structure of SiO2: (a) amorphous; (b) crystalline; (c) with sodium cations.
(Adapted from Ref. 15.)
318
Fig. 4
Schematic of electrobalance and vertical tube furnace used for oxidation studies.
fundamental oxidation behavior of these materials. These pure materials are also
important in microelectronic applications.
3. Oxidation of Silicon
Silicon oxidation has been studied in great detail due to its application in microelectronic devices. The model of Deal and Grove (18) has been used to successfully describe silicon oxidation behavior at all but the shortest times. This model
considers two possible rate-limiting steps for the oxidation of silicon. First, the
reaction of oxygen with silicon according to a linear rate law controls the oxidation rate at short times or for thin scales. Second, transport of oxygen through
Nonoxide Ceramics
319
320
Fig. 6 Arrhenius plot of log kp versus 1/T obtained in dry oxygen for Si, chemicallyvapor-deposited (CVD) SiC, CVD Si3N4, and Si3N4 with additives. (Data from Refs. 18,
20, and 35.)
the growing silica scale controls the oxidation rate according to a parabolic rate
law for long times and for thick scales. These two rate laws are combined into
a single expression that is valid for all times:
t
xo
1 2
1/2 A
A /4B
1/2
(8)
where xo is the oxide thickness, t is time, is the offset time which corrects for
the presence of an initial oxide layer, B is the parabolic rate constant, and B/A
is the linear rate constant. At short times, this expression simplifies to the linear
rate law xo (t )B/A; at long times, the expression simplifies to the parabolic
law x 2o Bt. For long-term applications of silica-forming materials, the parabolic
law usually describes the oxidation kinetics adequately.
Deal and Grove have derived the parabolic rate constant in terms of the
properties of silica; that is,
B
2DeffC*
N
(9)
Nonoxide Ceramics
321
and
C* kP n
(10)
where Deff is the effective diffusivity of the oxidant in silica, C * is the equilibrium
concentration of the oxidant in the oxide, N is the number of oxidant molecules
incorporated into a unit volume of the oxide layer, P is the oxidant pressure, n
is a power-law exponent, and k is a constant. The parabolic rate constant was
found to vary linearly with the oxidant pressure for both oxygen and water vapor
(i.e., n 1), indicating that the oxidant did not dissociate and molecular permeation through the silica is the rate-limiting step for oxidation in the parabolic
regime.
4. Oxidation of SiC
The oxidation of SiC is similar to the oxidation of silicon because a silica scale
forms in both cases. In this case, however, the gases generated, as shown in the
following oxidation reactions, will cause some differences:
SiC 3/2O2(g) SiO2 CO(g)
(11)
(12)
The linear reaction rate is different because the oxidation of C to CO(g) or CO2(g)
occurs in addition to the oxidation of Si. The parabolic oxidation rate could be
different if the outward transport of CO or CO2, rather than the inward transport
of oxygen, limited the oxidation rate. It has been shown that the oxidation rates
of SiC are about a factor of 2 slower than silicon (1921) due to the extra consumption of oxygen in the reaction with C, as predicted (22). In addition, the
activation energy for oxidation (20) is nearly identical to that of silicon (18) (see
Fig. 6) and the permeation of molecular oxygen through silica (16). Finally, the
oxidation rate of SiC is found to depend on the oxygen partial pressure (23).
Therefore, it is generally agreed that oxygen transport inward is the rate-limiting
step for parabolic oxidation of SiC. This issue has been discussed more fully in
Refs. 24 and 25.
The dependence of the parabolic rate constant on the oxygen partial pressure, given by the power-law exponent, n, in Eq. (10), gives information about
the type of oxygen transport occurring in the silica. For n 1, molecular permeation of oxygen occurs. For n 1, some dissociation of oxygen into a charged
species occurs and network diffusion of oxygen by a bond-breaking process is
likely. Zheng et al. (23) have determined the power-law exponent for the oxidation of SiC to vary between 0.6 and 0.3 at temperatures from 1200C to 1500C.
This implies that some combination of permeation and network diffusion limits
the oxygen transport through the silica scale grown on SiC, with network diffusion increasing with temperature. This is confirmed by 18O tracer diffusion stud-
322
ies, which show isotope exchange with network oxygen becomes increasingly
important as the temperature is increased (26). However, the similar activation
energies for molecular oxygen permeation of silica (16) and for oxidation of SiC
up to temperatures of 1500C (20) indicates that oxygen transport is dominated
by the permeation mechanism.
The effect of crystallization of silica on SiC oxidation is a complex topic.
Some investigators have used crystallization to explain observed rate changes
and/or activation energy changes. However, current evidence indicates that many
of these deviations in rate law and activation energies can be explained by impurities in either the environment or the SiC (17,20,27,28). Ogbuji (29) has shown
that silica scales fully crystallized during argon anneals at 1300C do result in
slower oxidation rates by about a factor of 30, but that fully crystalline scales
are not found in actual practice because amorphous silica is continually formed
during the course of SiC oxidation.
5. Oxidation of Si3N4
Like SiC, the oxidation of Si3N4 results in the formation of a silica layer and the
generation of gaseous products according to the following simplified reaction:
Si3N4 3O2(g) 3SiO2 2N2(g)
(13)
(14)
Nonoxide Ceramics
323
324
Fig. 7 Thermal expansion of SiC, Si3N4, amorphous SiO2, and crystalline SiO2 as a
function of temperature. (From Ref. 24.)
AlN is currently a candidate for electronic substrates due to its high thermal
conductivity (9). As an alumina former, it is expected to exhibit slow oxidation
kinetics. However, it appears to oxidize quite rapidly above about 1000C, with
rates quite dependent on grain size, porosity, the presence of second phases, and
impurity content (49). In many instances, linear kinetics have been reported, in
contrast to alumina formation on metal alloys (50). In addition, it is well established that water vapor in the oxidizing stream leads to extensive attack (51).
The reasons for this extensive alumina formation in dry and wet oxygen are not
Nonoxide Ceramics
325
Fig. 8 Cyclic oxidation kinetics of SiC and Si3N4 obtained in air at 1300C. (Adapted
from Ref. 46.)
clear. Clearly, nitrogen must escape through the alumina scale, which may lead
to micropore formation. It has been suggested that the injection of N3 leads to
excess vacancy formation in the alumina and more rapid diffusion rates (52). In
addition, the coefficient of thermal expansion mismatch between AlN and Al2O3
may lead to scale cracking (53).
326
spheres, enhanced solubility of water in the silica scale, alterations in the silica
scale viscosity and structure and, finally, formation of volatile silicon hydroxide
and oxyhydroxide species. Each one of these topics will be discussed separately
below. Again, references to the oxidation behavior of silicon in water vapor are
made for comparison.
Oxidation of silicon by water vapor occurs by the following reaction:
Si 2H2O(g) SiO2 2H2(g)
(15)
The enhanced growth rates of SiO2 by this reaction, relative to dry oxygen, has
commercial application in the more rapid fabrication of microelectronic devices.
Oxidation of SiC by water vapor occurs by the reactions (54)
SiC 3H2O(g) SiO2 CO(g) 3H2(g)
for T 1400K
(16)
for T 1400K
(17)
(18)
Note that in each case, the solid product SiO2 is formed with the generation of
additional gaseous products.
1. Impurity Transport
Impurities normally found in water-vapor-containing environment form M
OH(g) species, where M is the impurity element such as Na, K, Fe, and so forth.
Because MOH(g) species are so thermodynamically stable, the quantity of impurities transported to a silica scale forming on SiC or Si3N4 in a water-vapor
containing environment is increased. This increased contamination of the silica
scale results in faster transport rates of oxidant through the scale, and thus increased oxidation rates of SiC or Si3N4. This effect has been identified for both
Na (55) and K contamination (56) during the oxidation of SiC.
Nonoxide Ceramics
327
oxidation rate constants has been observed for SiC (55,57) and in some cases
for Si3N4 (58).
Deal and Grove (18) have also demonstrated that the parabolic oxidation
rate has a power-law dependence on the water-vapor partial pressure, with the
power-law exponent, n [Eq. (10)], equal to 1, indicating that molecular water
diffusion is the rate-controlling step in the oxidation of silicon. Power-law exponents for the oxidation of SiC of 0.67 (59) and 0.67 to 0.85 (54) have been
obtained. This indicates dissociation of water into a singly-charged species is
probable. The discrepancy between the findings for silicon and SiC is unexpected
because the transport properties of silica should be independent of the substrate
material. A complex relationship between parabolic rate constant and water-vapor
partial pressure has been observed for Si3N4 (58), but this may be explained by
silica volatility, as described in Sec. III.A.4.
Fig. 9 Bubbles formed in SiO2 due to oxidation of SiC in 90% water vapor/10% O2 at
1200C. (From Ref. 54.)
328
bubbles, in turn, create shorter transport paths for the oxidant to the SiC/SiO2
interface by decreasing the effective oxide thickness, thereby increasing the oxidation rate. In contrast, bubbles are not observed in the scales formed on SiC in
dry oxygen or on silicon in wet oxygen.
(19)
(20)
These volatile silicon hydroxides and oxyhydroxides have been identified experimentally using the transpiration technique (61) as well as mass spectrometry (62).
The reaction of water vapor with SiC or Si3N4 involves the oxidation reaction
[Eqs. (16)(18)] and the simultaneous linear volatilization reaction [Eqs. (19)
and (20)] resulting in overall paralinear kinetics (63). Paralinear kinetics, as measured by weight change, are shown in Fig. 10. At long times, the weight change
and recession can be approximated by the linear volatilization rate alone. At this
time, a steady state is achievedthe silica scale is consumed at the same rate it
is formed, leaving a constant oxide thickness. The volatilization rate of silica is
controlled by transport of the volatile species through a gaseous boundary layer
Fig. 10
Nonoxide Ceramics
329
v 1/2P(H2O)2
1/2
P total
(21)
where kl is the linear volatility rate, v is the gas velocity, and P is pressure. Thus,
in high-pressure and high-velocity applications, such as a gas turbine engine,
the recession of a SiC or Si3N4 component by the volatility mechanism can be
significant.
In summary, water vapor has deleterious effects on the durability of silica
formers through a number of different mechanisms: increased transport of impurities to the oxide surface, increased oxide formation rates due to impurity effects,
enhanced solubility of water vapor, short-circuit paths for oxidant transport, increased permeability of the oxidant, and, finally, consumption of the silica scale
and component recession due to silica volatility.
B. Carbon Dioxide
Although CO2 is a major component of combustion gases (Fig. 1), there are only
a few limited oxidation studies of SiC in CO2 (6467). It has been found that
the oxidation rates of SiC in CO2 are less than those in oxygen. Because the
oxide growth rates are so low, it is difficult to determine whether the kinetics
are linear, parabolic, or more complex. The oxidation weight gain for SiC in CO2
is shown compared to that observed in oxygen and a 50% water vaporoxygen
mixture in Fig. 11 (67). Thus, in a complex combustion environment, the effects
of CO2 as an oxidant are negligible.
C. Effects of Low P(O2), Reducing Gases, H2S, Cl2
1. SiO(g) Formation
A unique, but very important issue with silica formers is the highly stable volatile
suboxide, SiO(g). Consider the free energy of formation at 1500 K:
/2Si 1/2O2(g) 1/2SiO2(s), G 322 kJ/mol
(22)
G 227 kJ/mol
(23)
Si /2O2(g) SiO(g),
1
Note that these equations are normalized to 1 mol of oxygen atoms. The free
energies of formations are close, indicating that SiO(g) can readily form.
There are two conditions which lead to SiO(g) formation (68): active oxidation and oxidation in mixed oxidizing/reducing gases. In the active oxidation
330
Fig. 11 Comparison of SiC oxidation in 50% H2O/O2, O2, and CO2 at 1200C. (From
Ref. 67.)
case, the partial pressure of oxygen is too low to form a stable SiO2 film, but
sufficient to form SiO(g). This can occur in certain heat-treating environments
(69). Let us begin with a bare SiC surface. As the partial pressure of oxygen is
increased, SiO(g) will form in increasing quantities. Then, SiO(g) formation will
stop and a stable SiO2 film will form. This is the active-to-passive transition.
Wagner has derived this for pure silicon (70). His results can be easily extended
to SiC and Si3N4 (71). The active-to-passive transition occurs when sufficient
SiO(g), via Eq. (23), is generated to satisfy the SiC/SiO2 condition for equilibrium. There is some controversy about the exact equilibrium condition, but
reasonable agreement with measurements is obtained from the following equilibria:
2SiC SiO2 3Si 2CO(g)
(24)
The transition P(O2) for active-to-passive oxidation is calculated based on equilibrium conditions and diffusion through the gas boundary layer to the sample.
Calculated and measured active-to-passive transitions (7274) are shown in
Fig. 12.
Now, consider the case of passive-to-active transition. Beginning with a
stable SiO2 film on SiC; as the partial pressure of oxygen is lowered, the film
Nonoxide Ceramics
331
will become unstable and SiO2 will react to form SiO(g). The transition pressure
is calculated from the decomposition of the protective SiO2 film (70):
SiO2(s) SiO(g) 1/2O2(g)
(25)
332
with appreciable amounts of CO2, CO, H2O, and H2. It has also been observed
in laboratory experiments with CO/CO2 (73) and H2 /H2O (72). Thus, a sequence
of reactions might be
SiC 3H2O(g) SiO2 CO(g) 3H2(g)
(26)
(27)
This leads to paralinear kineticssimultaneous oxide growth and linear reduction kinetics.
(28)
(29)
These reactions can lead to material consumption (5); however, when the oxygen
potential is high enough to form SiO2, the corrosion rate drops substantially.
Chlorine, which may be found in certain chemical process environments,
also leads to volatile products (75). In the case of SiC, reaction occurs with the
silicon, leading to formation of various silicon chlorides with residual carbon.
As the oxygen potential is increased, attack becomes less severe (75).
Current interest focuses on chlorine as HCl in waste incineration applications (76). Here, the situation is quite complex, involving not only HCl, but also
a range of deposits.
D. Oxidation in the Presence of Impurities and Deposits
2. Na2SO4 Deposits
As impurity levels increase, actual deposits may form. The chemistry of these
depends on application, as shown in Table 1. Corrosive deposits have long been
Nonoxide Ceramics
333
(30)
The source of NaCl depends on the application: It may be from a marine environment in the case of a heat engine, or from the process chemicals in the case of
an industrial furnace. Under the appropriate conditions, Na2SO4 forms as a condensed phase, depositing on parts. Figure 15 shows the calculated dew points
334
Fig. 14 Optical micrographs of sintered SiC coupons with carbon and boron additives,
oxidized in a burner rig at 1000C (a) 46 h with no sodium and (b) 13.5 h with a sodiumchloride-seeded flame. (From Ref. 24.)
for Na2SO4 deposition (81). The rates of deposition are also critical and have
been treated in detail (82).
The most useful interpretation of this process is with the acidbase theory
of molten salts. Na2SO4 decomposes to Na2O, which is the key reactant:
Na2SO4 Na2O SO3(g)
(31)
Note that the overpressure of SO3 sets the chemical activity of Na2O. A high
activity of Na2O is a basic molten salt; a low activity of Na2O is an acidic molten
salt. Because SiO2 is an acidic oxide, it is readily attacked by a basic molten salt:
2SiO2(s) Na2O Na2O2(SiO2)(l)
(32)
Nonoxide Ceramics
335
Fig. 15 Calculated dew points for Na2SO4 deposition. (From Ref. 24.)
This describes the product observed in Fig. 14: A solid, protective SiO2 layer
has been replaced by a liquid nonprotective sodium silicate layer. The liquid layer
allows rapid diffusion of oxygen inward and carbon monoxide outward, leading
to accelerated oxidation. The accelerated oxidation provides additional SiO2 for
reaction (32) and can lead to rapid consumption of the SiC.
The conditions for reaction (32) to proceed can be calculated from basic
thermodynamics. We can assume unit activities for the SiO2 and Na2O2(SiO2)(l)
and, hence, the threshold Na2O activity [a(Na2O)] for SiO2 dissolution is given by
a(Na2O) exp
G
RT
(33)
Here, G is the free-energy change for reaction (32). At 900C, the minimum
a(Na2O) is 1011, meaning any activity greater than this will lead to silicate formation. Because the activity of Na2O is set by P(SO3) in reaction (31), this means
that a P(SO3) of 5 105 bar or less will lead to silicate formation.
These predictions hold in the actual burner, where the P(SO3) is set by the
sulfur content of the fuel. Figure 16 shows a series of quartz coupons treated in
a burner with a low- (0.05%) and high- (0.5%) sulfur fuel. The low-sulfur fuel
has a low P(SO3) (basic salt) and, thus, allows silicate formation, whereas, the
high-sulfur fuel has a higher P(SO3) (acidic salt), which suppresses silicate
formation.
336
Fig. 16 Quartz coupons treated in burner rig for 1 h with a 2 ppm sodium (as NaCl)
seeded flame: (a) No. 2 diesel fuel (0.5% sulfur), 5 h; (b) Jet A fuel (0.05% sulfur). (From
Ref. 81.)
Actual ceramic systems are more complex than the quartz coupons used
in the preceding experiment. Ceramic systems are silica films on SiC and Si3N4
and may contain additives such as refractory oxides and carbon. We have seen
that carbon, in silicon carbide and as an additive, leads to extensive salt corrosion.
Using an electrochemical sensor, it can be shown that carbon tends to drive so-
Nonoxide Ceramics
337
dium sulfate more basic. The exact chemical mechanism for this is not clear, but
it may involve a Na2S intermediate (81).
Fig. 17 CVD SiC oxidized for 1 h at 1800C. (Courtesy of D. Fox, NASA Glenn.)
338
Fig. 18 (a) Total pressure [P(SiO) P(CO)] generated from SiC/SiO2 interactions.
(b) Total pressure [P(SiO) P(N2)] generated from Si2N2O/SiO2 interactions. (From
Ref. 24.)
Nonoxide Ceramics
339
As in the Na2SO4 case, the basic slag tends to cause more rapid material degradation than the acidic slag. Under some conditions, metal silicides may form with
a basic coal slag (4).
E.
High-Temperature Effects
Silica melts at 1723C. Transport rates are very high in a liquid scale and rapid
oxidation occurs, as shown in Fig. 17. Some of the bubbles in this sample are
from escape of CO(g), but some are also from the interaction of SiC and SiO2
at high temperatures:
SiC(s) 2SiO2(s) 3SiO(g) CO(g)
(34)
Figure 18a is a plot of total pressures generated by this reaction (87). Note that
for carbon saturated SiC, the pressures are much higher, leading to a lower upper
use temperature. An analogous situation exists for Si3N4, but based on the Si2N2O
interaction with SiO2:
Si2N2O(s) SiO2(s) 3SiO(g) N2(g)
(35)
340
Fig. 19
match between SiC and mullite. Mullite coatings on SiC were first developed at
Solar Turbines (San Diego, CA) and further developed at NASA Glenn (88,89).
The critical processing issues are surface roughening of the SiC for adherence
and application of a fully crystalline mullite coating. A mullite coating on a SiC
coupon is shown in Fig. 19. There are clear advantages with mullite coatings
in Na2O-induced corrosion due to formation of high-melting sodium-aluminosilicates, as opposed to lower-melting sodium silicates. However, the chemical
activity of silica in mullite is only about 0.4, so reactions which volatilize SiO2
can still occur readily.
It may be possible to apply other refractory oxides to SiC. As Table 2
shows, the problem of CTEs must be considered. One approach is graded coatings
from SiC to a refractory oxide such as alumina or zirconia.
V.
Nonoxide Ceramics
341
The hydrothermal corrosion of SiC may form silica by the following reactions (91,92):
SiC 2H2O(g) SiO2 CH4(g)
(36)
(37)
Given nearly equal amounts of SiC and water vapor at high temperatures, free
carbon can form (92,93):
SiC 2H2O(g) SiO2 C 2H2(g)
(38)
This provides a method for synthesis of carbon films. However, in most applications, there is excess water and SiO2 is formed as a protective film. The hydrothermal corrosion of Si3N4 is similar and occurs by
Si3N4 6H2O(g) 3SiO2 4NH3(g)
(39)
(40)
In the case of Si3N4, this leads to attack of the grain-boundary phases and dislodging of the grains (94). Methodical studies have been done on the effect of additives in Si3N4 on its hydrothermal corrosion behavior (13,94). The morphology
of attack has been correlated with the type of additive and its resistance to hydrothermal attack.
Ceramics which form oxides other than silica behave differently under hydrothermal conditions. AlN forms AlOOH (95). Other ceramics such as TiC and
ZrC, which do not form protective oxides at high temperatures, also do not form
protective oxides under hydrothermal conditions (91).
342
tack of porous ceramics. More importantly it can lead to substantial grain-boundary attack, depending on the relation of the solidliquid interfacial energy and
the grain-boundary energy (12).
Nonoxide Ceramics
343
Fig. 20 Fracture origin for sintered -SiC reacted with Na2SO4 for 48 h at 1000C. The
area noted in (b) is enlarged in (c). (c) The preferential grain-boundary attack is observed
at arrows. (From Ref. 108.)
344
VIII. CONCLUSIONS
Nonoxide ceramics are a promising class of materials for a wide range of applications at high temperatures. Before application, their interactions with the hightemperature environment must be well understood. The silica-forming ceramics
(SiC and Si3N4) exhibit the best oxidation behavior and the focus of this chapter
is on them. However, some information is included on AlN and BN, which show
promise for several specific applications.
High-temperature oxidation is critical to most applications. The oxidation
of SiC is rate limited by the diffusion of oxygen inward through the growing
silica scale. The oxidation of Si3N4 is more complex: Here, the oxynitride scale
plays a role. Most engineering ceramics contain additives, which tend to increase
oxidation rates.
Actual applications involve complex gas mixtures. The effects of water
vapor, carbon dioxide, low oxidant pressures, and mixed oxidizing/reducing
gases are discussed. Water vapor enhances oxidation, whereas carbon dioxide is
a less effective oxidant. Low oxidant pressures and mixed oxidizing/reducing
gases lead to SiO(g) formation.
Corrosive deposits are also encountered in some applications. These include sodium sulfates, vanadate, and slags. Deposit-induced corrosion can be
described with the acidbase theory of oxides. Because silica is an acidic oxide,
it is readily attacked by a basic molten salt.
These problems may be minimized with the application of a refractory
oxide coating. Mullite shows a good thermal expansion match to SiC and is a
promising starting point. It may be possible to develop other more refractory
oxide coatings.
Some proposed applications involve hydrothermal conditions: water at high
temperatures and pressures. Hydrothermal corrosion is analogous to high-temperature oxidation in many ways. Ceramics which form protective high-temperature
oxides also form protective hydrothermal oxides. The form of corrosive attack
is also similar.
Finally, the interaction of corrosive attack and degradation of mechanical
properties is discussed briefly. It is essential to understand this before these materials can be applied. Microstructural changes such as pitting and grain-boundary
attack lead to a concurrent loss of mechanical integrity. The development of
tougher ceramics and coatings are necessary to minimize this problem.
Nonoxide Ceramics
345
REFERENCES
1. HB Probst. Structural ceramics in heat enginesThe NASA viewpoint. Proceedings of International Symposium on Ceramic Components for Heat Engines, Japan,
1983, pp. 4558.
2. AF McLean. Materials approach to engine component design. Proceedings of the
Second International Symposium on Ceramic Materials and Components for Engines, Bad Honnef, Germany, 1986, pp. 10231034.
3. JI Federer, TN Tiegs, DM Kotnick, D Petrak. Analysis of candidate silicon carbide
recuperator material exposed to industrial furnace environments. Report No.
ORNL/TM-9677, Oak Ridge National Laboratory, 1985.
4. MK Ferber, J Ogle, VJ Tennery, T Henson. Characterization of corrosion mechanisms occurring in a sintered SiC exposed to basic coal slags. J Am Ceram Soc
68:191197, 1985.
5. FA Costa Oliviera, RAH Edwards, RJ Fordham, JHW De Wit. Factors limiting the
application of silicon nitride ceramics in sulphur-containing environments of low
oxygen potential at high temperatures. In: K. G. Nickel, ed. Corrosion of Advanced
Ceramics Measurement and Modelling. Dordrecht: Kluwer, 1994, pp. 177188.
6. A Lacombe, C Bonnet. Ceramic matrix composites, key materials for future space
plane technologies. Proceedings of the AIAA Second International Aerospace
Planes Conference, Orlando, FL, 1990.
7. NS Jacobson, RA Rapp. Thermochemical degradation mechanisms for the reinforced carbon/carbon panels on the space shuttle. NASA Technical Memorandum
106793, 1995.
8. VE Chelnokov, AL Syrkin. High temperature electronics using SiC: Actual situation and unsolved problems. Mater Sci Eng B 46:248253, 1997.
9. PA Janeway. Aluminum nitride: Making the grade in demanding electronic applications. Ceram Ind 137:2832, 1991.
10. CH Henager Jr, RH Jones. Effects of an aggressive environment on the subcritical
crack growth of a continuous-fiber ceramic matrix composite. Ceram Eng Sci Proc
13:411419, 1992.
11. A Lepp, KA Schwetz, K Hunold. Hexagonal boron nitride: Fabrication, properties,
and applications. J Eur Ceram Soc 5:39, 1989.
12. R Sangiorgi. Corrosion of ceramics by liquid metals. In: K. G. Nickel, ed. Corrosion
of Advanced Ceramics Measurement and Modelling. Dordrecht: Kluwer, 1994, pp.
261284.
13. K Oda, T Yoshio, Y Miyamoto, M Koizumi. Hydrothermal corrosion of pure, hot
isostatically pressed silicon nitride. J Am Ceram Soc 76:13651368, 1993.
14. NJ Shaw, JA DiCarlo, NS Jacobson, SR Levine, JA Nesbitt, HB Probst, WA Sanders, CA Stearns. Materials for engine applications above 3000Fan overview.
NASA Technical Memorandum 100169, 1987.
15. MA Lamkin, FI Riley, RJ Fordham. Oxygen mobility in silicon dioxide and silicate
glasses: A review. J Eur Ceram Soc 10:347367, 1992.
16. FJ Norton. Permeation of gaseous oxygen through vitreous silica. Nature 171:701,
1961.
346
17. EJ Opila. Influence of alumina reaction tube impurities on the oxidation of chemically-vapor-deposited silicon carbide. J Am Ceram Soc 78:11071110, 1995.
18. BE Deal, AS Grove. General relationship for the thermal oxidation of silicon. J
Appl Phys 36:37703778, 1965.
19. JA Costello, RE Tressler. Oxidation kinetics of silicon carbide crystals and ceramics: I. In dry oxygen. J Am Ceram Soc 69:674681, 1986.
20. LUJT Ogbuji, EJ Opila. A comparison of the oxidation kinetics of SiC and Si3N4.
J Electrochem Soc 142:925930, 1995.
21. CE Ramberg, G Cruciani, KE Spear, RE Tressler, CF Ramberg Jr. Passive-oxidation kinetics of high-purity silicon carbide from 800 to 1100C. J Am Ceram Soc
79:28972911, 1996.
22. K Motzfeldt. On the rates of oxidation of silicon and silicon carbide in oxygen,
and correlation with permeability of silica glass. Acta Chem Scand 18:15961606,
1964.
23. Z Zheng, RE Tressler, KE Spear. Oxidation of single-crystal silicon carbide, Part
II. Kinetic model. J Electrochem Soc 137:28122816, 1990.
24. NS Jacobson. Corrosion of silicon-based ceramics in combustion environments. J
Am Ceram Soc 76:328, 1993.
25. KL Luthra. Some new perspectives on oxidation of silicon carbide and silicon nitride. J Am Ceram Soc 74:10951103, 1991.
26. Z Zheng, RE Tressler, KE Spear. Oxidation of single-crystal silicon carbide. Part
I. Experimental studies. J Electrochem Soc 137:854858, 1990.
27. WC Tripp, JW Hinze, MG Mendiratta, RH Huff, AF Hampton, JE Stroud, ET
Rodine. Internal structure and physical properties of ceramics at high temperatures.
Aerospace Research Laboratories Report No. 75-0130, 1975, pp. 116125.
28. CE Ramberg. PhD dissertation, University of Pennsylvania, Philadelphia, 1997.
29. LUJT Ogbuji. Effect of oxide devitrification on oxidation kinetics of SiC. J Am
Ceram Soc 80:15441550, 1997.
30. LUJT Ogbuji, DT Jayne. Mechanism of incipient oxidation of bulk chemical vapor
deposited Si3N4. J Electrochem Soc 140:759766, 1993.
31. LUJT Ogbuji, SR Bryan. The SiO2 Si3N4 interface. Part I: Nature of the interphase.
J Am Ceram Soc 78:12721278, 1995.
32. BW Sheldon. Silicon nitride oxidation based on oxynitride interlayers with graded
stoichiometry. J Am Ceram Soc 79:29932996, 1996.
33. H Du, RE Tressler, KE Spear, CG Pantano. Oxidation studies of crystalline CVD
silicon nitride. J Electrochem Soc 136:15271536, 1989.
34. D Cubicciotti, KH Lau. Kinetics of oxidation of hot-pressed silicon nitride containing magnesia. J Am Ceram Soc 61:512517, 1978.
35. D Cubicciotti, KH Lau. Kinetics of oxidation of yttria hot-pressed silicon nitride.
J Electrochem Soc 126:17231728, 1979.
36. JA Costello, RE Tressler, IST Tong. Boron redistribution in sintered -SiC during
thermal oxidation. J Am Ceram Soc 64:332335, 1981.
37. DM Mieskowski, WA Sanders. Oxidation of silicon nitride sintered with rare-earth
oxide additions. J Am Ceram Soc 68:C160C163, 1985.
38. P Andrews, FL Riley. The microstructure and composition of oxide films formed
Nonoxide Ceramics
39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
54.
55.
56.
57.
58.
347
348
59. JE Antill, JB Warburton. Oxidation of silicon and silicon carbide in gaseous atmospheres at 10001300C. In: PG Atkinson, DJ Dawton, O Lutz, R Monti, J Surugue, eds. Reactions between Solids and Gases, AGARD CP-52. Paris: Advisory
Group for Aerospace Research and Development, 1970, pp. 10-110-14.
60. G Hetherington, KH Jack, JC Kennedy. The viscosity of vitreous silica. Phys Chem
Glasses 5:130136, 1964.
61. A Hashimoto. The effect of H2O gas on volatilities of planet-forming major elements: I. Experimental determination of thermodynamic properties of Ca-, Al-, and
Si-hydroxide gas molecules and its application to the solar nebula. Geochim Cosmochim Acta 56:511532, 1992.
62. EJ Opila, DS Fox, NS Jacobson. Mass spectrometric identification of SiOH(g)
species from the reaction of silica with water vapor at atmospheric pressure. J Am
Ceram Soc 80:10091012 (1997).
63. EJ Opila, RE Hann Jr. Paralinear oxidation of CVD SiC in water vapor. J Am
Ceram Soc 80:197205, 1997.
64. AC Lea. The oxidation of silicon-carbide refractory materials. J Soc Glass Technol
33:2750, 1949.
65. E Fitzer, R Ebi. Kinetic studies on the oxidation of silicon carbide. In: RC Marshall,
JW Faust Jr., CE Ryan, eds. Silicon Carbide 1973. Columbia: University of South
Carolina Press, 1974, pp. 320328.
66. W Bremen, A Naoumidis, H Nickel. Oxidation behavior of pyrolytically deposited
-SiC in an atmosphere composed of COCO2 gas mixtures. J Nucl Mater 71:56
64, 1977.
67. EJ Opila, QN Nguyen. Oxidation of silicon carbide in carbon dioxide. J Am Ceram
Soc 81:19491952, 1998.
68. EJ Opila, NS Jacobson. SiO(g) formation from SiC in mixed oxidizingreducing
gases. Oxidation of Metals 44:527544, 1995.
69. DP Butt, RE Tressler, KE Spear. Durability of SiC materials in gaseous N2 H2
CO heat treat environments I. Ind Heat 58:4448, 1991.
70. C Wagner. Passivity during the oxidation of silicon at elevated temperatures. J Appl
Phys 29:12951298, 1958.
71. SC Singhal. Thermodynamic analysis of the high-temperature stability of silicon
nitride and silicon carbide. Ceram Int 2:123130, 1976.
72. HE Kim, DW Readey. Active oxidation of SiC in low dew-point hydrogen above
1400C. Proceedings of Silicon Carbide 87, Columbus, OH, 1989, pp. 301
312.
73. T Narushima, T Goto, Y Yokoyama, Y Iguchi, T Hirai. High-temperature active
oxidation of chemically vapor-deposited silicon carbide in COCO2 atmosphere.
J Am Ceram Soc 76:25212524, 1993.
74. T Narushima, T Goto, Y Iguchi, T Hirai. High-temperature active oxidation of
chemically vapor-deposited silicon carbide in an ArO2 atmosphere. J Am Ceram
Soc 74:25832586, 1991.
75. JE Marra, ER Kreidler, NS Jacobson, DS Fox. Reactions of silicon-based ceramics
in mixed oxidation chlorination environments. J Am Ceram Soc 71:10671073,
1988.
76. P Steinmetz, C Rapin. Corrosion of Metallic Materials in Waste Incinerators. In:
Nonoxide Ceramics
77.
78.
79.
80.
81.
82.
83.
84.
85.
86.
87.
88.
89.
90.
91.
92.
93.
94.
349
350
11
Oxide Ceramics
F. S. Pettit, G. H. Meier, and J. R. Blache`re
University of Pittsburgh, Pittsburgh, Pennsylvania
I.
INTRODUCTION
352
Pettit et al.
Not placing acid and basic refractories in contact is a major requirement in the
design of furnaces such as arc furnaces for steel-melting and glass-melting furnaces. Although reactions between solid oxide ceramics are important, such reactions will not be considered specifically in this chapter. However, the processes
discussed in the liquid environment section are relevant to solid reactions. Finally,
it is important to emphasize that environmental effects on ceramics can extend
from room temperature to extremely high temperatures. In this chapter, a very
wide range of temperatures will be used to illustrate the various environmental
effects.
(1)
2G CuO G Cu2O
2RT
(2)
Oxide Ceramics
353
354
Pettit et al.
The effect of the oxygen pressure change depends on the function the oxide
ceramic may be serving. In the case of Cr2O3, which is often used to develop
oxidation resistance on nickel-, cobalt-, and iron-based alloys, the following reaction must be considered:
Cr2O3 3/2O2 2CrO3(g)
(3)
which gives
PCrO3 P O3/42 exp
G Cr2O3 2G CrO3
2RT
(4)
Equation (4) shows that the pressure P CrO3 increases with oxygen pressure. Therefore, the formation of CrO3 causes Cr2O3 to be a less effective protective oxide
barrier and this condition increased with oxygen pressure and temperature increases (G Cr2O3 2G CrO3 is negative). The vapor pressures of oxide phases are
always a factor that must be considered. The vapor pressure of Al2O3 is relatively
low and the pressures of other oxides involving aluminum and oxygen are also
low, and so, Al2O3 is an oxide ceramic that can be used at temperatures up to
13001400C with little effects of vaporization. On the other hand, MoO3 has
high vapor pressures and cannot be used at temperatures above 400500C. Gulbransen and Jansen (3) have proposed that vaporization of oxides begins to cause
problems at pressures greater than 109 atm. An interesting situation arises in the
case of SiO2, as has been discussed by Wagner (4). As indicated by the equations
SiO2 SiO(g) 1/2O2
P SiO
(5)
(6)
Oxide Ceramics
355
gas. For example, when the gas is pure nitrogen and the oxide ceramic is Al2O3,
the following reaction can be used to determine what will occur:
Al2O3 N2 2AlN 3/2O2
PN2
P O3/22
exp
(7)
2G AlN G Al2O3
RT
(8)
At 1500 K, PN2 /PO3/22 3 1030, and, therefore, if the P N2 /P O3/22 ratio in the gas is
greater than this value, the Al2O3 will be converted to AlN. Of course, the rate
at which this will occur will be determined by kinetics, which, in turn, is determined by factors that usually require some experimentation to understand. For
example, the AlN will form as a layer covering the Al2O3 and the reaction will
involve nitrogen diffusing inward and oxygen diffusing outward. If the oxygen
cannot be removed, pressures could develop that could rupture the AlN layer and
cause the rate of the reaction to be changed.
In principal, reactions, such as those described for nitrogen with Al2O3,
could occur for sulfur and carbon, as well as numerous other gases, but it is
important to emphasize that almost all metals have much greater affinities for
oxygen than for these other reactants, and, therefore, the oxide ceramics are comparatively stable. At 1500 K, the ratio PN2 /P O3/22 for equilibrium between Al2O3
and AlN is 3 1030. At a total pressure of 1 atm, if the nitrogen gas contains
even a small amount of oxygen, the oxide phase will remain stable.
C. Mixed Gases Containing Oxygen
When oxide ceramics are exposed to gases containing oxygen and another reactant, as discussed previously, the oxygen usually is not displaced unless the
oxygen pressure is extremely low. Even though the oxygen is not displaced from
the ceramic, the second reactant can affect the oxide ceramic by reactions of the
following type, where sulfur is the second gaseous reactant:
Al2O3 9/2O2 3/2S2 Al2O3 3SO3 Al2(SO4)3
(9)
The conditions for which such reactions can occur can be described by constructing thermodynamic diagrams such as the one presented in Fig. 2 for the
aluminumoxygensulfur system. Again, the rate of conversion of Al2O3 to
Al2(SO4)3 (Fig. 2) will be determined by kinetic processes. Reactions such as that
described by Eq. (9) can occur for carbon and nitrogen, as well as other gases.
Usually, most gas environments encountered in industrial practice are not appropriate for reactions of this type to take place. However, the stabilities of these
phases do vary. An interesting example is CaO or MgO, which is often present
in Al2O3 as impurities. The sulfates of these two oxides are more stable than
356
Pettit et al.
Fig. 2 Stability diagram for Al2O3 as a function of O2 and SO3 pressures at 700C where
the boundaries for the CaO/CaSO4 and MgO/MgSO4 are also included.
Al2(SO4)3, and these sulfates have been observed (5) to form on Al2O3 without
any Al2(SO4)3 formation (Fig. 3).
In some gas mixtures, the second reactant may not enter directly into a
chemical reaction with the oxide ceramic, but it still exerts a profound effect on
the mechanical properties of the ceramic. An excellent example is the effect of
water vapor on crack propagation in soda-lime silica glass as has been described
in articles by Weiderhorn (6,7).
The strength of glass is known to deteriorate in atmospheres containing
moisture. Under constant tensile load below the fracture strength at room temperature, the glass may eventually fail in a process called static fatigue. This delayed
fracture is associated with preferential attack by water of stressed SiEO bonds
at the crack tip. This is usually considered a stress corrosion associated with the
reaction between the glass and water, as indicated schematically in Fig. 4, in
which an oxygen bridge of the glass structure is broken. This was proposed by
Orowan and many others.
Wiederhorn found that the velocity of crack propagation in soda-lime silica
glass depended on the humidity according to three regimes as shown in Fig. 5.
Over a range of stresses, the velocity of crack propagation depended on the relative humidity as expected from Fig. 4. However, at higher stresses, the velocity
of the crack propagation was limited apparently by the rate of transport of the
water to the crack tip. This transport was usually assumed to be by surface diffu-
Oxide Ceramics
357
358
Pettit et al.
Fig. 5 Dependence of crack velocity in silica on the applied force. The percent relative
humidity for each set of runs is given on the right-hand side. Roman numerals identify
the different regions of crack propagation. In region I, the crack growth rate is proposed
to be controlled by adsorption or reaction of water vapor at the crack tip. In region II,
the rate is supposedly controlled by diffusion of water vapor in the crack. Region III
involves a crack propagation mode not involving water vapor.
sion along the crack or by vapor transport. At higher stresses yet, the crack velocity does not depend on humidity.
Tomazoa (8) recently discussed that either crack initiation or crack growth
can dominate static fatigue. He emphasized the evidence for water penetration
into the glass. The water would diffuse into the glass as molecular water and
react with the glass structure as shown in Fig. 4.
The interaction of water with the strength of glass is well established and
is expected to affect the strength of many ceramics which contain a glassy phase.
Michalski and Freiman (9) proposed that such processes can take place in other
oxides and species containing polar molecules. It has been found that such an
Oxide Ceramics
359
effect can occur in Al2O3. Furthermore, similar processes have been proposed as
the cause of the excessive spalling of alumina scales from metallic alloys exposed
to gases containing water vapor (10).
1. Theoretical Considerations
The most important reactions between oxide ceramics and molten deposits can
be viewed as involving oxide ions, where considering the metal M, oxide ions
can be donated to the melt,
MO2 M4 202
(10)
(11)
360
Pettit et al.
Depending on the characteristics of the melt, the oxide ions may not exist
as major species but can be defined by reactions of the type
SO3 O2 SO42
(12)
2SO3 O2 S2O72
(13)
V2O5 O2 2VO3
(14)
In view of such reactions, at times it is useful to view the oxide ions as a basic
component in the melt where in the previous equations the acidic components
would be SO3 or V2O5.
Stability diagrams can be used to identify the reactions that may occur
between an oxide ceramic and a molten deposit. For example, in the case of a
Na2SO4 deposit, the diagram in Fig. 6 defines the stability range of Na2SO4 at a
fixed temperature in terms of oxygen and SO3 pressures. The SO3 pressure is the
acidic component in the melt and is related to the basic component, Na2O, or
O2, via the equation
Na2O SO3 Na2SO4
(15)
Fig. 6 Stability diagram for the NaOS system at 1000C; the dashed lines are sulfur
isobars.
Oxide Ceramics
361
1
a Na2OP SO3
(16)
(17)
The stability diagram for the oxide ceramic can now be superimposed
upon the stability diagram for Na2SO4. In Fig. 7, the stability diagram for the
AlOS system has been superimposed on Fig. 6. Inspection of Fig. 7 shows
that depending on the composition of the Na2SO4, aluminum may be present as
NaAlO2 (AlO2 ions), Al2O3, Al2(SO4)3 (Al3 ions), or Al2S3. It is important to
note that these are the phases of aluminum which exist at unit activity. Hence,
in the region of Fig. 7 where Al2O3 is stable, an important equilibrium is
Al2O3 2Al3 3O2
(18)
Fig. 7 Stability diagram showing the phases of aluminum that are stable in Na2SO4 at
1000C; the dashed lines indicate sulfur isobars and the arrows indicate compositional
changes of Na2SO4.
362
Pettit et al.
where K 18 a 2Al3a 3O2 and, therefore, as the activity of oxide ions decreases, the
activity of Al3 ions increases until a value of unity is reached at the Al2O3 Al3
boundary. Another reaction of importance is
Al2O3 O2 2AlO2
(19)
and the activity of AlO2 ions increases as the oxide ion activity increases until
a value of unity is reached at the AlO2 Al2O3 boundary.
Rapp (13) has experimentally determined the solubilities of a number of
oxides in Na2SO4, and some of the results from these studies are presented in
Fig. 8. The slopes of these solubility curves can be used to verify the reactions
by which the oxide dissolves. For example, in the case of Al2O3, basic reactions
can be written as
Al2O3 Na2SO4 2NaAlO2 SO3
(20)
Al2O3 O2 2AlO2
(21)
or
which both yield slopes of 1/2. The acidic dissolution reactions can be written
as
Al2O3 3Na2SO4 Al2(SO4)3 3Na2O
(22)
Oxide Ceramics
363
or
Al2O3 2Al3 3O2
(23)
(24)
where
K 24
(1 X Na2SO4)2PSO3
X Na2SO4a V2O3
(25)
and K 24 is the equilibrium constant, XNa2SO4 is the mole fraction of Na2SO4 in the
Na2SO4 NaVO3 solution, which is assumed to be ideal, PSO3 is the pressure of
SO3 and aV2O5 is the activity of V2O5 in equilibrium with the melt. In order to use
Eqs. (24) and (25), it is necessary to determine the major components in the melts
of interest. For example, the vanadium may exist as V2O5, NaVO3, Na4V2O7, or
Na3VO4. Equilibrium conditions for reactions such as
V2O5 4Na3VO4 3Na4V2O7
(26)
(27)
(28)
can be used to determine the proportions of the important species. Rapp (13)
has developed diagrams for solutions of Na2SO4 NaVO3 and typical results are
presented in Fig. 9 for a temperature of 900C. These results show that for SO3
pressures between 102 and 107 at 900C, Na2SO4 NaVO3 melts contain predominantly NaVO3. The amount of NaVO3 in the melt versus Na3VO4, Na3V2O7,
or V2O5 also depends on temperature. When Na2SO4 and NaVO3 are the major
species in Na2SO4 NaVO3 melts, Eq. (25) can be used to relate the activities of
the two acidic species, SO3 and V2O5.
The solubilities of oxides in Na2SO4 NaVO3 melts are influenced by both
the basic and acidic components. In Fig. 10, the solubilities of some oxides in
Na2SO4 and in a Na2SO4 30 mol% NaVO3 melt are compared using results obtained by Hwang and Rapp (15). Because the basic component is the same in
both melts, the solubility curves are identical for both melts in regions where
dissolution involves the basic component. The differences arise when the dissolution reactions involve acidic species with higher solubilities occurring for V2O5
than SO3.
364
Pettit et al.
Fig. 9 Mole fractions of V2O5, NaVO3, Na2V2O7, and NaVO3 in a Na2SO4 20 mol%
NaVO3 melt at 900C as a function of SO3 pressure.
Oxide Ceramics
365
Fig. 10 Solubilities of Al2O3 in pure Na2SO4 and in Na2SO4 30 mol% NaVO3 at 900C
as determined by Hwang and Rapp (15).
greatest at the oxidemelt interface. The oxide can, therefore, dissolve but precipitate out in the melt where the solubility is lower. Synergistic dissolution occurs
when two oxides react with melts in ways that the dissolution of one accelerates
the dissolution of the other. For example, if the solubility curves for Al2O3 and
Fe2O3 in Fig. 8 are considered at a value of log aNa2O 14, the dissolution of
Al2O3 occurs via
Al2O3 O2 2AlO2
(29)
As this reaction proceeds, oxide ions are consumed, and the melt becomes more
acidic. The Fe2O3 dissolution involves the reaction
/ Fe2O3 2/3Fe3 O2
13
(30)
where oxide ions are donated. The sum of these two reactions is
/3Fe2O3 Al2O3 2AlO2 2/3Fe3
(31)
The combined dissolution process of both oxides is such that the melt composition does not change with respect to oxide ion concentration and, hence, both
reactions proceed more rapidly when occurring concomitantly.
2. Alumina Corrosion
The corrosion of Al2O3 when exposed to deposits of molten salts is dependent
on purity (18,19). High-purity (99.9% pure) -Al2O3 single crystals lost weight
366
Pettit et al.
(32)
The alumina dissolves by basic dissolution. The SO3 pressure is greater at the
Al2O3 interface than the gas interface and SO3 is lost from the deposit. As dissolution of the alumina continues, the increased concentration of AlO2 ions causes
the SO3 at the Al2O3 interface to approach that of the gas interface, and, eventually, the dissolution process will stop unless fresh Na2SO4 is applied.
The weight-loss rates of Al2O3 increased as the activity of V2O5 was increased in Na2SO4 NaVO3 deposits. The dissolution reaction is proposed to be
Al2O3 3VO3 2Al3 3VO43
(33)
(34)
Oxide Ceramics
367
cate at triple points was evident (Fig. 13b). These results show that the Na2O
activity in these melts is high enough to promote reaction with silicates present
at grain boundaries of the Al2O3. As Na2O is removed from the deposit, the SO3
pressure is increased to levels at which Al2O3 can dissolve by acidic dissolution.
Synergistic dissolution has occurred.
3. Silica Corrosion
The solubility for vitreous silica in acidic melts is extremely low (20), but silica
does react with basic melts to form Na2SiO3 (21) and there are significant solubilities (22). Furthermore, the silica devitrifies under certain conditions and sodium
in melts accelerates this devitrification (23). In view of these conditions, silica
is very resistant to acidic melts but rather susceptible to hot corrosion by basic
melts. Typical reaction products developed on silica with a Na2SO4 deposit in
air for different times at 1000C are presented in Figs. 14 and 15. After 1 h, the
Na2SO4 wetted the silica and a layer of cristobalite spherulites formed under the
368
Pettit et al.
Fig. 12 Parabolic rate constants for growth of porous zones on polycrystalline alumina
as a function of activity of V2O5 in the deposit. Horizontal lines indicate the rate constants
for sulfate deposits.
(35)
and diffusion of sodium into vitreous silica. In some deposits, the SiO32 ions can
convert to silica out in the melt when the acidity is sufficiently high. Sodium
does diffuse into fused silica and may be incorporated in the glass as a network
modifier (24) via the reaction
Na2O SiEOESi 2SiEO 2Na
(36)
Oxide Ceramics
369
370
Pettit et al.
Fig. 14 Fused silica exposed to basic conditions at 1000C for 1 h developed globular
arrays of spherulites.
B. Molten Metals
The attack of oxide ceramics by molten metals has been discussed by McCauley
(1). This type of attack is encountered in the manufacture of steel and nonferrous
metals such as aluminum and copper. A primary cause of the corrosion process
is reaction of the molten metals with oxygen in the oxide ceramic. Consequently,
the thermodynamic stability of the oxide ceramic compared to that of the metal
oxide is an important factor. In the case of alumina exposed to some metal M,
the important reaction is
Al2O3 M (molten metal) 3MO 2Al (molten metal)
(37)
a 2Al
aM
eq
(38)
Oxide Ceramics
371
Fig. 15 Cross section of fused silica exposed to basic conditions at 1000C for 100 h
with thermal cycling; three layers cover the smooth glass: sodium silicates, tridymite, and
cristobalite, which is in contact with the glass.
(38) can be a large positive number and so conditions will be favorable for Eq.
(37) to proceed to the right. Even when MO is less stable than alumina, in principle some aluminum must be in the molten metal, and so some reaction can occur.
The extent of oxide ceramicmetal reactions is determined by the distribution
of the reaction product MO. When MO forms as a continuous layer over the
alumina, the reaction rate will be controlled by transport through MO and can
be expected to be small. However, if the MO is discontinuous or as particles in
the molten metal, then reaction rates can be rapid.
C. Corrosion in Thick Melts
Corrosion of oxide ceramics in deep melts involves many of the same reactions
that have been discussed previously, but the gas phase usually does not play a
significant role. In cases where a component in the gas phase is involved in the
corrosion process, this component can be substantially depleted from the liquid
phase, and, consequently, the corrosive characteristics of the liquid can change
substantially. Such conditions usually are established during the corrosion of metals and alloys in thick melts where, for example, oxygen is depleted from molten
salts. To illustrate the effects of corrosion of oxide ceramics, only examples where
the gas phase does not play a significant role will be considered. Many of the
372
Pettit et al.
Fig. 16 Schematic to show the products developed during exposure of sapphire in CaO
MgOAl2O3 SiO2 slags.
factors that affect the corrosion or dissolution of oxides in liquids have been
described by Cooper and Kingery (25) in studies concerned with the dissolution
of sapphire in CaOAl2O3 SiO2 slags, where it was shown that rates were controlled by mass transport in the liquid. The dissolution of sapphire in this slag
was controlled by mass transport in the liquid even when transport rates were
increased by rotating the immersed specimens. These investigators also analyzed
the effects of molecular diffusion, natural convection, and forced convection with
self-consistent results. Sandage and Yurek (26,27) studied the corrosion of sapphire and (Al, Cr)2O3 in CaOMgOAl2O3 SiO2 slags and showed that a solid
spinel reaction product (MgAl2O4) could be formed upon the surface of the sapphire depending on the experimental conditions. At times, the spinel did not form
a continuous layer, but it did affect the dissolution of the sapphire. It was shown
that in cases where the spinel formed as a continuous layer upon the sapphire,
a steady-state condition was achieved for which a constant spinel thickness was
established. The proposed mechanism for this process is shown schematically in
Fig. 16. The net effect of this process is to transfer Al2O3 from the sapphire
to the melt, but because the dissolution process involves transport through an
intermediate phase, it is called indirect dissolution.
Oxide Ceramics
373
REFERENCES
1. RA McCauly. Corrosion of Ceramics. New York: Marcel Dekker, 1995, pp. 194
197.
2. K Kitazawa, RL Coble. Electrical conduction in single-crystal and polycrystalline
Al2O3 at high temperatures. J Am Ceram Soc 57:245250, 1974.
3. EA Gulbransen, SA Jansen. Thermochemistry of gasmatal reactions. In: DL Douglass, ed. Oxidation of Metals. Materials Park, OH: American Society for Metals,
1971, pp. 6386.
4. C Wagner. Passivity during the oxidation of silicon at elevated temperatures. J Appl
Phys 29:1295, 1958.
5. HR Kim. Gaseous hot corrosion of oxide ceramics. MS dissertation, University of
Pittsburgh, Pittsburgh, PA, 1983.
6. SM Wiederhorn. Moisture assisted crack growth in ceramics. Fracture Mech 4:171
177, 1968.
7. SM Wiederhorn. Influence of water vapor on crack propagation in soda-lime glass.
J Am Ceram Soc 50:407444, 1967.
8. M Tomozawa. Fracture of glasses. Annu Rev Mater Sci 12:4374, 1996.
9. TA Michalski, SW Freiman. A molecular mechanism for stress corrosion in vitreous
silica. J Am Ceram Soc 66:284288, 1983.
10. RJ Janakiraman, GH Meier, FS Pettit. The effect of water vapor on the oxidation
of alloys that develop alumina scales for protection. Met and Mat Trans 30A:2905
2913, 1999.
11. KL Luthra. Low temperature hot corrosion of cobalt-base alloys: Part I. Morphology
of the reaction product. Part II. Reaction mechanism. Met Trans A 13A:18431864,
1982.
12. FS Pettit, CS Giggins. Hot corrosion. In: C Sims, N Stoloff, W Hagel, eds. Superalloys II. New York: John Wiley & Sons, 1987, pp. 327358.
13. RA Rapp. Hot corrosion of materials. In: O Johannesen, A Andersen, eds. Selected
Topics in High Temperature Chemistry. New York: Elsevier, 1989, pp. 291329.
14. FS Pettit, GH Meier, JR Blachere. Hot corrosion of oxide ceramics. In: KG Nickel,
ed. Corrosion of Advanced Ceramics. Dordrecht, The Netherlands: Kluwer, 1994,
pp. 235248.
15. YS Hwang, RA Rapp. Thermochemistry and solubilities of oxides in sodium sulfatevanadate solutions. Corrosion 45:933937, 1989.
16. RA Rapp, KS Goto. The hot corrosion of metals by molten salts. In: J Braunstein,
JR Selman, eds. Proceedings of the Electrochemical Society Symposium on Molten
Salts. The Electrochemical Society, Pennington, NJ, 1981, pp. 159177.
17. YS Hwang, RA Rapp. Synergistic dissolution of oxides in molten salts. In: T
Grobstein, J Doychak, eds. Oxidation of High Temperature Intermetallics. Warrendale, PA: TMS, pp. 257270.
18. MG Lawson, FS Pettit, JR Blachere. Hot corrosion of alumina. J Mater Res 8:1964
1971, 1993.
19. BM Warnes. The influences of vanadium on the sulfate induced hot corrosion of
thermal barrier coating materials. PhD dissertation, University of Pittsburgh, Pittsburgh, PA, 1990.
374
Pettit et al.
12
Metal Matrix Composites
Russell H. Jones
Pacific Northwest National Laboratory, Richland, Washington
I.
INTRODUCTION
375
376
Jones
377
potential of the core region was about 100 mV anodic as compared to the MA87 material. Paciej and Agarwala (6) found that AA 7091 reinforced with particulate SiC (20 vol%) did not exhibit the same behavior as the whisker-reinforced
material. Clusters of particles were observed to serve as regions for localized
corrosion, but there was no difference between the skin and the core. The pitting
potential in 3.5% NaCl ranged from 100 mV to no difference for the composite
material relative to the MA-87, depending on the heat treatment. The small 100mV shift is similar to that observed by Trzaskoma and co-workers (2,3) for SiCwhisker-reinforced AA 2024.
The effects of variable volume fraction (1540%) of SiC particles on the
corrosion behavior of AA 6061 was studied by Sun et al. (7). Studies were conducted in aerated and de-aerated solutions of NaCl. These authors found no shift
in the corrosion potential with respect to the volume fraction of SiC, but they
did see increased corrosion rates and less stable films with increasing volume
fraction of SiC. A significant density of small pits was observed for NaCl concentrations ranging from 0.035% to 3.5% for potentials beyond the pitting potential,
and these pits led to exfoliation-type cracking.
Nunes and Ramanathan (8) compared the corrosion behavior of SiC (5 and
10 vol%) and Al2O3 (5 and 20 vol%) particle-reinforced aluminum matrix material. The matrix alloys were Al7.5%, Si1% Mg, and AA 2014 and the composite was produced by a melt-stirring process rather than a powder process as that
reported by Trzaskoma and co-workers (2,3), Paciej and Agarwala (5,6), and Sun
et al. (7). The matrix composition of the Nunes and Ramanathan (8) material is
also different from the previous studies, with the exception of the AA 2024 reported by Trzaskoma and co-workers. Both types of reinforcement caused increased corrosion during immersion testing in 3.5% NaCl. The increased corrosion rate was associated with pits or microcrevices near the particlematrix
interface and from particle dropout. Pits in the SiC-reinforced material were
deeper than in the Al2O3-reinforced composite. The pitting potential of the composite materials was shifted to lower potentials compared to the alloy, with this
shift ranging from 95 mV in the aerated solution to 119 mV in the de-aerated
solution. These shifts were similar for both particles, although there is some uncertainty in the data, and were similar in magnitude to that observed by
Trzaskoma and co-workers (2,3) and Paciej and Agarwala (5,6).
Further information on the effect of SiC particles on the interfacial corrosion of an Al 2024 alloy in a NaCl solution was presented by Yao and Zhu (9).
They noted that the interfacial preferential dissolution (IPD) zone size was similar
to the plastic accommodation size around the particles. They concluded from this
that IPD was caused by the low integrity of the particle surface oxide film and
not by chemical, metallurgical, or galvanic coupling effects.
The corrosion characteristics of a cast AlSi alloy containing 3 wt% graphite was evaluated by Saxena et al. (10). They measured the weight change follow-
378
Jones
ing immersion in 3.5% NaCl and a salt spray and noted that the composite showed
a factor of 16 for the immersion test and 6 for the salt spray test at 25C. Potentiodynamic polarization tests in 3.5% NaCl showed considerably higher anodic and
cathodic current densities for the composite relative to the base material. The
authors concluded that the galvanic coupling of the graphite to the aluminum
could explain the increased corrosion rates of the composite relative to the matrix
material. They also observed localized corrosion at the interface between the
graphite particles and the aluminum matrix as observed for composites with SiC
and Al2O3 reinforcement.
In summary, the primary affect of ceramic-phase reinforcement of aluminum alloys on corrosion is that of providing sites for pitting and crevice corrosion.
This affect is manifested in decreased pitting potentials and increased corrosion
rates from localized corrosion at particlematrix interfaces and microcavities.
Graphite particles have a much larger affect on the corrosion rate of aluminum
composites than SiC or Al2O3 particle reinforcement because of the galvanic coupling and electrical conductivity of graphite.
379
passive film at cathodic potentials in both the reinforced and unreinforced condition and there is no evidence for galvanic corrosion between the matrix and SiC
particles. The electrochemical polarization response of the reinforced and unreinforced material were nearly identical.
Reinforcement of AZ91, the most corrosion-resistant magnesium alloy,
with Al2O3 (13) fibers produced a shift in the electrochemical polarization response for high-pH (10.5) solutions. There was a shift of about 150 mV of the
open-circuit potential in the anodic direction for the reinforced material relative
to the unreinforced material with and without NaCl addition (3.5%). The shift
in the open-circuit potential with Cl concentration was similar for both reinforced and unreinforced material, whereas the corrosion rates were similar up to
about 1% NaCl. The corrosion rate of the reinforced material increased much
more rapidly than the unreinforced material for Cl ion concentrations greater
than 1%. The corrosion rate for the composite was three times greater than the
matrix material at 3.5% NaCl.
B. Continuous Reinforcement
380
Jones
composites where this intermetallic formed at the fibermatrix interface. Pohlman (17) associated the localized corrosion observed at the fibermatrix interface
with the presence of the BAl intermetallic phase. Sedriks et al. (19) measured
the corrosion behavior of B-fiber-reinforced 2024 aluminum alloy. They reported
the loss in tensile strength of 2024 aluminum reinforced with 15% and 40% B
following heat treatment for various times at 190C and exposed with and without
stress to a solution of 53 g of NaCl per liter of water. The maximum strength
loss occurred for aging treatments of 12 h, with the matrix losing 45%, the
202415% B composite losing 50%, and the 202440% B composite losing 60%.
Specimens exposed to this same environment but with a stress of 90% of the
yield strength again showed an effect of the volume fraction of B reinforcement
where specimens with the matrix material did not fail in 4000 h, 202415% B
specimens failed in 1000 h, and 202440% B specimens failed in 40 h. Evans
and Braddick (20) noted a preferential attack at the fibermatrix interface for Bfiber-reinforced aluminum composites. They concluded that this localized corrosion was likely the result of an Al boride that formed at the interface during hot
pressing. Weight-loss results show a rapid increase with time over the first 80
h, reaching a limiting weight loss of about 5 103 g/cm2. Similar results were
obtained for a C-fiber-reinforced aluminum composite.
One of the earliest reports of the corrosion behavior of CAl composites
was that by Dull et al. (21) for 6061 aluminum reinforced with carbon fibers.
These materials were made by laying up alternating layers of fibers and foils and
hot pressing to achieve the desired density. Corrosion tests were conducted in
3.5% NaCl at temperatures between 25C and 75C. The corrosion rate of the
composite was 15 times greater than that of the alloy when tested at 25C and
increased substantially for temperatures exceeding 50C. A graphite fiber6061
Al composite was also evaluated for its corrosion performance by Aylor et al.
(22), although the material was produced by infiltrating graphite fiber tows with
6061 aluminum alloy and hot pressing these tows. The graphite6061 Al tows
were hot pressed between 0.3-mm-thick alloy foils so that the outer skin had no
exposed fibers, although the edges did have exposed graphite as in the Type I
configuration discussed by Trzaskoma (16). Aylor et al. (22) also found that the
composite exhibited considerably faster corrosion than the matrix material in
splash and spray tests, as did Dull et al. (21). Pitting of the 6061 Al outer foil,
which was the same on the composite and unreinforced 6061 Al, preceded the
rapid corrosion of the underlying composite. Galvanic effects of the graphite
Al couple were suggested as the primary cause of the accelerated corrosion.
Shimiziu et al. (4) noted that the pitting potential for graphite-fiber-reinforced
6061 Al was the same as the unreinforced alloy but that pit initiation was suggested as being easier because of the galvanic couple between graphite and the
6061 Al. They also noted a 100-mV shift of the open-circuit potential in the
381
anodic direct and a much higher cathodic current density than the unreinforced
material or material reinforced with SiC whiskers or Al2O3 fibers.
In studies to evaluate the corrosion mechanism in fiber-reinforced aluminum composites, Hihara and Latanision (23,24) found residual chloride in a
graphite6061 Al composite and measured galvanic corrosion for SiC, C, and
TiB2 fibers coupled to 6061 Al. The chloride was traced to the aluminum infiltration process and was suggested as a factor in accelerated corrosion of these composites. The galvanic current density for the graphite fiberaluminum couple was
30 times that of SiC or TiB2 fiberaluminum couples. The galvanic current density was controlled by the rate of O2 reduction on the graphite and, therefore, the
galvanic current density was greater in aerated solution than nonaerated solutions.
In an effort to develop a corrosion-resistant graphitealuminum composite,
Wendt et al. (25) evaluated the corrosion performance of graphite-reinforced Al
Mo alloys. Alloys with 1123% Mo were produced by sputter deposition and
the corrosion behavior measured in 0.1M NaCl at a pH of 8 and a temperature
of 25C. Galvanic current densities for the Al18% Mo and Al23% Mo alloys
coupled to graphite were three orders of magnitude less than for pure Al coupled
to graphite.
A directionally solidified eutectic of Al reinforced with lamina of Al4Ca
intermetallic (15) can be considered in the same category as continuous-fiberreinforced metal matrix composites because of the potential for continuous crevice corrosion in the direction of the Al4Ca intermetallic. Corrosion testing was
conducted in 3.5% NaCl at pH of 7 and a temperature of 25C. The electrochemical response of the eutectic was essentially equal to that of the extruded alloy
while being better than an AlCu alloy.
382
Jones
to the fibers. The open-circuit potential shifted with time in a very similar manner
for the reinforced and unreinforced AZ91C material. A galvanic corrosion rate
for the exposed edges of this composite was determined to be 1.8 mg/cm2 /day,
which is very fast for structural materials.
A potentially more stable Mg matrix composite reinforced with SiC fibers
was evaluated by Hihara and Kondepudi (27). The composite was a unidirectionally aligned SiC monofilaments in a matrix of ZE41 aged at 329C for 2 h to a
T5 temper. The volume fraction of SiC monofilaments was about 50%. Corrosion
studies were conducted in near-neutral 0.5M NaNO3 solutions at 30C, de-aerated
with high-purity N2. The magnesium alloy ZE41 had the following composition:
4.2% Zn, 0.7% Zr, 1.2% rare earth. Hihara and Kondepudi (27) found that the
presence of the SiC monofilament caused the MMC to corrode faster in aerated
than de-aerated solutions, unlike the matrix alloy in which the corrosion behavior
was insensitive to oxygen concentration. This difference in corrosion behavior
was evidenced by a 100-mV shift in the open-circuit potential of the composite
material in aerated solutions relative to de-aerated solutions. The effect of O2
was explained by the reduction of O2 on the SiC. The corrosion rate of both the
matrix material and the composite were very high at small voltages anodic to
the open-circuit potential. A corrosion current density of about 3.2 103 A/
cm2 was reported for the composite.
383
384
Jones
cycles to failure. Also, in contrast to the observations of Yau (29), the cycles to
failure of the reinforced material tested in moist salt air exceeded the cycles to
failure of the unreinforced material tested in laboratory air.
Corrosion fatigue results of 6061 Al reinforced with 10% Al2O3 and tested
in air and 3.5% NaCl have recently been reported by Bertolini et al. (31). They
observed that there was little difference in the da/dn versus K curve between air
and 3.5% NaCl for material extruded and forged 17%, but there was an increase in
da/dn for material extruded 8.3%. However, the increase in da/dn was only about
a factor of 5. Pitting at particlematrix interfaces was noted following potentiodynamic corrosion tests, as noted by others. The authors concluded that the pitting
corrosion contributed to the corrosion fatigue crack growth results; however, they
did not evaluate an unreinforced 6061 Al alloy to determine if the particleinduced localized pitting was a significant factor in the da/dn versus K behavior
of the composite.
B. Continuous Reinforcement
385
h for 15% B fibers and 40 h for 40% B fibers. The failures were intergranular
in the matrix and interfacial between the matrix and fibers. No failures were
observed for times up to 4000 h in the 2024 alloy tested in the longitudinal
direction. Stress-corrosion tests of samples loaded in the longitudinal direction
showed that the composite outperformed the 2024 Al in NaCl solution. No failures occurred in the composite for times up to 1000 h, whereas the matrix failed
in less than 10 h. The inferior stress-corrosion properties of the composite in the
transverse direction resulted from the exposed transverse cross sections.
A unique composite comprised of metal matrix and polymer composite
laminates has been evaluated by Wanhill (16). Corrosion fatigue tests were conducted on 2024 Al-T3 with carbonepoxy laminates and 7475 Al-T761 with
carbonepoxy laminates in air and air with a water spray. The specimens were
a symmetrical laminate of Al face sheets over a unidirectional carbon fiberepoxy
composite core. These materials were being evaluated for aerospace applications
so they were loaded with a flight simulation loading program with a mean load
of about 20% of the ultimate load capacity of the composite. The possibility of
a galvanic potential between the carbon fibers in the epoxy composite and the
aluminum alloys existed, but there was no evidence for accelerated corrosion
from galvanic corrosion. The results are reported in number of flights to produce
a crack extension of 40 mm. In the air plus water spray tests, the 2024-T3 carbon
epoxy composite was superior to the 2024-T3 material and the other composites.
In an environment of air plus water spray, the 2024-T3 carbonepoxy composite
required 8300 flights for 40 mm of crack extension, whereas the matrix required
only 4800 flights and the 7475-T761 carbonepoxy composite only 1800 flights.
386
Jones
the 0 composite. However, the K th for the 90 composite was only 6.5 MPam1/2
while it was 16.2 and 11.2 MPam1/2 for the 0 composite and the matrix, respectively.
387
(l/d) ratio and volume fraction of reinforcing phase and the matrix stress exponent. The model is based on the assumption that several stress-corrosion mechanisms are controlled by crack tip strain rate and that the presence of a reinforcing
phase reduces the crack tip strain rate. Processes such as the passive film rupture
rate, dislocationsurface interaction rate, surface reactivity rate, and crack opening effects on transport of species into and out of the crack and crack wall corrosion rates are all factors that depend on the crack tip strain rate.
The resulting predictions of this model are presented in Figs. 1 and 2, along
with data by Yau (29), Hasson et al. (30), and Jones (37). The model predictions
given in Fig. 1 show that the value of R, the ratio between the stress corrosion
velocity of the composite to that of the unreinforced matrix, is a strong function
of the l/d ratio and moderately dependent on n, the matrix creep stress exponent.
Increasing values of the matrix stress exponent causes a decrease in the value of
R. The predictions given in Fig. 2 show that R has only a small dependence on
the volume fraction of the reinforcing phase over the range 540%. This model
assumes that the reinforcement does not accelerate the corrosion rate and induce
other damage. As discussed earlier, discontinuous reinforcement with ceramic
phases often cause localized corrosion around clusters of particles or the particle
matrix interface. This localized corrosion very likely influences the quality of the
Fig. 1 Calculated and experimental values of R versus l/d and n for 7090 Al/SiC and
6061 Al/SiC.
388
Fig. 2
Jones
comparison between the model and experiment. The results of Jones (37) and
Yau (29) are consistent with the model and matrix stress exponent of 36. Hasson
et al. (30) results are plotted at an l/d of 100, which is based on the whisker
dimensions prior to processing the composite, but they reported that the l/d was
reduced to about 5 after processing. Therefore, their results should be shifted
substantially to the left as the arrow indicates.
389
ized layer that did not impact the corrosion resistance for times up to 180 days.
Anodization was judged to be a useful method for protecting metal matrix composites. Mansfeld et al. (39) concluded that chromate conversion coatings and
immersion in CeCl3 both produced stable, protective coatings on a SiCAl composite. Cathodic protection is another route for protecting these composites, but
Hihara and Latanision (40) have demonstrated that cathodic overprotection can
lead to accelerated corrosion for aluminum matrix composites. This occurs because of the production of OH and the instability of aluminum in high-pH environments. The effect of cathodic overprotection was the most severe in the 6061
Algraphite composite and less so in the 6061 Al/SiC composite and 6061 Al.
Modifying the matrix corrosion behavior through alloy compositional changes
is a route to improving the corrosion performance of Algraphite composites
considered by Wendt et al. (25). Aluminum and AlMg alloys with 11%, 18%,
and 23% molybdenum were prepared by sputter deposition and the galvanic corrosion response of these alloys with graphite determined from galvanic diagrams
constructed from the polarization data. The authors concluded that an Al18%
Mo/graphite composite would exhibit good galvanic corrosion behavior.
REFERENCES
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
390
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
Jones
SL Pohlman, Corrosion 34:156, 1978.
AJ Sedriks, JAS Green, DL Novak. Met Trans A 2:871, 1971.
JM Evans, DM Braddick. Corros Sci 11:611, 1971.
DL Dull, WC Harrigan, MF Amateau. In: FH Meyer, ed. Proceedings of the 1974
Triservice Corrosion of Military Equipment Conference, Vol. 1 (Dayton, OH, Air
Force Materials Laboratory Report, AFML-TR-75-42, 1975), p. 399.
DM Aylor, RJ Ferrara, RM Kain. Mater Perform 23:32, 1984.
LH Hihara, RM Latanision. Mater Sci Eng A126:231, 1990.
LH Hihara, RM Latanision. Corrosion 48:546, 1992.
RG Wendt, WC Moshier, B Shaw, P Miller, DL Olson. Corrosion 50:819, 1994.
PP Trzaskoma. Corrosion 42:609, 1986.
LH Hihara, PK Kondepudi. Corros Sci 34:1761, 1993.
LH Hihara, C Tamirisa. Mater Sci Eng A198:119, 1995.
S Yau. Crack propagation in corrosion-fatigue of metal matrix composites. PhD
thesis, North Carolina University, Raleigh, 1983.
DF Hasson, CR Crowe, JS Ahearn, DC Cooke. In: Proceedings of Failure Mechanisms in High Performance Materials, Gaithersburg, Maryland, May 13, 1984.
New York: Cambridge University Press, 1985, p. 147.
L Bertolini, MF Brunella, S Candiani. Effect of chloride environments on fatigue
behavior of AA6061Al2O3 particle composite. In: Corrosion/98. Houston, TX: National Association of Corrosion Engineers, 1998, Paper 738.
DW Berkeley, HEM Sallam, H Nayeb-Hashemi. Corros Sci 40:141, 1998.
AJ Sedriks, JAS Green, DL Novak. Metal Trans. 2:781, 1971.
JT Evans. Acta Metall 34:2075, 1986.
D Mahulikar, HL Marcus. Metal Trans A 15A:209, 1984.
S Floreen, HW Hayden, N Kenyon. Corrosion 27:519, 1971.
RH Jones. In: RH Jones, RE Ricker, eds. Proceedings of Environmental Effects on
Advanced Materials. Warrendale, PA: TMS, 1991, p. 283.
DM Aylor, RJ Ferrara, RM Kain. Mater Perform 32, July 1984.
F Mansfeld, S Lin, K Kim, H Shih. Corros Sci 27:997, 1987.
LH Hihara, RM Latanision. Scripta Metall 22:413, 1988.
13
Ceramic Matrix Composites
Russell H. Jones, C. H. Henager, Jr., Charles A. Lewinsohn, and
Charles F. Windisch, Jr.
Pacific Northwest National Laboratory, Richland, Washington
I.
INTRODUCTION
392
Jones et al.
(1a)
(1b)
32
(2)
(3)
393
Fig. 1 Transition pressures for SiC activepassive oxidation versus temperature, according to Gulbransen and Jansson (1).
(4)
(5)
where M is an alkali element such as Na and Li. A phase diagram for the Na2O
SiO2 system is shown in Fig. 2. Alkali elements such as Na and Li cause a breakdown of the passive SiO2 film by the formation of low-melting alkali silicates
such as those which occur at 800C in the Na2 SiO2 system and 1024C in the
Li2OSiO2 system. The eutectic temperature in the Li2OSiO2 system is about
250C higher than that in the Na2OSiO2 system and, therefore, Li is expected
to have less effect on the passive film on SiC at temperatures below 1000C than
Na.
A summary of the behavior of SiC in gasmolten-salt environments as
presented by McKee and Chatterji (2) is shown in Fig. 3. Passivation occurs at
high pO2, and active oxidation (formation of gaseous SiO) occurs at low pO2. A
basic salt or salt melt with a low pO2 at the saltSiC interface will cause active
corrosion, as depicted by reaction scheme (4) and (5). McKee and Chatterji suggest that SiC will not react with H2; however, more recent analysis by Herbell
et al. (3) indicates that the equilibrium partial pressure of CH4 [Eq. (2)] is 104
394
Fig. 2
Jones et al.
395
396
Jones et al.
Fig. 4 Weight change versus exposure time to Na2SO4 at 1000C for several ceramic
composites. (From Ref. 6.)
product was removed prior to measuring the weight change. An in situ measurement would probably exhibit a weight gain up to the time at which reaction
product spallation occurs. Corrosion data obtained on SiCSi3N4 and SiCSiC
composite materials by Henager and Jones (7) is also presented in Fig. 4. The
corrosion rate of the composite material was found to exceed that of the monolithic material for a NicalonSiC-reinforced, hot-pressed Si3N4 and a Nicalon
SiC reinforced CVI SiC. A SiC-whisker-reinforced, hot-pressed Si3N4 exhibited
a similar corrosion rate to the unreinforced hot-pressed Si3N4. The activation energy for the corrosion of the fiber-reinforced material was 880 kJ/mol, whereas
the whisker-reinforced matrix material had an activation energy of 1280 kJ/mol.
The difference in corrosion rate and activation energy for the fiber-reinforced
material was associated with preferential corrosion of the Nicalon fiber. The fiberreinforced material had a carbon interfacial layer between the fiber and matrix,
and the whisker-reinforced material had a thin amorphous glass layer. However,
the difference in corrosion behavior is not thought to be associated with the interfacial layer but with the structure of the Nicalon fiber following hot pressing.
As produced, Nicalon has an amorphous/microstryalline structure while the hotpressed structure was crystallized with evidence for a Mg silicate in the grain
boundaries. The source of the Mg was the MgO sintering aid used for the hotpressing process. Corrosion of the Nicalon fiber along the glass-enriched grain
397
boundaries probably accounted for the high corrosion rate of the Nicalon-fiberreinforced composite material.
The corrosion rate of a SiCSiC composite produced by the CVI process
was also greater than that of monolithic SiC (Fig. 4). Microstructures of CVI
SiCSiC composites made by the ORNL process and the Du Pont process both
exhibit porosity, which exists in materials made by this process. Composites containing fibers coated with C, BN, CB4CBN, and no coating were evaluated.
The corrosion rate was highest for the material made with fibers with no coating
and the BN coating, whereas composites made with C and CB4CBN coatings
exhibited smaller corrosion rates. However, the corrosion rate of these materials
at 900C was considerably greater than that of monolithic material at 1000C.
Matrix corrosion contributed the majority of the weight change, and while some
fiber corrosion was noted, although this was a relatively small factor compared
to the matrix corrosion. The open porosity and penetration of the molten salt into
the composite are considered the primary factors causing the high corrosion rates
for the SiCSiC composite.
Fox et al. found similar results in burner rig corrosion studies of CMCs to
those obtained in molten salt (8). The burner rig tests utilize a hot-combustion
gas and would be expected to differ from a molten Li or He gas environments;
however, molten deposits are one mechanism for corrosion in hot-combustion
gas studies. Fox et al. evaluated Si3N4 reinforced with 30 vol% SiC whiskers and
fibers. The method of procesing for the whisker-reinforced materials was not
reported; however, it was probably hot pressed, whereas the fiber-reinforced material was reaction bonded and had a residual porosity of 30%. Tests were conducted for 40 h at 1000C in an environment containing 2 ppm Na added as
NaCl. Monolithic Si3N4 exhibited a 10% decrease in the fracture stress and the
whisker-reinforced material a 35% decrease. The fiber-reinforced material exhibited an excessive corrosion rate, but no fracture studies were conducted. The high
residual porosity in the reaction-bonded material and the free C from the carboncoated fibers were postulated as the cause for the high corrosion rate in the fiberreinforced material. In the molten-salt corrosion studies (8), the more porous CVI
composite material also exhibited a higher corrosion rate than did the hot-pressed
Si3N4 /fiber- or whisker-reinforced-material.
Effects of corrosion on the mechanical integrity of SiCSiC structural components is an important factor in the durability of components constructed of
SiC/SiC. Decreases in the fracture strength would be possible by corrosion penetration along grain boundaries. Smialek and Jacobson (9) also found that moltensalt-induced pits were responsible for up to 50% of the observed strength reduction of SiC after 48 h at 1000C in Na2SO4 SO3. A relationship of fracture stress
versus (pit depth)2 illustrated the flaw-induced fracture relationship. Henshall et
al. (10) demonstrated that combustion gases containing alkali salts also contribute
to accelerated subcritical crack growth of monolithic SiC (Fig. 5). The crack
398
Jones et al.
Fig. 5 Crack velocity versus stress intensity for monolithic SiC exposed to air and hotgas environments. (From Ref. 10.)
399
kJ/mol found for Si. Silicon nitride has a much higher activation energy of 486
kJ/mol because of the formation of a silicon oxynitride phase between the Si3N4
and the SiO2 phases. A growth rate of 1.1 nm/min was determined for SiC at
1300C in 1 atm of O2. Oxygen diffusion was postulated as the rate limiting
process up to 1350C, above which ionic oxygen diffusion was rate controlling.
The protective properties of the passive SiO2 film are very sensitive to the
impurities of the SiC from which the film is formed. Comparison between the
oxidation behavior of several batches of SiC to that of very high-purity SiC produced by chemical vapor deposition was made by Fergus and Worrell (12). They
observed an activation energy of 142 kJ/mol for CVD SiC and 217 kJ/mol for
sintered -SiC. This compares with the value of 120 kJ/mol reported by Spear
et al. for single-crystal SiC platelets. Both the CVD and SiC platelets are of much
higher purity than the sintered -SiC to which sintering aids were added. Vaughn
and Maahs (13) in a review of the activepassive transition for SiC showed that
this transition was quite variable depending on the source of the SiC and other
experimental conditions such as the gas flow velocity. The Gulbransen data (Fig.
1) was very close to that determined theoretically, but most other materials exhibited much higher transition temperatures, presumably due in part to variable impurity concentrations in the materials.
Results of oxidation studies of ceramic composites have not been reported,
although Luthra (14) has presented some theoretical concepts on the oxidation of
ceramic composites. Luthra assessed the gas-phase diffusion of oxygen through
microcracks, solid-state diffusion of oxygen through protective oxide layers, formation of gaseous reaction products, and the combined reaction of the matrix
reinforcement and oxygen. In particular, the diffusion of oxygen through microcracks to react with the reinforcing fiber or the interface between the fiber and
matrix is a definite possibility for SiCSiC composites produced by the CVI
process because they have at least 10% residual porosity. Thermal cycling is also
likely to induce matrix microcracking because of the thermal expansion mismatch
between the reinforcing fiber and the matrix. For SiC fibers in a CVI SiC matrix,
this mismatch will be small, but it is not zero. Luthras analysis is primarily for
oxide matrix composites, where the oxygen will not react with the matrix,
whereas for SiC where the oxygen could react with the matrix, the transport
would be considerably slower.
The structural behavior of CMCs in an inert gas environment will be dependent on (a) whether the material is undergoing active or passive oxidation, (b)
the stability of the fibermatrix interfacial layer in O2, and (c) the stability of the
fiber in O2. Kim and Moorhead (15) have recently evaluated the flexural strength
of -SiC at room temperature following a 10-h exposure to ArO2 at 1400C.
Flexural strengths following exposure to ArO2 mixtures with O2 partial pressures of 7 107 to 2 104 MPa showed a strength reduction with increasing
pO2 up to about 2 105 MPa and a complete restoration of the as-polished
400
Jones et al.
strength at pO2 2 105 MPa (Fig. 6). The minimum strength (60% of unexposed material) at 2 105 MPa was the result of active corrosion creating
strength-reducing flaws along grain boundaries and at pits. Easler et al. (16) found
that the room-temperature flexure strength both increased and decreased upon
exposure to air at 1370C with and without an applied load. The strength increases
were thought to result from oxidative blunting, whereas the decreases which occurred under load were due to subcritical crack growth producing larger flaws.
Minford et al. (17) also found that stress enhanced the uptake of O2 into SiC.
Oxygen penetration occurred along cation-enriched grain boundaries. A stress
intensity threshold of about 1 MPa m1/2, below which crack blunting occurred
and above which crack extension occurred was found.
Oxidation can alter the dynamic crack growth behavior of SiC as well as
reduce the fracture strength due to corrosion-induced flaws. A reduction in the
K th of -SiC after being loaded to different stress intensities for 4 h at 1200C
and 1400C was reported by Minford et al. (18). They reported a reduction in
the K th from 2.25 in the unoxidized to 1.75 MPa m1/2 in the oxidized conditions
at 1200C and from 1.75 to 1.25 MPa m1/2 at 1400C for samples tested in
air or preoxidized. They concluded that the O2 caused a change from diffusioncontrolled crack growth to viscous cavity growth and linkage with a corresponding reduction in K th. The K th was defined as the transition between crack blunting
401
and growth as determined from the fracture strength following the 4-h static loading at various K values. McHenry and Tressler (19) found that the K th and subcritical crack growth rate was independent of the pO2 for pressures ranging from 108
to 104 atm and temperatures of 9001100C. The crack growth rate exhibited
an Arrhenius temperature dependence with an activation energy of 20 kcal/mol,
which is consistent with viscous flow of a grain-boundary glass phase. The lack
of a dependence on pO2 does not necessarily indicate that O2 did not induce crack
growth but that the effect of O2 was saturated at pressures above 108 atm.
C. Oxidizing Environments
A key feature of the oxidation behavior of SiCSiC in O2 at pressures of 2
104 Pa (atmospheric pressure of O2) observed by Windisch et al. (20) is that only
a weight loss was observed. No SiO2 formation occurred in any of the materials
with graphite interphases, although some boron-containing glass phase was observed for the material with a BN interphase. The kinetics of mass loss is shown
as a function of pressure and temperature in Figs. 7 and 8. Complete burnout of
the graphite interphase occurred in less than 104 s in the small test samples at a
Fig. 7 Mass loss versus exposure time for SiCSiC with a C interphase exposed to
various O2 partial pressures at 1100C.
402
Jones et al.
Fig. 8 Mass loss versus time for SiCSiC with a C interphase exposed at 2.5 103 Pa
O2 at various temperatures.
pressure of 2.4 104 Pa and a temperature of 1373 K. It is also clear that the
reaction rate increases with increasing temperature. An activation energy of about
50 kJ/mol was reported by Windisch et al. (20), which could be explained as
diffusion controlled through a boundary layer or as being reaction rate controlled.
An interphase recession rate was determined from the weight-loss measurements
and by direct physical measurement. Both methods gave very similar recession
rate equations with the physically measured equation as follows:
log(RR) 0.9 log(pO2) 9.9
(6)
It should also be noted that only weight loss was observed over the temperature
range 10731373 K, which borders on the temperature range (8731073 K) suggested by Evans et al. (21) for the pest phenomena.
Oxidation of SiCSiC composites with carbon interphases can also result
in the formation of SiO2 and a weight gain following an initial weight loss (22,23)
or a reduced weight loss with increasing temperature (24). Tortorelli et al. (23)
observed an initial weight loss followed by a weight gain in a Nicalon-fiberreinforced SiCSiC composite with a 0.3-m-thick graphite layer exposed to dry
air (pO2 of 2 104 Pa) at 1223 K. Following the initial weight loss from oxidation
403
of the carbon, they found that SiO2 formation occurred within the interfacial
region previously occupied by the carbon. For the Nicalon-reinforced composite
material, complete carbon depletion occurred within 15 min at 1223 K, followed
by weight gain from SiO2 formation. Unal et al. (24) observed their largest weight
loss (5%) at 1223 K for an exposure of 50 h in dry O2 and a decreasing weight loss
with increasing temperature up to 1673 K (2%). Their material was a Nicalonreinforced SiCSiC with a 0.5-m-thick fibermatrix carbon interphase. Kleykamp et al. (25) observed the following reactions of air with SiC-fiber-reinforced
SiC composites: (a) oxidation of free carbon at temperatures of 800965 K and
(b) a fast exothermic reaction and weight gain beginning at 1073 K and containing
up to 1773 K and up to times of 1 h followed by (c) the diffusion-controlled
oxidation of bulk SiC. Sebire-Lhermitte et al. (26) identified the presence and
location of SiO2 formation in SiCSiC composites using transmission electron
microscopy. They noted the presence of 15-nm-thick SiO2 layers at both the fibercarbon and matrixcarbon interfaces following an exposure of 1 h at 1123
K in air.
That Windisch et al. (20) only observed a weight loss while others (22
24) observed a weight gain following the weight loss could be the result of lower
O2 pressures, exposure time, and perhaps, carbon-layer thickness. The lower O2
pressure would reduce the SiO2 formation rate and therefore the chance for a
measurable weight gain during the 5-h exposures. Tortorelli et al. (22,23) used
exposures of up to 150 h, whereas Unal et al. (24) used 50-h exposures. Measurable SiO2 formation at pO2 2 103 Pa would need a much greater exposure
time than that used by Windisch et al. (20) and even greater than the time used
by Tortorelli and More (22). The existence of subcritical crack growth, as described in the next section, that coincides with only weight loss or interphase
removal without the embrittling effect of SiO2 or other solid-product formation
is the primary difference between an interphase removal mechanism (IRM) and
oxidation embrittlement mechanism (OEM) of crack growth. The oxidation results of Windisch et al. (20) demonstrate that the results of Henager et al. (27,28)
at temperatures ranging from 1073 to 1473 K and pO2 2 103 Pa occurred
by IRM only. The OEM is dependent on SiO2 formation, which depends on O2
pressure, temperature, and time. The interfacial layer thickness may also impact
this regime if a solid product can seal off the interface from further reaction. A
clear demonstration of this possibility has not been presented but remains as an
open issue needing further evaluation.
The effect of oxygen on the subcritical crack growth velocity of SiCSiC
is clearly demonstrated by the data given in Fig. 9. Oxygen has little effect on
the midpoint displacement (i.e., crack velocity) for about 2 104 s, but a marked
increase in the crack velocity is noted for longer times. These tests were performed in the O2 pressure, temperature, and time regime where only weight loss
was observed during oxidation studies (20). Therefore, the embrittling effect of
404
Jones et al.
a solid reaction product should not be a factor, but only the effect of fiber creep
and interfacial removal should have contributed to the crack growth rate. However, even if SiO2 or other glassy phases were present, they would have low
viscosity at very high temperatures and would not likely affect the crack growth
behavior or cause brittle fracture.
The dependence of the crack velocity on O2 partial pressure up to 10% that
of air is given in Fig. 10 for tests at 1373 K. There is a sharp increase in the
crack velocity at low pressure and a slower increase from pressures of about 0.25
102 Pa up to 2.5 103 Pa. Material with a BN interface exhibited about a
factor-of-10 slower crack velocity. Some glass-phase formation was noted in this
material as a result of these exposures, but there was no evidence that the crack
growth behavior was affected by the presence of this glass phase.
The IRM has been observed to cause crack growth in SiCSiC at temperatures of 10731473 K at pressures ranging from 2 102 to 2 103 Pa for stressed
samples (29,30). Weight-loss measurements suggest that the IRM operates over
these same temperatures and pressures and at 1373 K and a pressure of 2 104
Pa (20). An example of the crack velocity versus crack length for tests on specimens with Hi-Nicalon-reinforced material tested in gettered Ar and Ar 2
102 Pa of O2 is given in Fig. 11. The acceleration in the crack velocity induced
by the presence of O2 is clearly shown by these data, whereas the effect of temperature on the Ar O2 test is only apparent after a crack extension of 3.5 mm.
405
Fig. 10 Minimum crack velocity versus O2 partial pressure for SiCSiC with a C or
BN interphase exposed at 1100C.
Fig. 11 Crack velocity versus crack length for SiCSiC with a C interphase exposed
to 202 Pa of O2 Ar (dashed curves) or gettered Ar (solid curves) at 1175C (1448 K)
and 1200C (1473 K).
406
Jones et al.
Dynamic (sample stressed during exposure) OEM has been observed (31
33) to occur in SiCSiC at temperatures up to 1073 K, and 1223 K in air (O2
pressure of 2 104), whereas static (sample unstressed during exposure) OEM
was also observed in room-temperature tests following elevated-temperature exposures at 1223 K in air (23) and up to 1673 K in dry O2 at a pressure of 105
Pa (24). Heredia et al. (31) reported an upper pest temperature of 1073 K for
SiCSiC tested at elevated temperature in air, whereas Lin and Becher (32) and
Raghuraman et al. (33) observed OEM in air at 1223 K. Tortorelli et al. (23) and
Unal et al. (24) found the OEM to operate in room-temperature tests following
elevated-temperature exposure tests conducted in air without the application of
stress. In summary, the upper temperature limit for the dynamic operation of
OEM is between 1073 and 1223 K, and the formation of a glass phase at temperatures greater than 1223 K can cause OEM to occur when specimens are tested
at lower temperatures. Because the OEM results from the formation of a brittle
glass phase, this mechanism must depend on the growth rate and viscosity of the
glass phase. The growth rate will increase with increasing temperature and pO2,
but the viscosity decreases with increasing temperature. Therefore, there must
be a temperature at which the effectiveness of OEM is maximum.
The results of Heredia et al. (31), Lin and Becher (32), and Raghuraman
et al. (33) appear to define the upper temperature and O2 pressure limits for OEM
in SiCSiC at 10731223 K and O2 pressures of 2 104 Pa and above. However,
the IRM appears to operate over temperatures of at least 10731473 K at O2
pressures of 2 103 Pa and below. It may also operate at temperatures below
1073 K, within the OEM range, at low pressures, but this has not been observed
because the crack velocities are too low for experimental measurements. Bouchetou et al. (34) and Frety and Boussuge (35) observed a degradation of the mechanical properties of SiCSiC containing cracks produced by stress or thermal gradients and exposed to an oxidizing atmosphere at 773 K. The authors did not
provide sufficient details to identify the strength loss as OEM; however, OEM
is expected at temperatures below about 1223 K, although 773 K would be the
lowest reported occurrence to OEM. The low activation for carbon oxidation (50
kJ/mol), as reported by Windisch et al. (20), would result in a small decrease in
the oxidation rate of a carbon interphase with decreasing temperature.
The transition from OEM to IRM is not only a function of temperature and
oxygen pressure but also of the thickness of the carbon interphase layer between
the fibers and matrix. Filipuzzi et al. (36) and Cawley (37) noted that interphases
with a thin carbon layer (e.g., 0.1 m) were quickly sealed by SiO2, whereas the
carbon was totally oxidized before glass formation in material with a thicker
interphase region (e.g. 1 m). The interphase thickness would not alter the temperature dependence of the OEM to IRM transition but would lower the pO2 and
shorten the time for OEM to occur in favor of IRM. A schematic of the matrix,
interphase, and fiber oxidation process shown in Fig. 12 from Cawley (37) shows
407
Fig. 12 Schematic of O2 reaction with the C interphase and interphase recession and
with the SiC matrix and fiber to form SiO2.
the competition between interphase oxidation and SiO2 formation on the matrix
and fiber that leads to sealing off of the interphase region. Huger et al. (38)
measured the oxidation rates for Nicalon NLM202 fibers exposed to air at temperatures ranging from 700C to 1200C. After 100 h, the weight change was about
five times greater at 1200C relative to 700C with a 1-m-thick glass layer
forming after 100 h at 1000C. Clearly, the oxide layer can grow sufficiently
thick to seal the interphase region and to perhaps to act as a crack initiator which
could reduce the fiber strength.
The environmental stability of composites with silicon-based matrices,
such as blackglass, nitrides, and carbides have also been evaluated by several
authors. In studies of material with Nextel 312 fibers in a matrix of Allied Signal
Blackglass with a BN interphase, Campbell et al. (39) measured the weight
change and bend strength of material exposed to dry air, air 3% H2O, dry air
80 ppm KCl vapor, and air containing 3% H2O 80 ppm KCl vapor. The
exposures were conducted for variable lengths of time at 700C and 5 h at 900C.
The composites show only continuous weight loss with time for exposures to
dry air at 700C, but a weight gain following a small initial weight loss in KCl
containing environments. The weight gain was identified as resulting from the
formation of alkali silicates. Vaidyanathan et al. (40) also studied the effects of
408
Jones et al.
oxidation on the mechanical properties of the Nextel 312/BN/blackglass composite but for times of 201000 h at 600C. They observed a 50% reduction in
strength after 500 h at 600C and concluded that times greater than 200 h under
these conditions were a concern for the durability of the composite. Oxidation
resulted in increased fiber pullout, consistent with IRM, but at a much lower
temperature than IRM in a SiCCSiC composite. The authors concluded that the
degradation may have resulted from the weakening of the fibermatrix interface.
The instability of the fibermatrix interphase in SiCSiC composites has
led to the evaluation of coatings for these materials. Fox (41) evaluated the oxidation resistance of three coatings on SiCSiC composites: (a) CVD SiC, (b) a
particulate-based sealant with a CVD SiC outer layer, and (c) a boron-rich inner
layer and CVD SiC outer layer. Oxidation studies were conducted in dry oxygen
at 9811316C. All three provided protection for up to 100 h, and the CVD SiC
was the most protective. Oxygen diffusion through the SiO2 that formed on CVD
SiC was considerably slower than through the glass layer that formed on the
other two coatings which contained B. Lee and Miller (42) evaluated the stability
of mullite coatings with a refractory oxide barrier coating on SiCSiC exposed
to air at 12001400C. The sample temperatures were cycled every 1, 2, or 20
h from 1200C and 1300C to room temperature and every 1 h from 1400C
to room temperature. The mulliterefractory oxide composite coating exhibited
improved adherence and oxidation resistance relative to a straight mullite coating.
D. Hydrogen-Containing Environments
Herbell et al. (43) have evaluated the thermodynamic stability of SiC in pure H2
at 1 atm, as shown in Fig. 13. The primary gaseous reaction produce is CH4 as
described by Eq. (2), whereas other reactions which produce SiH4 and SiH are
also possible at temperatures as low as 900C.
A small amount of H2O can alter the phase stability such that at 1400C
and about 1000 ppm of H2O, the dominant gaseous reaction products become
SiO and CO (Fig. 14). Results for lower temperatures were not reported, but the
reaction of H2O with SiC occurs at much lower temperatures, so similar reaction
products would be expected at lower temperatures. No loss in the room-temperature flexural strength of sintered SiC was noted by Herbell et al. (43) for samples
exposed to H2 saturated with H2O for 100 h at temperatures from 800C to
1400C. In dry H2 (25 ppm H2O), Hallum and Herbell (44) noted a 33% decrease
in the fracture strength of sintered SiC after exposures of 500 h at 1100C and
1300C. A statistically significant decrease in flexural strength was also observed
after 50 h at 1000C. The stability of SiC in an ArH2O5% H2 mixture was
calculated by Jacobson et al. (45) in the same manner as the H2 H2O mixtures,
with the result shown in Fig. 15. Except for the lower gas pressures and the shift
in the relative activities of SiO and CH4 in region III, the results are essentially
409
Fig. 13 Gases for equilibrium partial pressures of reaction products for reaction of SiC
with pure H2 at 1 atm.
410
Jones et al.
411
factors that encourage the development of oxide matrix composites. Much of the
high-temperature corrosion data for oxide matrix composites exists for particulate-, whisker-, or platelet-reinforced material, which have not been optimized
for strength and toughness. Oxide fiber development has progressed to the state
where continuous-fiber composites are being produced; however, there is no
high-temperature oxidation data for these materials. Examples are alumina and
aluminaYAG matrix composites reinforced with Nextel 610 and 720, as reported by Goettler (48). Interphase layers of ErTaO4 or CaWO4 are being evaluated for producing fiber pullout and fracture resistance. However, the stability
of these interphase materials in oxidizing, reducing, or salt environments has
not been evaluated. Even though the matrix and fiber may exhibit excellent behavior in oxidizing environments, uncertainty about the composite chemical stability remains. Also, the high-temperature strength of oxide fibers is less than
SiC fibers, such that further improvements in strength must occur before continuous-fiber oxideoxide composites are attractive for high-temperature applications.
Borom et al. (49) have examined the oxidation behavior of Al2O3 reinforced
with SiC and MoSi2 particles and SiC whiskers. The particulate volume fractions
ranged from 10% to 30% and tests were conducted at 12001500C in air; the
oxidation rate was determined by weight change and reaction layer thickness.
Both SiC and MoSi2 form protective SiO2 layers when oxidized as bulk materials.
Borom et al. (49) reported a 15-fold increase in the oxidation rate of these phases
when incorporated into an Al2O3 matrix. This increase was postulated as resulting
from the volume change of the reaction product that forms on the composite and
the thermal expansion mismatch of the reaction product with the composite. Both
of these factors were less favorable for the composite as compared to bulk SiC
and MoSi2. Larger volume fractions of these phases produced a large volume
fraction of mullite in the reaction scale and this was favorable because the silica
in the mullite will produce a more viscous scale that will allow more stress relaxation and accommodation for mismatch stresses. These authors suggested that a
mullite matrix is preferred because the reaction product will contain aluminosilicate plus mullite, which will flow and relax thermal mismatch stresses.
The bend strength of SiC-whisker-reinforced (28 vol%) Al2O3 was found
by Leaskey et al. (50) to increase by 33% when oxidized in air at 1600C for
15 min. Composites with SiC particle reinforcement showed a 66% improvement
in the bend strength following an oxidizing treatment of 2 h at 1600C. The
authors suggested that the improved properties are the result of the oxidation of
the SiC reinforcement to produce a compressive surface layer. The following
conditions were necessary for this improvement: (a) a sufficiently large SiC content to produce a continuous oxide surface layer, (b) oxidation conditions that
produce a low porosity layer with a critical thickness, and (c) elimination of large
flaws in the bulk of the material.
412
Jones et al.
The tensile strength of Nicalon-fiber-reinforced Al2O3 following heat treatment in air at 750C has been reported by Heredia et al. (31). This material
contained 0/90 fiber orientation with a BNSiC interphase. The room-temperature tensile strength was found to decrease from about 250 MPa to about 120
MPa following exposure to air at 750C for 24 h. The formation of a glass phase
on the Nicalon fiber was suggested as the cause of the observed oxidation embrittlement.
Corrosion studies of SiC-reinforced Al2O3 have been conducted in coal
combustion environments by Watne et al. (51) and Breder et al. (52). In the
study reported by Watne et al. (51), the material produced by the Lanxide Corp.
contained 50% SiC with 10% residual Si. Following a 100-h pilot-scale combustion test at about 1350C in a radiant zone of the furnace, the composite, in the
form of a tube, was intact but had a 0.85-mm reduction in the wall thickness.
This loss was suggested as resulting from erosion from slag flow on the tube.
The original wall thickness was 5.25 mm. A smaller amount of loss was found
for a tube placed in the convective pass region of the combustor where the temperature was about 1200C. Breder et al. (52) exposed a similar tube made by Lanxide Corp. to coal slag obtained from two coal combustion plants. Exposures
were conducted in a box furnace with the tube and coal slag at temperatures of
1090C, 1260C, and 1430C. Fracture tests were conducted on samples removed
from the tubes following a 500-h exposure. The tube strengths were reduced by
2045% at 1260C depending on the type of slag.
Although Al2O3 is the most commonly used matrix for oxide matrix composites, composites with other oxides such as MgO, ZrO2, and mullite have also
been evaluated. The oxidation kinetics of SiC particulate-reinforced MgO has
been examined by Hallum (53) and Camey and Readey (54). Hallum studied
MgO reinforced with 5, 10, and 15 vol% SiC particles or whiskers over the
temperature range of 11001500C. The reaction-product thickness increased
with the square root of time and was a function of the volume fraction of SiC
in the composite. Mg cation diffusion was proposed as controlling the growth
rate with a reaction layer formed by Mg cation diffusion through the reaction
layer to the atmosphere where oxidation produced a columnar growth region.
Camey and Readey (53) identified three oxidation-product layers unlike the single
layer observed by Hallum; however, they agreed with Hallum regarding the
growth rate being controlled by Mg cation diffusion through the product layer.
Luthra and Park (55) evaluated the oxidation of SiC in mullite and alumina
matrices and found parabolic rate constants that were three orders of magnitude
larger than SiC. Xu et al. (56) measured the effects of adding ZrO2 to mullite on
the oxidation of mullitezironiaSiC composites. They found that the addition
of ZrO2 to mulliteSiC composites increased the reaction rate with oxygen. They
rationalized this as being due to the increased diffusion rate of oxygen in the
zirconia phase. A rapid mode II type of oxidation, where oxygen can penetrate
413
deep into the sample before the outer region is completely oxidized, occurred at
12001400C and with the volume percent of ZrO2 greater than 20%.
414
Jones et al.
tests where the stress was below the proportional limit. However, dynamically
loaded samples loaded above the proportional limit, or matrix cracking stress,
exhibited limited fibermatrix interface oxidation. Oxygen diffusion along matrix
microcracks created by stresses exceeding the proportional limit was thought to
be the primary cause for the fibermatrix interfacial oxidation. This effect was
most pronounced under cyclic loading compared to static or quasistatic loading.
Embrittlement of a MASNicalon composite during fatigue loading at 500C in
air was also reported by Heredia et al. (31). A reduction in fatigue life was noted
after only 1000 cycles at 500C relative to room-temperature tests. Heredia et
al. (31) related this loss to a pest process where the Nicalon is embrittled
and they suggest that the compressive matrix stress in the glass-ceramic matrix
Nicalon composites requires a cyclic stress to reveal this process. Sorensen et al.
(60) studied the effect of environment and frequency on the fatigue properties
of a CASNicalon composite. They concluded that fiberinterfacial wear processes play a significant role in the loss of fatigue life of these composites and
that the environment enhances this wear-induced loss of strength.
B. Hot Corrosion Environments
High-temperature salt environments will occur in engine components on a Navy
gas turbine engine and heat exchangers in coal-fired power plants. There is a
strong emphasis on increasing the trust-to-weight ratio of Navy planes, and the
low-density and high-temperature performance of CMCs are needed to achieve
these goals. CASNicalon and LASNicalon composites have been evaluated
for this application by Wang et al. (61,62). They examined the reaction of sodium
sulfate with these composites by coating specimens and heating them to 900C
in either air or argon atmospheres for up to 100 h. The CASNicalon composites
exposed in air showed surface cracking and extensive reaction between the salt
and the Nicalon fibers. The surface fibers were completely attacked and were
totally removed. X-ray diffraction was used to identify the presence of CaSiO3
and NaAlSiO4. The unreinforced CAS glass exposed to the same conditions reacted to form NaAlSiO4 but not CaSiO3. Therefore, the SiC fibers contributed to
the reaction products and altered the corrosion reaction. The tensile strength and
strain to failure of the CASNicalon composite exposed to sodium sulfate in air
was reduced relative to the as-received properties and those for material annealed
at 900C for 100 h but without the presence of the salt. However, the properties
of material exposed to salt in an argon atmosphere showed no degradation in
properties. The authors concluded that oxidation is the primary reaction responsible for the strength degradation of the composite. In contrast, the LASNicalon
composites did not form additional phases, although there were surface cracks
and interdiffusion of Na into the composite and Mg outward diffusion. A 30%
415
strength reduction was noted for the LASNicalon composite exposed to the salt,
presumably a result of the surface cracking and Na and Mg interdiffusion.
A thermodynamic evaluation was conducted by Kowalik et al. (62) for the
CASNicalon composite exposed to sodium sulfate at 900C, as reported by
Wang et al. (61). This study suggested the following reaction path: (1) SiC oxidizes to form SiO2, (2) the silica reacts with the Na2O in Na2SO4(Na2OSO3), (3)
the result of reaction 2 may lead to a liquid oxide (or soda slag) phase which
may attack the CAS matrix, (4) the SO3 from reaction 2 may combine with the
CaO in the matrix to form CaSO4, and (5) this lost CaO from the matrix is replaced by Na2O to yield NaAlSi3O8. These thermodynamic predictions closely
match the experimental results reported by Wang et al. (61). Step 1 shows the
significance of oxidation in the high-temperature corrosion of these materials.
High-temperature corrosion studies of CASNicalon and BMASTyranno
exposed to sodium and magnesium salts have also been conducted by Scott et
al. (63). The environments were 3.5% NaCl, 3.5% magnesium salts, and a mix
of 3.5% of both sodium and magnesium salts. The samples were coated with
these solutions by immersion and then heated to 600C, 800C, or 1100C for
up to 60 h. Reaction occurred primarily between the Ca and Mg ions, and the
Nicalon in the CASNicalon composite, but the Na ions penetrated the glassy
phase and lowered its viscosity in the BMASTyranno composites. The authors
concluded that both reactions were of concern for the stability of these composites
in high-temperature salt environments.
V.
SUMMARY
Ceramic composites are being considered for a variety of high-temperature applications in which their corrosion properties will be important for their performance. Examples include combustor liners and blade shrouds for gas turbines,
heat exchangers in a coal-fired power plants, burner nozzles, gas injection lances,
sensor shields, tundish nozzles for molten Al and steel plants, and furnace/reformer tubes. Each of these applications involves some form of corrosion.
The corrosion of ceramic composites is more complicated than that of a
monolithic ceramic because composites are chemically and microstructurally heterogeneous. The high-temperature corrosion of CMCs are often affected by the
fiber, fibermatrix interphase, or the method used to produce the matrix of the
composite. High-temperature oxidation of the C or BN interphase in SiCSiC
composites is a clear example where the interphase causes the corrosion performance of the composite to be less than that of monolithic SiC. The presence of
the SiC in mullite or alumina matrix composites were also found to increase the
parabolic rate constants for oxidation by several orders of magnitude, whereas
416
Jones et al.
the presence of the SiC fiber resulted in a different reaction product in a CAS
Nicalon composite than in the unreinforced matrix when reacted with a hightemperature salt environment. Therefore, the corrosion performance of CMCs
differs from the unreinforced monolithic ceramic and must, therefore, be carefully
evaluated for each application.
REFERENCES
1. EA Gulbransen, SA Jansson. Oxid Met 4:181, 1972.
2. DW McKee, D Chatterji. J Am Ceram Soc 59:441, 1976.
3. TP Herbell, AJ Eckel, DR Hull. Effect of hydrogen on the strength and microstructure of selected ceramics. In: RH Jones, RE Ricker, eds. Proceedings of the Symposium on Environmental Effects on Advanced Materials. Warrendale, PA: TMS,
1990.
4. NS Jacobson. J Am Ceram Soc 69:74, 1986.
5. NS Jacobson, JL Smialek. J Electrochem Soc 133:2615, 1986.
6. RE Tressler, MD Meiser, T Yonushonis. J Am Ceram Soc 59:278, 1976.
7. CH Henager Jr and RH Jones. A molten salt corrosion of hot-pressed Si3N4 /SiCreinforced composites and effects of molten salt exposure on slow crack growth of
hot-pressed Si3N4. Proceedings of the International Symposium on Corrosion and
Corrosive Degradation of Ceramics (1989). First International Ceramic Science and
Technology Congress, October 1989, Anaheim, California, and J. Am. Ceramics
Society, Westerville, OH, 1990, p. 197.
8. DS Fox, NS Jacobson, JL Smialek. In: RE Tressler, MJ McNallen, eds. Ceramic
Transactions, Corrosion and Corrosive Degradation of Ceramics, Westerville, OH:
J. American Ceramics Society, 1990, p. 227.
9. JL Smialek, NS Jacobson. J Am Ceram Soc 69:741, 1986.
10. JL Henshall, DJ Rowcliffe, JW Edington. J Am Ceram Soc 62:36, 1979.
11. KE Spear, RE Tressler, Z Zheng, H Du. ibid. Ref. 12, p. 1.
12. JW Fergus, WL Worrell. ibid. Ref. 12, p. 43.
13. WL Vaughn, HG Maahs. J Am Ceram Soc 73:1540, 1990.
14. KL Luthra. ibid. Ref. 12, p. 81.
15. HE Kim, AJ Moorhead. ibid. Ref. 12, p. 81.
16. TE Easler, RC Bradt, RE Tressler. I Am Ceram Soc 64:731, 1981.
17. EJ Minford, JA Costello, IST Tsong, RE Tressler. In: RC Bradt, AG Evans, DPH
Hasselmann, FF Lange, eds. Fracture Mechanics of Ceramics. New York: Plenum,
Vol. 6, 1983.
18. EJ Minford, DM Kupp, RE Tressler. I Am Ceram Soc 66:769, 1983.
19. KD McHenry, RE Tressler. I Am Ceram Soc 63:152, 1980.
20. CF Windisch Jr, CH Henager Jr, GD Springer, RH Jones. Oxidation of the carbon
interface in nicalonfiber-reinforced silicon carbide composite. J Am Ceram Soc
80(3):569574, 1997.
417
21. AG Evans, FW Zok, RM McMeeking, ZZ Du. Models of high temperature, environmentally assisted embrittlement in ceramic-matrix composites. J Am Ceram Soc
79(9):23452352, 1996.
22. PF Tortorelli, KL More. Time dependence of oxidation-induced microstructural
changes in Nicalon- and Nextel-reinforced SiC. In: V Greenhut, ed. Proceedings of
20th Annual Conference on Composites, Advanced Ceramics, Materials and Structures. Westerville, OH: American Ceramic Society, 1996.
23. PF Tortorelli, S Nijhawan, L Riester, RA Lowden. Influence of fiber coatings on
the oxidation of fiber-reinforced SiC composites. In: D Cranmer, ed. Proceedings
of the 17th Annual Conference on Composites and Advanced Ceramic Materials.
Westerville, OH: American Ceramic Society, 1993.
24. O Unal, AJ Eckel, FC Laabs. Mechanical properties and microstructure of oxidized
SiC/SiC composites. In: V Greenhut Proceedings of the 20th Annual Conference
on Composites, Advanced Ceramics, Materials and Structures. Westerville, OH: The
American Ceramic Society, 1996.
25. H Kleykamp, V Schauer, A Skokan. Oxidation behavior of SiC fiber reinforced SiC.
J Nucl Mater 227:130137, 1995.
26. I Sebier-Lhermitte, M Gomina, J Vicens. TEM observations of SiCSiC composites
with a carbon interphase layer annealed in air at high temperatures. J Micros 169:
197205, 1993.
27. CH Henager Jr, RH Jones. Subcritical crack growth in CVI silicon carbide reinforced
with Nicalon fibers: Experiment and model. J Am Ceram Soc 77(9):23812394,
1994.
28. RH Jones, CH Henager Jr, CF Windisch Jr. High temperature corrosion and crack
growth of SiC/SiC at variable oxygen partial pressures. Mater. Sci Eng A198:103
112, 1995.
29. CH Henager Jr, RH Jones, CF Windisch Jr, MM Stackpoole, R Bordia. Time-dependent, environmentally assisted crack growth in Nicalonfiber-reinforced SiC composites at elevated temperatures. Metals Mater Trans A 27A:839949, 1996.
30. CA Lewinsohn, CH Henager Jr. Microstructural and environmental parameters influencing subcritical crack growth in CVI SiC/SiC composites. J Am Ceram Soc in
press.
31. FE Heredia, JC McNulty, FW Zok, AG Evans. Oxidation embrittlement probe for
ceramic-matrix composites. J Am Ceram Soc 78(8):20972100, 1995.
32. H-T Lin, PF Becher. Effect of fiber coating on lifetime of Nicalon fibersilicon
carbide composites in air. Mater Sci Eng A, in press.
33. S Raghuraman, MK Ferber, JF Stubbins, AA Wereszcak. Stress-oxidation tests in
SiC/SiC Composites. Ceram Trans 46:10151026, 1999.
34. MF Bouchetou, T Cutard, M Huger, D Fargeot, C Gault. In: R. Naslain, J Lamon,
D Doumeingts, eds. High-temperature ceramic-matrix composites-I. Abington,
Cambridge: Woodhead Publishing, Ltd. 1993, p. 81.
35. N Frety, M Boussuge. Comp Sci Technol 37:177189, 1990.
36. L Filipuzzi, G Camus, R Naslain. J Am Ceram Soc 77:459466, 1994.
37. JD Cawley. In: AG Evans, R Naslain, eds. High-Temperature Ceramic-Matrix Composites I: Design, Durability and Performance. Westerville, OH: American Ceramic
Society 1995, p. 377.
418
Jones et al.
14
Issues in Predicting Long-Term
Environmental Degradation of
Fiber-Reinforced Plastics
Aaron Barkatt
The Catholic University of America, Washington, D.C.
I.
INTRODUCTION
Several major unresolved issues are involved in predicting the effects of environmental degradation on the long-term behavior of fiber-reinforced plastics in construction applications. Many types of change in mechanism in the course of exposure of fiber-reinforced plastics (FRPs) to the surrounding environment are
possible. Such changes limit the applicability of extrapolation from short-term
test data and, in certain cases, cause the degradation rate to rise in the course of
the exposure. Such increases in degradation rate may be gradual or abrupt, and
their effects may rapidly disappear or persist for long periods of time. Changes
in the degradation mechanism may involve the degradation of the fibers, the
matrix, or the interphase. Changes in mechanism also affect the dependence of
the degradation rate on temperature and moisture content and, thus, limit the
range of conditions over which temperature may be used as an accelerating factor
in predictive tests. In addition to these scientific issues, uncertainties concerning
the effects of scale, limitations of existing test procedures, and, in particular, the
variability of FRPs produced on a large scale and the lack of information regarding the effects of such variability on the chemical properties of the materials
constitute problems that require solution as a prerequisite for extensive use of
FRPs in large-scale structures, such as the bridges, highways, and buildings.
This chapter does not attempt to present an updated, comprehensive overview of the area of environmental degradation of fiber-reinforced plastics. In419
420
Barkatt
stead, it attempts to use published data to focus on some of the serious issues
that complicate the development of long-term predictions of the effects of the
environment on the behavior of FRPs in construction applications. These issues
include scientific complications involving changes in degradation mechanism, in
particular changes that make it possible for supralinear behavior to take place,
as well as engineering problems such as scaling, test design, and, in particular,
the variability of existing materials and the lack of information about the effects
of such variability on the reliability of test-derived predictions of material performance.
421
FRPs in such structures requires addressing issues that include (a) cost as a major
factor, (b) a very broad range of environments, often subject to numerous and
frequent changes, and often involving the presence of highly corrosive media
such as water previously in contact with cement, and (c) the need to maintain
the performance of the FRP-supported structures not far below initial levels over
long periods of time, covering several decades. For instance, service life of 75
years is required for highway structures (7). The combination of these factors
requires extensive work in the areas of testing and predictive modeling to guide
efforts in FRP fabrication, quality control, and structural design in order to ensure
adequate safeguards against premature degradation of performance under service
conditions.
A. Structure of FRPs
Fiber-reinforced plastics consist of a combination of fibers, which impart strength
and stiffness to the material, with a polymeric matrix, that serves to hold the
fibers in the desired configuration (random or aligned), to transfer applied stresses
to the fibers, and to protect the fibers against abrasion and corrosion. The boundary region between the fibers and the matrix is the interphase, consisting of the
bonding layer interfacing with both the fiber and the matrix (8). This region is
of critical importance in controlling the adhesion of the fibers to the matrix and
the ability of the composite as a whole to maintain its strength in moisture-containing environments. The adhesion of the fibers to the matrix results from a
combination of mechanical fitting and chemical adhesion. The relative importance of the physical and chemical factors, respectively, to fibermatrix adhesion,
as measured, for instance, in fiber pullout tests, is variable and is not yet fully
understood in many cases. Empirically, however, it is well established that the
use of silane coupling agents to treat the fibers, known as sizing, improves the
properties of the interface. Sizing just prior to impregnation of the fibers with
resin improves the adhesion between the inorganic fiber and the organic resin
and enhances the resistance to moisture. [Such permanent sizing should be distinguished from pretreatment of the fibers with organic lubricants to minimize
fiberfiber abrasion and to facilitate maintaining the fibers in their desired configuration. These organic lubricants are usually removed by heating before the
treatment with the permanent sizing agent (i.e., the silane coupling agent)]. In
addition to the treated fibers and the matrix, FRPs often contain additional components such as fillers (typically clay or hydrated alumina) and mold-release agents
(lubricants).
Polymeric matrices can be divided into two broad classes consisting of
thermosetting polymers and thermoplastic resins, respectively. Thermoplastic
resins such as polyetherimide (PEI), polyethersulfone (PES), polysulfone (PSU),
and polyetheretherketone (PEEK) generally possess higher tensile strength, better
422
Barkatt
impact resistance, and much higher heat-distortion temperatures than thermoplastic polymers. However, the cost of thermoplastic matrices is much higher than
that of thermosetting matrices (9). Accordingly, thermosetting matrices have a
greater potential at the present time for use in civil construction applications,
whereas thermoplastic matrices are more suitable for high-performance engineering applications such as aerospace components (8), especially in combination
with high-performance carbon fibers. Major thermosetting matrices include vinylester, epoxy, polyester, and phenolic resins. Vinylester and epoxy matrices
are somewhat more expensive than the other two types of polymers, but they
possess higher tensile strength as well as better resistance to attack by water
and by various chemical solutions (9). Of the four types of thermosetting resins
mentioned, phenolic resins are the least expensive, but their relatively low tensile
strength and their susceptibility to void formation during curing, which results
in large capacity for absorption of moisture (10) render them less promising for
civil structural applications than the other three types of polymeric matrix material. Nevertheless, phenolic composites are suitable for a variety of applications
in consumer products, electrical equipment, and automotive components. Other,
more advanced types of thermosetting resins, developed for high-temperature
applications, consist of polyimides and polybismaleimides (BMIs). Because of
their brittleness, high cost, and the requirement for sophisticated techniques during fiber incorporation (10), the use of polyimides and BMIs is largely confined
to the aerospace industry. Accordingly, the most promising candidates for use in
large-scale civil applications are polyesters, epoxies, and vinylesters. Vinylesters
combine the desirable thermal, mechanical, and chemical characteristics of epoxies with the rapid curing and ease of processing of polyesters (11).
The three major types of fiber used in the fabrication of FRPs are glass
fibers, carbon fibers, and aramid fibers, respectively. Carbon fibers exhibit the
highest strength and stiffness. They have excellent chemical resistance as well
as high-temperature performance, which makes it possible to use these fibers
in conjunction with carbon, metal, or ceramic matrices in demanding aerospace
applications. In civil construction applications, however, such high-temperature
resistance is not required, and the cost of carbon fibers poses major hindrance
to their use. In addition, carbon fibers are moderately brittle, resulting in a low
capacity to absorb impact energy. They have low abrasion resistance and are
subject to galvanic corrosion in aqueous media in the presence of metals and
alloys (1214). Aramid fibers, produced from poly(paraphenylene terephthalamide) have lower strength than carbon or glass fibers, but they can absorb large
impact energies. Their transverse strength and compressive strength are relatively
low and they exhibit a strong tendency to absorb water. In addition, their adhesion
to polymeric matrices is sometimes weak and composites based on such fibers
are difficult to cut and machine. Glass fibers have high tensile strength, although
not as high as that of carbon fibers. They have a high capacity to absorb impact
423
energy and a good temperature resistance. Unlike aramid fibers, glass fibers do
not tend to absorb water and they have excellent temperature resistance. Glass
fibers are less stiff than carbon fibers. The low cost of glass fibers is a major
consideration that makes them prime candidates for use as reinforcements in
composites intended for large-scale construction applications. In addition to the
three major types of fiber described, advanced fibers, such as boron fibers and
metal fibers, have found use in the aerospace industry and in certain sporting
goods, but the excellent temperature resistance that such fibers possess is usually
unnecessary when used in conjunction with polymeric matrices, and their excellent mechanical and chemical stability is offset by their high cost in large-scale
applications. Polyethylene fibers have high tensile strength and stiffness, but their
compressive strength and modulus are low and they have very low softening
temperatures.
B. Environments
The environments that FRPs are expected to encounter in civil construction applications can be divided into those expected in everyday use and those encountered
only under accident conditions. Routine service environments involve a range of
temperatures between about 30C and 60C, depending on geographical location and subject to seasonal and diurnal fluctuations. The amount of moisture in
the environment may vary between a relative humidity of approximately 30%
and fully water-saturated air. Contact with liquid water will occur as a result of
rain, melting snow, or runoff. The water may be slightly acidic (ordinarily with
a pH no lower than 4.55) in the case of acidic precipitation or extraction of
acidic components from the ground. On the other hand, very high pH values may
be encountered in cases where FRP tendons are used in direct contact with cement
environments, as the pH of cement pore water is as high as 1314 (15). The salt
content of the water may range from very low in the case of fresh rainwater to
very high in the case of slush containing large amounts of road salts (NaCl or
CaCl2). Both wetdry cycling and freezethaw cycling may aggravate the effects
of aqueous environments. FRPs in civil engineering applications can be expected
to have some exposure to gasoline and motor oil fumes, and occasionally to
gasoline and oil in liquid form, but it has been found that such nonpolar media
have smaller effects on the mechanical properties of FRPs than aqueous media
(16). Some exposure to solar radiation in the ultraviolet (UV) and visible ranges
may take place, but at levels much lower than those encountered, for instance,
in sails and boat surfaces. While exposed to the environments described earlier,
FRP reinforcements in civil structures such as bridges, roads, and buildings are
subject, of course, to large and variable mechanical stresses. Although discussion
of the effects of such stresses by themselves is outside the scope of this review,
the interaction between environmental and mechanical stresses has to be taken
424
Barkatt
into account in situations involving phenomena such as fatigue and stress corrosion.
ka
ks
(1)
ka
exp
ks
ER (T
a
1
s
T a1)
(2)
where T a and T s are the temperatures of the accelerated test and the temperature
corresponding to the service condition, respectively, E a is a constant activation
energy characteristic of the degradation of the particular material under evalua-
425
tion, and R is the gas constant. In the case of silicate glasses, for instance, Arrhenius-type behavior has been observed over a broad range of temperatures (up to
at least 90C), and the values of E a were generally observed to range between
20 and 25 kJ with respect to leaching of alkalis and between 15 and 20 kJ with
respect to dissolution of the silicate matrix (18,19).
Accelerated tests based on the use of elevated temperatures have been used
in the evaluation of FRPs for structural applications in the aircraft industry to
predict the effects of moisture as well as of temperature within the range of
expected service environments (20).
The use of accelerated tests is very useful in predicting the performance
of materials over periods of time longer than those available for service-condition
testing. However, the use of this approach is limited to the range of conditions
over which Eq. (1) is valid. This range is generally limited to those conditions
under which the mechanism controlling the kinetics of the degradation of the
material is the same as the controlling mechanism under service conditions. The
use of accelerating conditions that cause a different degradation mechanism to
become predominant generally results in inapplicability of the accelerated test
to service conditions. It is known, for instance, that in the case of metals, the
structure of the oxide layer and, consequently, the power law characterizing the
oxidation kinetics are different at high temperatures than they are at the relatively
low temperatures used in many applications (21,22). Likewise, the nature of the
alteration products that control the course of glass hydration depends on the temperature (23). Similar precautions (i.e., assuring that the acceleration method does
not cause a change in the nature of the controlling degradation mechanism) are
required upon the use of other acceleration methods such as employing more
corrosive environments or the application of mechanical stresses.
B. Degradation Kinetics and Predictive Modeling for FRPs
In the case of composite materials, the role of model development in establishing
predictions of long-term performance is extremely important. As detailed earlier,
in certain cases involving single-phase materials the kinetics of degradation follows a simple, uniform-rate law characterized by a single rate constant that exhibits a simple Arrhenius-type behavior. In such cases, the extent of long-term degradation can be readily correlated with the results of short-term accelerated tests
using simple extrapolation. However, such simple behavior is not very common
even in the cases of homogeneous materials, and the probability of encountering
such behavior in the cases of complex material systems such as composites is
very low. The development of more sophisticated predictive models to evaluate
long-term performance and the use of test data to verify such models are therefore
the key for the use of composites in applications involving long-term exposure
in service environments. Considerable efforts have been made to model the degra-
426
Barkatt
427
428
Barkatt
Fig. 2 Changes in ultimate tensile strength, short beam shear strength, tensile modulus,
shear modulus, and weight of a glass/polyester sheet molding compound in humid air as
a function of immersion time: ( ) baseline data; () specimens tested after 6
months immersion followed by 3 weeks of drying; () specimens tested without postdrying. Bars indicate spread in data. Left: 23C; Right: 93C. (From Ref. 16.)
exhibit a distinct rise at a certain time after the beginning of the test. For instance,
upon subjecting sheet-molding compounds to a combination of mechanical loading and hygrothermal exposure to induce creep, strain was observed to exhibit
stepwise jumps (see Fig. 4) (34). The phenomenon of microcrack nucleation
and propagation, which can increase the rate of degradation of composites, was
identified both in cycling tests and in constant immersion tests (3436). In agreement with the above discussion, changes in mechanism were observed to affect
429
Fig. 4 Creep of SMC-R57 glass/epoxy sheet molding compound at 90C and 50% relative
humidity under different loads (percent of static ultimate tensile strength). (From Ref. 34.)
not only the degradation rate as a function of time at a given temperature but
also the temperature dependence of the degradation process, causing deviations
from simple Arrhenius-type behavior (37). Such limitations and complexities
must be addressed in order to establish a basis for reliable prediction of the extent
of environmental degradation of FRPs in long-term service. Further discussion
of specific cases of supralinear behavior is given in the following sections.
430
Barkatt
aramid fibers) (see Fig. 5) (38). It has been recognized that the conditions under
which samples are held between the end of the exposure period and the mechanical test have significant effects on the test results (43). During the intervening
period, samples often undergo partial drying that causes further changes (decrease
or increase) in the extent of degradation of strength resulting from the preceding
exposure to moisture (43). This is one example of the difficulties in comparing
results obtained under different test procedures and of the need to standardize
such procedures. The procedures for testing of composites developed by the
American Society for Testing and Materials (44) provide a useful starting point
in controlling the test parameters.
Although the results of mechanical tests provide data that are closely related
to the properties that determine the performance of composites in service, such
results are, in general, insufficient to provide a full basis for predictive modeling.
As detailed earlier, the degradation kinetics has been found in many cases to be
complicated and to involve the possibility of supralinear behavior at some stage
of the exposure. This complexity requires a more thorough understanding of the
microscopic processes of environmental degradation. Furthermore, such understanding is required in order to overcome other limitations of most mechanical
tests. One such limitation is the difficulty in extrapolating results obtained for
test coupons to full-size structural components. Another limitation involves tak-
431
ing into consideration the presence of other materials used along with FRPs. The
presence of cement, for instance, is known to have significant effects on the
degradation of FRPs based on glass fibers, and the presence of metals affects the
degradation of FRPs based on carbon fibers. Due to all of these reasons, it has
been found necessary, in many studies, to supplement the results of mechanical
tests on FRP samples exposed to environmental degradation by measurements
of various physical and chemical characteristics indicative of changes of the structure of the composite at the microscopic level.
B. Thermal and Thermogravimetric Methods
Gravimetric analysis of changes in sample weight following exposure to aqueous
or humid environments has been widely used in monitoring the consequences of
the exposure. In general, such exposure leads to a gain in weight, reflecting the
uptake of water by the composite. Exceptions are observed in high-pH environments, where fiber dissolution offsets water absorption by the polymeric matrix
(45). However, even in the absence of extensive fiber dissolution, interpretation
of weight-gain data is not always straightforward. Ideally, such weight gain is
expected to be linear and reversible. If the capacity of the matrix for uptake of
water is constant, then the weight gain observed upon exposing the FRP for a
long period of time should be linearly dependent on the relative humidity. Furthermore, upon bringing the sample back to an environment, which has the temperature and humidity at which the sample was held before the exposure, it is
expected to return gradually to its original weight. In fact, sharp increases in
moisture content have been observed upon exposure at high relative humidities
(46,47) and the process of moisture uptake is not entirely reversible. These effects
have been attributed to the formation of microcracks in the polymeric matrix
(31). An even more serious limitation on the applicability of weight-gain measurements is the observation, based on 2-year immersion tests, that there is no
direct correlation between the quantity of water absorbed and the loss of mechanical properties. Hence, absorption curves were concluded to be of limited use to
material selection (48).
Thermogravimetric analysis (TGA) has proven to be a very useful technique in evaluations of the effects of environmental exposure. TGA measurements at a heating rate of 10C/min on a variety of polymeric composites previously exposed to various aqueous media showed that the weight loss observed
upon heating from room temperature to 150C reflected the expulsion of absorbed
water. The weight loss observed between 150C and 300C was the most useful,
reflecting the amount of monomer volatilization and thus of polymer degradation
during the exposure. This weight loss also gave a good correlation with the extent
of degradation of the mechanical properties such as the tensile, flexural, and shortbeam strength (42,45). TGA analysis can be supplemented by evolved gas analy-
432
Barkatt
sis (EGA). Such analysis has confirmed that enhanced weight loss from FRPs
previously exposed to aqueous environments is due to hydrolytic depolymerization (32).
Dynamic mechanical analysis (DMA) and dynamic mechanical thermal
analysis (DMTA) have been used to monitor damping properties, including the
changes of these properties that occur at the glass transition (T g). These methods
have been applied mostly to the polymeric matrix (49) but also to polymeric
(aramid) fibers (50). As in the case of mechanical measurements, techniques have
been developed to perform DMA measurements during immersion in order to
avoid uncertainties due to changes upon sample storage following the exposure
(51). The results of DMA and DMTA have been used to establish correlations
between changes in T g and in mechanical properties, respectively
(24,31,37,49,52).
Very clear and quantitative measurements of changes in T g as a result of
environmental exposure have been provided by differential scanning calorimetry
(DSC).
Determinations of T g by means of DSC have shown that exposure to water
at elevated temperatures causes a decrease in T g due to partial hydrolysis and
plasticization (42,45,53) of the polymeric matrix, which are also the factors involved in the enhanced TGA loss (see above). However, in the cases of commercial FRP matrices that are usually not fully cured, the decrease in T g due to
depolymerization has been observed to be offset, in part or in full, by an increase
due to additional cross linking (42). DSC measurements have also been used to
measure changes in free volume upon thermal exposure (31).
C. Microstructural, Spectroscopic, and Electrochemical
Methods
Optical spectroscopy has been extensively used to characterize major damage
caused to FRPs by environmental exposure, including fiber buckling and matrix
cracking (37,54). For a more sensitive characterization of microstructural damage, scanning electron microscopy (SEM) has been widely used to study failure
modes and changes in failure mechanisms (33,54). Conventional SEM techniques
have the disadvantage of requiring exposure of the samples to high vacuum,
causing rapid and extreme drying that can enhance the actual damage by promoting shrinkage, cracking, and fragmentation. These artifacts can be minimized
through the use of environmental SEM (E-SEM), which permits SEM measurements to be made in the presence of humid air at pressures only slightly below
ambient. E-SEM measurements of polymeric composites have directly shown
that the extent of the damage increases in aqueous media with high pH and at
elevated temperatures (29), in agreement with the results of mechanical and thermochemical measurements.
433
V.
A. Glass
Because of their low cost, silicate glass fibers have the greatest potential of serving as reinforcements in FRPs to be used in large-scale civilian applications.
The structure of such glasses is based on a three-dimensional network of silicate
tetrahedra, modified through the introduction of alkali oxides (e.g., Na2O) and
alkaline earth oxides (e.g., CaO) to reduce melt viscosity and permit processing
at moderate temperatures (typically around 12001300C). Multivalent oxides
(e.g., Al2O3, B2O3) are also added in many cases in order to improve chemical
resistance, and they may also improve the mechanical, thermal, or optical proper-
434
Barkatt
ties. Thus, E-glass, the most common glass used in fiber reinforcements, consists
of 54.3% SiO2, 17.2% CaO, 15.2% Al2O3, 8.0% B2O3, 4.7% MgO, and 0.6% Na2O
by weight. The interaction between silicate glasses and aqueous environments has
been extensively investigated since the beginning of the 20th century. It has been
established that glasses have a very high chemical resistance to oxidants, reductants, and organic solvents (except amines, which have a basic character). Silicate
glasses, unless they have a sizable content of oxides of heavy and multivalent
metals such as Pb, Fe, and Al, have excellent resistance toward attack by acids
(except HF). Commercial silicate glasses are very durable in aqueous media with
near-neutral pH. The only environments in which glasses are subject to rapid
attack are basic media (i.e., solutions with pH levels in excess of approximately
9) and hydrofluoric acid solutions. Recently, it has been shown that concentrated
aqueous solutions of salts of alkali metals or calcium also attack glasses at moderately high rates (59).
The cause of the high reactivity of alkali toward glass is the ability of the
hydroxide ion, OH, to break the siloxy bonds that hold together the glass structure. (In the case of HF, the fluoride ion fulfills a similar role.) If the starting
medium is water, OH ions can be generated as a result of selective leaching of
alkali or alkaline earth ions from the glass. This mechanism has been formulated
in the following scheme (60):
DSiEOENa H2O DSiEOEH Na OH
(3)
(4)
(5)
The attack on Si sites in the surface regions of the glass continues until all four
siloxy bonds are hydrolyzed and the resulting Si(OH)4 monomer passes into the
solution (59,61).
Remarkably, upon considering the interaction of water with glass, the simple mechanism consisting of reactions (1)(3) can give rise to various types of
kinetic behavior. As an illustration, several cases of glasswater interaction may
be considered:
1. The glass is exposed to attack by water containing low levels of solutes
with the volume of the water being sufficiently large and/or the water
flowing at a sufficiently high rate to prevent significant accumulation
of glass-dissolution products in the water. Leaching of alkali and alkaline earth species according to reaction (1) first proceeds rapidly, resulting in the buildup of a high-silica surface layer, which slows down
further interaction of the water with the glass. Eventually, the depth
of this layer reaches steady state and further attack on the glass proceeds slowly at a constant rate determined by the hydrolysis of the
2.
3.
4.
5.
435
siloxy network. The overall kinetics of glass corrosion is initially sublinear and later becomes linear with time (18).
The volume of the aqueous environment in contact with the glass is
limited, flow is very slow or nonexistent, and the solution is buffered
against significant changes in pH. Under these conditions, dissolved
silica levels eventually approach the saturation limit, resulting in almost complete retardation of further attack (62). The sublinear and
linear stage of Case 1 are followed by another sublinear stage.
In the cases of complex glasses, the saturation effect described in Case
2 may be followed by the formation of highly insoluble crystalline
silicates. This results in lowering of the concentration of dissolved silica in the solution, elimination of the saturation effect, and renewed
increase in the rate of glass dissolution (63).
The volume of water is limited and flow is very slow, as in Case 2,
but the aqueous medium is not buffered. As a result, the concentration
of OH ions increases as reaction (1) proceeds OH ions consumed
during attack on the siloxy bond network in reaction (2) are regenerated
in reaction (3). Progressively higher pH results in increasing rate of
attack on the glass. The sublinear and linear stages of Case 3 are followed by a supralinear stage (64).
The glass is exposed to water, resulting in the formation of dealkalized,
hydrated surface layer. Hydration is enhanced by factors that may be
related to the glass composition [low silica content (65)], to the composition of the solution [high concentrations of alkali ions or Ca2 (59)],
or to environmental factors [high temperature]. This result in buildup
of large hydration stresses. Eventually, the surface layer cracks and
spalls off, destroying the retarding effect of this layer and opening up
a large area of fresh glass for the surrounding water to attack. This
results in a sharp increase in the effective rate of glass corrosion. This
effect may be further promoted by certain exposure conditions such
as wetdry cycling.
436
Barkatt
B. Carbon
Carbon fibers are more chemically resistant than silicate glass fibers. In general,
their solubilities and dissolution rates in both aqueous media and organic solvents
are very low. In addition, they possess excellent mechanical and thermal properties. The main obstacle for the use of composites reinforced by carbon fibers in
large-scale construction applications is economic. It should be noted, however,
that despite their high resistance to chemical dissolution, composite materials
reinforced with carbon fibers are susceptible to galvanic corrosion and other
modes of electrochemical degradation when they are in contact with metals in
electrolyte solutions such as seawater (13,14,66).
C. Aramid
Aramid fibers are known to be more susceptible in aqueous environments because, unlike silicate glass and carbon fibers, they absorb appreciable amounts
of water. This can result in significant effects on their mechanical properties
(36,39,67), causing a decrease in strength by as much as 4070% depending on
the temperature of the exposure (36). The degradation of the mechanical properties of polymeric fibers such as aramid fibers can be attributed to hydrolytic processes similar to these that take place with polymeric matrices (see Sec. VI). The
effect of such processes on polymeric fibers is particularly noticeable because of
the primary role that the fibers have in bearing the mechanical loads applied to
the composite materials under service conditions.
ECECEOECE ECEOH HOECE
(6)
437
O
O
RCEOR RC HOR H
|
OH
HEOH H
(7)
O
O
RCEOR RC HOR OH
|
OHHEOH OH
(8)
438
Barkatt
Fig. 7 Master relaxation modulus curves for [45] s T300/934 and GY70/339 carbon/
epoxy laminates. Horizontally shifted T300/934 data shown. (From Ref. 73.)
439
It is interesting to note that polymeric matrices exhibit viscoelastic behavior. At near-ambient temperatures, polymeric composites exhibit elastic or
glassy behavior, but upon exposure to moisture at elevated temperatures, they
experience transition from glassy to leathery behavior with a corresponding
change in the time dependence of the modulus (see Fig. 7) (73).
440
Barkatt
441
6 months, despite the fact that the absorption of moisture (as reflected in weight
gain) was completed in 1 month at 23C and in a few days at 93C.
In aqueous environments, the pH and the presence of solutes have a significant effect on fiber corrosion and, as a result, on the kinetics of degradation
of the composite as a whole. A recent extensive study on vinylesterglass and
polyesterglass composites (42,77) has shown that at high-pH fiber dissolution
is much faster, as expected on the basis of Eqs. (2) and (3). This agrees with the
well-known incompatibility of glass-based composite components with cement
environments (27,78) which are characterized by highly basic pore water (15).
Strong acid solutions also cause degradation of the strength of glass fibers due
to ion exchange of protons for metal ions on the glass surface (79). As mentioned
earlier, strong acids and strong bases also catalyze depolymerization of the polymeric matrix and hydrolysis of siloxy bonds in the interphase region. Aqueous
solutions of strong acids, such as sulfuric acid, were shown to cause accelerated
degradation as a result of interface corrosion (35) and formation of fractures (80).
In the cases of aramid fibers, the fibers as well as the matrix undergo accelerated
depolymerization in strongly acidic or strongly alkaline solutions. Weakly acidic
solutions, such as acetic acid solutions buffered at pH values of 35, do not cause
significantly more extensive corrosion than deionized water (77). Carbon fibers
are the most resistant to acidic or alkaline environments, but higher concentrations of electrolytes accelerate galvanic corrosion in the presence of metals (see
above).
B. Effects of Temperature and Moisture Content
Exposure of polymeric composites to temperatures exceeding 300C results in
massive depolymerization and volatilization (42,81) together with rapid degradation of the mechanical properties (24,82). Such temperatures, however, are encountered only in the vicinity of fire and are outside the range of normal service
environments. At lower temperatures, simple Arrhenius-type behavior is observed only over limited temperature ranges. In particular, the degradation kinetics of FRPs can be generally expected to change once the temperature significantly exceeds the glass transition range of the polymeric matrix. In the case
of epoxyaramid composites, for instance, it was observed that even under dry
conditions, mechanical properties such as flexural strength and stiffness were
complicated functions of temperature. In the presence of moisture, the deviations
from simple Arrhenius-type behavior are even more noticeable, with the strength
exhibiting a sharp drop with increasing temperature below room temperature,
then only a small dependence between room temperature and T g, and then a sharp
drop above T g (83). An important factor involved in the complex degradation
patterns was the observed strong dependence of T g itself on the moisture content
442
Barkatt
of the composite, amounting to a drop of 60C observed when the moisture content increased from 0% to 2%. These changes in T g are themselves a function of
the time of exposure to elevated temperatures and moisture, and the time dependence often exhibits a supralinear dependence, with rapid variation following a
period of little change (71). Because of the importance of changes in transport
mode which take place at T g, the extent of degradation of mechanical properties
is strongly dependent on the fraction of the material exhibiting the glass transition
(i.e. on the degree of noncrystallinity) (69). In tests on laminates, large differences
in the sensitivity with respect to exposure to elevated temperatures and to moisture were observed among laminates of different orientations among the fiber
directions in adjacent laminas (84). In this study, it was noted that large scatter
(in many cases, 2060%) existed in the data obtained for the degradation of
the mechanical properties. In addition, it was noted that the loading rate in the
mechanical tests and that interactions among this rate, the temperature, and the
moisture content were also expected to be significant. The importance of considering the ply and the laminate as well as individual fibermatrix elements in
degradation studies has been stressed (24).
Even more complex behavior was observed when FRPs were exposed to
periodically varying environmental conditions. Thermal cycling at relatively low
temperatures was shown to result in cracking of the polymeric matrix (85), especially when it involved freezethaw cycles. Thermal cycling between long exposures at 70C in humid air and short exposures at 150C was shown to give rise
to very prominent supralinear behavior, with the mechanical properties starting
to change rapidly after approximately 1 year (33). This was attributed to microcracking due to the alternation between a high moisture content at the lower
temperature and a low moisture content at the higher temperature.
C. Gaseous and Radiation Environments
Exposure of aramidepoxy composites to gaseous environments containing air
pollutants such as nitrogen oxides was shown to result in degradation of mechanical strength as a result of the formation of free radicals capable of attacking
the polymeric fibers (86). At elevated temperatures, the diffusion of atmospheric
oxygen and its attack on the matrix become important (24).
In addition to chemical environment and temperature, short-wavelength radiation is also known to affect the properties of polymeric materials. Degradation
of marine fabrics containing polyester or nylon fibers was observed upon exposure to ultraviolet radiation under dry or wet conditions (87). The mechanism of
degradation is likely to involve the formation of free radicals, similar to the case
of attack by nitrogen oxides. Bond scission and the formation and propagation
of free radicals also constitute the mechanism of thermal degradation and depolymerization of polymeric materials in general under dry conditions (68).
443
D. Stress Corrosion
A number of studies were carried out on the degradation of FRP materials under
a combination of environmental exposure and mechanical loads (35,37,77,80,86).
In general, moderate mechanical loads were observed to cause a moderate increase in the severity of the effects of environmental attack. However, definite
dependence of the rate of crack growth on stress intensity was observed. It was
concluded that the extent of stress corrosion reflects the susceptibility of the fibers
to corrosion and to shrinkage in the corroding environment. Stress corrosion was
observed to depend on the temperature, the acidity of the medium, and the type
of glass fibers used as reinforcements (88).
The dependence of crack growth velocity in glass on stress intensity and the
effects of the environment on this dependence have been thoroughly investigated
(89,90). The velocity of crack propagation was found to be controlled in its initial
and most important stage by the rate of stress corrosion at the crack tip, according
to the equation
V
Ax0n exp(bK I)
n
(9)
CD(H2O)x0
n
(10)
where D(H2O) is the diffusivity of water in the environment, 0 is the boundarylayer thickness, and C is a constant. In the third and final stage (region III),
the crack-propagation velocity again exhibits a strong dependence on the stress
intensity, but is no longer dependent on the relative humidity. This mechanism
has to be taken into account when glass fibers are used as FRP reinforcements.
In the cases of FRP materials, delamination crack propagation has been
shown to exhibit classical supralinear behavior leading to failure upon exposure
to elevated temperatures under conditions of either static creep (see Fig. 8) or
cyclic creep (see Fig. 9) (91). Measurements of crack growth in sheet-molding
compounds have shown that the dramatic increase in crack growth occurs once
a critical crack length is achieved. The critical crack length (see Fig. 10) is a
constant characteristic of each sheet-molding compound. Both the time to achieve
the critical crack length and the subsequent rate of crack growth are functions
of the applied stress intensity, K I (92). As noted earlier, under conditions conduc-
444
Barkatt
445
tive to creep, stepwise jumps in strain may occur at random times (34). The
sharp increase in creep observed in bonded FRP joints (see Fig. 11) was found
to occur somewhat sooner at higher loads (see Fig. 12) (93).
E.
446
Barkatt
447
448
Barkatt
B. Matrix-Based Mechanisms
Matrix degradation usually does not lead to failure directly, but opens the way
to environmental attack on the fibers (36,9799) or on the interphase (67), resulting in ultimate failure. The principal mechanisms of matrix degradation consist of microcracking at relatively low temperatures (below T g) and low moisture
content and of softening at elevated temperatures (above T g) and high moisture
content (which tends to cause plasticization, resulting in a lower T g) (54). As detailed
in the previous paragraph, these modes of matrix degradation are followed by
the onset of fiber degradation and eventual failure. Of course, at the time when
the rate-determining step changes from matrix degradation to fiber corrosion, the
rate of overall degradation is expected to exhibit a very significant change. This
greatly complicates extrapolation of short-term results to longer times. Furthermore, the change in failure mechanisms with increasing temperature also complicates the development of models intended to account for the temperature dependence of the degradation process. As noted earlier, the controlling failure
mechanism at low temperatures and low moisture content is often matrix cracking
and tensile fiber failure, whereas at elevated temperature and high moisture content, the dominant mechanism becomes matrix plasticization and softening opening the way to fiber buckling and failure in shear. This change makes it impossible
to use a simple Arrhenius-type model to predict the degradation behavior over
a broad temperature range or to evaluate the behavior of FRPs with high moisture
content from measurements performed while the moisture content is low.
C. Interface-Based Mechanisms
As noted earlier, the interphase is a region that is particularly prone to moisture
ingress and to degradation because of the relative weakness of the bonds in this
region and their susceptibility to hydrolysis. The mechanisms of interphase degradation involve loss of adhesion due to such hydrolytic attack (33,35,69,72). This
explains why pretreatment of the fibers has an important effect on the strength
of FRPs (e.g., aramidepoxy composites), both before and after exposure to
moisture and to elevated temperatures. However, increasing moisture concentration eventually reduces the beneficial effects of such pretreatment (67).
D. Multistage Mechanisms
A recent detailed investigation of the chemical steps involved in the degradation
of glassvinylester composites (29) showed that ingress of water leads to leaching
of alkali ingredients from the fibers and to a local rise in pH around the fibers,
which was identified by means of the use of acidbase indicators. Elevated pH
results in enhanced corrosion of the fibers and leads to degradation and perforation of the interphase, allowing further ingress of water. Hydrolytic attack on the
449
X.
A. Scale Problems
In the cases of homogeneous, dense materials such as metals and glasses, the
size of structural components affects only the extent of corrosion which may be
safely tolerated, but not the kinetics of the corrosion process itself. In the cases
of FRP materials, on the other hand, size is an important consideration, because
degradative processes involve internal surfaces as well as external surfaces. Obviously, saturation of a thicker slab or tendon of a composite with respect to moisture takes longer than the penetration of moisture through a thinner structural
component. This would extend the period required for hydrolytic processes to
affect the entire thickness of the material. It has been noted that laboratory coupons are not satisfactory for determining structural performance, as the load path
is not always obvious because of the variation in the material properties due to the
anisotropic nature of the composite (101). These considerations limit the range of
long-term extrapolations and require coupon tests to be supplemented by tests
on full-size structural components (102). Measurements on the transverse tensile
strength of graphiteepoxy composites have shown that the strength decreased
as the volume of the materials under stress increased (103).
B. Materials Variability
In the characterization of various properties of FRPs related to engineering applications, in particular resistance to environmental degradation, the issue of uniformity and standardization poses very serious problems. The fabrication of large-
450
Barkatt
451
452
Barkatt
453
Another extremely important area, in which the progress made to date has
been inadequate, is the issue of representative sampling and testing. On one hand,
FRPs used for large-scale applications at the present time are not well specified
with regard to their composition and fabrication technique. On the other, there
is almost no information as to the effect of moderate variations in composition
and in fabrication technique on the resistance to environmental degradation. Accordingly, efforts to quantify such effects have to be undertaken, along with attempts to introduce tighter control and better specification of FRP products and
to develop suitable reference materials. Although a promising start has been made
in the development of standard test methods (44), such methods have to be further
developed to make them applicable to large-scale structural components. For this
purpose, availability of suitable reference materials is indispensable. Until these
requirements are met, it is necessary to limit the use of predictive testing [e.g.,
to extrapolation by no more than one decade in time (102)]. Such limitations are
likely to continue to slow down the introduction of FRPs into use in large-scale
structural applications until the scientific and engineering issues related to longterm prediction are better resolved.
ACKNOWLEDGMENTS
The author is extremely grateful to Catherine R. Lang and Christina L. Monkres
for very valuable help in preparation and organization of the manuscript.
REFERENCES
1. GS Springer, ed. Environmental Effects on Composite Materials, Vol. 1. Westport,
CT: Technomic Publishing Co., 1981.
2. GS Springer, ed. Environmental Effects on Composite Materials, Vol. 2. Lancaster,
PA: Technomic Publishing Co., 1984.
3. GS Springer, ed. Environmental Effects on Composite Materials, Vol. 3. Lancaster,
PA: Technomic Publishing Co., 1988.
4. CE Harris, TS Gates, eds. High Temperature and Environmental Effects on Polymeric Composites (STP1174). Philadelphia: ASTM, 1993.
5. TS Gates, A-H Zurieck. High Temperature and Environmental Effects on Polymeric Composites, Vol. 2. (STP1302) West Conshohocken, PA: ASTM, 1997.
6. LC Bank, TR Gentry, A Barkatt. J Reinf Plast Composites 14:559587, 1995.
7. Standard Specifications for Highway Bridges, 15th ed. Washington, DC: American
Association of State Highway and Transportation Officials, 1992.
8. DP Jones, DC Leach, DR Moore. Polymer 26:13851393, 1985.
9. CS Smith. Design of Marine Structures in Composite Materials. London: Elsevier
Applied Science, 1990.
454
Barkatt
455
456
Barkatt
63. P Van Iseghem, B Grambow. Mat Res Soc Symp Proc 112:631639, 1988.
64. X Feng, JK Bates, CR Bradley, EC Buck. Mat Res Soc Symp Proc 294:207214,
1993.
65. Aa Barkatt, SA Olszowka, W Sousanpour, MA Adel-Hadadi, R Adiga, Al Barkatt,
GS Marbury, S Li. Mat Res Soc Symp Proc 212:6576, 1991.
66. FE Sloan, JB Talbot. Corrosion 48:10201026, 1992.
67. LE Doxsee, Jr, W Janseens, I Verpoest, P DeMeester. J Reinf Plastics Composites
10:645655, 1991.
68. W Schnabel. Polymer DegradationPrinciples and Practical Applications. Munich: Hanser Publ., 1992.
69. RI Brady, RS Porter. J Appl Polym Sci 39:18731885, 1990.
70. EG Wolff. SAMPE J 29:1119, 1993.
71. J Mijovic. Interplay of Physical and Chemical Aging in Graphite/Epoxy Composites. In: G.S. Springer, ed. Environmental Effects in Composite Materials, Vol. 3.
Lancaster, PA: Technomic Publ., 1988, pp. 150162.
72. AY Lou, TP Murtha. J Composite Mater 21:910924, 1987.
73. DL Flaggs, FW Crossman. Analysis of the viscoelastic response of composite laminates
during hygrothermal exposure. In: G.S. Springer, ed. Environmental Effects on Composite Materials, Vol. 2. Lancaster, PA: Technomic Publ., 1984, pp. 363380.
74. EP Plueddemann. Interfaces in Polymer Matrix Composites, New York: Academic
Press, 1974.
75. P Ehrburger, JB Bonnet. Trans Roy Soc A 294:495, 1980.
76. K Jayaraman, KL Reifsnider, RE Swain. J Composites Tech Res. 15:313, 1993.
77. TR Gentry, LC Bank, A Barkatt, L Prian, F Wang. Development of Accelerated
Test Methods to Determine the Durability of Composites Subject to Environmental
Loading (STP1323), West Conshohocken, PA: American Society for Testing and
Materials, 1998.
78. A Nanni, S Matsubara, S Hasuo. Durability of braided epoxy-impregnated aramid
FRP rods. In: KW Neale, P Labossiere, eds. Advanced Composite Materials in
Bridges and Structures. Montreal: CSCE, 1992, pp. 101110.
79. AG Metcalfe, GK Schmitz. Glass Technol 12:1523, 1971.
80. MA French, G Pritchard. Composites Sci Technol 45:257263, 1992.
81. JB Henderson, JA Wiebelt, MR Tant. In: GS Springer, ed. Environmental Effects
in Composite Materials, Vol. 3. Lancaster, PA: Technomic, 1988, pp. 213229.
82. CA Griffis, JA Nemes, FR Stonesifer, CI Chang. In: GS Springer, ed. Environmental Effects in Composite Materials, Vol. 3. Lancaster, PA: Technomic, 1988, pp.
361380.
83. HR Phelps, ER Long Jr. In: GS Springer, ed. Environmental Effects in Composite
Materials, Vol. 2. Lancaster, PA: Technomic, 1984, pp. 2742.
84. C-H Shen, GS Springer. In: GS Springer, ed. Environmental Effects in Composite
Materials, Vol. 1. Westport, CT: Technomic, 1981, pp. 7993.
85. HW Lord, PK Dutta, J Reinf Plast Composites 7:435457, 1988.
86. MC Perry, MA Vail, KL Devries. Effect of NOx environments and stress on the
degradation of mechanical properties of Kevlar 49 and Nylon 6. Polym Eng Sci
35:411418, 1995.
87. MA Moore, HH Epps. J Testing Eval 20:139143, 1992.
457
88. SN Sapalidis, PJ Hogg, DH Kelley, SJ Youd. Stress corrosion of fiberglass-reinforced (FRP) composites. In: Corrosion/97. Houston, TX: NACE International,
1997, Paper 356.
89. SM Wiederhorn. J Am Ceram Soc 50:407414, 1967.
90. SW Freiman. J Am Ceram Soc 57:350353, 1974.
91. R Ohtani, T Kitamura, Y Uematsu, M Hojo. Mode I delamination of carbon fiber
reinforced thermoplastic polymer under static and cyclic creep at elevated temperatures. In: CE Harris, TS Gates, eds. High Temperature and Environmental Effects
on Polymeric Composites (STP1174). Philadelphia, PA: American Society for
Testing and Materials, 1993, pp. 2338.
92. Y Arslanian, PJ Hogg. Measurement of stress corrosion crack growth in sheet molding compounds. In: CE Harris, TS Gates, eds. High Temperature and Environmental Effects on Polymeric Composites (STP1174). Philadelphia, PA: American Society for Testing and Materials, 1993, pp. 722.
93. GS Springer, TK Wang. Creep, strength and moisture absorption of adhesive
bonded FRP joints. In: GS Springer, ed. Environmental Effects on Composite Materials, Vol. 3. Lancaster, PA: Technomic, 1988, pp. 171184.
94. GC Wicks, WC Mosley, PG Whitkop, KA Saturday. J Non-Cryst. Solids 49: 413
428, 1982.
95. GS Springer. In: GS Springer, ed. Environmental Effects in Composite Materials,
Vol. 2. Lancaster, PA: Technomic, 1984, pp. 5978.
96. GS Springer. In: GS Springer, ed. Environmental Effects in Composite Materials,
Vol. 3. Lancaster, PA: Technomic, 1988, pp. 134.
97. K Katawaki, I Nishizaki, S Sasaki. Evaluation of durability of advanced composites
for applications to prestressed concrete bridges. In: KW Neale, P Labossiere, eds.
Advanced Composite Materials in Bridges and Structures. Montreal: CSCE, 1992,
pp. 119128.
98. R Sen, D Marischal, M Shahawy. ACI Struct J 90:525533, 1993.
99. S Birger, A Moshonov, S Kenig. Composites 20:341348, 1989.
100. KE Hofer, GN Skaper, LC Bennett, N Rao. J Reinf Plast Composites 6:5365, 1987.
101. A Green. Strengths of Panels for Structural Applications, 39th Annual Conference.
Society of the Plastic Industry, 1984.
102. A Green, T Bisarnsin. Predicting the performance of composites experimentally.
In: RC Talbot, ed. Managing Corrosion with Plastics, Vol. IX. Houston, TX: NACE
International 1990, pp. 114.
103. TK OBrien, SA Salpekar. In: ET Camponeschi Jr, ed. Composite Materials: Testing and Design, Vol. 11. (STP1206) Philadelphia, PA: American Society for Testing and Materials, 1993, pp. 2352.
104. DE Poland. Field fabrication technology for FRP equipment under real plant conditions with epoxy vinyl ester resins. In: Corrosion/97. Houston, TX: NACE International, 1997.
105. TJ Reinhart, LL Clements. Introduction to composites. In: TJ Reinhart, ed. Engineered Materials Handbook, Vol. 1 Composites. Metals Park, OH: ASM International, 1987, pp. 2734.
106. RS Bauer, SL Stewart, HD Stenzenberger. Composite materials (polymer-matrix):
Thermosets. In: JI Kroschwitz, M Howe-Grant, eds. Kirk-Othmer Encyclopedia of
458
Barkatt
Chemical Technology, 4th ed., Vol. 7. New York: WileyInterscience, 1991, pp.
2960.
107. CD Shirrell, JC Halpin, CE Browning. Moisture: An assessment of its impact on
design of resin based advanced composites. NASA Technical Report, NASA-44TM-X-3377, April 1976.
108. L Prian, A Barkatt. Chemical, thermochemical, and mechanical studies of FRP
degradation in corrosive environments. In: Corrosion/98, Houston, TX: NACE International, 1998, Paper 455.
15
Amorphous and Nanocrystalline
Alloys
Koji Hashimoto
Tohoku Institute of Technology, Sendai, Japan
I.
INTRODUCTION
460
Hashimoto
461
Fig. 1 Changes in corrosion rates of amorphous Fe-, Co-, and Ni-based metalloid alloys
containing chromium in 1M HCl at 30C as a function of alloy chromium content. (From
Ref. 14.)
462
Hashimoto
passive films formed on amorphous alloys is far higher than that on crystalline
alloys, as shown in Table 1.
The passive films formed by additions of sufficient amounts of valve metals
to amorphous nickelvalve metal alloys are exclusively composed of valve metal
oxyhydroxides such as TaO 2(OH) and NbO 2(OH) (16).
Consequently, amorphous alloys containing strongly passivating elements,
such as chromium and tantalum, have the very high ability to concentrate the
beneficial ions in their passive films and have high corrosion resistance based on
spontaneous passivation.
463
Fig. 3 Corrosion rates of amorphous NiNb, NiTa, and NiNbTa alloys in boiling
9M HNO3 with 100 ppm Cr6 ions. (From Ref. 16.)
464
Hashimoto
Amorphous alloy
Fe10 Cr13 P7 C
Fe3 Cr2 Mo13 P7 C
Co10 Cr20 P
Ni10 Cr20 P
Stainless steel
Fe30 Cr(2 Mo)
Fe19 Cr(2 Mo)
Passivation
Ref.
0.97
0.57
0.95
0.87
Spontaneous
Anodic polarization
Spontaneous
Spontaneous
18
11
19
20
0.75
0.58
Anodic polarization
Anodic polarization
22
23
465
trix, the precipitation of nanocrystalline phase is not always detrimental but sometimes increases the corrosion resistance.
466
Hashimoto
Fig. 5 Change in corrosion rate of melt-spun CrNiP alloys in 6M HCl at 30C. The
alloys were prepared changing the ratio of eutectic Cr13 P to Ni19 P. (From Ref. 29.)
Fig. 6 Potentiodynamic polarization curves of sputter-deposited Cr67 Zr alloy specimens measured in 6M HCl at 30C. Specimens were heated at a rate of 4C/min to the
temperature indicated in the figure and kept at the temperatures for 30 min. (From Ref.
30.)
467
film results from selective dissolution of alloy constituents unnecessary for the
passive film formation. When an alloy is able to passivate, fast active dissolution
of the alloy results in rapid enrichment of beneficial ions. The passivating ability
is, therefore, closely related to the activity of the alloy.
Because amorphous alloys are thermodynamically metastable, unless passive films are formed, the amorphous alloys dissolve more rapidly than the crystalline counterparts. This can be seen from an example shown in Fig. 7 (32). The
amorphous Co25 at% B alloy does not contain elements forming a stable passive
film in acids; hence, the dissolution rate of the alloy in an acid reflects the reactivity of the alloy. When heat treatment was conducted for 5 h, the amorphous Co
25 at% B alloy begins to crystallize at temperatures higher than 523 K, forming
a Co3 B phase in the amorphous matrix. The Co3B phase has the same composition as the amorphous matrix. The formation of the crystalline phase whose composition is the same as that of the amorphous matrix results in a decrease in
anodic dissolution current. This reveals that the amorphous phase is more active
than the crystalline counterpart. At 623 K, where the amorphous phase almost
468
Hashimoto
disappears, the single Co3B phase alloy shows the lowest activity for alloy dissolution. At even higher temperatures, the Co3B phase is decomposed to Co and
Co2B phases by the disproportionation reaction. The formation of the heterogeneous three-phase structure leads to an increase in the alloy dissolution rate.
The high reactivity of amorphous alloys based on the thermodynamically
metastable nature is effective in enhancing the accumulation of beneficial passivating elements on the alloy surface as a result of the fast dissolution of unnecessary elements into environments and, hence, is responsible for fast passivation
by the formation of the film in which the beneficial ions are highly concentrated,
as shown in Table 1.
The influence of the activity of alloys on the passivating ability can also be
seen in Fig. 1. Among iron-, cobalt-, and nickel-based alloys, when the chromium
content of alloys is not high enough to passivate spontaneously, the most active
iron-based alloys dissolve most rapidly and the most noble nickel-based alloys
dissolve most slowly. However, the fast active dissolution of iron-based alloys
is effective in concentrating chromic ions on the surface and, accordingly, the
iron-based alloys passivate spontaneously with the addition of the lowest amount
of chromium. By contrast, the slowly dissolving most noble nickel-based alloys
require the addition of the largest amounts of chromium for spontaneous passivation.
In this connection, the passivating ability and, hence, the corrosion resistance based on spontaneous passivation sometimes increase when an environment
becomes more aggressive. For instance, sputter-deposited WNb alloys are spontaneously passive in both 6M and 12M HCl, forming significantly tungsten-enriched double oxyhydroxide films of W4 and Nb5 ions, and the corrosion rates
of the alloys with less than 52 at% niobium in 12M HCl are significantly lower
than those in 6M HCl (33). Faster dissolution of niobium in 12M HCl leads to
faster formation of the stable and protective passive films than in 6M HCl; hence,
the corrosion damage in 12M HCl is significantly less than that in 6M HCl.
4. Effects of Metalloids
As can be seen in Fig. 1, the corrosion resistance of amorphous alloys changes
with additive metalloids, and the beneficial effect of metalloid in enhancing the
corrosion resistance based on passivation decreases in the order phosphorus, carbon, silicon, and boron, as shown in Fig. 8 (34). The effect of metalloids on the
corrosion resistance of alloys is dependent on the stability of the polyoxyanions
contained in their films. Phosphorus and carbon contained in ironchromium
metalloid alloys do not generally constitute passive films, as do phosphate and
carbonate in strong acids, and they do not interfere with the formation of the
passive hydrated chromium oxyhydroxide film (18,35). By contrast, as can be
seen in Fig. 1, boron-containing alloys require larger amounts of the chromium
469
Fig. 8 Average corrosion rates of amorphous Fe10 Cr10 B7 X and Fe10 Cr13
P7 X alloys in 0.1N H 2SO 4 at 30C. X is the metalloid denoted in the abscissa. (From
Ref. 34.)
470
Hashimoto
(38). A special feature of silicon addition can be seen in Fig. 9 (39). Increasing
the silicon content of amorphous Fe10 Cr5 MoBSi alloys leads to a decrease
in current densities in both the active and passive regions in 6M HCl at 25C
without changing the open-circuit corrosion potential. This is due to the formation
of a SiO2-like substance covering the alloy surface along with the hydrated oxyhydroxide film, and the amount of the SiO2-like substance increases with increasing
silicon content of the alloys.
471
evaporaled iron, chromium, and iron phosphide multilayer, they showed remarkable corrosion resistance to sulfuric acids with and without chloride ions (40).
The corrosion rate of amorphous NiP alloys is significantly lower than
that of crystalline nickel, as shown in Fig. 5. Kinetic and analytical investigations
(41) reveal the role of phosphorus as follows: Immersion of the amorphous Ni
P alloys in strong acids gives rise to preferential dissolution of nickel, leaving
elemental phosphorus on the alloy surface. The elemental phosphorus is accumulated and forms the surface layer, which can act as the barrier layer to diffusion
of nickel through the layer to dissolve into the solution. Accordingly, dissolution
of nickel leads to thickening of the elemental phosphorus barrier layer and to the
decrease of the dissolution current. In this situation, the decrease of the dissolution current with time at a constant potential follows Ficks second law. A typical
example is shown in Fig. 10. The equation in the figure is based on Ficks second
law, and the reciprocal of the current density squared changes linearly with polarization time.
Phosphorus has another beneficial effect in enhancing corrosion resistance.
Chromium contained in an alloy dissolves in the form of Cr 2 ions in acids without oxidizing power, and if the oxidizing power is high, chromium is oxidized
to Cr 3 ions. Cr 3 ions constitute a stable solid film with O 2 and OH ions even
in strong acids in the form of CrO x (OH)32x . This is the passive film of chromium.
Fig. 10 Change in the reciprocal of the current density squared as a function of time
of polarization at 100 mV versus saturated calomel electrode (SCE), where it is slightly
anodic from the open-circuit potential. (From Ref. 41.)
472
Hashimoto
Under natural conditions where no electricity is supplied by the outer circuit, the
higher oxidizing power corresponds to the faster cathodic reactions, particularly
oxygen reduction. The reduction rate of dissolved oxygen is mostly dependent
on the activity of the surface for the oxygen reduction. The elemental phosphorus
layer is especially active for the oxygen and proton reduction (42). The beneficial
effect of metalloids in enhancing spontaneous passivation of amorphous and nanocrystalline CrP and CrB alloys due to acceleration of cathodic proton reduction is also pointed out by Moffat et al. (43,44). Accordingly, for amorphous
alloys containing phosphorus and a strongly passivating element such as chromium, the formation of the elemental phosphorus layer not only prevents alloy
dissolution acting as the diffusion barrier but also accelerates the passive film
formation; that is, spontaneous passivation owing to the high activity for cathodic
reactions. The beneficial effect of phosphorus in enhancing passivation in strong
acids has been confirmed not only for NiCrP alloys but also for FeCrmetalloid alloys (45) and NiTaP alloys (46).
It has been found from these investigations that phosphorus-containing
phases and/or metal phosphides are generally corrosion resistant even in strong
acids and active for cathodic reductions. An interesting fact has been found in
the role of phosphorus in the corrosion resistance of crystalline CrP alloys in
a 47% HF solution at 30C in which the concentration of dissolved oxygen is
considerably higher than that in hydrochloric acids (47). The CrP alloys containing 0.7 at% or less phosphorus exhibit higher corrosion rates than that of
chromium in the HF solution showing slightly higher open-circuit potentials than
chromium in the active region of chromium, but the alloys with 0.8 at% or more
phosphorus are spontaneously passive showing more than four orders of magnitude lower corrosion rates than chromium. The addition of small amounts of
phosphorus enhances the cathodic oxygen reduction and hence accelerates active
dissolution of the bcc chromium matrix phase. Accordingly, extended immersion
results in complete dissolution of matrix leaving butterflylike flaky Cr 3P residues
which previously exist in grain boundaries. This is the detrimental role of phosphorus in inducing intergranular corrosion and stress-corrosion cracking of crystalline alloys. By contrast, an increase in the phosphorus addition to CrP alloys
leads to further enhancement of cathodic activity and to passivation of the matrix.
C. Recent Efforts in Tailoring Extremely CorrosionResistant Alloys
One serious restriction for the practical use of corrosion-resistant amorphous
alloys prepared by melt-spinning methods is their limited thickness of several
tens of microns, because the thickness of the melt which can be rapidly quenched
from the liquid state for the formation of the amorphous structure is limited.
Furthermore, conventional welding techniques cannot be applied, because heat-
473
474
Fig. 11
Hashimoto
475
476
Hashimoto
solid-solution structure over wide composition ranges. They were all new amorphous alloys. Their corrosion resistance to concentrated hydrochloric acids is
remarkably high.
Figure 14 shows the change in the corrosion rates of CrTi and CrZr
alloys in 6M HCl solution at 30C and CrNb and CrTa alloys in 12M HCl
solution at 30C as a function of the valve-metal content of the alloys (6568).
In a 6M HCl solution, chromium and titanium dissolve actively, whereas the Cr
Ti alloys show very low corrosion rates, which are several orders of magnitude
lower than those of the alloy components. Binary CrZr alloys also show low
corrosion rates. In spite of the fact that the corrosion rate of chromium metal is
five orders of magnitude higher than that of zirconium metal, the corrosion rate
of the CrZr alloys decreases with increasing chromium content of the alloy.
Amorphous CrNb and CrTa alloys show very high corrosion resistance, which
is higher than that of the alloy components. These results indicate that if both
components of binary alloys have a strong passivating ability, the alloys are able
to possess better corrosion resistance than the alloy components. The corrosion
rate of CrTi, CrZr, and CrNb alloys tends to decrease with chromium content.
Fig. 14 Corrosion rates of sputter-deposited CrTi (65) and CrZr (67) alloys in 6M
HCl at 30C and CrNb (68) and CrTa (68) alloys in 12M HCl at 30C.
477
The corrosion rates of CrTa alloys are extremely low and are lower than the
level measurable by inductively coupled plasma (ICP) spectrometry. It can, therefore, be said that the amorphous CrTa alloys are immune to corrosion even in
12M HCl and that the CrTa alloys possess the highest corrosion resistance
among known all metallic materials in strong acids. The chromiumvalve metal
alloys are all passivated spontaneously, forming the passive film consisting of
both chromium and valve-metal cations.
An interesting fact has been found with regard to the binding energy of
core electrons of elements constituting the passive film. Figure 15 shows the
correlation of the binding energies of the Cr 3 2p3/2 and Ta 5 4f 7/2 electrons with
the cationic fraction of tantalum in the spontaneously passivated film formed on
the CrTa alloys (69). Their binding energies change with film composition,
indicating the electronic interaction (i.e., charge transfer from chromium ion to
tantalum ion in the film). Similar charge transfer from chromic ions, Cr 3, to
valve-metal cations has been found for CrTi (65), CrZr (67), and CrNb (69)
Fig. 15 Change in binding energies of the Cr 3 2p3/2 and Ta 5 4f 7/2 electrons with the
cationic fraction of tantalum in the spontaneously passivated films formed on the CrTa
alloys in 12M HCl at 30C. (From Ref. 69.)
478
Hashimoto
alloys. The charge transfer between different cations indicates that these cations
are located very closely to show the electronic interaction. This means that the
passive film does not consist of a simple mixture of chromium oxyhydroxide and
valve-metal oxide, but is composed of a double oxyhydroxide of chromium and
valve-metal cations. The resultant double oxyhydroxide films are more protective
than chromium oxyhydroxide and valve-metal oxide films in these aggressive
solutions, and increasing chromium content of the alloys increases the chromium
content of the double oxyhydroxide film and increases the corrosion resistance
of the binary chromiumvalve metal alloys.
479
Fig. 17 Potentiodynamic polarization curves of MoNb (73) and MoTa (72) alloys
measured in 12M HCl at 30C.
480
Hashimoto
polarization curves of molybdenumvalve metal alloys and MoCr alloys measured in concentrated hydrochloric acids are similar to those shown in Fig. 17,
although high zirconium alloys suffer pitting by anodic polarization at about 1.5
V (SCE) in 12M HCl. These molybdenumcorrosion-resistant element alloys are
spontaneously passive and their open-circuit potentials are close to or higher than
the open-circuit potential of molybdenum. Molybdenum dissolves actively from
about 0.8 to 0.2 V (SCE) and passivates from about 0.2 to 0.2 V (SCE),
forming the film consisting of Mo4 (76). However, molybdenum suffers transpassive dissolution at further higher potentials. As can be seen in Fig. 17, the
cathodic activity of passive molybdenum for both proton and oxygen reductions
is very high. Accordingly, the open-circuit potential of molybdenum in 12M HCl
is very high and slight anodic polarization gives rise to a sharp current increase
due to transpassive dissolution.
Figure 18 shows cationic fractions of films and atomic fractions of underlying alloy surfaces of MoNb (73) and MoTa (72) alloys. Major anions in the
films are O2 ions and the balance is OH ions. The results shown in Fig. 18 are
almost the same as those obtained for MoTi (74), MoZr (71), and MoCr (75)
Fig. 18 Cationic fractions of films and atomic fractions of underlying alloy surfaces of
MoNb (73) and MoTa (72) alloys before and after immersion in 12M HCl at 30C.
481
alloys either air-exposed or spontaneously passivated in concentrated hydrochloric acids. Therefore, both spontaneously passivated films and air-formed films
are composed of oxyhydroxides containing cations of both alloy components.
The air-formed films are generally rich in cations of alloying elements, because
their affinity to oxygen is higher than that of molybdenum. Spontaneous passivation leads to the formation of molybdenum-enriched passive films on molybdenum-rich alloys and to the formation of the films with higher concentrations of
alloying element cations on the alloys containing higher concentrations of alloying elements.
There is a concentration gradient of cations in the films. Figure 19 shows
the change in analytical results as a function of the photoelectron take-off angle
(71,72). The measurement at a low photoelectron take-off angle emphasizes information from the exterior of the surface, and that at high photoelectron takeoff angle gives information not only from the exterior but also from the interior
Fig. 19 Apparent cationic fractions of films and atomic fractions of underlying alloy
surfaces of MoZr (71) and MoTa (72) alloys before and after immersion in 12M HCl
at 30C as a function of photoelectron take-off angle.
482
Hashimoto
483
Fig. 20 Changes in surface composition and oxidized molybdenum species for Mo57
Cr alloy specimens polarized in 12M HCl as a function of polarization potential.
484
Hashimoto
time that WNb and WTa alloys were found to form single amorphous phase
structures in spite of the complete miscibility of tungsten, with niobium or tantalum forming a continuous series of solid solutions at equilibrium. In general, if
two alloy components form a continuous series of solid solutions at equilibrium,
amorphous alloys are hardly formed, as can be seen in examples of MoCr, Mo
Ti, MoNb, MoTa, and WCr. Furthermore, because sputter deposition is a
very effective method for the formation of amorphous structure, when the composition is not suitable of forming the amorphous structure, their structure becomes
nanocrystalline. Figure 22 shows corrosion rates of WTi alloys (77) in 6M HCl
and WZr (78), WNb (33), WTa (80), and WCr (79) alloys in 12M HCl. In
very aggressive hydrochloric acids, the binary alloys show lower corrosion rates
than those of alloy component metals. The change in the corrosion rates of tungstenvalve metal alloys is very similar to that of molybdenumvalve metal alloys
shown in Fig. 16. The corrosion rate decreases with increasing alloying elements
and tantalum-containing alloys show particularly high corrosion resistance. W
Cr alloys also show higher corrosion resistance than tungsten and chromium,
although the corrosion resistance of MoCr alloys cannot exceed that of molybdenum.
Figure 23 shows polarization curves of WZr alloys measured in 6M and
12M HCl at 30C (78). Polarization curves of tungsten alloys are almost the same
485
Fig. 22 Corrosion rates of WTi alloys (77) in 6M HCl at 30C and WZr (78), W
Nb (33), WTa (80), and WCr (79) alloys in 12M HCl at 30C.
486
Hashimoto
487
Fig. 24 Sulfidation (solid lines) and oxidation (dotted lines) rate constants for amorphous AlMo and AlMoSi alloys as well as several high-temperature alloys. (Data
from Refs. 82, 83, and 85.)
488
Hashimoto
sulfidation (solid lines) and oxidation (dotted lines) rate constants for amorphous
AlMo and AlMoSi alloys as well as several high-temperature alloys
(82,83,85). As can be seen from the comparison between solid and dotted lines,
the sulfidation rate of conventional oxidation-resistant crystalline alloys is generally many orders of magnitude higher than the oxidation rate. By contrast, the
sulfidation rates of AlMo and AlMoSi alloys are significantly low and comparable to the oxidation rate of oxidation-resistant alloys. Furthermore, the sulfidation rate constants of these alloys are more than one order of magnitude lower
than those of molybdenum, and the molybdenum-containing Fe30 Mo9 Al
alloy. This result is of particular importance because this is the first time in the
history of corrosion science that a metallic material has been obtained with a
corrosion rate in highly sulfidizing atmosphere comparable to the oxidation rate
of oxidation-resistant alloys. The steady-state sulfidation rates of AlNb (86
90) and AlTa (91) alloys are also lower than those of the corresponding refractory metals.
The sulfide scales on these alloys consist of two layers: the Al 2S3 outer layer
and the inner refractory metal sulfide layer (84). The high sulfidation resistance of
the AlMo alloys is attributed to the formation of the MoS 2 phase, constituting
the major part of the inner barrier layer of the scale. The better protective properties of the sulfide scale formed on the AlMo alloys in comparison with those
of the MoS 2 scale on molybdenum is attributed to a lower defect concentration
in the aluminum-doped MoS 2 phase.
The oxidation rate of AlMo alloys is comparable to that of chromia-forming alloys, although it is higher than that of alumina-forming alloys. However,
the oxidation rate at temperatures higher than 900C is very high. The scale
consists mostly of alumina, but because of the high molybdenum contents of the
alloys, molybdenum is also oxidized, forming volatile MoO 3. Because the melting point of MoO 3 is 793C, the formation of low-melting point MoO 3 is responsible for the relatively low oxidation resistance of the AlMo alloys. Accordingly,
an attempt to improve the oxidation resistance has been made by adding silicon to
the AlMo alloys (83,85). The ternary AlMoSi alloys have a high sulfidation
resistance similar to that of the AlMo alloys and have a higher oxidation resistance than AlMo alloys. This is attributable to the formation of molybdenum
silicide, which is stable to oxidation. During sulfidation and oxidation, amorphous
alloys are crystallized, forming intermetallics. AlMo alloys form Al 8Mo 3 and
Mo 3Al phases. The molybdenum-rich Mo 3Al phase is readily oxidized, forming
volatile MoO 3. Accordingly, when the alumina scale surface on the AlMo alloys
is analyzed, a low concentration of molybdenum is always found. By contrast,
AlMoSi alloys are crystallized to Al 8Mo 3 and Mo 5Si 3 phases without forming
the easily oxidizable molybdenum-rich Mo 3Al phase. The Mo 5Si 3 phase is very
stable to oxidation. Accordingly, any molybdenum and silicon are not detected
489
in the topmost surface of the alumina scale. The oxidation resistance of AlNb
and AlTa alloys is also improved by the silicon addition (90,92).
The chromium-refractory metal alloys also have a high sulfidation resistance, and the sulfidation rates of the alloys containing 50 at% or more refractory
metals are almost comparable to those of corresponding refractory metals (93).
The sulfide scales formed on these alloys consist of two layers: the Cr 2S 3 outer
layer and the inner refractory metal sulfide layer. CrNb and CrTa alloys possess high oxidation resistance nearly comparable to typical chromia-forming
alloys, although CrMo alloys show poor oxidation resistance due to the formation of volatile MoO 3.
490
Hashimoto
Fig. 25 Schematic drawings of laser (a) and electron beam (b) processing of an amorphous surface alloy on a conventional metal substrate.
491
492
Hashimoto
steel tubes of 38.4 and 54.9 mm inner diameter was performed. The sputter deposits showed good performance in boiling 8M HNO 3 solutions containing various
oxidizing cations such as Cr 6.
V.
CONCLUDING REMARKS
493
REFERENCES
1. M Naka, K Hashimoto, T Masumoto. Corrosion resistance of amorphous alloys
with chromium. J Japan Inst Metals 38:835841, 1974.
2. K Hashimoto, M Naka, T Masumoto. Effect of nickel addition on corrosion resistance
of amorphous iron base alloys. Sci Rep Res Inst Tohoku Univ A-26:4854, 1976.
3. M Naka, K Hashimoto, T Masumoto. Effect of chromium addition on corrosion
resistance of amorphous FeBC and FeBSi alloys. Sci Rep Res Inst Tohoku
Univ A-26:283289, 1976.
4. M Naka, K Hashimoto, T Masumoto. High corrosion resistance of amorphous Fe
Mo and FeW alloys in HCl. J Non-Cryst Solids 29:6165, 1978.
5. K Hashimoto, K Asami, M Naka, T Masumoto. Effect of molybdenum on the corrosion behavior of amorphous FeMoC alloys in 1 N HCl. Sci Rep Res Inst Tohoku
Univ A-27:237245, 1979.
6. T Masumoto, K Hashimoto, M Naka. Corrosion behavior of amorphous metals.
In: Proceedings of 3rd International Conference on Rapidly, Quenched Metals.
London: The Metals Society, 1978, pp. 435448.
7. M Naka, K Hashimoto, A Inoue, T Masumoto. Corrosion-resistant amorphous Fe
C alloys containing chromium and/or molybdenum. J Non-Cryst Solids 31:347
354, 1979.
8. P Cadet, M Keddam, H Takenouti. Electrochemical behavior of amorphous Fe
CrP alloys in sulfuric acid. In: Proceedings of 4th International Conference on
Rapidly Quenched Metals. Sendai: The Japan Institute of Metals, 1982, Vol. II,
pp. 14471451.
9. K Kovacs, J Farkas, L Kiss, A Lovas, K Tompa. Electrochemical behavior of
Fe80TM3B17 amorphous alloys. In: Proceedings of 4th International Conference on
Rapidly Quenched Metals, Sendai: The Japan Institute of Metals, 1982, Vol. II,
pp. 14711474.
10. K Kobayashi, K Hashimoto, T Masumoto. Spontaneously passivating amorphous
FeCrMometalloid alloys in 6 N HCl at room temperature and 80C. Sci Rep
Res Inst Tohoku Univ A-29:284295, 1981.
11. K Hashimoto, M Naka, J Noguchi, K Asami, T Masumoto. Effects of alloying
elements on corrosion resistance of amorphous iron-base alloys. In: RP Frankenthal, J Kruger, eds. Passivity of Metals. Corrosion Monograph Series. Princeton,
NJ: The Electrochemical Society, 1978, pp. 156169.
12. M Naka, K Hashimoto, T Masumoto. Corrosion behavior of amorphous Co-base
alloys. In: Proceedings of 3rd International Conference on Rapidly, Quenched Metals. London: The Metals Society, 1978, pp. 449456.
494
Hashimoto
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
495
496
46.
47.
48.
49.
50.
51.
52.
53.
54.
55.
56.
57.
58.
59.
Hashimoto
of phosphorus on the passivation behavior of amorphous Fe8Cr13P7C alloy
in 9 M H2SO4 solution. Mater Sci Eng A181/A182:11191122, 1994.
H-J Lee, E Akiyama, H Habazaki, A Kawashima, K Asami, K Hashimoto. The
effect of phosphorus addition on the corrosion behavior of amorphous Ni30Ta
P alloys in 12 M HCl. Corros Sci 37:321330, 1995.
B-P Zhang, H Habazaki, A Kawashima, K Asami, K Hashimoto. The corrosion
behavior of CrP alloys in hydrofluoric acid. Corros Sci 34:599613, 1993.
H Yoshioka, A Kawashima, K Asami, K Hashimoto. The corrosion behavior of
sputter-deposited amorphous aluminum-valve metal alloys. In: RB Diegle, K Hashimoto, eds. Corrosion, Electrochemistry and Catalysis of Metallic Glasses.
Princeton, NJ: The Electrochemical Society, 1988, pp. 242253.
H Yoshioka, A Kawashima, K Asami, K Hashimoto. The corrosion behavior of
sputter-deposited amorphous aluminum-refractory metal alloys. In: Proceedings of
MRS International Meeting on Advanced Materials, Tokyo. Pittsburgh, PA: Materials Research Society, 1988, Vol. 3, pp. 429434.
H Yoshioka, Q Yan, H Habazaki, A Kawashima, K Asami, K Hashimoto. Passivity
and its breakdown on sputter-deposited amorphous Al-early transition metal alloys
in 1 M HCl at 30C. Corros Sci 31:349354, 1990.
Q Yan, H Yoshioka, H Habazaki, A Kawashima, K Asami, K Hashimoto. Passivity
and its breakdown on sputter-deposited amorphous Al-Ti alloys in a neutral aqueous solution with Cl. Corros Sci 31:401406, 1990.
H Yoshioka, H Habazaki, A Kawashima, K Asami, K Hashimoto. An XPS study
of corrosion behavior of sputter-deposited amorphous AlW alloys in 1 M HCl.
Corros Sci 32:313325, 1991.
H Yoshioka, H Habazaki, A Kawashima, K Asami, K Hashimoto. Anodic polarization behavior of sputter-deposited AlZr alloys in a neutral chloride containing
buffer solution. Electrochim Acta 36:12271233, 1991.
H Yoshioka, H Habazaki, A Kawashima, K Asami, K Hashimoto. The corrosion
behavior of sputter-deposited AlZr alloys in 1 M HCl solution. Corros Sci 33:
425436, 1992.
K Hashimoto, N Kumagai, H Yoshioka, J H Kim, E Akiyama, H Habazaki, S
Mrowec, A Kawashima, K Asami. Corrosion-resistant amorphous surface alloys.
Corros Sci 35:363370, 1993.
BA Shaw, TL Fritz, GD Davis, WC Moshier. The influence of tungsten on the
pitting of aluminum films. J Electrochem Soc 137:13171318, 1990.
GD Davis, BA Shaw, BJ Rees, M Ferry. Mechanisms of passivity of nonequilibrium AlW alloys. J Electrochem Soc 140:951959, 1993.
GD Davis, BA Shaw, BJ Rees, EL Principle, CA Pecile. Electrochemical behavior
and surface chemistry of nonequilibrium aluminum-tantalun alloys: Solute-rich interphase mechanism. In: CR Clayton, K Hashimoto, eds. Corrosion, Electrochemistry and Catalysis of Metastable Metals and Intermetallics. Princeton, NJ: The Electrochemical Society 1993, pp. 125.
EL Principle, BA Shaw, CA Pecile, AS Lyengar, GD Davis, BJ Rees. Nonequilibrium alloying studies on passivity in chloride environments. In: Proceedings of 12th
International Corrosion Congress. Houston, TX: National Association of Corrosion
Engineers, 1993, Vol. 3B, pp. 21872206.
497
498
Hashimoto
92.
93.
94.
95.
96.
97.
98.
99.
100.
101.
499
Index
[Aluminum]
corrosion (see Corrosion and Environments for corrosion)
Amorphous (see Alloys)
Anneal (heat treatment):
amorphous alloys, 462465
ceramic matrix composites, 412
copper, 130, 139
iron, 23
FeAl, 289
magnesium, 254258, 264, 270
metal matrix composites, 380
nickel, 61, 103, 104
stainless steel, 45
Antenna, 375
Atmosphere (see Environments for Corrosion)
Austenitic (see Steel, types of)
Automotive:
fiber-reinforced materials, 420
metal matrix composites, 375
Bainitic (see Steel, types of)
Bases (see Environments for corrosion)
Battery, high temperature, 103
Bearings, ball, 314
Blisters (formation on aluminum), 205
207
501
502
Boiler:
coal or oil fired, 88, 100
waste heat, 91
Boron (suppresses environmental embrittlement), 277283
Brass:
single-phase alloy, 118
two-phase alloy, 119
Bridge (fiber-reinforced materials), 419,
420
Bridging, crack wake, 391
Buildings (fiber-reinforced materials),
419, 420
Cavitation, 126
Cell, fuel, 103
Cement (see Environments for corrosion)
Ceramics, BN, AIN, silica forming,
316323
Ceramics, non-oxide, 311344:
corrosion (see Environments for corrosion)
effects of oxidation/corrosion, 342
344
oxygen transport in silica, 315, 316
Ceramics, oxide, 351416:
corrosion (see Corrosion and Environments for corrosion)
Chlorination, 9599
Chromium (amorphous alloys), 474
478
Coatings (see Corrosion, mitigation)
Composites:
ceramic matrix, 391, 392
glass, 413415
oxides, 410413
silicon carbide, 392410
metal matrix, 375
aluminum, 376381, 384, 385
iron, 386
magnesium, 378, 379, 381, 382,
385, 386
titanium, 382, 386
metal matrix corrosion (see Corrosion)
Index
Composition:
grain-boundary (steel), 4144
equilibrium processes, 42, 44
nonequilibrium processes, 45
Fe3Al, 294
solution (see Environments for corrosion)
Concrete (see Environments for corrosion)
Conductivity, solution, 48, 49
Containment, 103, 109
Contamination (see Environments for
corrosion)
Copper:
corrosion (see Corrosion and Environments for corrosion)
general, 115, 116
Corrosion:
causes of (see Environments for corrosion)
cracking (see stress corrosion cracking)
crevice corrosion:
aluminum, 177, 234, 238
copper, 118
metal matrix, 379
steel, 6
titanium (saltwater), 156
dealloying:
copper, 118, 119
steel, 6
erosion-corrosion:
aluminum, 243245
copper, 125
steel, 7
exfoliation of aluminum, 178, 180
fatigue:
aluminum, 180
copper, 121123
FeAl (crack growth), 292
Fe3Al (crack growth), 294, 295
magnesium, 267270, 385
plastics, fiber reinforced, 433
reinforced aluminum, 384
static (oxide ceramics), 356359
steel, 6
titanium (crack growth), 386
Index
[Corrosion:]
galvanic:
aluminum, 177, 178
copper, 126
magnesium, 272
metal matrix composites, 379, 380,
382
plastics, fiber reinforced, 422
steel, 7, 20, 21
titanium (saltwater), 156
general:
aluminum, 174, 175
copper, 116, 117
magnesium, 272
steel, 2, 3, 5, 2129, 45, 75109
hydrogen-stress cracking (steel), 7,
10, 11, 16, 17
intergranular:
aluminum, 175, 230, 241, 244
copper, 123125
steel, 6, 32
localized (aluminum), 175
mechanisms, fiber-reinforced plastics:
fiber based, 447
interfaced based, 448
matrix based, 448
multistage, 448, 449
pitting:
aluminum, 175, 208212, 221
225, 228231, 233
amorphous alloys, 473, 474
composites, metal matrix, 376
copper, 117, 118
magnesium, 265
metal matrix composites, 377, 379
steel, 610, 32
zirconium, 163, 166
problems, fiber-reinforced plastics:
scale, 449
variability of materials, 449, 450
resistance, amorphous Fe-Cr-metalloid alloys, 460
stress corrosion cracking:
aluminum, 180, 238, 241
composite materials, model of,
386388
503
[Corrosion:]
copper, 119121
intermetallics, 285, 296, 383
385
iron matrix composite, 386
magnesium, 266271
plastics, fiber reinforced, 443445
steel, 7, 1019, 32, 4144, 47, 48
Corrosion, mitigation:
anodic and cathodic protection:
amorphous alloys, 472, 480
FeAl, 290291
metal matrix composites, 381, 388,
389
steel, 3, 5, 21
coatings:
amorphous alloys, 489492
on ceramics (refractory oxide),
339
on intermetallics, 304
magnesium, 271
on steel 3, 2729
Cracking:
hydrogen-stress (see Corrosion)
matrix, 391
stress-corrosion (see Corrosion)
Crevice (see Corrosion)
Dealloying (see Corrosion)
Debonding, interface, 391
Deflection, crack, 391
Dezincification, 118, 119, 132, 139
Discontinuities, microstructural (steel):
grain boundaries, 9
second-phase particles, 9
sulfide inclusions, 9
Duplex (see Steel, types of )
Electron beam (producing amorphous
alloys), 489491
Electronics, high temperature, 311
Embrittlement:
alleviation, 303307
alloy additions (B, Cr, Zn), 306
grain size and shape, 304, 305
prestrain, 306
504
[Embrittlement:]
processing techniques, 306
surface conditions, 303, 304
elevated temperatures, 298303
L12 alloys (Ni3Al, Ni3Si), 299301
other intermetallics, 301, 302
environmental, suppressed by:
boron, 277283
zirconium, 283
hydrogen induced, 296, 297
ceramic matrix composites, 408410
intermetallics (see Intermetallics)
metallic alloys, 162, 170, 172
titanium, 158
mechanisms, 302, 303
moisture induced, 284, 297, 307
oxygen induced, 307
Energy, solar, 103, 107
Engine:
gas turbine, 414
heat, 311, 323
Environments, aqueous (aluminum),
193232
alloys in water, 215, 216
binary alloys in water:
Al-Cu, 216220
Al-Mg, 216221
Al-Zn, 220
unalloyed alloys in water, 195
adsorption process, 195201
chemical reaction, 201204
direct attack of exposed metal,
205207
thinning of oxide film, 204, 205
Environments for corrosion:
acids, effect on:
amorphous metalloid alloys, 460
copper, 146, 147
plastics, fiber reinforced, 441, 447
steel, 40
titanium (oxidizing and reducing),
157
alkalis, effect on:
copper, 147
plastics, fiber reinforced, 433, 441
titanium, 158
Index
[Environments for corrosion:]
ammoniacal solutions, effect on copper, 145, 146
antifreeze, effect on fiber-reinforced
plastics, 440
aqueous, effect on aluminum (see Environments, aqueous)
ash- and salt-deposits, effect on
nickel (high temperature), 75, 76,
99103
atmosphere, effect on:
copper, 130134
steel, 26, 38
base, effect on steel, 40
carbon dioxide, effect on silicon carbide, 329
carburization and metal dusting, effect on nickel (high temperature),
75, 76, 8891
cement, effect on fiber-reinforced plastics, 423, 431, 441, 447
chloride concentration, effect on aluminum, 213
concrete, steel rebar, 26, 27
contamination, 25, 45
deionized water, effect on steel, 39
diesel fuel, effect on fiber-reinforced
plastics, 440
environmental degradation, effect on
fiber-reinforced plastics, 419, 453
erosion, flow assisted, effect on
steel, 7
freshwater, effect on:
copper, 139141
steel, 25, 26
gas, effect on:
fiber-reinforced plastics, 440
titanium, 158, 159
halogens, effect on:
nickel (high temperature), 75, 76,
9599
zirconium, 167, 168
inorganic acids, effect on titanium,
157, 158
lubricating oil, effect on fiber-reinforced plastics, 440
Index
[Environments for corrosion:]
melts, thick, effect on oxide ceramics,
371
metals, liquid (molten):
effect on oxide ceramics, 370
372
nickel (high temperature), 75, 76,
109
titanium, 159
neutral solutions, effect on copper,
147
nitrogen, effect on oxide ceramics,
355
nitridation, effect on nickel (high temperature), 75, 76, 9195
organic compounds and solutions, effect on:
copper, 147
titanium, 158
oxidation, effect on:
aluminum, 221, 232, 233 (See
Films, surface)
amorphous alloys (high temperature), 486489
ceramic matrix composites, 413
ceramics, non-oxide, 314325,
332339
copper, 128130, 146
magnesium, 265266
nickel (high temperature), 75, 76
85
silicon carbide, 398408
steel, 2
oxygen, effect on oxide ceramics,
352355
pH, effect on:
aluminum, 214, 237
steel, 2, 25
pollution, effect on copper, 144, 145
potable water, effect on:
copper, 137139
steel, 39
pressure, hydrostatic, effect on fiberreinforced plastics, 445
radiation, effect on fiber-reinforced
plastics, 442
505
[Environments for corrosion:]
saline solution (saltwater, seawater),
effect on:
aluminum, 218, 222, 225, 231,
235, 241245
copper, 141144
magnesium, 261265
plastics, fiber reinforced, 433
steel, 39, 40
titanium, 156, 157
salts, molten, effect on:
nickel (high temperature), 75, 76,
103109
oxide ceramics, 359371
salt spray, effect on magnesium matrix, 378, 379
soil, effect on:
copper, 136, 137
steel, 26, 27
solution composition, 2
steam, effect on copper, 145
sulfidation, effect on:
amorphous alloys (high temperature), 486489
nickel (high temperature), 75, 76,
8588
temperature, effect on:
aluminum, 214, 215
amorphous alloys, 486489
ceramic matrix composites, 402
407
ceramics, non-oxide, 339
glass matrix composites, 413415
nickel, 75109
plastics, fiber reinforced, 425, 426,
432, 440442, 451
steel, 2
velocity, fluid, effect on steel, 2, 23,
24
water, effect on:
aluminum (see Environments,
aqueous)
ceramics, non-oxide, 325329,
340, 341
ceramics, oxide, 356359
copper, 134136
506
Index
Hafnium, 169
Hardening, precipitation (see Steel,
types of)
Hardware, military, 375
Heat treatment (see Anneal)
Highways (fiber-reinforced materials),
419
Homogeneity (amorphous alloys), 462
465
Humidity, relative (steel), 25
Hydrostatic (see Pressure, hydrostatic)
Index
Liners:
combustor, 311
fiber-reinforced material, 420
Lines, groundwater (copper), 115
Magnesium, 253272:
Mg-Al, 254258
Mg-Zn, 254258
Marine (see Vessels, marine)
Martensitic (see Steel, types of )
Mechanisms, toughening, 391
Melt-spinning (aluminum), 225
Metalloids, effects on amorphous alloys,
468470
Metals:
liquid, 341, 342
reactive, 151, 152
refractory, 151, 152
Microelectronics, silicon, 318
Modes of corrosion (see Environments
for corrosion)
Molybdenum (amorphous alloys), 478
483
Motors, rocket (fiber-reinforced materials), 420
Nanocrystalline (see Alloys)
Nickel:
advantages, 55, 56
corrosion (see Corrosion and Environments for corrosion)
processing chemicals:
chlorides, 71, 72
hydrochloric acid, 6568
hydrofluoric acid, 68, 69
hydroxides, 72, 73
nitric acid, 70
phosphoric acid, 69
sulfuric acid, 6365
resistance to high-temperature corrosion (see Environments for corrosion)
systems:
Ni, 5658
Ni-Cr, 58
Ni-Cr-Si, 62
507
[Nickel:]
Ni-Cu, 5860
Ni-Fe-Cr, 63
Ni-Mo, 60, 61
Niobium, 170171
Nonequilibrium (aluminum), 225
Optics, precision, 375
Organic compounds (see Environments
for corrosion)
Oxidation (see Environments for corrosion)
Oxygen, dissolved (steel), 25
Particulate, 391
Passivation (amorphous alloys):
high activity, 465468
high passivating ability, 461462
Pearlitic (see Steel, types of )
pH (see Environments for corrosion)
Phenolic (see Resins)
Phosphorus in amoprhous alloys, 470
472
Pipelines (fiber-reinforced materials),
420
Pitting (see Corrosion)
Plastics, fiber reinforced:
corrosion (see Corrosion and Environments for corrosion)
environments, 423
interphase degradation, 439440
mechanical properties, 429431
porosity, 450
predicting long-term performance,
451453
structure, 421423
Platforms (fiber-reinforced materials),
420
Poisoning (see Toxicity of cuprous
oxide)
Polarization (aluminum), 209212
Polybismaleimides (see Resins)
Polyester, 422
Polyether (see Resins, polyether ketone)
Polyetherimide (see Resins)
Polyethersulfone (see Resins)
508
Polyimides (see Resins)
Polymers:
degradation, 436
thermoplastic, 437
thermosetting, 421, 437
Polysulfone (see Resins)
Potential:
critical pitting (aluminum), 208, 209
protection (aluminum), 209
Pressure, hydrostatic (aluminum), 242,
243
Protection, passivation (steel):
factors that hinder:
concentration changes, 5
impurities, 5
temperature increase, 5
velocity changes, 5
factors that promote:
hydrofluoric acid, 3
sodium hydroxide, 3
sulfuric acid, 3
Pullout, fiber, 391
Rail, light (fiber-reinforced materials),
420
Rainfall (see Environments for corrosion)
Reactor, fusion, 171
Recuperators, 102
Reinforcement:
ceramic matrix composites (fiber),
396, 397
metal matrix composites
continuous fiber, 379382, 384
388
discontinuous, 376379, 383384
Repassivation (aluminum), 223, 230
Resins:
phenolic, 422
polybismaleimides, 422
polyetherimide (thermoplastics), 421
polyether ketone (thermoplastics),
421
polyethersulfone (thermoplastics), 421
polyimides, 422
polysulfone (thermoplastics), 421
Index
Resistance to corrosion (see Corrosion,
mitigation)
Roads (fiber-reinforced materials), 420
Rocket (fiber-reinforced materials), 420
Semiconductor, high temperature, 313
Shield, reentry (space vehicle), 311, 313
Silane (see Agents, silane coupling)
Silicon nitride, oxidation of (see Ceramics, non-oxide)
Silicon, oxidation of (see Ceramics,
non-oxide)
Silicon carbide, oxidation of (see Ceramics, non-oxide)
Sizing (fiber-reinforced materials), 421
Soil (see Environments for corrosion)
Solar (see Energy)
Solidification, rapid (aluminum), 225
Solutions, neutral (copper) (see Environments for corrosion)
Sputter-deposition:
aluminum, 225, 226, 229, 230
amorphous alloys, 473, 491, 492
magnesium, 271
metal matrix composites, 389
Stairs (fiber-reinforced materials), 420
Steel:
carbon, 2, 6
corrision (see Corrosion and Environments for corrosion)
elements of stainless steel:
carbon, 37
chromium, 36
manganese, 36
molybdenum, 36
nickel, 37
nitrogen, 37
low-alloy, 2
stainless steel, types of:
austenitic, 3133, 4850
bainitic, 23
duplex, 32, 33, 51
ferritic, 1, 23, 32, 33, 50, 51
martensitic, 1, 23, 32, 33, 51, 52
pearlitic, 23
precipitation hardening, 32, 33, 36
Index
[Steel:]
strength:
high-strength (see Corrosion, stress
corrosion)
low-strength (see Corrosion, stress
corrosion)
weathering, 22
Stiffeners, cargo bay, 375
Stress, threshold (steel), 13
Substrate, electronic, 313
Superaustenitics, 50
Superduplex, 51
Supermartensitics, 52
Tanks:
chemical storage (fiber-reinforced materials), 420
oil storage (fiber-reinforced materials), 420
Tantalum, 171172
Tarnishing, 135, 136
Temperature (see Environments for corrosion)
Tests:
accelerated, 424, 425
service condition, 424
Titanium, 154159:
applications, 154, 155
alloys (alpha, alpha-beta, beta), 155
509
[Titanium:]
corrosion resistance (see Environments for corrosion)
Toxicity of cuprous oxide, 144
Tubes:
copper:
condenser, 115
plumbing, 115
heat exchanger, 311
Tungsten (amorphous alloys), 483485
Vanadium, 169170
Velocity, fluid (see Environments for
corrosion)
Vessels:
marine (fiber-reinforced materials),
420
reaction (fiber-reinforced materials),
420
Vinylester, 421, 422
Water (see Environments for corrosion)
Whiskers, 376, 391, 396
Zirconium:
general, 159170
nuclear applications, 160, 161, 167
suppresses environmental embrittlement, 283