You are on page 1of 12

Technical aspects of

ORTHO-PARAHYDROGEN CONVERSION
A. H. SINGLETON

G. E. SCHMAUCH

The existence of two forms of hydrogen has been of

theoretical interest to the physicist and chemist for


over 35 yeon, providing the physicist with a relatively
simple system for testing postulates of quantum theory
ond the chemist with a choice subjecf for kinetic
studies. The nature of these forms of hydrogen
remained something of a curiosity until the demands
e'~thed&ictlon' behveen
of cryogenic technbios/.
.,
ortho- and pomhydrogen more than academic. Aerorpace and commercial demo.nds'rerequire o more exteiuive understanding of fhe conversion and hove stimulated renewed interest in the ortho-p&ohydrogen conilable methods for altp+iw the

' '

he interest in cryogenic hydrogen technology has


Tdeveloped
rapidly, over the past several years
very

as a result of the applicatiop of liquid hydrogen in the


rockets and space programs. One important facet of
this growing hydrogen technology is concerned with the
ortho-parahydrogen conversion phenomenon since the
conversion is of extreme importance when considering
the production, storage, and utilization of liquid hydro$en.
Figure 1 s h c y the equilibrium parahydrogen
concentration as a function of temperature. Wha,,one
is concerned with hydrogen technology at ambient
temperature and above, the conversion phenomenon
need not^ be considered since the equilibrium compo&
or ortho-paca ratio remains constant at 25% para&rdrogm. This ambient equilibrium composition is
quUaUy referred to as normal hydrogen. ' When one is
cerned with hydrogen technology at temperatures
fi~antlylower than ambient, the ortho-para cond o n becomes important since the equilibrium ortho#am ratio varies with temperahlre'in thecryogenicregion.

most promising
.. .

,,

, .

.:

. .

..

llkrcaun u.vi.w
As early as 1927, Heisenberg (27), Hund (29), and
Dmnison (17) postulated the existence of two m o d i i a tions of molecular hydrogen. In 1929, McLuu~anand
M U (34)and Bonhoeffer and Worteck (4)S U ~
tiated these postulates with experimental evidence.
The two fprm were n a m d orthohydrogen and parahydrogen. The existence or the ortho and para modifications of molecular hydrogen is a direct result of the
rmdear spin associated with the hydrogen atom. I n hy4rogen molecules, the nuclear spins of the individual
atom are either oriented in the same direction (parallel)
whieh corresponds to the ortho modification, or in opposite directions (antiparallel) which c m p o n d s to the

.;,..:. ..r.jr..,:r..,.

wa modification. These two different orientations of


nuclear spins in the hydrogen molecule are responsible
for the difference in magnetic, optical, and thermal
properrieS of the two modifications. The resultant
nuclear spin of the molecule contributes to the total
rotational energy of the molecule which can exist only
in certain discrete quantites or energy levels. These .
rotational energy levels ace designated by rotational
quantum numbers, J. which can have integral values
such as 0, 1, 2, 3, 4 . . .corresponding to increasing
discrete amounts of rotational energy. It happens that.
cum rotational quantum numbers (0, 2, 4
~

energy leve
oicupied only by
molecules and add rotational quantum numbers (1, 3,
5 . . . .) refer to levels which can be occupied only by
) molecules.
:equilibrium, the molecules are distributed throughout these various energy levels as a function of temperature. The ortho-para ratio is thus determined by the
number of molecules occupying odd or cum energy states,
respectively. If the temperature is changed, the equilibrium energy dutribution is also changed. However,
under normal conditions, the transition probability
between ortho and para states is practically zero.
..

..~

...,..

'1

5:

.
f

.. .:... .:.

.,

.,:

'

concluded, as a result of their work on p


surfaces, that the mechanism must involve
of hydrogen at the catalyst surface.
nt of the catalyst upon th
section, there are two

hydrogm .@ one boitnd~.atom


.aud'Saeege(20).
was Eon-

to

*
s

?I

able velocity in the presence of molecular oxygen.


Other paramagnetic molecules such as nitric oxide were
found to be effective. T h e conversion is caused by the
iduence of the inhomogeneous magnetic field of the
paramagnetic molecule upon the nuclear magnetic
field of the hydrogen molecule during collion. I t is a
simple bimolecular mechmism in the gas phase. The
conversion velocity is independent of the hydrogen
pressure, is proportional to the oxygen pressure, and is a
direct function of the temperature.
A theoretical treatment of the paramagnetic conversion
which was developed by Wigner (60) and improved by
Kalckar and Teller (32) showed agreement with experimental data on the magnetic conversion. The catalytic
activity is proportional to the square of the magnetic
moment of the paramagnetic molecule. This was substantiated by Far& and S a c k (20) who studied the
effect of paramagnetic ions in solution and dissolved
oxygen on the conversion of hydrogen. Wigners
theory included a factor involving the distance of closest
approach of the hydrogen to the paramagnetic molecule.
The theoretical absolute rate of conversion depends on
the inverse 6th power of this factor. By assuming an
approach factor betweem one and two angstrom units,
experimental convenion rates were in agreement with
theory.
The heterogeneous catalysis of the ortho-parahydrogen
conversion by a magnetic mechanism was studied by
such early workers as Bonhoeffer and Harteck (5),
Bonhoeffer and Farkas (6), and Rummel (38). They
shoived that the rate of conversion of para- to orihohydrogen on charcoal, as a function of temperature,
passed through a definite minimum. This,suggested
that two mechanisms were operating, one increasingly
effective as the temperature was lowered, and responsible for a negative temperature coefficient between liquid
air and mom temperatures; the other increasingly
effective as the temperature \*as raised, and responsible
for the positive temperature coefficient above 20 C.

Taylor (47)showed that two types of adsorption occurs


in these temperature ranges. He showed that van der
Waals or physical adsorption occurs at low temperatures
and decreases as the temperature increases while activated or chemisorption takes place above room temperatures and increases as the temperature increases.
The low temperature conversion was believed to involve
the magnetic effect of impurities in the charcoal.
Rummel(38) demonstrated that the low temperature
(molecular) adsorption of oxygen on the charcoal
strongly enhanced the conversion of parahydrogen.
I t was also shown that the adsorption of oxy-gen at high
temperatures (activated adsorption) failed to enhance
the conversion, but in fact hindered it by covering part
of the active surface. Gould, Bleakney, and Taylor
(24) showed that outgassed charcoals at liquid air temperatures catalyzed hydrogen conversion but not hydrogendeuterium exchange. This is indicative of a
magnetic mechanism. In 1933, Taylor and Diamond
( 4 4 9 )showed that paramagnetic subtancen are better
low-temperature conversion catalysts than diamagnetics.
They also showed that the catalytic activity was a direct
function of the magnetic properties of the catalyst.
Farkas and Sandler (27) and Turkevich and Selwood
(50) showed that low-temperature heterogeneous catalysis of the para- to orthohydrogen conversion could be
ascribed to the influence of the inhomogeneous magnetic
field of the paramagnetic comp-ent
of the catalyst
-upon the physically adsorbed hydrogen. The work
of Harrison and McJhwell (25, 26) showed that diamagnetic zinc oxide physically adsorbed hydrogen but
was not very effective as a conversion catalyst. They
also showed that para-magnetic m-diphenyl-&picryl
hydrazyl, a solid free radical, physically adsorbed hydrogen very slightly and was only slightly effective as a conversion catalyst. However, a mixture of the two materials proved to be more effective a8 a conversion
catalyst than expected by their individual characteristics.
In view of thii, an effective low-temperature catalyst
bas three primary requirements. I t must have a high
physical adsorptive capacity for hydrogen, a high concentration of active or magnetic specie, and a high
paramagnetic susceptibility. However, the conversion
rate is also influenced by such variables as pressure,
temperature, feed composition, and space velocity of
the hydrogen. The effects of these-parameters iduence
the kinetics of the conversion and perhaps determine
whether the conversion is controlled by adsorption,
desorption, diffusion, or the surface reaction. Detailed
studies of the kinetics of the heterogeneous catalytic
para-orthohydrogen conversion a t low temperatures on
zinc oxide, aa-diphenyl-&picryl hydrazyl and a mixture
of the two were performed by Harrison and McDowell
(25,26). These studies show that the conversion follows
fimt-order kinetics since the expression

uI -- u&- k t
VOL 56

NO. 5 M A Y 1 9 6 4

23

An important consideration in the ortho-parahydrogen conversion is the


is applicable where U t
Plots of

X,- X , and U,

X, - X,.

versus time produced straight lines, the slope of which


equals the rate constant ( k H ) . The rate constant for
the conversion was shown to be independent of pressure
at pressures less than an atmosphere and to decrease with
increasing temperature.
A theoretical treatment is given by Harrison and
McDowell (25, 26) for a heterogeneous conversion
occurring during the collision of the hydrogen molecule
with the catalyst surface both in the presence and the
absence of a physically adsorbed layer. Only the
theory based on a physically adsorbed layer is in agreement with the experimental results. Therefore, it is
concluded that the catalysis is due to the interaction
bet\veen the hydrogen molecules in a physically adsorbed layer and the inhomogeneous magnetic field
at the surface of the catalyst. The perturbation resulting
from this interaction enables the otherwise forbidden
para-ortho transition to take place. Harrison and
McDowell (25, 26) also showed, theoretically, that the
conversion rate constant is independent of pressure and
exhibits a negative temperature effect only in the case
of low surface coverage by the physically adsorbed
hydrogen. Their theory states that, for high surface
coverage (approaching a complete monolayer) the conversion rate constant will vary inversely with pressure
and will increase slowly with temperature. I t also
states that for both high and low surface coverages, the
rate is proportional to the number of active sites. The
theory of Harrison and McDowell (25,26) for high
surface coverage was substantiated by the work of
Chapin and Johnston (13) on the conversion of hydrogen
under pressure at low temperature by heterogeneous
catalysis with chromia-alumina catalysts. The rate
data were first order with respect to composition. The
rate constant varied linearly with the number of gramatoms of chromium in the catalyst and linearly with the
reciprocal of the pressure The temperature dependency of the rate constant was positive as predicted for
the case of high hydrogen surface coverage.
The majority of the work done on the ortho-parahydrogen conversion catalysis has been with static
systems. However, several papers by Weitzel et al.
(3, 55-59) are concerned with catalyst evaluation and
kinetic studies of the conversion in dynamic systems.
The early studies ( 3 ) involved the characterization of
conversion catalysts for use in the large quantity
production of liquid parahydrogen.
Chromium oxide
supported on alumina was found to be effective for rapid
ortho-parahydrogen conversion under streaming conditions. A systematic study (55,59) was undertaken
to find materials which might be more effective
catalysts than chromia on alumina.
Materials were
24

INDUSTRIAL A N D E N G I N E E R I N G CHEMISTRY

chosen, as suggested by the work of Taylor and


Diamond (49),on the basis of magnetic susceptibility
and adsorptive capacity. Various metallic oxides such
as cerium oxide, neodymium oxide, manganese dioxide,
ferric oxide, and chromium oxide, both supported and
unsupported, were studied. The comparative effectiveness of the various catalysts were determined by
measuring the space velocity of each, in STP volumes
per minute per unit volume of catalyst, when converting
normal hydrogen (25% parahydrogen) to 90% parahydrogen with the catalyst at 20' K. and at 15 p.s.i.g.
hydrogen pressure.
The unsupported hydrous ferric oxide proved to be the
most effective conversion catalyst with respect to those
studied (55,59). A space velocity of 330 STP/min.
was achieved with l/g-inch pellets of the material.
Cutting these pellets into about four pieces increased
the space velocity to 750, and grinding into a fine powder
resulted in a STP space velocity of almost 2000 per
min. Further investigation of the effect of particle
size revealed that 40- to 50-mesh catalyst was 15yomore
effective than 10- to 20-mesh but beyond 50-mesh there
was little or no further improvement. An excellent
correlation was found between catalytic activity and
surface area as determined from low temperature nitrogen adsorption by application of the Brunauer, Emmett,
and Teller theory (8). The catalytic activity proved to
be a linear function of the surface area. The factors
that affect the surface area of the catalyst were not
investigated in this study, but it is felt, by the authors,
that variations in the degree and extent of heating during
formation and/or activation are of prime importance.
Based on considerations of the paramagnetic nature of
the materials, the increased activity of the iron oxide
catalyst over the chromium oxide was expected. However, the actual improvement realized was manyfold
greater than expected. I t is suggested that this increased activity might be attributed to the influence of
ferromagnetism since the hydrous ferric oxide catalysts
which exhibited ferromagnetism were the very samples
which showed the highest activity. However, no correlation was attempted among activity, surface area, and
magnetic character for the same sample of catalyst.
Perhaps the ferromagnetic samples had greater surface
areas than those which were free of ferromagnetism.
The importance of proper catalyst activation was also
illustrated in these papers (55,59). I t was shown that
the activity was a function of activation temperature and
time. The optimum conditions for activation were
found to be evacuation at 1 mm. of mercury or less
for 16 hours at a temperature of 120' C. These conditions were maintained immediately prior to an activity
test and the vacuum was broken with pure hydrogen gas.
The catalyst activity was found to decrease if the catalyst
was heated much above 140' C.
A paper by Keeler and Timmerhaus (31) treated the
poisoning of the hydrous ferric oxide catalyst by im-

heat transfer between the catalyst pellets and the walls of the chamber
purities in the hydrogen stream. They showed that
certain impurities such as methane, ethylene, and carbon
monoxide caused temporary poisoning which could be
remedied by reactivation, but others such as hydrogen
sulfide, butyl mercaptan, and chlorine caused permanent
poisoning. It is believed that both the permanent and
the temporary poisons initially reduced the activity by
merely decreasing the hydrogen adsorptive capacity,
but upon reactivation, the permanent poisons actually
react chemically with the catalyst and permanently
alter the catalytic sites while the temporary poisons do
not.
The importance of efficient heat transfer between the
catalyst and the walls of the catalyst chamber is illustrated in a paper by Weitzel (56). Since the paraorthohydrogen conversion at low temperature is endothermic, it is desirable to supply heat to the system so
that the driving force for conversion remains constant.
Otherwise, the catalyst temperature decreases as conversion proceeds and the equilibrium parahydrogen concentration, as determined by the catalyst temperature,
approaches that of the feed. At extremely high conversion rates, this heat transfer problem may influence
the conversion rates.
Weitzel (56) also presents some experimental data on
the effect of pressure on catalyst efficiency. These
data show that, over the pressure range of 20 to 400
p.s.i.g., there is a small but definite increase in the
conversion with increase in pressure. For example,
the rate of conversion of normal hydrogen (25% para)
to 48% para at 76' K. was achieved with an STP space
velocity of 790 per min. at 20 p.s.i.g. and 1200 per min.
at 400 p.s.i.g. This effect has been attributed to the
increased residence time and/or increased physical
adsorption at higher pressures. These data are consistent with the theory of Harrison and McDowell
(25, 26) which predicts that the conversion rate constant will vary inversely with pressure under these conditions. Even though the rate constant decreases with
increases in pressure, the conversion rate can actually
increase since it is the product of the concentration and
the rate constant. This same effect can be seen in the
work of Chapin and Johnston (73). They show that
the conversion rate constant decreases as the pressure
increases but further analysis of the data indicates that
the quantity of hydrogen converted per unit time (conversion rate) actually increases as a function of pressure.
This effect may be better illustrated by an examination

AUTHORS George E. Schmauch is Manager of Contract Research and Alan H. Singleton is Project Manager, both in the
Research and Development Department of Air Products and
Chemicals, Inc., Allentown, Pa. The work reported in
this paper was sponsored by the Research and Technology
Division of thc U. S. Air Force, Wright-Patterson Air Force
Base, Ohio.

of the mathematical expression for a first order reaction


rate.
dc

- dt- = kc

moles/liter/min.

According to this expression, the reaction rate at any


instant is equal to the product of the rate constant, which
is constant at a definite temperature for a given reaction,
and the concentration of the reactant a t that instant (23).
Therefore, an increase in pressure of the system decreases
the reaction rate constant but also increases the concentration of reactants per unit volume. The net
effect is an actual increase in reaction rate.
Two papers by Weitzel (57, 58) are concerned with
the actual kinetics of the flow conversion of ortho- and
parahydrogen. They follow the suggestions by Hougen
and Watson (28) and Yang and Hougen (61) for dealing
with the various mechanisms which may be encountered
in gaseous reactions catalyzed by solid surfaces in flow
reactors. In heterogeneous reactions, the mechanism
consists of several consecutive steps. For the ortho-para
conversion of hydrogen, these steps are:
Di$usion of the hydrogen from the main stream to the surface
of the catalyst.
Di$usion from the surface into the pores of the catalyst.
Adsorption of the hydrogen on the catalyst surface.
Magnetic interaction resulting in the otherwise forbidden paraortho transition.
Desorption of the hydrogen.
Dz$usionfrom the pores to the surface of the catalyst.
Diffusion from the surface to the main stream.

One of these steps will normally be slower than any of


the others and will therefore control the rate of conversion. Weitzel (57) studied the effect of diffusion on the
rate of conversion by following a procedure suggested
by Corrigan (75) which consists of varying the linear
velocity while holding a constant space velocity or
contact time. The efficiency of diffusion between the
main gas stream and the catalyst surface is favored by a
high linear velocity because high velocity results in a
greater turbulence and a thinner laminar film around
the particles than is present at low linear velocity.
For a given space velocity, the linear velocity varies
directly with the length of the catalyst bed. Weitzel
(57) found that for catalyst chambers of the same
diameter but different length, there is no significant
difference in conversion rate for various linear velocities
at constant space velocity. This is evidence that diffusion is not rate-determining under the conditions studied.
However, for catalyst chambers of different diameter and
length, there is a significant difference in conversion
rate for various linear velocities at constant space
velocity. The smaller diameter catalyst chamber tube
(l/d-inch 0.d. as opposed to a/,-inch 0.d.) produced the
VOL. 5 6

NO. 5

MAY 1964

25

high- conversion rate. A RXMOM&


explanation
is that the d e r tube has a higher heat transfer efficiency so that the catalyst is more nearly an isothermal
dollvcrz~r. If the convm&n 3a nut isothcmal, the
equilibrium ortho-para ratio changes as the convertex
temperature changes and tends to reduce the conversion
interval or driving force of the conversion.
The importance of pore diffusion was assessed by

developed by .We&
(57, 58)
etics df the ov&-aU conversion
Tression. for the ortha-parahydqen
..

n e a v i s o t h d cxmditiom.,
.

ThisdeparihKefcom h e a i ~ ~ , e o u lFc ld t from a rate


controlling unimolecillar surface reaction in which one
component is adsorbed more stm&y,.than the other.
.

any decrease in k by increase in pressur, YI GUUWZacted by the increase in the concentration of c which
is a function of pressure. Thus, although the conversion
rate constant is a function of p-e,
the rate of wnversion is essentially independent df prassure.
Thc convemion of normal hydrogen to the equilibrium
farm a t -195O C. ovm y-FerOa, arFe& and a series
of iron qick-zinc &e mixtwrcs heated at several
a m p a a h l n a to produce varying degrees of rmation to

iameinc~iteiareported~Gva~ellPLa2pdSeotttrU).
The m a l b show that -pFetOI is the most efficient of
thcaeaataly~ar
mda-Fe&is uext. Themixedcatalysts
d a w with moderate d v i t i e s . From a study uf t h e
m a p d c propaties of these mamia4 it is concluded
that furomagnetic materials are more effective than
aatiterrotnagn@ticmaterials for inducing an +para
emwemion and
am mon,&ective than paranraepaic materials. Since little io Locayn abotlt the
affscts of fatolaagnctism on catalpiq theaedart,Nggast
that fiazpemagnetic aratlrials ata0 b@ 4 d m d as pa~iorthohgdrogcn converaion c a f a l y a Also, the wonk of
J d and Veith (30)m d Veith (52) illuahate thar as
e
d *tic
field appeared capable of increasiag
the activity of fmomagnetic catalpa for the conversion
of para- to orthohydrogen at ambient tempexaturn.
The e f k t of a magno& M on the law temperature
conversion is not known and presents an area for futwe
investigah.
Several d e s dealing with the hydrogen shift have
a p e d &Iy recently. In a &s
of papers (977). Jiiuyarnm reports his shldies d bothgel# and oxides

I
i
i.

1 '

98

MID U S T R M 1 A N.0'E NO kcEE L 111 N G C H E M I S T R Y

--

I
I

when the reaction proceeds at 10% temperature. The


relative enthalpies of para-, equilibrium, and normal
hydrogen are shown on Figure 2. The differences in
the enthalpies of hydrogen of the various compositions
represents the enthalpy change due to conversion.
Commercial Applications

Figure 3. Storoga losr of w a t a l y u d l i p d hydrogen

of tramition and rare earth elements. The experimental data for magnetically dilute heterogeneous
t that the activiq is proportional to the
ratio &/
as predieted by Wigners theory (60) for
dilute homogeneous catalysis. kff.in the above ratio,
is the effective magnetic moment and r is the distance
of closest approaeh of a hydrogen molecule and the
paramagnetic catalyst ion. I t was both predicted and
observed that the gels of the first transition group elements make the best catalysts for this reaction.
Wakao (54) performed ortho-parahydrogen conversion rate studies at liquid nitrogen temperature using
a NiO on Ab08 multiple impregnation catalyst. After
an evaluation of the diffusion resistances, the surface
reaction rate was correlated on the basis of a mechanism
involving physically adsorbed hydrogen. Separate
evaluation of the catalyst for the hydrogendeutefium
exehange reaction suggested that the dissociative mechanism was not significant at liquid nitrogen temperature.
Wakao (54) studied alumina-supported oxides of
copper, nickel, manganew, and terbium and aluminasupported copper, nickel, and palladium metals and
zirconia-supported terbium oxide. The supported
nickel m e d catalysts showed the highest activity:
a sample p r e p a d by multiple impregnation and hence
containing highly dis-d
nickel was 90 times more
active than commercial 19% chromia-alumina. I n
relation to -atomic radius and magnetic moment, the
supported nickel metal was over 10 times more active
than the oxides. This was explained on the basis of the
oxidation of Ni+z on the catalyst surface to Ni+L by
chemisorbed oxygen, resulhg in a drastic increase in
r, the distance of closest approach of the hydrogen
molecule to the surkce nickel ion. The results of
hydrogendeuterium exchange experiments at liquid
nitrogen temperature indicated that the conversion
mechanism probably does not involve a dissociative
mechanism.

Y,=

Purpose of Conversion

The commercial, military, and aerospace

hydrogen conversion stem from the heat of conversion

Liquid hydrogen is produced commercially to supply


the rocket engine testing programs and the hydrogen
fueled rocket vehicles presently being developed by the
Air Force and by NASA. Smaller quantities are
required for low temperature research. The succesnful
transportation and storage of liquid hydrogen depends
on limiting the heat addition to the transportation or
storage system, since the only feasible means of dissipating such heat is by boiling and venting a portion of the
stored fluid. When equilibrium hydrogen a t room
temperature (normal hydrogen -Z5yo para) is cooled
and liquefied, its ortho-para composition remains
essentially unchanged. After liquefaction the conversion
reaction proceeds at a significant rate, resulting in the
release of the heat of the conversion reaction within the
liquid. The rate of the uncatalyzed reaction was
measured by Scott (40). Results of calculations by
Weitzel reported in (47) are shown in Figure 3. The
upper curve shows the composition of the residual liquid
as a function of time, while the lower curve indicates
the quantity of liquid which has been boiled off due to
conversion. For example, after 100 hours of storage,
40y0 of the original charge to the tank has evaporated
due to the dissipation of conversion energy withiin the
tank. This loas is an order of magnitude higher than the
typical loss that would be incurred by heat leak through
insulation and tank supports.
In order to limit the boiil-off to a lower level, it is
n-ry
to fill the tank with liquid hydrogen which
has already been converted to the equilibrium composition. The first commercial plants to produce tonnage
quantities of hydrogen (51) used chromium oxide on
alumina catalyst to produce liquid hydrogen at close to
equilibrium compwition. Conversion was accomplished
by divertinp he hydrogen stream through packed beds
of the catallst at various temperature levels during the
liquefaction process, thereby extracting the heat of
conversion at as high a temperature as possible.
Later plants have used hydrous ferrous oxide gel as a
catalyst. This material is approximately five times
more effective as used than the chromium oxide, and
permits substantially smaller quantities of catalyst to be
used.
Many of the next generation of aerospace vehicles will
use hydrogen as a fuel or propellant. The use of hydrorren &el c&s for rreneratina eiectricitv is fast ba~o&
a
reality for both b i m a r y &d a&
power syste&
(36,44). The use of hydrogen fuel cells for powering
electrically driven military vehicles is under serious conk sideration a t this time (2). Many of the vehicles. or
systems in which hydrogen is used will have refrigeration
requirements. In those cases where hydrogen is used

V O L 5 6 NO. 5 M A Y 1 9 6 4

27

Improved catalyst for ortho-parahydrogen conversion is available

at a higher temperature than the storage temperature,


it is desirable to utilize the refrigeration capability of the
hydrogen.
In general, where relatively large quantities of hydrogen are required, it will be stored as a liquid with a
composition close to 100% para. If the hydrogen is
used in a refrigeration circuit which does not contain a
catalyst, the temperature-enthalpy path will follow the
para- curve of Figure 2. However, if a catalyst is included in the system, causing the temperature-enthalpy
path to follow the equilibrium curve of Figure 2, a
significantly greater quantity of refrigeration czn be
provided by the hydrogen at low temperature.

E X P E R I M E N T A L PROGRAM
Catalyst Properties

As pointed out in the literature review, the physical


approach utilizing heterogeneous catalysis appeared to
be the most promising approach to speeding up the conversion reaction. It was concluded on the basis of theoretical considerations that the following properties are important : high physical adsorptive capacity for hydrogen ;
high concentration of a para-magnetic component on the
surface of the catalyst; and a large magnetic moment
associated with the para-magnetic component. The resulting catalyst which has been developed, APACHI-1~
is the successful combination of these properties.
Measurement of Catalyst Activity

Catalyst activity is measured in a flow process at


constant temperature. An equation describing the
conversion has been derived (74)and is shown below :

The factor ktF/(l - C,)is a function of temperature and


pressure. Thus, if the quantity converted, F(Ci - Co),
is plotted against the log mean driving force,

c, - e,
a straight line with slope k t F / ( I - C,) is obtained. The
activity of the catalyst for hydrogen conversion is directly
proportional to this slope.
The experimental apparatus for measuring conversion
activity has been described in detail elsewhere (14, 39).
The basic approach involves measuring C, and C, at
various flow rates while maintaining catalyst temperature and the hydrogen pressure constant.
28

I N D U S T R I A L A N D E N G I N E E R I N G CHEMISTRY

Factors Affecting Catalyst Activity

I n addition to the basic measurements of catalyst


activity, a number of measurements were made of
factors related to the mechanism of the reaction. Experiments were made to determine the importance of the
following on catalyst activity : hydrogen dissociation ;
catalyst particle size; catalyst surface area; and contaminant nitrogen adsorption.
Hydrogen Dissociation

As pointed out in the literature review, dissociation of a nonequilibrium mixture of hydrogen will
cause it to recombine in an equilibrium condition. I n
order to determine whether the APACHI-1 catalyst was
capable of dissociating hydrogen, the hydrogendeuterium exchange reaction on the catalyst surface
was studied. If the catalyst was effective in dissociating
molecular hydrogen and molecular deuterium, the
products of the recombination would include hydrogen
deuteride according to the reaction shown below :

Hz+Dza2HD
If the dissociation reaction were not catalyzed, no
hydrogen deuteride would be detected. Thus, the
results of this exchange reaction study were to indicate
whether the low-temperature conversion was effected
by the physical (magnetic) mechanism or a combination
of the physical and chemical (dissociation) mechanisms.
This study required an analytical system for the
detection and estimation of the components, hydrogen,
deuterium, and hydrogen deuteride. This was accomplished by using a gas-solid chromatographic technique
as suggested by Moore and Ward (35). This technique
permits the separation of hydrogen, deuterium, and
hydrogen deuteride on a specially treated alumina
column at liquid nitrogen temperature using helium as
a carrier gas. A thermal conductivity detector was
used to indicate the separated components.
The experiments were performed by flowing a mixture
of 80yoHz and 20% Dz at a constant flow rate through a
catalyst chamber containing the APACHI catalyst at
various temperatures. These experiments indicated
that the exchange reaction was catalyzed by this improved catalyst even at liquid nitrogen temperature.
The exchange reaction was observed at the following
temperatures, -320 F., -297 F., and 212 F. The
exchange experiments showed that the exchange reaction
exhibited a positive temperature coefficient-the exchange
rate increased as the temperature increased. This was
illustrated by experimental data Jvhich showed an
increase in the HD concentration in the effluent stream
as the temperature was increased while maintaining a
constant H2-Dzflow rate. For example, a flow rate of
450 cc./min. of an 807G H z and 2070 Dz mix through a

I(

0.49-gram sample of catalyst produced 3.3y0 HD in the


effluent at -320' F. and 29.5% H D in the effluent at
212' F.
These experiments indicate that the low temperature
ortho-para conversion of hydrogen as effected by this
improved catalyst is the result of both a physical and a
chemical mechanism. However, the contribution of the
chemical mechanism to the over-all conversion activity
is considered to be quite small. This conclusion was
reached as a result of the ease with which the exchange
activity of the catalyst at cryogenic temperatures could
be poisoned. Exposure to trace quantities of moisture
as contaminant in the hydrogen gas completely inhibited
the ability of the catalyst to exchange hydrogen and
deuterium at -320' F. but did not seriously alter the
ortho-para conversion activity at this temperature.
This suggests that at -320' F., only a small fraction of
the "active sites" are capable of effecting conversion by
hydrogen dissociation and that the majority of the sites
are converting as a result of magnetic interactions with
the molecular hydrogen.
Similar exchange studies were performed with a
hydrous ferric oxide gel catalyst and there was no indication of hydrogen-deuterium exchange even at 212'
F. This suggests that the iron gel catalysts are not
capable of dissociating molecular hydrogen at temperatures up to 212' F. and that the ortho-parahydrogen
conversion effected by this material is strictly the result
of a physical (magnetic) mechanism. These conclusions
are compatible with the fact that the iron gel catalysts
are not effective conversion catalysts or normalizers at
ambient temperature while the APACHI catalysts are
effective normalizers.
Effect of Catalyst Particle Size on Conversion Activity

A study of the effect of catalyst particle size on the


conversion activity for the APACHI catalyst was performed. The catalyst was separated into various mesh
size ranges by sieving. A U. S. Standard Sieve Series
was used to accomplish this sizing. Samples of each
size were activated by purging with high-purity, dry
nitrogen at 150' C. for two hours, and the conversion
activity was determined at -320' F. and 100 p.s.i.g.
T h e results are tabulated in Table I.
These data indicate that the activity increases as the
particle size decreases and this effect is attributed to
internal or pore diffusion limitations in the catalyst.
By reducing the particle size, the average pore length is
reduced and the diffusion characteristics are improved
until a point is reached beyond which no further improvement is observed. This particle size is considered
to be that which exhibits no limitation to conversion
activity as a result of diffusional restrictions. In effect,
the total surface is available for conversion, whereas in
larger particles, the interior surface cannot function
efficiently because of mass transfer limitations.
Effect of Surface Area on Conversion Activity

A special technique for increasing the surface area,


and therefore the activity of a catalyst, has been qualita-

tively examined. The effects of this special technique


are illustrated in Table I1 where the B.E.T. surface
area and the activity, relative to the standard iron gel,
are presented for both specially treated and untreated
samples for a given activation temperature. The special
samples are designated as (- 2) while the control samples
are marked (- 1).

TABLE I.

EFFECT

O F PARTICLE S I Z E ON A C T I V I T Y
Activity
Relative to
Iron Gel (70to 80-Mesh)

M e s h Sire
Range

Particle Size
Range (Microns)

25-30

7TO to 590

30-50

590 to 297

3.58

50-70

297 to 210

4.11

3.25

yoIncrease
in Activity

+ 10
4-15
-0

70-80
TABLE I I .

210 to 177

4.08

EFFECT OF SURFACE AREA ON A C T I V I T Y

Catalyst
Sample No.

Surface
Area,
m2/Gm.

Activation
Temp., C.

Relative
Activity

52- 1

479

145

4.60

52-2

542

145

5.73

19-A- 1

32 1

150

1.90

19-A-2

273

150

1 .56

26- 1

256

130

4.59

26-2

239

130

5.13

26- 1

256

155

5.60

26-2

239

155

6.53

28- 1

309

250

4.44

28-2

376

250

5.27

30- 1

555

150

2.19

30-2

...

150

4.26

yo Change
in Activity
+24.6
-17.9
+11 .8
+16.6
+i8.7
+94.5

This tabulation represents four samples of supported


catalysts with different concentrations of promoter and
one unsupported catalyst. The catalyst systems represented by numbers 52 and 28 show both an increase in
surface area and an increase in activity as a result of this
special treatment. Samples of number 26 exhibit an
increase in activity but a paradoxical decrease in surface
area. In the case of 19-A which was the unsupported
sample, the treatment resulted in a decrease in both
surface area and activity. Number 30 exhibited a
very significant increase in activity as a result of the
special technique but the surface area of the treated
sample is not yet available for comparison with the
untreated sample.
Although the catalyst data resulting from special
treatment do not show a clear-cut correlation between
VOL. 5 6

NO. 5

MAY 1964

29

uusian

surface area and activity, only the unsupported catalyst


(19-A) experienced a decrease in both. The four supported catalysts showed definite increases in activity
after treatment, although the surface area did not
d e c t this gain in every case. It appears that this
special preparation technique influences the pore size
distribution as well as the surface area. Further examination of this procedure is expected to broaden our
understanding of this effect and ultimately provide
additional improvements in catalyst activity.
ERM of Nih0g.n Adsorption

Knowledge of the influence of adsorbed impurities


upon catalyst electiveness is of theoretical and practical
interest. The ortho-parahydrogen conversion mechanism in heterogeneous catalysis involves the adsorption
of hydrogen. Any adsorbate which competes with
hydrogen may be expected to reduce the catalyst effectiveness unless this competing adsorbate is also a catalyst.
The extent and character of the reduction due to adsorption of an impurity may illuminate the catalyst behavior.
Experiments were carried out in which hydrogen containing trace quantities of nitrogen were passed through
a catalyst bed. Simultaneous measurements were
made of outlet hydrogen composition and nitrogen
concentration. The experimental method and the
detailed mults are reported in (33). Typical results
are shown in Figure 4. Outlet parahydrogen concenhation at constant flow rate is plotted as a function of the
aotal quantity of hydrogen which has passed through the
bed. Plotted on the same graph is the outlet nitrogen
,concentration. The outlet parahydrogen concentration
dmeasea from the initial equilibrium concentration
until the catalyst is saturated with nitrogen, as indicated
by the sudden increase in nitrogen concentration in the
&ent
at six cubic feet of hydrogen. At this point,
steady state is achieved, and the performance of the
Ftalyxt becomes constantat the reduced level of activity.
The relationship between the portion of the original
catalyst B.E.T. surface not covered with adsorbed
nitrogen and the residual catalytic activity is presented
in Figure 5. If the reduction in orbparahydrogen
30

I N D U S T R I A L AND ENGINEERING C H E M I S T R Y

conversion activity were diroctly related to the reduction


in B.E.T. surface due to nitrogen adsorption, the diagonal
straight line would represent the nitrogen adsorption
effect. It is seen that the falloff in activity is much
greater than such a linear effect. A coverage of 10%
of the catalyst with nitmgen reduces the activity to
about 30y0 of the original; a coverage of 25% reduces
the activity to 20% of the original; a coverage of 50%
of the surface reduces the activity to about 10% of the
original. It may also be noted that the curve of Figure
5 empirically correlates the effect of adsorbed nitrogen
upon three different catalysts which vary greatly in
B.E.T. surface area and in intrinsic activity.
These experiments show that a relatively small portion
of the catalyst surface area is responsible for most of the
catalytic activity. The active sites which bring about
most of the ortho-parahydrogen conversion are
apparently also the sites which preferentially adsorb
nitrogen. Such adsorbed nitrogen drastically reduces
the catalyst activity. The line of Figure 5, drawn
through the data points and extended, rep-ts
the
viewpoint that the loss in catalytic activity is exponentially related to the fraction of the catalyst surface which
is covered with adsorbed nitrogen. It appears that the
loss in catalytic activity as a function of surface coverage
by nitrogen is similar to the decrease in differential
heats of adsorption as a function of coverage, both
phenomena being related to site energy distribution.
Further theoretical and experimental work is in progress
for the examination of these relationships.

PRESENT STATE OF THF ART


Through the experimental program herein described,
it has been p s i b l e to develop an understanding of some
or the important variables of a hydrogen conversion
catalyst. Through proper attention to the significant
variables, a catalyst has been fabricated which is about
10 times more active than the iron gel catalyst on a
weight basis. Table If1 shows the relative activities of
chromium oxide, iron gel, and APACHI catalyst as
determined in our laboratory.

t
C
t

= static reaction rate w m m t (fin.-')


= male fraction of parahydmgen.
= residence t h e in catalyst chambcr(min.)

hbseripti:
at time t.

= DFC+

= accuring at time ecm


e, m = equilibrium d u e
= condition at catalyst inlet
i
o

condition at'eatalystoutkt

REFERENCES

(1) HarkDem%, J. &.Llk. Sr, Ir, 2850 (1932).


(2) Aui. -Air
h m ! a~
& Qb'mkds, I=.. I h f N m h e , Fcb

196%
(3) Budct, P. L..Wcitnl, D.W..C o d y ,T. W.. Pnr. -8.

Wr.&f.,

1. 285

(1954).

TABLE 111.

ORTHO-P4RA CONVERSION CATALYST


ACTIVITIES

(4) Ilamhoc8cr,K.F.. Fktek, P.. N d . ,17,181 (1929).


(5) Bomhodcr,K.P., H m d , P., E. w.Chm., M, 113 (1929).
(6) ~ , K K . F . F u t . * A . W , D I Z l l (1931).
(3 ~ ~ ~ , R F . , F u t u . A . , T n v . ~ ~ S r . , U , U 2 . 5 6 1 ( 1 9 3 2 ) .

(8) ~ ~ , S . , E ~ ~ t , P . H . , T ~ , K . J . h . ~ ~ . , ~ ~ ( 1
(9) BU~MDS,R A, X&i& i X
&,
1, No. 2, SO6 (1960).
(lo) Iu.,1. No. 3 (IWO).
(11) IM., 1. No. I(1960):
(12) C.pmn.P.C.,63.Ss=.~.Bn;d*r.~,'~(1995).'
(13) Chapin, D.
H. I*. J . h.s*,YS, 2406 (1957).

%,.a.

APACHI
kon gel

2.4
0.27
0.054

Chromio on alumina

100

IO0
IO0

160
160

(14) CW R.G.. Kuei;t.. I.F.,


of tbc Panonbo 8hift dH-"

A,, &hmawh, 0. E., " I n ~ t b a t i w


.$
Roduca
ITd -.I=.,
Sum

160

Figure 6 shows the outlet concenhation of an isothermal bed as a function of .catalyst weight for three
catalysts : APACHI, iron gel, and chromia on alumina.
.

. :,

vc literan& reV& has provided a thorough


omenon and has indicated that heterogened.
is the most promising technique
?#he rate at which this conversion can be effec
:.,,
&e+l applicationqof hydrogen.coryzsion are
e a t e r e d k d ortho to paraconves8ion f& the purpose
d producing a product with high para concentration
in liquefaction plants. Military and F c s p a c e applications prescndy
the
of
low-

*
a
etas which iud

WOMEMCUTURE
.X = fraction parahydmgcn
c
= gmerpl WncentratiOIl ofrcacting+
E
= hydrogen flow ntc (g./min./g. of catslyrt)
VOL 66

NO. 5

M A Y 1964

31

You might also like