Professional Documents
Culture Documents
257᎐265
S.J. Bull a,U , R.I. Davidsona , E.H. Fisher a , A.R. McCabe b, A.M. Jones b
a
Department of Mechanical, Materials and Manufacturing Engineering, Uni¨ ersity of Newcastle, Newcastle-upon-Tyne, NE1 7RU, UK
b
AEA Technology plc, Harwell, Oxfordshire, OX11 0RA, UK
Abstract
Glass-filled polymers are known to produce considerable wear on the screws and barrels of injection moulding machines and
several coatings and surface treatments have been used to combat this ranging from chromium plating to High Velocity Oxy Fuel
ŽHVOF. WCrCo. However, quantification of the benefits of such coatings and treatments has proceeded on an ad hoc basis and
there is little data available to aid selection. In this study we have developed a novel wear tester to simulate the conditions of
wear which occur in the barrel of an injection moulding machine and used it to rank the coatings and surface treatments which
are often used in this application. The tester concept is similar to that of the ASTM rubber wheel abrasion test except that the
rubber wheel is replaced by a steel wheel heated to a fixed temperature Žup to 200⬚C. by hot air and the sand is replaced by
pelletised plastic. The plastic sticks to the hot wheel and is softened and dragged past the sample which is also heated to the same
temperature. The sample is pressed against the plastic-coated wheel with a small load and a wear scar is produced at the contact
point by a combination of abrasive wear from the glass filler and sliding wear from the plastic. In the case of glass filled nylon the
best surface treatments were based on nitriding. Coatings tend to not perform very well in this test but are still better than
uncoated steel. This paper discusses these results and the mechanisms of failure observed in the test. 䊚 2000 Elsevier Science
S.A. All rights reserved.
0257-8972r00r$ - see front matter 䊚 2000 Elsevier Science S.A. All rights reserved.
PII: S 0 2 5 7 - 8 9 7 2 Ž 0 0 . 0 0 6 9 7 - 6
258 S.J. Bull et al. r Surface and Coatings Technology 130 (2000) 257᎐265
surface just below the sample. During set-up, the tem- Table 1
perature of the wheel is measured just above the sam- Coatings and surface treatments investigated
ple to ensure that adequate heating occurs. The tem- Sample 100 g Vickers Treatment thickness
perature variation between these two positions was 2⬚C hardness Žm.
which is much smaller than the accuracy of control of Žkgfrmm2 .
the disc surface temperature. The sample is held in a
Untreated 391 ᎐
holder connected to a pivot loading system which can Nitrided 1284 20 m white layer
press it against the surface of the hot disc in an almost 250 m diffusion layer
vertical orientation. The sample and its holder can be Nitridedq nitrogen 1759 20 m white layer
heated independently of the disc up to 200⬚C by car- ion implantationa 250 m diffusion layer
tridge heaters and its temperature is controlled using a Laser hardened 477 Not measured
WCrCo coated ŽAPS. 510 120 m
thermocouple on the rear side of the sample to "2⬚C. Chrome plated 1125 25 m
A hopper system was designed to contain the pel- a
letised plastic feedstock. The heated wheel passes Ion implantation took place in the Blue Tank at Harwell. The
typical ion beam conditions were 76 keV N2q for a dose of 4 = 10 17
through the hopper where it can heat the plastic charge ionsrcm2 .
but the walls of the hopper remain cold. The hot
polymer sticks to the wheel and is dragged through the
tact provides only a very small contribution to the total
contact causing wear to take place. A scraper is
wear; for nearly all the samples tested in this study
mounted at the opposite side of the rig to the sample
metal᎐metal contact early in the test leads to the
to remove excess polymer that remains adhered to the
transfer of small patches of disc steel to the coated
rotating disc. The flow rate of the polymer feedstock is
surfaces which are subsequently removed by abrasion
controlled by the size of the hole at the bottom of the Ži.e. by the glass filler..
hopper, the surface roughness of the disc, the disc
temperature and the length of the hot disc passing
through the hopper which is in contact with the feed-
stock. A number of tests were undertaken to determine 4. Experimental
the feedstock flow rate in the test by capturing all the
material which has passed through the contact in a 4.1. Test samples
fixed time and it was found that the flow rate was
relatively uniform at ; 3 kgrh. If the pellet size is P20 steel ŽCarrs Tool Steel. was machined into 50-
increased above 5-mm diameter there is a high chance mm diameter, 10-mm thick discs and ground flat to give
of blockage and the test becomes unreliable. It is a surface roughness of less than 0.5 m Ra prior to
therefore likely that a different polymer feed system coating. A range of coatings and surface treatments
will be necessary. To avoid this problem the polymer were obtained from commercial suppliers or produced
feedstock was reduced to less than this size using a in-house at AEA Technology ŽTable 1.. Three discs
mechanical grinder of the type used for recycling plas- were supplied for each treatment to give an indication
tic materials. of the variability in performance of a given treatment.
Loading of the coatings and surface treatments dur- Laser treatment was used to harden the surface of
ing the test is intermittent in that individual polymer the barrel steel disc but since this had a relatively low
pellets travel through the contact at randomly, closely hardenability the increase in hardness is low ŽTable 1..
spaced, times Žapprox. every 0.5᎐1 s.; there is a small The barrel steel is suitable for nitriding and a typical
delay between them and in extremes there will be nitriding treatment was used to produce a case depth
metal᎐metal contact. This occurs for less than 5% of of 0.25 mm with a 20-m white layer which consists of
the total test time in the early stages of the test but a mixture of ␥⬘ and iron nitrides. The ion implanta-
after a short time a layer of polymer adheres to the tion treatment injects nitrogen into the top ; 0.2 m
complete surface of the disc and metal᎐metal contact of the nitrided sample surface. The implanted layer is
ceases until this is removed. This can also occur between confined within the white layer and the treatment leads
the screw and the barrel in an injection moulding to some densification of this layer increasing its hard-
machine as mentioned previously. The contact stress ness ŽTable 1.. The sprayed WCrCo coatings were
varies between 40 MPa when an unmelted polymer deposited by Air Plasma Spraying ŽAPS. from
pellet passes through the contact and 28 MPa when the WCr6%Co powders.
polymer-coated metal disc is in contact with the sam-
ple. The wear process will thus include some sliding 4.2. Test conditions
wear and cyclic loading Žwhich could lead to contact
fatigue failure. as well as the abrasive and chemical Test parameters used for comparing all of the sam-
wear assessed by previous tests. The metal᎐metal con- ples are summarised in Table 2. It is essential to run
260 S.J. Bull et al. r Surface and Coatings Technology 130 (2000) 257᎐265
Table 2 for all coatings and surface treatments ŽFig. 3.. The
Test parameters used in the comparison testing
samples which show the lowest weight loss are nitrided
Parameter Setting steel with the nitrogen implanted nitrided samples close
behind. The worst performance was observed for the
Disc material Steel laser hardened, APS WCrCo-coated and chrome-
Disc diameter 203 mm
plated samples which all performed worse than the
Disc width 10 mm
Disc rotation speed 84 rev.rmin uncoated material. However, there was considerable
Disc surface temperature 200 " 10⬚C scatter in the mass loss values between samples with
Sample diameter 50 mm the same coating which obscures the variations between
Sample thickness 10 mm samples with different coatings and surface treatments.
Sample temperature 200 " 2⬚C
The large scatter in the mass loss data is caused by a
Load at sample 21.4 N
Polymer feedstock Black plastic Žglass-filled nylon. number of factors. Firstly, the samples typically weigh
Pellet size - 3 mm ; 200 g and the mass loss is ; 1 mg which requires a
Feed rate 3 kgrh very accurate balance with a high load capability for its
Test duration 30 min assessment. A five decimal place balance was used
here. Secondly, contact of polymer occurs over a wide
stripe across the sample and not just in the wear scar.
the rotating disc until its temperature reaches 200⬚C
At the end of the test the polymer that is in contact
before any polymer is placed in the hopper. A flow of
with the test sample becomes adhered to it. Measure-
material through the hopper and onto the disc was
ments taken immediately after the test without clean-
established before the sample was loaded against it. A
ing thus show large weight gains for all samples. Ultra-
load of 10 N was used on the end of the load hanger
sonic agitation in a solvent such as acetone can remove
which corresponds to a load of 21.4 N at the discrsam-
the majority of this superficial polymer and the weight
ple contact. Lower loads led to excessive vibration as
gains now become weight losses but the extent of
almost solid polymer pellets were dragged through the
weight loss depends on the effectiveness of the clean-
contact. Higher loads showed an increasing fraction of
ing process. Cracked coatings such as hard chrome
steel-on-steel contact as the softened polymer was
plate and porous sprayed coatings trap polymer in the
squeezed out of the contact. To minimise vibration of
coating which is almost impossible to remove by this
the loading arm and high amplitude oscillations caused
method. Thus, the scatter in the test data is sufficiently
by severe impacts between disc and sample in the early
large to change the ranking order so a different way of
stages of the test, a simple backstop behind the sample
assessing wear is necessary.
holder was used to limit the loading arm motion. The
Surface profiles were measured across the wear scar
backstop was set at a distance just less than the diame-
in the sliding direction and along the centre line of the
ter of the feedstock pellets behind the sample holder
wear scar perpendicular to the direction of sliding, as
ᎏ the precise position was not found to affect the
this is the place where wear is expected to be greatest.
measured results.
Traces were long enough to ensure that unworn mate-
To measure the worn volume it was necessary to use
rial was traced at each edge of the scar. Measurements
profilometry. Mass loss measurements were made for
taken across the scar were very variable, reflecting the
all the samples but these are very inaccurate since
some plastic sticks to the surface of the materials and
the wear scars Žand hence the mass changes they
create. are very small Žsee later.. In many cases there
was insufficient reflectivity to allow optical profilometry
and thus no direct measurements of wear volume were
made by this technique. However, stylus profilometry
was used to measure the change in roughness of the
samples before and after testing and an estimate of the
worn volume has been made based on these values Žsee
later..
5. Results
After the 30-min test period, a mass loss was recorded Fig. 3. Mass loss date for all coatings and treatments.
S.J. Bull et al. r Surface and Coatings Technology 130 (2000) 257᎐265 261
Table 3
Roughness parameters for all the coatings and surface treatments investigated in this study
fact that some samples wore in a uniform fashion but fined wear scars of ) 10 m maximum depth were
others become grooved after wear testing. The precise observed. In this case stylus profilometry across the
shape of the profiles in the direction of sliding thus scar in the direction of sliding confirms that the shape
depends on whether the stylus of the profilometer of the scar is a cylindrical cap with the same radius as
interacts with such a groove. Also, in many cases it was the disc used to make it. However, it was not practical
almost impossible to identify the wear scar in the traces to achieve so much wear on the coated samples in a
parallel to the sliding direction as the amount of wear reasonable test time. Furthermore, the extent of wear
was too small. in this case is much greater than is usually observed for
Some very long wear tests Žgreater than 10 h. were injection moulding machine components.
undertaken on untreated material where very well de- Close examination of the wear scar as a function of
Fig. 4. Schematic comparison of roughness traces before and after testing; Ža. low wear samples Ž R a Žinitial . ) R a Žfinal. and Žb. high wear
samples Ž R p Žinitial ) R p Žfinal...
262 S.J. Bull et al. r Surface and Coatings Technology 130 (2000) 257᎐265
test time for several different surface treatments shows levelling depth, is a measure of the displacement of the
that the following series of events is often observed mean profile position from a line drawn through the
which is very similar to what is seen on the screws and highest peaks in the profile ŽFig. 4b.. If we assume that
barrels of injection moulding machines. The samples the highest points on the profile are effectively unworn
were generally quite rough and machining marks or Že.g. outside the scar at each end. then as wear pro-
coating features were clearly visible on the surface of gresses the mean line will move further away from
the test sample prior to testing. In the early stages of these points and the value of R p will increase. Thus,
the test, wear is confined to the peaks of this rough- we have
ness, these are polished flat and the surface roughness
is reduced. Scratch marks from the filler are clearly ds R p Ž final . y R p Ž initial . Ž2.
visible with reflected light microscopy in the smoothed
regions. The change in surface roughness thus gives a Fig. 5 shows plots of the measured d value for all the
good indication of the amount of material which is samples calculated using Eqs. Ž1. and Ž2.. It is clear
removed. However, as the test progresses wear be- that when d is negative as determined from Eq. Ž1. it is
comes much more severe in localised regions of the positive from Eq. Ž2. and we thus have a method for
scar. This is due to the detachment of larger wear calculating the wear depth for all samples. The wear
particles, perhaps by fracture or fatigue failure of the depths presented in Fig. 6 can be used as a ranking
surface layer. These large debris particles can be method for all the samples. The ranking order is not
dragged through the contact making deep scratches or the same as that given by weight loss. In general, the
just leave deep holes where they are removed. In this nitrided samples give the best performance, followed by
case the roughness of the surface will increase. Again the laser hardened material. Of the coatings there is
the change in surface roughness can be used as a guide little difference between the chromium-plated samples
to the progress of wear.
Accordingly, wear measurements were made from
the profiles perpendicular to the sliding direction which
were taken along the centreline of the scar for a total
scan length of 4 mm. Scans were previously taken in
the same place prior to testing.
Roughness data for the tests is presented in Table 3.
In cases where there was not very much wear the
asperities in the wear scar are smoothed when com-
pared to their untested shape and the R a value tends
to decrease ŽFig. 4a.. In this case a measure of the wear
depth, d, is given by
Fig. 6. Measured wear depths for all the samples investigated. Fig. 7. Calculated wear factors for all samples.
Table 4
Measured wear depths and calculated wear volumes and wear factors for all the coatings and surface treatments investigated
Fig. 8. Scanning electron micrographs of the hard chrome plated sample after wear testing Ža. wear scar showing scratches from abrasion by
ehglass filler Žb. higher magnification image of the edge of the wear scar showing that polymer has been forced into the microcracks in the plating
and that the microcracks have become opened during wear.
264 S.J. Bull et al. r Surface and Coatings Technology 130 (2000) 257᎐265
Fig. 9. Scanning electron micrographs of the nitrogen implanted nitrided steel sample Ža. general view of the wear scar showing polymer
deposition on the surface and some wear by microfracture in the centre of the scar and Žb. higher magnification image showing pits formed by
fracture together with some grooves formed by abrasion.
S.J. Bull et al. r Surface and Coatings Technology 130 (2000) 257᎐265 265
surface of asperities and propagate into the material at likely performance of other coatings and surface treat-
a shallow angle. After propagating a small distance the ments in the application.
cracks turn back towards the surface and a pit is
formed. Such micropitting is known to be influenced by
the relative hardness of the asperities, so increasing 7. Conclusions
their hardness to prevent plastic deformation is an
advantage. Increasing the coating hardness will also A wear test has been developed to simulate the
help to improve the abrasion resistance of the softer conditions inside the barrel of an injection moulding
coatings. However, the process is not controlled by machine. The test has been successfully applied to a
hardness alone and surface roughness, surface trac- range of coatings and surface treatments which are
tions Žfriction. and surface chemistry are all likely to be widely used for barrels and screws of such machines.
important. Further work is underway to see if these The best performance was achieved by a combination
factors are also important in dictating the wear be- of nitriding and ion implantation. This is probably due
haviour of all the coatings tested in this study. to the fact that the surfaces undergo a combination of
The fact that the nitrided and laser-hardened sam- abrasive wear and contact fatigue during the test. This
ples perform so well in the test confirms that failure is expected to be similar to what occurs in the injection
occurs by a microfracture mechanism Žperhaps with a moulder barrel and so the results of the test provide
fatigue component.. The more brittle hard coatings are reasonable data on which to base coating or surface
easier to fracture and create hard wear debris which treatment selection.
can do abrasive damage to the sample surface as well
as the glass filler in the polymer. These coatings will
thus offer some wear protection but are not as good as Acknowledgements
the nitrided surfaces which combine relatively high
hardness with good ductility. If the implantation treat- The authors would like to thank Chris Aylott and
ment can increase the hardness of a nitrided layer Jon Tuck for producing some of the experimental data.
without embrittling it then it will further improve per-
formance as is observed. Ion implantation treatments References
also introduce a compressive stress which is known to
close surface cracks and improve fatigue performance. w1x G. Mennig ŽEd.., Wear in Plastics Processing, HanserrGardner,
There remains the question of whether the test re- Munich 1995.
w2x W.D. Mahler, PhD. Thesis, TU Darmstadt, 1975.
sults reflect what is likely to be seen in service. The w3x G. Paller, B. Matthes, W. Herr, E. Broszeit, Mat. Sci. Eng.
wear damage certainly looks very similar but whether A140 Ž1991. 647.
the balance of the observed mechanisms is correct is w4x M. Heinze, G. Mennig, G. Paller, Surf. Coat. Technol. 74r75
difficult to say. Relative wear performance of the com- Ž1995. 658.
w5x M. Heinze, Surf. Coat. Technol. 105 Ž1998. 38.
plete range of coatings and surface treatments for
w6x G. Berg, C. Freidrich, E. Broszeit, M. Cremer, Int. J. Polym.
injection moulder screws and barrels in service is not Process. 14 Ž1999. 122.
known. However, tests at AEA Technology have shown w7x S. Kalpakjian, Manufacturing Engineering and Technology,
that nitrogen ion implantation can increase the wear 3rd, Addison-Wesley, Menlo Park CA, 1995.
life of injection moulder screws by a factor of 4. Inter- w8x G. Paller, M. Schwab, Kunststoffe-Plast Europe 86 Ž1996. 792.
w9x ASTM G65-94. Standard test method for measuring abrasion
estingly, the wear rate of nitrogen implanted nitrided
using the dry sandrrubber wheel apparatus, ASTM 1994.
steel is only a quarter of that observed for the nitrided w10x S.J. Bull, J.T. Evans, B.A. Shaw, D.A. Hoffman, J. Eng. Tribol.
steel alone. This strongly implies that the test results ŽProc. Inst. Mech. Eng. J.. 213 Ž1999. 305.
are valid and can give useful information about the w11x S.J. Bull, T.F. Page, J. Mater. Sci. 26 Ž1991. 3086.