wear characteristics of uhmw polyethylene: a method for accurately measuring extremely low wear...
TRANSCRIPT
Wear Characteristics of UHMW Polyethylene: A Method for Accurately Measuring Extremely Low
Wear Rates
H. McKELLOP, I. C. CLARKE, K. L. MARKOLF, and H. C. AMSTUTZ, Division of Orthopedics, Department of Surgery,
School of Medicine, University of California, Los Angeles, California 90024
Summary
The wear of UHMW polyethylene bearing against 316 stainless steel or cobalt chrome alloy was measured using a 12-channel wear tester especially developed for the eval- uation of candidate materials for prosthetic joints. The coefficient of friction and wear rate was determined as a function of lubricant, contact stress, and metallic surface roughness in tests lasting two to three million cycles, the equivalent of several years’ use of a prosthesis. Wear was determined from the weight loss of the polyethylene specimens corrected for the effect of fluid absorption.
The friction and wear processes in blood serum differed markedly from those in saline solution or distilled water. Only serum lubrication produced wear surfaces resembling those observed on removed prostheses. The experimental method provided a very accurate reproducible measurement of polyethylene wear. The long-term wear rates were proportional to load and sliding distance and were much lower than expected from previously published data. Although the polyethylene wear rate increased with in- creasing surface roughness, wear was not severe except with very coarse metal sur- faces.
The data obtained in these studies forms a basis for the subsequent comparative evaluation of potentially superior materials for prosthetic joints.
INTRODUCTION
In the majority of the total joint replacements currently in use, one component of ultra-high molecular weight (UHMW) polyethylene bears against a second component of either 316 LVM stainless steel or cobalt chrome alloy. The low friction and high wear resistance of these material combinations have been a major factor in the overall clinical success of prosthetic joints. However, a number of mate- rial-related problems remains to be solved. One study has reported
Journal of Biomedical Materials Research, Vol. 12,895-927 (1978) 0 1978 John Wiley & Sons, Inc. 0021-9304/78/0012-0895$01.00
896 MCKELLOP ET AL.
wear rates in acetabular sockets of total hip prostheses apparently as high as 0.6 mm/yr.l Even though the vast majority of prostheses wear a t a much lower rate, such that gross wearing out of the compo- nents is not a problem, experience with other polymers has indicated that the gradual accumulation of debris particles can generate adverse reactions in the tissues surrounding the prosthesis. This may include bone resorption leading to loosening failure of the prosthesis.2 Poly- ethylene components are also subject to deformation due to creep (cold flow) and ordinary plastic yielding. Severe deformation may result in instability of the prosthesis and eventual loosening.3
Although the metal component of a polymer-metal prosthesis is generally not subject to wear, plastic yielding and even fracture of metal femoral stems are significant problems with stainless steel and cobalt chrome alloy total hip prostheses. These shortcomings have led to a search for higher strength polymers and metals with accept- able friction and wear properties for use in prosthetic joints. The wear properties of a great variety of potential material combinations have been evaluated on machines ranging from simple pin-on-disk devices to elaborate joint simulators. However, to date, no poly- mer/metal combination has been consistently and reliably shown to have wear properties equal or superior to those materials already in use. Furthermore, there is no general agreement on the proper lab- oratory conditions for evaluating prosthetic joint materials. Two polymers, PTFE and polyester, exhibited apparently adequate wear resistance in preliminary testing but were found completely unac- ceptable in actual clinical ~ ~ 8 . 4 . ~ Prostheses using these polymers were subsequently removed from hundreds o i patients.
The unreliable nature of much of the wear data obtained in earlier laboratory wear studies can generally be attributed to four factors: (a) the test conditions did not adequately simulate the in uiuo wear environment, (b) wear was not accurately measured, (c) the tests were too short to establish the long-term wear properties of the material, or (d) conclusions were based on the results of tests with a single specimen of each material.
The purpose of this study was to establish a laboratory test protocol for accurately and reliably assessing the wear properties of candidate prosthetic bearing materials. The wear of UHMW polyethylene against 316 stainless steel and cobalt chrome alloy was determined in long-term multispecimen tests. A twelve channel wear machine was developed especially for the purpose of testing materials under conditions which duplicated the essential features of the physiological
WEAR OF UHMW POLYETHYLENE a97
wear environment. In the tests reported here, polyethylene wear was measured as a function of contact stress, lubricant and counterface surface finish. The results of these tests provided valuable insight into the nature of polyethylene/metal wear and established a basis for the comparative evaluation of alternate materials.
PROCEDURE
Wear Screening Device
The polymer specimen consisted of a 12-mm-diam cylinder with one end tapered at 45' to form a contact area of 64 mm2 (0.1 in.2). When mounted in the wear machine the polymer specimen was pressed end-wise against a flat metal counterface (Fig. 1). Constant axial load up to 445N (100 lb) was applied to each set of specimens by a pneumatic cylinder mounted above the wear chamber. The individual chambers were clamped to an oscillating table which was driven through a 25-mm (1-in.) stroke at a frequency of 100 cycles per minute. The polymer specimen was held stationary while the counterface oscillated against it. The wear chambers were made of Plexiglas to allow the use of potentially corrosive fluids such as blood serum or saline solution. Each chamber, containing 25 ml of lubri- cant, was connected to a reservoir of distilled water through siphon tubes which automatically compensated for evaporation from the chambers, maintaining the lubricant at a constant level and concen- tration. A thermocouple inserted into one chamber was used to record the temperature of the lubricant bath during the test. The chambers could be individually removed for inspection of the speci- men surfaces while testing continued on the remaining channels. Frictional force between the polymer specimen and its counterface was monitored by strain gauges attached to the upper specimen holder. The absolute value of the friction on each set of specimens was electronically averaged over several cycles and continuously plotted on a 12-channel chart recorder. In addition, the full friction signal from any two sets of specimens could be displayed on a separate two-channel recorder.
898 MCKELLOP E T AL.
Fig. 1. (a) Twelve-channel wear testing machine and friction recorder, (b) poly- ethylene specimen in friction transducer loaded against polished metal counterface, and ( c ) diagram of specimens and loading arrangement.
WEAR OF UHMW POLYETHYLENE 899
PNEUMATIC
POLYMER SPECIMEN
L OR CERAMIC PLATE
AR STROKE 25mm (1 in)
(C)
Fig. 1 (continued from previous page)
Wear Measurement
For the purpose of evaluating prosthetic materials, wear can be described as the removal of material from the polymeric specimen through a combination of mechanical and chemical processes. While wear naturally results in a reduction in the thickness of the specimen, this reduction can also be due to creep and plastic yielding. In wear tests of prosthetic materials it is important to measure wear inde- pendently of deformation since a prosthesis may fail from tissue re- action to excess debris long before the components have “worn out.” Earlier studies have commonly used three methods to attempt to quantify wear of polymers: (a) recording dimensional changes, (b) weighing wear debris filtered from the lubricant bath, or (c) weighing the specimen directly. Each technique has its inherent difficul- ties.
With polyethylene specimens, recording the dimensional changes alone may be inadequate since deformation due to creep and plastic yielding can be many times larger than the height loss due to actual wear, making an accurate measurement of the latter virtually im- possible.
Weighing of wear debris filtered from the lubricant can be useful when it is not practical to weigh the specimens themselves, as in joint simulator tests of full-scale prostheses. However, this method is not feasible in tests with serum lubrication because of the difficulty of
900 MCKELLOP ET AL.
separating minute polymer wear particles from the various semi-solid components of the serum. Furthermore, a significant portion of the worn polymer material may adhere to the metal component and therefore not be recovered as debris.
In this study the wear rates were determined by direct weighing of the polyethylene specimens. However, it was soon discovered that the weight gain due to fluid absorption could actually be greater than the loss due to wear, causing a net increase in the weight of the spec- imens. Two different methods were used to attempt to correct for this effect of fluid absorption.
Method A: “Dry” Weighing
Prior to wear testing, the polyethylene specimens and a set of controls were washed in an ultrasonic cleaner, vacuum-desiccated for three days and weighed. During the wear test the control specimens were soaked separately. Finally, the wear specimens and controls were recleaned and desiccated for 2 weeks to remove as much of the absorbed lubricant as possible. The results of the control tests were highly variable. In general it was found that the control specimens had gained from 300 to 400 pg during the soak period (about 1 week). This was reduced to 100-200 pg during 2 weeks of desiccation, with little change thereafter. Since some of the wear specimens still showed a net gain in weight, even after final desiccation, it was as- sumed that total weight loss due to wear was less than 100-200 pg. The very low amount of wear combined with the large variation in the control soak specimens indicated that a more accurate weighing method was needed.
Method B: “Wet” Weighing
The wear specimens and controls were presoaked in serum for several weeks to minimize fluid absorption during the wear tests. After soaking the specimens were washed, rinsed, dried with alcohol, and then weighed. At intervals during the wear test the wear and control soak specimens were recleaned and weighed. The average net gain (or loss) in weight of the control specimens relative to the start of the wear test was added to (or subtracted from) the apparent weight loss of each wear specimen to correct for fluid absorption. In long- term tests the weight of the control specimens eventually stabilized at a fixed value. While the error in this method was estimated to be
WEAR OF UHMW POLYETHYLENE 901
f50 pg on any one weighing, the sequential weighings provided a very accurate indication of wear as a function of sliding distance. The overall wear rate was taken as the slope of a best fit straight line through the wear data, calculated by using the method of least squares linear regression. Since this assumes a linear relationship of wear to sliding distance, the correlation coefficient was also calculated as a measure of the actual linearity of the wear graph. (A correlation coefficient of 1.0 would indicate an exactly linear wear graph, that is, a constant wear rate.)
A special jig fitted with a dial indicator accurate to f 1 pm (40 pin.) was used to measure 5 points on the wear surface of each of the poly- ethylene specimens. The same points were measured before and after the wear test and the total height loss was taken as the average loss for the 5 points.
A control test was performed with three polyethylene specimens to indicate the amount of unrecovered creep that could be expected to occur on the wear specimens. The creep specimens were placed under a constant ioad of 6.9 MPa for one week without wearing. The height of the specimens was measured immediately upon load removal and periodically over several days until no further creep recovery occurred. The permanent height loss (unrecovered creep), measured 12 days after load removal, was 50 pm, or 0.4 percent of the original specimen height (Fig. 2).
The worn surfaces of the polymer specimens and counterfaces were examined visually and under a light microscope at intervals during the test. At the completion of the test polyethylene specimens worn in each of the lubricants were vacuum-coated with gold palladium and examined in a scanning electron microscope.
Materials*
Polyethylene specimens were machined from a 25-mm-( l-in.)-dim bar of extruded Hercules 1900@ ultra-high molecular weight poly- ethylene.
The 316 LVM stainless steel counterfaces were machined in our laboratory from 41.2-mm-(1.64-in.)-diam wrought bars. Each specimen was lapped, polished in a slurry of 0.05 pm alumina, cleaned and degreased ultrasonically, and finally passivated in nitric acid before being used in the wear test. The surface roughness was typical of prosthetic components, about 0.05 pm rms (2 pin.).
* Materials provided by Zimmer, Inc., Warsaw, Indiana.
902
SPEC HEIGHT
AJM t INCHES)
125 (005)
100 (.ma
75 (.OW
,a 25
(.an1
MCKELLOP ET AL.
4EN IHANGE
I A : I 0 20 4 0 60 80 100 120 140 v 2 6 0 2 8 0 HOURS
TME AFTER LOAD REMOVAL
Fig. 2. Residual height loss (creep) of polyethylene specimens as a function of time after load removal. A constant load of 6.9 MPa (1000 psi) was applied for one week. Creep recovery was rapid during the first day; however, the specimens never regained their original height but retained about 50 km permanent deformation.
The cobalt chrome alloy counterfaces were prepared from disks of Zimallop cast cobalt-chrome-molybdenum alloy. These were treated similarly to the stainless steel counterfaces except that four grades of surface finish were prepared, 0.03-0.05 pm rms (prosthesis-quality), 0.08-0.13 pm (produced using 600 grit silicone carbide paper), 0.20-0.30 pm (240 grit paper) and 0.75-0.86 pm (60 grit paper). Again, the surfaces were passivated after final polishing,
RESULTS
Polyethylene Wear in Various Lubricants
The first test compared the wear of UHMW polyethylene bearing against 316 stainless steel in distilled water, physiological saline (Ringer’s solution) and bovine blood serum, three lubricants com- monly used in earlier studies. Specimens were run for one million cycles under a load of 3.45 MPa (500 psi). The weight loss of the polyethylene specimens was measured using Method A, that is, the specimens and soak controls were vacuum desiccated and weighed before and after the wear test.
WEAR OF UHMW POLYETHYLENE 903
The friction, amount of wear, and, most importantly, the nature of the wear process varied in each of the lubricants. The three specimens lubricated with distilled water showed an increase in weight of 50,60, and 120 pg. Subtracting 150 pg, the average gain of the soak controls, gave a net loss of 100,90, and 30 pg, respectively. This same correction was applied to the serum and saline lubricated specimens. Weight loss was converted to equivalent wear depth by dividing by the density and contact area of the polyethylene specimens (0.936 gm/cm3, 64.5 mm2). Friction and wear for the three lubricants are compared in Table I.
Transfer layers of polyethylene gradually formed on the surface of the saline and water lubricated counterfaces (Fig. 3), accompanied by an increase in the coefficient of friction. The transfer layers on the saline lubricated counterfaces had an orange cast suggestive of corrosive action, and would occasionally break up into flakes of debris, at which time the coefficient of friction dropped to its initial lower value. This process was repeated many times during the test over a period of about 104 cycles. The transfer layers on the distilled water
TABLE I Wear of Polyethylene in Different Lubricants
Average Number of Wear Rate
Speci- pm/106 cycles Friction Summary Lubricant mens (range) ( p = coefficient of friction)
Serum 4 0.65 (rt 17%)
Distilled 3 0.08 water (5 60%)
Saline 3 5.2 solution (* 17%)
p = 0.07-0.12 normally, p = 0.35 during temporary high friction. Polymer transfer onto metal counterfaces oc- curred only during the high friction phase.
p = 0.07-0.13 a t start. A heavy polymer transfer layer formed by 0.3 million cycles, p then ranged from 0.14 to 0.18. The transfer layer remained intact for the duration of the test.
p = 0.07-0.10 a t start. Heavy, orange- colored transfer layers formed as p in- creased to 0.27. These layers occa- sionally broke up and I.I dropped to the . -.
initial level.
Polyethylene specimens were run against 316 stainless steel counterfaces at 3.45 MPa (500 psi) nominal contact stress for one million cycles. Wear was measured by Method A.
904 MCKELLOP ET AL.
Fig. 3. Stainless steel counterfaces after one million wear cycles against UHMW polyethylene, lubricated with (a) physiological saline solution, (b) distilled water, and (c) bovine blood serum. The white tracks in (a) and (b) are polyethylene transfer layers.
lubricated counterfaces lacked this orange cast and were stable over the duration of the run.
Transfer layers normally were not present on the serum lubricated counterfaces. However, a temporary period of unusually high friction often occurred about 1 hour after a test was restarted with fresh serum (Fig. 4). A sudden rise in the coefficient of friction, to as high of 0.35, was accompanied by the rapid formation of a dense layer of poly- ethylene on the metal counterface similar to the ones which formed in distilled water and saline solution. The bulk lubricant tempera- ture, normally 28-32OC, rose to about 50°C. After reaching a peak value, the coefficient of friction and lubricant temperature gradually returned to their initial normal levels. During the recovery period (usually several hours) the transfer layer on the counterface was progressively rubbed away, eventually leaving a clean metal sur- face.
Although the high friction states were of short duration compared to the overall length of the test, wear during these periods was very rapid and tended to increase the overall wear rate significantly. Since it did not occur with every change of serum, a number of tests were performed in an attempt to isolate the cause of this anomalous high wear mode. Several factors were found to influence its occurrence. 1. It occurred more often with serum that had been stored in the refrigerator, as compared to freshly thawed serum. However, even serum which was allowed to stand at room temperature for several days did not consistently induce high friction. 2. It was apparently accompanied by a permanent change in the bulk properties of the serum since it never occurred more than once in a
WEAR OF UHMW POLYETHYLENE 905
COEFFICIENT OF FRICTION
0. 30
0.0 4 1 I 8 I r 1 1 1 T 1 0 1 2 3 4 5 6 7 8 9 1 0
HOURS
Fig. 4. Two graphs illustrating the sudden rise in the coefficient of friction that sometimes occurred after the serum in the wear chambers was changed. The friction rise was accompanied by the rapid formation of a polyethylene transfer layer on the metal counterface. This transfer film was gradually rubbed away as the friction re- turned to the normal low level.
given charge of serum, even if the specimens were removed, cleaned and replaced. 3. It occurred more often with serum having a noticeable red color due to hemolyzed blood cells. 4. It occurred more often, and was more severe, with tests run at 6.90 MPa than at 3.45 MPa contact stress. In addition, it occurred more often with specimens running at 100 cycles per minute than at 60 cycles per minute.
The tests described in the following sections of this report were performed using commercially prepared serum* that was stored frozen until needed. In addition, 3 ml of 1 %sodium azide were added to each wear chamber when the serum was changed to inhibit bacte- riological degradation. Tests with a control polyethylene specimen indicated that the sodium azide had no measurable effect on friction or wear rate.
The appearance of the worn polyethylene surfaces examined in the scanning electron microscope varied for each of the lubricants. Ini- tially the surfaces were covered with ridges formed in the machining
* Pel-Freeze Biologicals, Inc., Rogers, Arkansas.
906 MCKELLOP ET AL.
process (Fig. 5a). After wearing in serum (Fig. 5b), these machine tracks 'were flattened but were still visible over the entire surface, indicating that very little wear had occurred. With distilled water the machine tracks were discernible only in a few areas and had a more roughened appearance (Fig. 5c). In most areas on this specimen and over the entire contact area of the saline lubricated polyethylene (Fig. 5d), the machine tracks were worn away entirely or obscured by
Fig. 5. Scanning electron micrographs of the wear surfaces of polyethylene speci- mens. (a) A new specimen with the surface grooves formed during the original ma- chining process. Worn surfaces after one million cycles against stainless steel lubricated with (b) serum, (c) distilled water, and (d) saline solution. The patches of redeposited polymer on the saline lubricated surface were also observed in some areas on the specimen lubricated with distilled water.
WEAR OF UHMW POLYETHYLENE 907
patches of polymer debris adhering to the surface. These islands of redeposited polyethylene were not observed on serum lubricated specimens.
Polyethylene Wear against 316 Stainless Steel and Cobalt Chrome Alloy: Method B
Polyethylene specimens were run against stainless steel and cobalt chrome counterfaces lubricated with commercial serum to compare the wear rate for these two alloys under the more accurate Method B. Tests were run at 3.45 MPa (500 psi) and 6.90 MPa (1000 psi) to examine the influence of contact stress on polymer wear rate over the load range thought to occur in total hip pros these^.^.^ Since the measurement of polyethylene wear rate in distilled water and serum ufider Method A was inconclusive due to the effect of fluid absorption, this comparison was repeated using Method B. Five polyethylene specimens were run against 316 stainless steel counterfaces in distilled water at 6.90 MPa contact stress.
With serum lubrication, wear was very low for both counterface materials, such that two to three million cycles were necessary to clearly establish the pattern of wear in each test (Fig. 6). The long- term wear rates and coefficient of friction are compared in Table 11. In many of the tests the wear rate did not stabilize until after the first 0.5 million cycles. The wear rate and correlation coefficient were therefore calculated excluding the initial data point (0,O) to minimize the effect of this wearing-in period on the values obtained for the long-term wear rates. The correlation coefficients ranged from a low of 0.93 (for specimen C-1) to a high of 0.998 (for specimen S-6), the latter value indicating a very constant wear rate over the duration of the test.
Polyethylene wear against stainless steel increased with load, the average wear rate at 6.90 MPa being double that at 3.45 MPa. The average wear rate against cobalt chrome alloy was less than that with stainless steel at the same load (Table 11); however, the individual values overlapped considerably.
Each of the metal specimens had surface scratches running the length of the contact area (Fig. 7). This scratching was more exten- sive on the stainless steel than cobalt chrome and tended to increase as the test progressed. As in the previous tests with serum, poly- ethylene transfer layers did not form on these counterfaces.
The total frictional force in serum increased with load; however,
908 MCKELLOP ET AL.
TABLE I1 Wear of Polyethylene with Serum Lubrication
Polyeth- ylene Counter- Load Wear Rate Average Speci- face MPa (pm/106 Correlation Rate Friction
men No. Material (psi) Cycles) Coefficient (Range) Coefficient
s-1 316 3.45 1.2 0.96 1.6 0.04-0.16 S-2 Stainless (500) 1.7 0.994 ( f20%) s-3 steel 1.7 0.97
s-4 316 6.90 2.6 0.98 3.1 0.03-0.09 S-5 Stainless (1000) 3.1 0.98 (f15 %) S-6 steel 3.6 0.998
C-1 Cobalt 6.90 2.3 0.93 2.6 0.05-0.11 C-2 chrome (1000) 2.3 0.98 ( f23 %) c-3 alloy 3.3 ~ ~ ~ 0.97 ,
These wear rates, obtained using Method B, are the slopes of the best-fit straight lines indicated on the wear graphs of Fig. 6. The correlation coefficient indicates the linearity of the wear graph; a value of 1.0 would correspond to an exact fit (a constant wear rate). The coefficients of friction were initially very low but increased through the range indicated as the test progressed.
the coefficient of friction decreased, that is, frictional force did not increase in proportion to load. Several of these specimens experi- enced a temporary slight increase in the coefficient of friction, to about 0.15, when the serum bath was changed, but this did not appear to affect the overall wear rate. In the earlier distilled water lubricated tests run at 3.45 MPa contact stress the polymer transfer layers that formed on the stainless steel counterfaces were gray in color. In contrast, at 6.90 MPa the transfer layers in distilled water had an orange color similar to that observed with saline solution lubrication. Although these transfer layers were not rubbed off of the counterfaces during the test, large plaques of orange colored debris were deposited on the contact surface of several of the polyethylene specimens. In addition, the entire inner surface of the Plexiglas wear chambers had a faint orange-colored stain that was removed only by vigorous scrubbing in a detergent solution.
The polyethylene wear in distilled water was erratic (Fig. 8). Several specimens gained weight early in the test, and one specimen (DW-4) continued to gain weight until nearlyl.8 million cycles. Three
WEAR OF UHMW POLYETHYLENE 909
WEAR mmL bm
-1 0 0.6 - - 8 ST. STEEL, 3.45 MPa 0.4-- 6
0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
t
0.5 1.0 1.5 2.0 2.5 3.0 3.5
0.6 8 VS CoCrMo. 6.90 MPa ,-$;
,.r.".-
d.5 110 115 210 ;5 310 2 5 CYCLES (mi II i on$
Fig. 6. Wear of UHMW polyethylene as a function of sliding distance (one cycle = 50 mm) using serum lubrication and highly polished counterfaces. (a) 316 stainless steel a t 3.45 MPa, (b) 316 stainless steel at 6.90 MPa, (c) cobalt-chrome alloy a t 6.90 MPa. The dashed lines are the best-fit straight lines for each set of data, calculated using linear regression. The slope of'this line is the wear rate for each specimen (Table 11).
specimens, DW-5, DW-6 and DW-7, eventually stabilized a t similar wear rates of 5.4,6.7, and 5.6 pm per million cycles, respectively, as indicated by the dashed lines on Fig. 8. This is about three to four times the average wear rate in serum (at 6.90 MPa). The friction and wear properties with distilled water and serum, determined by Method B, are summarized in Table 111.
910 MCKELLOP ET AL.
Fig. 7. Metal counterfaces after wearing against polyethylene at 6.90 MPa with serum lubrication, highlighted to show surface scratches. (a) 316 stainless steel after 3.6 million cycles. (b) cobalt chrome after 2.1 million cycles.
7
3.5 + CYCLES
(millions)
Fig. 8. Wear of polyethylene against 316 stainless steel counterfaces lubricated with distilled water, 6.90 MPa contact stress. A gain in weight is plotted as negative wear.
Effect of Counterface Surface Finish on Polyethylene Wear Table IV lists the wear rates for polyethylene against cobalt chrome
counterfaces having surface roughness ranging from “Grade A” (prosthesis quality) to “Grade D” (very rough). These specimens were also subject to the temporary increase in friction which some- times occurred after the serum was changed. The magnitude of the increase in friction was greater for the rougher surfaces, as indicated by the peak values in Table IV. The friction on the roughest Grade D counterfaces was very high throughout the test. The polymer specimens wore very rapidly and were removed at less than 0.25 million cycles.
TA
BL
E I
11
Com
pari
son
of P
olye
thyl
ene
Wea
r Pr
oper
ties
with
Ser
um a
nd D
istil
led
Wat
er L
ubri
catio
n T
este
d ag
ains
t 316
Sta
inle
ss S
teel
at 6
.9 M
Pa
(100
0 ps
i):
Met
hod
B
Lub
rica
nt
Bov
ine
Seru
m
Dis
tille
d W
ater
Wea
r ra
te
Stea
dy a
t an
aver
age
of 1
.6 p
m p
er m
illio
n cy
cles
. R
e-
Err
atic
, som
e sp
ecim
ens
gain
ed w
eigh
t. St
eady
sta
te
valu
es r
ange
d fr
om 5
.4 to
6.7
pm p
er m
illio
n cy
cles
, th
ree
to fo
ur ti
mes
that
in s
erum
. pr
oduc
ible
with
in f
20
%.
Coe
ffic
ient
of
0.03
-0.0
9.
0.10
-0.2
0.
fric
tion
area
Po
lym
er c
onta
ct
Smoo
th, g
loss
y, w
ith s
ome
ridge
s pa
ralle
l to
the
slid
ing
dire
ctio
n co
rres
pond
ing
to th
e sc
ratc
hes o
n th
e m
etal
co
unte
rfac
e. N
o di
scol
orat
ion.
su
rfac
e.
Cle
an a
nd s
hiny
with
som
e sc
ratc
hes
runn
ing
the
full
Smoo
th, g
loss
y ba
ckgr
ound
ove
rlaid
with
smea
red
lum
ps
of p
olym
er a
nd o
rang
e-co
lore
d fl
akes
adh
erin
g to
the
Com
plet
ely
cove
red
with
a g
ray-
oran
ge-c
olor
ed tr
ansf
er
Met
al c
ount
erfa
ce
cont
act a
rea
leng
th o
f the
con
tact
are
a.
laye
r. So
me
scra
tche
s par
alle
l to
wea
r di
rect
ion.
C !2 8
TA
BL
E I
V
Wea
r of
Poly
ethy
lene
for C
ount
erfa
ces
with
Var
ying
Sur
face
Rou
ghne
ss
Cou
nter
face
Sp
ecim
en
Rou
ghne
ss
Coe
ffic
ient
of
Fric
tion
Wea
r R
ate
(+m
/106
cycl
es)
Num
ber
(pm
rms)
St
eady
Ran
ge
Peak
Val
ue
0-0.
5 m
illio
n cy
cles
1-
2 m
illio
n cy
cles
c-1
c-2
c-3
c-4
c-5
C-6
c-7
‘2-8
c-
9
Gra
de A
0.
06-0
.11
0.15
0.
03-0
.05
Gra
de B
0.
07-0
.12
Gra
de C
0.
20-0
.30
0.07
to 0
.12
0.11
to 0
.15
0.20
0.23
3.6
7.6 1.2
5.3
1.4
9.5
22.3
73
.8
86.8
2.3
2.2
2.9
5 14
.6
12 r 0
v
r
3.6
3.9
M
P
7.8
c3 r
12.9
10
.5
c-10
G
rade
D
0.22
to 0
.30
0.30
3,
000
Gra
de D
test
s st
oppe
d c-
11
0.75
-0.7
6 20
,000
at
less
than
0.2
5 c-
12
8,70
0 m
illio
n cy
cles
due
to
exce
ssiv
e w
ear
Initi
al a
nd fi
nal v
alue
s of
the
wea
r rat
es w
ere
calc
ulat
ed.
For G
rade
s A, B
, and
C, t
he lo
ng-t
erm
wea
r rat
es w
ere
com
para
tivel
y lo
w.
Ver
y ro
ugh
surf
ace
scra
tchi
ng (G
rade
D) w
as n
eces
sary
to
prod
uce
seve
re p
olym
er w
ear.
WEAR OF UHMW POLYETHYLENE 913
Since the wear rates with the Grade B and C surfaces tended to decrease significantly over the period from 0.5 to one million cycles, both early and late values of the wear rates were calculated. Exam- ination of the contact area on these counterfaces revealed that the rubbing action of the polyethylene had caused a general smoothing of the initial surface texture (Fig. 9). These specimens exhibited long scratches running parallel to the wear direction similar to those pro- duced on the Grade A surfaces.
Fig. 9. Micrographs of the surfaces of cobalt-chrome alloy counterfaces. Left: original surface. Right: contact area at end of wear test. (a) Grade A surface with virtually no damage other than isolated scratches. (b) Grade B surface showing con- siderable smoothing of the original scratches. (c) Grade C surface, also showing evi- dence of smoothing of the scratches. (d) roughest, Grade D surfaces that generated severe polyethylene wear. Magnification M O O .
914 MCKELLOP E T AL.
Fig. 9 (continued)
The wear rate with one specimen (C-4) increased over the test run. This specimen went into very high wear (450 pm per million cycles) during the period from 1 to 1.2 million cycles, with the coefficient of friction as high as 0.24 and a heavy transfer layer covering about 50 percent of the contact area. Eventually the transfer material was rubbed away and the friction returned to its normal level. The wear rate for the remainder of the run was about 15 pm per million cycles. There was no obvious difference in the quality of the counterface surface compared to the other Grade B specimens to account for this unusual behavior.
WEAR OF UHMW POLYETHYLENE 915
DISCUSSION
Effect of Creep and Fluid Absorption on Wear Measurements
Knowing the density and geometry of the polyethylene specimens, it was possible to convert the measured weight loss into an equivalent height loss due to wear (h). The deformation due to unrecovered creep (D) was then found by subtracting wear from the total measured height loss ( H ) , i.e.,
Creep deformation, D = H - h
For our wear specimens the unrecovered creep calculated in this manner ranged from 58 to 75 pm. This is comparable to the 50 pm deformation of the control creep specimens (Fig. 2).
Theoretically, it should be possible to determine the wear of poly- ethylene specimens by subtracting the height loss (due to creep) of control specimens from the total height loss (due to creep and wear) of the wear specimens. In practice, however, this method is subject to considerable error, as shown by a sample calculation for specimen c-1:
Total height loss = 75 pm Creep of control specimens = 50 pm
Difference = calculated “wear” = 25 pm Actual wear (from Table V), = 4 pm
Thus, the calculated “wear” is over six times the actual wear. There are two sources of error in this example. First, the control
creep specimens were under constant load for 1 week, whereas the wear specimen was under load for about 4 weeks with periodic un- loading for weighing. Second, the creep specimens were at room temperature (25OC), while the wear specimen bath temperature ranged from 28 to 32OC. Both factors would tend to increase the amount of unrecovered creep of the wear specimen compared to the controls. However, even if a separate temperature-controlled creep specimen could be maintained for each wear specimen, the overall dimensional change corrected by creep is not likely to give an accurate measure of wear in the case of low-wear high-creep polymers such as polyethylene, where the wear value obtained depends on the differ- ence between two relatively large numbers.
Some investigatorsSl0 have attempted to eliminate the error caused by polyethylene creep by continuously recording the specimen de-
TA
BL
E V
Com
pari
son
of W
ear
and
Cre
ep o
f Pol
yeth
ylen
e Sp
ecim
ens
Hei
ght
due
to
Hei
ght
Cre
ep,
Cre
epIW
ear
Loss
T
otal
U
nrec
over
ed
Wea
r, h
Lo
ss, H
D
=H
-h
Rat
io
K s-
1 3.
45 M
Pa
4 61
51
14
R
s-3
6 69
63
11
s-4
6.90
MPa
11
74
63
6
Spec
imen
Lo
ad
(wm
) (w
) (p
m)
Dlh
0
s-2
(5
00 p
si)
6 61
55
9
F 3 r
0
cd
P
s-5
(1
000 p
si)
12
71
59
5 S-
6 12
58
46
4
r c-
1 6.
90 M
Pa
4 75
71
18
C
-2
(100
0 psi
) 7
64
58
7 C
-3
6 65
59
10
The
hei
ght l
oss
due
to w
ear w
as c
alcu
late
d by
mul
tiply
ing
the
wea
r rat
e (T
able
IV) b
y th
e to
tal n
umbe
r of
cyc
les.
T
he to
tal h
eigh
t w
as d
eter
min
ed f
rom
dir
ect m
easu
rem
ent o
f th
e sp
ecim
ens
bero
re a
nd a
fter
the
wea
r te
st.
The
dif
fere
nce
in th
e tw
o is
an
indi
catio
n of
the
unr
ecov
ered
cre
ep.
The
val
ues
are
com
para
ble
to th
e 50
pm o
btai
ned
with
con
trol
spec
imen
s (F
ig. 2
).
WEAR OF UHMW POLYETHYLENE 917
formation or wear tract depth while the specimen is under load. Since the creep rate decreases exponentially with time after the load is ap- plied, eventually a point should be reached where additional defor- mation is due primarily to wear. This method may be satisfactory provided that the test is of sufficient duration and the instrumenta- tion is capable of accurately measuring wear depths as small as 1-2 pm per million cycles.
Wear rates determined from specimen weight loss are accurate only if the effects of fluid absorption are controlled. For example, the highest wear rate with serum lubrication (specimen S-3) corresponded to a weight loss of only 210 pg over one million cycles. If a polyeth- ylene specimen is not presoaked, the gain due to absorption during the same period can be as high as 400 pg, making weight measurement useless for determining wear.
Polyethylene Wear in Different Lubricants Although polyethylene wear has been examined previously with
each of the lubricants used in this study, there is currently no agree- ment as to which lubricant is most suitable for use in determining the wear properties of candidate prosthetic materials. Distilled water or saline solution have been used in earlier studies primarily because they were easily obtainable in large quantities, they were not subject to bacterial degradation and they did not tend to corrode metal components of the test apparatus as readily as did serum or synovial fluid. However, the results of our tests indicated that these fluids have fundamentally different lubricating properties.
Wear in saline solution and distilled water was characterized by the formation of orange-colored transfer layers on the stainless steel counterface. Transfer occurs whenever the shear strength of the polymer-metal junctions is greater than the bulk shear strength of the polymer. As sliding occurs, lumps or sheets of polymer adhere to the metal surface and eventually may form a nearly complete covering of the contact area such that further sliding occurs between the polymer specimen and the transfer layer. In the tests with saline solution, corrosive action a t the metal surface apparently loosened the transfer layers such that they were sheared off, forming flakes of debris that were visible floating the lubricant bath. Although the transfer layers that formed in distilled water were stable, polymer mixed with corrosion products was readily retransferred to the polyethylene specimen, a process which could account for the weight gain exhibited by several of these specimens.
918 MCKELLOP ET AL.
The fact that polyethylene transfer layers did not form in serum may have been due to proteins acting as a boundary lubricant layer on the metal surface, thereby reducing the polymer-metal interface shear strength. The formation of protein-metal lubricating com- plexes was suggested by Duff-Barclay and Spillman12 as an expla- nation for the lubricating advantage of blood plasma over saline so- lution in tests with all-metal prostheses. Later studies have shown that the active component might be either albuminl3 or gamma g10bulin.l~ In the case of polyethylene bearing against metal,boundary lubrication by serum proteins could serve to eliminate the adhesion necessary for the formation of transfer layers. Since polyethylene transfer layers are not observed on the surfaces of metal prosthetic components removed from patients,15 it is apparent that serum-type lubrication also occurs in uiuo.
It might be argued that even though wear in distilled water is qualitatively different than that in serum, the results of distilled water tests might be useful in establishing the relative wear rates of various combinations of materials. This does not appear to be a safe as- sumption. In addition to the marked differences in wear properties demonstrated in this study with polyethylene, other materials have been shown to have wear rates strongly dependent on the lubricant used. Charnley16 reported that the wear rate of PTFE was several hundred times greater in serum compared to distilled water. Shen and Dumbleton17 found that the opposite was true of Delrin 150, the wear rate in plasma being one-tenth that in distilled water. In view of these results, we have concluded that wear tests of materials for prosthetic joints should be conducted with blood serum as a lubricant to insure that the wear processes adequately simulate those occurring in uiuo.
Although lubrication with synovial fluid might provide an even closer replication of the physiological environment, it is difficult to obtain in other than minute quantities. Furthermore, there appears to be little difference in the lubricating properties of serum and syn- ovial fluid, probably due to their similar protein content.
The high friction state that often occurred just after the serum was changed appeared to be due to a temporary failure of the lubricating mechanism, with a subsequent rapid buildup of a dense transfer layer (adhesive wear) and a corresponding increase in friction. A similar effect may have been encountered by Seedhom and colleagues6 in their wear tests with synovial fluid lubrication. They reported that heavy polyethylene transfer layers occasionally formed on the
WEAR OF UHMW POLYETHYLENE 919
stainless steel counterface, and that wear was greatest in those tests when transfer occurred.
Since the high friction phase in our tests occurred only if the chambers were rinsed and fresh lubricant introduced, it seems un- likely that this type of wear occurs in uiuo. Investigators using blood serum in wear tests to evaluate prosthetic bearing materials should therefore take care to insure that their results are not unduly in- fluenced by this anomalous wear mode.
Effect of Surface Finish on Polyethylene Wear
It is generally believed that a high-quality, mirror-like surface finish on the metal component is essential for minimizing the polyethylene wear in a pro~thesis .~.~J~J8 Presumably, fine polishing removes the sharp metal asperities that otherwise might abrade the soft polymer. The efforts of implant manufacturers are typically directed at ob- taining surfaces with “a level of perfection beyond normal quantitative measurement.” l9 However, several investigators have suggested that a properly textured metal surface could serve to reduce the polymer wear still further. Such a surface might be less prone to wear caused by adhesion of the polymer to the smooth metal. In addition, the surface depressions would serve as lubricant reservoirs while simul- taneously allowing abrasive debris particles to escape from between the close-fitting bearing surfaces.
A few experiments have been done to test this theory. Oonishi20 compared the wear of polyethylene cups in a joint simulator using prosthetic heads with standard polishing as well as a variety of surface treatments. The least wear was obtained with a steel ball having a set of 1-mm-wide relief bands chamfered onto the surface. In a similar study by Swikert and Johnson,21 it was found that a prosthetic head with a liquid honed surface (0.11 pm RMS) produced less polyethylene wear than a highly polished head (0.06 pm). Unfortu- nately, both of these studies compared wear under dry conditions, where the amount of wear was apparently related to the ability of the polyethylene to form a transfer film on a particular surface. It is questionable whether the same relative wear rates would be obtained with physiological lubrication.
Two earlier studies did include an attempt to correlate polyethylene wear with counterface roughness in lubricated sliding. Seedhom and colleagues6 ran polyethylene against a stainless steel disk with 5 deg of roughness ranging from 0.05 to 0.27 pm, with synovial fluid as a
920 MCKELLOP ET AL.
lubricant. They reported that the polyethylene wear rate generally increased with metal surface roughness. However, the results are difficult to interpret since other conditions such as sliding speed, test duration, and specimen temperature (due to dry running) varied in the different tests. Miller and colleaguesg examined polyethylene wear against titanium alloy counterfaces with surface roughness specified as 0.05, 5 , and 15 pm, lubricated with Ringer’s solution. Although the initial wear was lowest with the smoothest counterface, the final rates were similar for each of the specimens. Since heavy polymer transfer layers formed on the metal counterfaces, it is possible that the transferred polyethylene tended to mask the underlying roughness of the metal, accounting for the similarity of the long-term wear rabes.
The tests reported here were intended to determine the sensitivity of polyethylene wear to a wide range of surface roughnesses. With serum lubrication no transfer layers formed on the metal counterfaces. Rather, the decrease in the polyethylene wear rate with sliding dis- tance of the Grade B and C surfaces was apparently due to a gradual polishing of the surface scratches. These results indicate that surface finish studies in particular should be performed with physiological lubrication to avoid the masking effect of the transfer layers that form on water-lubricated specimens. In addition, the tests should be of sufficient duration to detect gradual alterations of the initial surface texture.
Although roughened counterfaces produced higher polymer wear rates than the polished specimens, the long-term wear rates with Grades B and C were still very low, less than 15 pm per year. It ap- pears that super-smooth, mirror-like finishes on prostheses are not critical for acceptably low polyethylene wear. The possibility remains that a specially textured counterface surface might be developed that would produce even less wear than with the mirror-finished speci- mens. However, it is questionable whether the attainment of a wear rate less than 2-3 pm per year would constitute a significant im- provement in the overall performance of a prosthesis.
Lubricant Bath Temperature
In the tests with polyethylene bearing against stainless steel at 6.9 MPa (1000 psi) the temperature of the serum bath normally ranged from 28 to 32OC but increased as high as 50°C during the periods of abnormally high friction. The question arises as to whether or not
WEAR OF UHMW POLYETHYLENE 921
the lubricant bath temperature should be controlled at the physio- logical level of 37°C. This would be a difficult achievement with the 12-channel screening device. Since the lubricant is not circulated, separate heating and cooling apparatus and temperature sensors would be required for each wear chamber. Furthermore, it is not apparent that the bulk lubricant temperature has a direct effect on the nature of the wear occurring between the contacting specimen surfaces.
Friction between the polymer and metal specimens is a result of the shearing of junctions and plowing of asperities of one material through the other. In polymer-metal sliding, local temperatures a t these junctions and asperities may approach the melting point of the polymer22 (for polyethylene about 130°C). The average surface temperature of the polymer and of the lubricant between the con- tacting surfaces can be expected to fall somewhere between this high “flash” temperature and that of the surrounding bulk lubricant. It will depend on the rate of heat generation (friction force multiplied by sliding speed) and the rate of conduction through the specimens, specimen holders and bulk lubricant. Regulating the bulk lubricant temperature alone at 37°C instead of 28-32°C would not insure that the temperature of the contact surfaces was the same as that occurring on a prosthesis in uiuo. An exact simulation would require duplica- ting both the rate of heat generation and the conductivity of the prosthesis and surrounding tissues. While this may be attempted with a joint simulator, again it is not practical in the case of a multi- channel screening device.
Significance of Screening Test Results with Respect to Clinical Wear
In addition to providing a measure of the relative wear resistance of various candidate materials, it is desirable that laboratory screening tests give some indication of the potential wear life of a prosthetic material in actual use by a patient. For many combinations of ma- terials, the depth of wear (D) is directly proportional to the contact stress ( P ) and sliding distance (S), such that
D = KPS where K , the wear factor, is constant over a wide range of loads.23 If one assumes a typical value for the contact stress and sliding distance in a prosthesis, the wear factor measured under laboratory conditions
922 MCKELLOP E T AL.
for a given pair of materials can be used to calculate the approximate wear depth per year of clinical use. In Table VI, polyethylene wear rates are compared for several previous studies. In each case the annual wear for an acetabular cup was calculated from the wear factor assuming a contact stress of 3.45 MPaand a sliding distance of 50 X lo6 mm as typical for a total hip prosthesis.l* The equivalent in uiuo wear rates calculated in this manner ranged over four orders of mag- nitude. The results of our tests corresponded to the low end of this range, giving a predicted polyethylene wear depth of 1.6 pm and 1.3 pm per year against stainless steel and cobalt chrome alloy, respec- tively.
There are many possible explanations for the disparity in the results of the different studies. While the great variety of experimental apparatus, test conditions and methods of wear measurement used in these studies makes it difficult to draw specific conclusions, some general trends are apparent. Most experimenters have relied on measurements of dimensional changes such as specimen height or wear track depth to determine wear. Our control creep tests with polyethylene have shown that this can lead to an overestimate of the wear rate, especially in short-term tests where the ratio of creep de- formation to wear is very large. For example, Rostoker and Galante24 determined the wear of polyethylene by intermittently measuring the depth of a wear groove on a flat polyethylene plate. In a long-term test of 50 X lo6 mm sliding distance, the apparent wear rate decreased by a factor of 50, from 1 X 10-9 mm/mm to 2 X mm/mm. The latter value is the equivalent of only 1 pm wear per year in an aceta- bular cup. It seems possible that the decrease in the apparent wear rate in this experiment was due to a decrease in the rate of creep de- formation rather than a reduction in wear.
Wear rates calculated on the basis of specimen weight loss have generally indicated much less wear than the corresponding dimen- sional changes. Amstutzg found that two out of three polyethylene specimens tested in mineral oil showed a net gain in weight. Even when the weight change was corrected for fluid absorption, the ap- parent wear was less than one-tenth the value indicated by specimen height loss. Walker and colleagues25 and Pawluk and Eftekhar26 also reported a gain in weight of polyethylene wear specimens. Presum- ably the weight loss due to wear was less than the gain due to fluid absorption, indicative of very low wear rates. Homsy and King27 reported a specimen height decrement the equivalent of 0.08 mm per year (Table VI), even though wear was “of such low magnitude in terms of weight loss as to preclude precise measure.”
WEAR OF UHMW POLYETHYLENE 923
One method of assessing the validity of laboratory wear tests of prosthetic materials is to compare the resultant wear rates to those exhibited by the same materials in actual use in patients. Charnley and Halleyl reported that direct measurement of radiographs of 72 polyethylene acetabular cups in s i tu indicated an average wear rate of 0.15 mm per year over a 10-year period. A number of investigators have observed a comparable wear rate for polyethylene in laboratory tests and have taken this correspondence as evidence that their par- ticular wear apparatus effectively reproduced the in uiuo wear pro- cesses.6s7,27,28,29 However, this type of comparison can be misleading. The average value of 0.15 mm per year represents the numerical mean of a wide range of wear rates. The acetabular cups examined by Charnley, either radiographically or after removal from patients, had wear rates ranging from as much as 0.6 mm per year to zero wear in as long as 10 years, with no correlation with patient weight or activity level. Twenty-eight percent of the cups had less than 0.5 mm total wear in 9-10 years.' It is apparent that wear in uiuo under ideal conditions can be negligible but that certain factors can combine to accelerate it beyond this minimal rate. Charnley suggested that large differences in the acetabular cup wear rates under otherwise similar conditions could be due to differences in the quality of the polyeth- ylene. In addition, examination of removed prostheses has shown that severe abrasion of the polyethylene component can occur whenever acrylic cement particles become trapped between the bearing surface^.^^,^^,^^ Thus, reproducing Charnley's average wear rate under laboratory conditions does not necessarily imply that the same wear mechanisms are active, but only that the in uitro tests generate polyethylene wear in the range between the two possible extremes occurring in uiuo. More consideration needs to be given to those particular in uiuo conditions that result in negligible wear over many years.
It is suggested that the very low polyethylene wear rate (1.6 pm per year) observed in this study, with highly polished serum lubricated surfaces, corresponds to the minimal wear observed on a large per- centage of acetabular cups after years of use in uiuo. A comparative evaluation of a candidate material under the conditions used in this study will therefore provide an indication of the minimum wear rate that can be expected in uiuo. Other factors, such as variations in the quality of the materials and resistance to abrasion by acrylic debris, will need to be evaluated as separate test parameters.
The twelve channel wear machine and the experimental protocol
TA
BL
E V
I Su
mm
ary
of P
olye
thyl
ene
Wea
r Rat
es O
btai
ned
in V
ario
us S
tudi
es
Spec
i- W
ear
men
C
ount
er-
Tes
t M
easu
re-
Wea
r St
udy
Con
fig-
face
D
urat
ion
men
t R
ate
(Ref
eren
ce)
urat
ion
Mat
eria
l L
ubri
cant
(y
ears
) M
etho
d (p
mly
ear)
McK
ello
p, e
t al
Wal
ker,
et a
l. (2
5)
Am
stut
z (8
)
Hom
sy &
Kin
g (2
7)
Seed
hom
, et a
l. (6
)
Gal
ante
&
Ros
toke
r (3
11
Ros
toke
r &
Gal
ante
(24)
I. Pi
n on
dis
k
Pin
on d
isk
Jour
nal
Was
her
on fl
at
Pin
on d
isk
Edg
e of
dis
k on
fl
at
Edge
of d
isk
on
flat
Stai
nles
s ste
el
Cob
alt c
hrom
e
Stai
nles
s ste
el
Car
bon
stee
l
Cob
alt a
lloy
Stai
nles
s ste
el
Stai
nles
s ste
el
Stai
nles
s ste
el
Seru
m
Syno
vial
flui
d
Min
eral
oil
Pseu
do-S
ynov
ial
fluid
Syno
vial
flui
d
Dis
tille
d w
ater
Dis
tille
d w
ater
3 2 0.01
0.35
0.8
0.5
? 0.5
2.5
A W
eigh
t
A W
eigh
t
A H
eigh
t A
Wei
ght
A H
eigh
t A
Wei
ght
A H
eigh
t
Wea
r tra
ck
dept
h
Wea
r tra
ck
dept
h
1.6
1.3
Gai
n
700 68
80
Too
smal
l to
mea
sure
100
1600
50 1
Wei
ghtm
an, e
t al.
Join
t sim
ulat
or
(29)
Beu
tler
, et a
l. Jo
urna
l (3
2)
Mill
er, e
t al.
Was
her
on fl
at
(9)
Dum
blet
on, e
t al.
Was
her
on fl
at
(33)
Dum
blet
on a
nd
Was
her
on fl
at
Shen
(10
)
Cha
rnle
y Pi
n on
dis
k (1
5)
Stai
nles
s ste
el
Cob
alt c
hrom
e
Cob
alt c
hrom
e St
ainl
ess s
teel
Stai
nles
s ste
el
Stai
nles
s ste
el
Stai
nles
s ste
el
Seru
m
9 Sa
line
solu
tion
Seru
m
Dis
tille
d w
ater
Pl
asm
a
Dis
tille
d w
ater
or
D.W
. + 5%
ge
latin
e or
sv
novi
al fl
uid
1
Cup
thic
knes
s
0.6
A W
eigh
t
0.13
W
ear t
rack
de
pth
0.4
Wea
r tra
ck
dept
h
? W
ear t
rack
0.
6 de
pth
1.5
Dep
th o
f w
itnes
s gr
oove
in
oolv
mer
150 20
160
160 79
29
29
13
Thi
s tab
le c
ompa
res a
repr
esen
tativ
e sa
mpl
ing
of th
e nu
mer
ous l
abor
ator
y st
udie
s of w
ear o
f pro
sthe
tic jo
int m
ater
ials
and
illu
stra
tes
the
grea
t var
iety
of t
est c
ondi
tions
and
ran
ge o
f re
sults
obt
aine
d.
For p
urpo
ses
of c
ompa
rison
, the
wea
r val
ues
give
n in
the
indi
vidu
al
stud
ies w
ere
conv
erte
d to
com
mon
uni
ts a
s fo
llow
s:
Tes
t dur
atio
n:
50 X
lo6
mm
slid
ing
dist
ance
equ
als
1 ye
ar o
f use
of a
pro
sthe
sis.
1°
Wea
r ra
te:
Thi
s was
det
erm
ined
by
firs
t cal
cula
ting
the
wea
r fac
tor (
wea
r per
uni
t loa
d pe
r un
it sl
idin
g di
stan
ce) f
rom
the
publ
ishe
d da
ta a
nd th
en m
ultip
lyin
g by
a ty
pica
l con
tact
str
ess (
3.45
MPa
) and
yea
rly s
lidin
g di
stan
ce (5
0 X
lo6
mm
) for
a p
olye
thyl
ene
acet
abul
ar
sock
et.1
0 In
gen
eral
, the
low
est w
ear v
alue
s w
ere
obta
ined
in c
ompa
rativ
ely
long
-ter
m te
sts
(Ros
toke
r and
Gal
ante
; Cha
rnle
y) o
r in
test
s whe
re
the
spec
imen
wei
ght l
oss w
as re
cord
ed (
A W
eigh
t). T
he w
ear v
alue
s ba
sed
on sp
ecim
en d
imen
sion
al c
hang
es (
A H
eigh
t, et
c.) t
end
to b
e hi
gher
, pos
sibl
y du
e to
cre
ep d
isto
rtio
n of
the
poly
mer
.
926 MCKELLOP ET AL.
developed in this study provide a valuable method for consistently and accurately determining the wear properties of prosthetic mate- rials. It is now possible to examine polymer wear as a function of variables such as lubricant, load, counterface material, surface finish, sliding speed and number of wear cycles, in a reasonably short period of time. Wear tests are in progress with a number of new combina- tions of materials which may provide an improvement in the overall durability of total joint prostheses.
This work was supported by the National Institute of Health, Grant No. AM 18988. The authors would like to express their sincere appreciation to research machinist Edward Moran for his valuable contributions to the design and construction of the wear test apparatus.
References
1. J. Charnley and D. Halley, Clin. Orthop. and Rel. Res., 112,170 (1975). 2. H. Willert and M. Semlitsch, in Tissue Reactions to Plastic and Metallic Wear
Products of Joint Endoprostheses, Williams and Wilkins, Baltimore, 1976, pp. 205-239.
3. F. Reckling, M. Asher, F. Mantz, and D. Helton, J. Bone J t . Surg., 57A, 1 (1975).
4. J. Charnley and A. Kamanger, Med. & Biol. Engng., 7,31 (1969). 5. M. Semlitsch, in Technical Progress in Artificial Hip Joints, Williams and Wilkins,
6. B. Seedhom, D. Dowson, and V. Wright, Wear, 24,35 (1973). 7. J. Galante and W. Rostoker, Acta Orthop. Scand. Suppl., (145) Musk. Copenhagen,
8. H. C. Amstutz, J. Biomed. Mater. Res., 3,547 (1968). 9. D. Miller, J. Dumbleton, R. Ainsworth, and D. Page, Wear, 28,207 (1974):
Baltimore, 1976, pp. 256-278.
1973.
10. J. H. Dumbleton and C. Shen, Wear, 37,279 (1976). 11. J. K. Lancaster, Wear, 20,315 (1972). 12. I. Duff-Barclay and D. Spillman, Proc. Inst. Mech. Eng., 181, Part 35,90 (1966-
13. B. Weightman, S. Simon, I. Paul, R. Rose and E. Radin, Trans. ASME J . Lub.
14. P. S. Walker and M. J. Erkman, Wear, 21,377 (1972). 15. J. Charnley, in Plastics ~ n . Medicine and Surgery, (conference proceedings),
University of Strathclyde, Glasgow, Scotland, The Plastics and Rubber Institute, London, 1975, Chap. 3.
1967).
Tech., 94,131 (1972).
16. C. Shen and J. H. Dumbleton, Wear, 38,291 (1976). 17. H. McKellop, I. Clarke, K. Markolf, and H. Amstutz, in Polyethylene Wear i n
Prosthetic Joints-A New Interpretation, Proceeding of the Conference on the Eoaluation of Artificial Joints, The Biological Engineering Society, University of Leeds, Leeds, England, Jan. 1976 (to appear).
18. J. T. Scales, P. Kelly, and D. Goddard, Ann. Rheum. Dis., 28 Suppl., 30 (1969). 19. “Tec Sheet” Charnley-Type Total Hip Prosthesis, Meditec, Inc., Fairlawn, New
Jersey (1973-1974).
WEAR OF UHMW POLYETHYLENE 927
20.
21.
22.
23.
24. 25.
26.
27.
28.
29.
30. 31. 32.
33.
H. Oonishi, in Studies on the Shape and Contour of the Metal Prosthetic Head in Total Hip Prosthesis, Proc. 12th Congress Int . SOC. Orthop. Surg. and Trau- matology, American Elsevier, New York, pp. 107-123. M. A. Swikert and R. L. Johnson, in Simulated Studies of Wear and Friction in Total Hip Prosthesis Components with Various Ball Sizes and Surface Finishes, NASA Technical Note, NASA TDN D-8174. E. Rabinowicz, Friction and Wear of Materials, Wiley, New York, 1966, pp.
S. Swanson and M. Freeman, Eds., in The Scientific Basis of Total Joint Re- placement, Wiley, Philadelphia, 1977, Chap. 3, pp. 46-85. W. Rostoker and J. Galante, J . Biomed. Mat. Res., 10,303 (1976). P. Walker, D. Dowson, M. Longfield, and V. Wright, in Friction and Wear of Ar-
86-89.
tificial Joint Materials, Research Report 5, Proc.Inst. Mech. Engr., 196647,181, Pt. 35. R. Pawluk and N. Eftekhar, in I n Vitro Performance of a Low FrLctLon Hinged- Knee Prosthesis, presented at the 22nd Annual Meeting of the Orthopedic Res. Society, New Orleans, Jan. 1976. C. Homsy and J. King, “Friction and Wear Behavior of Prosthetic Cartilage Ma- terials under Simulated I n Viuo Conditions,” presented at the 18th Ann. Meeting of Orthopedic Res. Society, Washington, D.C., 1972. B. Weightman, I. Paul, R. Rose, S. Simon and E. Radin, J . Baomech., 6, 299 (1973). M. Semlitsch, M. Lehman, H. Weber, E. Doerre, and H. Willert, J . Biomed. Mat. Res., 11,537 (1977). P. Trent and P. Walker, Wear, 36,175 (1976). J . Galante and W. Rostoker, The Hip, Mosby, St. Louis, 1973, pp. 67-79. H. Beutler, M. Lehmann, and G. Stahli, in “Wear Behavior of Medical Engineering Materials,” Sulzer Research No. 1974, Sulzer Bros, Ltd., Winterthur, Switzer- land. J. Dumbleton, C. Shen, and E. Miller, Wear, 29,163 (1974).
Received September 29,1977 Revised February 5,1978