J. BIOMED. MATER. RES.
VOL. 9, PP. 611-621 (1975)
The Influences of Electrical Potential and Surface Finish on the Fatigue Life of Surgical Implant Materials M. S. BAPNA,* E. P. LAUTENSCHLAGER, and J. B. MOSER, Department of Biological Materials, Northwestern University, Chicago, and P. R. MEYER, JR., Department of Orthopaedic Surgery, Northwestern University, Chicago, Illinois 60611
Summary The influence of both imposed anodic potential in Ringer’s solution and surface finish on the fatigue lives of annealed 316 type stainless steel and annealed pure titanium were measured and statistically compared to fatigue data run in air. The applied potentials in simulated extracellular fluid approximated conditions existing within the body while also producing the types of surface defects actually found on removed long time implants within the time interval of the accelerated R. R. Moore fatigue tests. Differentiating tests were run a t single levels of applied cyclic stress well above the endurance limits. I n Ringer’s solution, the fatigue life of the 316 stainless steel decreased with increasing applied potential, and a t +SO0 mV was significantly shorter than when run in air. At each condition, the 316 stainless steel was independent, of initial surface finish. I n contrast, the fatigue life of titanium improved rapidly with increasingly fine surface finishes. Furthermore, compared to air, the application of +500 mV in Ringer’s solution improved the life of the rough surface finished material and markedly increased the number of cycles to failure for the electropolished specimens.
INTRODUCTION Certain metal alloys are now routinely used as surgical implant materials’ for prostheses and fixation devices. While having demonstrated acceptable strength and corrosion resistance in many biological applications, metallurgical analyses of broken orthopaedic implants indicate fatigue under cyclic loading as a major cause of s2
*Present address: University of Illinois Dental School, Chicago, Illinois. 611 @ 1975 by John Wiley & Sons, Inc.
612
BAPNA ET AL.
f a i l ~ r e . ~Moreover, whenever new designs are contemplated, particularly thinner devices requiring less removal of natural tissue, fatigue life should be a major design criterion. Unfortunately for the surgeon and the designer, with the exception of some fatigue data run in air,4-6there is a dearth of statistically significant information concerning the behavior of surgical implant metals during repeated cyclic loading in vivo or in simulated body environments. I n addition, little or no attention has been systematically focused upon those factors which are known to influence the fatigue life of metals under certain circumstance^,^ namely : compositional inhomogeneity, state of cold work, phase distribution, impurity content, nature and amplitude of applied stress, surface finish, etc. Although Wheeler and Jamess had demonstrated that crack growth rate in 316 type stainless steel was faster in Ringer’s solution than in air, preliminary studies in our laboratory revealed no significant decrease in the actual fatigue life of that material when tested in those two different environments. Also, simply running fatigue tests for several days in Ringer’s solution did not produce the type of crevice attack or surface pitting often observed in removed long time 316 type stainless steel implants>JO Anodic potentiostatic measurements” have successfully differentiated between the electrochemical stability of passivating oxide films of the common surgical implant alloys,12as well as demonstrated the capability of producing such oxide film breakdown phenomena aa pits.I3 With the harsh environment presented by the body, plus the ever present millivoltage signals and electrical potential differences existing in implants within the body,I4 it would appear that testing in air or in static simulated extracellular body fluid is not a sufficiently good approximation of the actual in vivo situation. Moreover, since the standard fatigue test is an accelerated life test, the environmental effects must also be speeded up to comply within the testing time. Therefore, we have chosen to conduct our fatigue experiments in a manner similar to that implied by Zarek,15with an external electrical potential imposed on a specimen being cyclically loaded in a simulated body fluid environment. The purpose of this investigation will be to determine the effects of different magnitudes of anodic polarization upon the fatigue life of
ELECTRICAL POTENTIAL AND SURFACE FINISH
613
two types of surgical implant metals; namely, 316 stainless steel (type A),* and pure titanium (grade 4),t when run in Ringer’s solution and possessing a number of different surface finishes. These results will be compared t o similar groupings tested in air.
METHODS The basic technique employed in this investigation was t o determine the number of cycles t o failure a t a selected applied stress for a given material of specific surface condition (transverse 60 grit, longitudinal 600 grit, or electropolished) in a particular environment (air or Ringer’s solution, plus imposed levels of anodic potential from 0 t o 500 mV). The testing apparatus (shown schematically in Fig. 1) was a n R. R. Moore type16 rotating beam fatigue machine modified with a chamber around the specimen t o permit immersion in a liquid and application of a n electrical potential. The applied weights set up a uniform bending movement on the specimen which is transformed into a continuous, completely reversing bending stress during cycling. The machine was run a t 3100 f 100 rpm (approximately 4.5 million cycles per day). Specimens were machined from single rods of purchased 12.7 mm round stock in a n annealed condition. The standard machined specimens were hour-glass shaped16 with a continuous decrease in circular cross section to the center. The Rockwell 15-N Superficial Hardness of all finished specimens was very uniform on both the surfaces and the later sectioned interiors, and found t o be 66 f 2 for the 316 stainless steel and 56 k 2 for the pure titanium. Each machined specimen was given one of the following surface finishes: (a) coarse mechanical, 60 grit transverse t o specimen long axis; (b) fine mechanical, 600 grit longitudinal; and (c) electropolished. The mechanical finishes were imparted with proper grit size emery paper being applied t o a rotating specimen. The electropolishing conditions were as follows: for the 316 stainless steel, electrolyte of 600 ml H3P04and 400 ml HzSOr a t 25°C run for 6 min a t *0.08 C max, 2 . 0 Mg max, 0.025 P max, 0.010 S max, 0.75 Si max, 17.00-20.00 Cr, 12.00-14.00 Ni, 2.00-4.00 Mo, remainder Fe; annealed condition; RN = 6 6 f 2 ; grain size 8. to .07 N max, 0.15C max, 0.0125 H max, 0.50 Fe max, 0.40 0 max, remainder Ti; annealed condition; RN = 56 f2; grain size 7.
BAPNA ET AL.
614
7 To Potentiostoi
\
\
\ \
%
Specimen (Working Electrode) Platinum (Counter Electrode) Ringer's Solution
1
L S o t u r a ted KCI Solution
Fig. 1. Schematic of R. R. Moore type fatigue machine equipped with chamber for physiological saline and application of anodic potentials.
3 V and 3 A; for the titanium, electrolyte of 943 ml glacial acetic acid and 57 ml of perchloric acid at -2OOC run for 3 min at 30 V. I n cases of applied potential, the electrochemical cell operation was similar to that described elsewhere."J7 A surrounding acrylic box (see Fig. 1) allowed the central 25 mm of the specimen to be completely immersed in Ringer's solution (9.000 g NaC1, 0.425 g KC1, 0.119 g CaC12, 0.100 g NaHC03, made up to 1 1. with deionized water and buffered to pH 7). Passing through stationary circular rubber seals in the ends of the box, the specimen could rotate freely but solution leakage was minimized. Also placed in the solution in close proximity to the rotating fatigue specimen was the open tip of a Luggin-Haber type salt bridge leading to a saturated calomel reference electrode (SCE) and a platinum wire counter electrode. The specimen became the working electrode (anode) and was electrically insulated from the rest of the fatigue machine by specially machined
ELECTRICAL POTENTIAL AND SURFACE FINISH
615
polytetrafluorethylene grips. The entire apparatus was housed in a wire mesh Faraday cage grounded a t a Wenking Potentiostat (Wenking 70TS1 Electronic Potentiostat, Brinkman Instruments, Westbury, New York). The specimen was placed in the apparatus, surrounded by Ringer’s solution and rotation begun with no weight applied. For the first hr the millivoltage generated between the specimen, and the reference cell was allowed t o come to a steady state known as the open circuit potential (OCP). A specified anodic millivoltage over and above the OCP was then chosen and set on the potentiostat which, via its feedback system, transmitted sufficient current between the counter electrode and the specimen in order t o maintain the selected anodic potential between the specimen and the reference cell. The current, measured on a potentiometric recorder (Fischer Recordall Service 5000, Fisher Scientific, Chicago, Illinois) reached a steady state within 5 min, a t which time the weight was applied creating the applied fatigue stress and the specimen was run to failure with the potentiostat continually maintaining the selected potential.
RESULTS Figure 2 shows the S-N data (S, applied reversing stress, versus N , number of cycles to failure) for some mechanically polished 316 stainless steel and titanium specimens fatigued in air. I n the early stages for both materials, as the applied stress decreases the fatigue life increases. I n the later stages, limiting stress levels (endurance limits) are reached, below which specimens are capable of withstanding greater than lo7 cycles. Superimposed on the figure is data of others indicating good agreement for specimens tested in air.4,5 It was anticipated that the variables of surface finish and applied potential might significantly alter the character of the S-N curve (e.g., shift N a t a given level of S; reduce or obliterate the endurance limit). Since in this preliminary study only about 50 machined specimens of each material were used, the present investigation was limited t o the change in N produced by a given set of variables when testing a t a single level of S well above the endurance limit. To this end, all subsequent fatigue tests were conducted a t 360 MPa for 316 stainless steel and 270 MPa for titanium. Five specimens were run at each selected condition.
616
rf
H -400In In
r
-
0
0
0
316 Stainleas 000 0 0 0 0
3002
Eidl
W
.-mc L
al
Steel
e+
~zzzzzzz,
0 0
Titanium
200-
w
N
, Number
of
Cycles to
Failure
Fig. 2. Fatigue of 2 common annealed implant materials. Open points represent number of cycles t o failure at a given stress level for specimens of 600 grit finish tested at 3100 rpm in air. Closed points indicate that specimen has not yet failed. The line through the 316 stainless steel results and the band of endurance limits shown for the titanium represents data of others. 4 . 5
Figure 3 presents the anodic potentiostatic curves of both the 316 stainless steel and titanium measured with the specimen rotating at 3100 rpm. The curves were generated by beginning from OCP and then increasing the potential in 50 mV steps. After 5 min the current per surface area of metal in electrolyte was read and the next incre-
-u W v,
-
.-0
-500
316 Stainless
0
c
al +
+500 U
.-al
g+1000 a
.
~
~ l a d
d
10' Current Density ( mo/cmL)
IO-~
IO-~
lo-'
10'
Fig. 3. Anodic polarization curves of annealed titanium and 316 stainless steel in Ringer's solution while rotating at 3100 rpm. Data gathered by 50 mV increments of applied potential over and above the open circuit potential (a).
~
~
ELECTRICAL POTENTIAL AND SURFACE FINISH
617
ment initiated. Note that the 316 stainless steel displays a transition from a passive to a transpassive state around +350 mV (versus SCE), while the titanium remains passive up to a t least +SO0 mV, the highest potential applied. From these curves it was elected to employ 0, +200 mV (passive state for 316), and +500 mV (transpassive for 316) as the applied potentials upon the specimens submerged in Ringer’s solution. These millivoltages appear to be within the ranges of potentials actually observed on metallic prostheses within the body.14 Figure 4 is a bar graph showing the average and the standard deviation of the fatigue life for the 5 specimens cycled a t 3100 rpm to failure a t 360 MPa for each of the selected conditions imposed upon 316 annealed stainless steel. Data was compared using the Student’s statistical t-test, and all subsequent statements are a t a confidence level of 95% or greater. I n any one particular environment (i.e., air or any given potential), there does not appear to be any significant difference in fatigue life resulting from the various extremes in surface finishes. However, there are significant changes in the number of cycles to failure depending upon the specific environment. 316 Stainless Steel 150,000Air
W 0:
II T1
Cycled
at
t360MPlr
Solution
3
-I
2 0
w
I1
loop00
v)
W J
0
>. 0
7
50,000
z
0
Fig. 4. Fatigue life of 316 stainless steel cycling at 3100 rpm through f 360 MPa. The bar for each particular condition represents the average N for 5 specimens while the inner line indicates f S.D.
BAPNA ET AL.
618
In Ringer's solution, the fatigue life was drastically lowered by increasing millivoltage. With respect to specimens fatigued in air, 0 mV (i.e., simply submerging the specimen in Ringer's solution) increased fatigue life, +a00 mV produced no significant change, but +500 mV was certainly of shorter life. Definite surface pitting was found in all specimens run a t +500 mV. Figure 5 indicates the results for pure annealed titanium cycled a t 3100 rpm to failure at an applied stress level of 270 MPa. In sharp contrast to the 316 stainless steel, the titanium was very sensitive to its surface finish with eleCtropolishing apparently producing the best results. Also, completely different from the 316 stainless steel results was the fact that the applied potentials (always in the passive region) enhanced the fatigue life of titanium rather than decreasing it. Particularly, beginning with an electropolished surface, the fatigue life was markedly improved by the application of +500 mV.
DISCUSSION Clearly, the most important finding of this preliminary investigation is the recognition that a harsher and/or accelerated simulated biological environment can drastically influence the fatigue life of the two common implant materials tested. Moreover, in the case of 500,000~ W
Titanium
Cycled at
K
-
~ 2 7 MW 0
T Ringer's Solution
400,000
c
0 Iv)
W
z
50,000 0
0
Fig. 5. Fatigue life of pure titanium cycling at 3100 rpm through f 270 MPa. The bar for each particular condition represents the average N for 5 specimens while the inner line indicates f S.D.
ELECTRICAL POTENTIAL AND SURFACE FINISH
619
pure titanium, it appears that there are statistically significant differences in fatigue cycles to failure depending upon the surface finish employed. For statistical comparisons, one must know the particular distribution function of the data. Fatigue data often exhibits a rather large amount of scatter about the mean and the distribution is often skewed. The observed cumulative frequency of the 5 values of N for each group in Figures 4 and 5 were plotted on Gaussian probability paper and were found in all instances to produce a straight line of correlation coefficient greater than 0.90 when subjected to linear regression analysis. Therefore, having demonstrated that the distributions of N were normal, it was justified to analyze averages, standard deviations, and t values directly in terms of N rather than log N or having to utilize other common fatigue statistical functions such as extreme-value's or Weibull di~tributions.'~ The present fatigue tests were run a t a single speed of 3100 rpm. The very nature of an accelerated mechanical test precludes that specimens remain in an environment only for a relatively short time. Since corrosion is a time dependent phenomenon, the amount of corrosion would tend to decrease with increasing speed of testing. Also, a t very high rotational speeds (the apparatus is capable of rotating a specimen at 10,000 rpm) the electrolyte solution is hurled away from the interface thereby altering surface kinetics. Yet, most fatigue machines are meant to produce accelerated tests, and one could not reasonably reduce cyclic speeds to only several per minute. Because current densities during anodic polarization stabilized rapidly in the apparatus when cyclic speeds were below 5000 rpm, a margin of safety was introduced and 3100 rpm was chosen as the standard test speed. It should be noted (see Fig. 2) that the current densities observed under these dynamic test conditons were larger than those normally observed under static conditions but the breakdown potent i a l ~ were ' ~ essentially unchanged. The application of potential in Ringer's solution is apparently capable of producing or assisting more than one mechanism. I n the case of the 316 stainless steel the pronounced shortening of fatigue life a t +500 mV is possibly associated with a fatigue corrosion phenomenon. I n such an instance the development and growth of microcracks during cycling a t high stress is enhanced by the corrosive, environment, which in the +200- +500 mV range is near or in excess
620
BAPNA ET AL.
of the breakdown potential. I n contrast, pure titanium, whose notch sensitivity is evidenced by its degradation with coarser finishes, seems to have its surface oxide layer repaired by the applied potential, which at +500 mV is still in the titanium passive region. The effect of continual surface oxide film replenishment is to counteract any polishing grooves and to heal cracks formed by stress cycling. Thus with applied +500 mV the fatigue life of all the titanium specimens was extended. Comparing the no potential applied data, there appears to be an improvement in the fatigue life of the 316 stainless steel in Ringer’s solution over and above that in air. This is thought by the authors to be due to an improved heat transfer situation with the liquid bath providing enhanced specimen cooling during testing. I n this initial study not all the necessary fatigue variables have been explored. Still to be carefully studied are the various states of cold work, grain size, impurity contents, different alloys, lowering or disappearance of endurance limits, along with changes in applied potentials, current densities, and nature (composition, pH, and aeration) of electrolyte. It would also be of great interest to prestrain specimens, to give them a trauma type loading prior to fatigue, or to change the cyclic frequency and alter the pattern from reversed bending to a closer simulation of actual gait. Also, as suggested by the present experiments, it might be extremely useful, particularly in the case of titanium, to give a prepolarizing treatment before subjecting to fatiguing in the body environment. However, be that as it may, even from the limited amount of data from the present work it is certain that the biomaterials scientists and engineers must not be content with fatigue data in air. Materials for implantation must be tested under conditons which simulate or accelerate the types of environmental influences experienced in the body. This work was supported in part by the Social and Rehabilitation Services Research Grant No. 23-P-55898.
References 1. ASTM Specification F 55, Type A: Stainless Steel Bars and Wire for Surgical Implants. American Society for Testing and Materials, Philadelphia,
Pennsylvania, 1973. 2. ASTM SpecificationF 67, Gradek: Titanium for Surgical Implants, American Society for Testing and Materials, Philadelphia, Pennsylvania, 1973.
ELECTRICAL POTENTIAL AND SURFACE FINISH
621
3. J. R. Cahoon and H. W. Paxton, J. Biomed. Muter. Res., 2, 1 (1968). 4. H. J. Grover, J. Muter., 1, 413 (1966). 5 . Metals Handbook, 8th Ed., Vol. 1, American Society for Metals, Novelty, Ohio, 1961, p. 529. 6. R. Earnshaw, Br. Dent. J., 110,341 (1961). 7. G. E. Dieter, Mechanical Metallurgy, McGraw-Hill Book Company, New York, 1961, p. 296. 8. K. R. Wheeler and J. A. James, J. Biomed. Muter. Res., 5, 267 (1971). 9. V. J. Colangelo and N. D. Greene, J. Biomed. Muter. Res., 3, 247 (1969). 10. E. J. Kaminski, R. J. Oglesby, N. D. Wood, and J. Sandrik, J . Biomed. Muter. Res., 2, 81 (1968). 11. ASTM Specification G5: Standard Reference Method for Making Potentiostatic and Potentiodynamic Anodic Polarization Measurements, American Society for Testing and Materials, Philadelphia, Pennsylvania, 1973. 12. N. K. Sarkar and E. H. Greener, Biomut., Med. Dev., Art. Org., 1,121 (1973). 13. H. J. Mueller and E. H. Greener, J. Biomed., Muter. Res., 4, 29 (1970). 14. T. P. Hoar and D. C. Mears, Proc. Roy. Soc., A , 294,486 (1966). 15. J. M. Zarek, Metals Muter., 139 (1967). 16. Metals Handbook, 1948 Ed., American Society of Metals, Cleveland, Ohio, 1948, p. 118. 17. N. D. Greene, Experimental Electrode Kinetics, Rensselaer Polytechnic Institute, Troy, New York 1965. 18. A. M. Freudenthal and E. J. Gumbell, J . Amer. Statist. Ass., 9, 575 (1954). 19. W. Weibull, J. Appl. Much., 18, 293 (1951).
Received November 27, 1974 Revised February 11, 1975
VOL. 9, PP. 611-621 (1975)
The Influences of Electrical Potential and Surface Finish on the Fatigue Life of Surgical Implant Materials M. S. BAPNA,* E. P. LAUTENSCHLAGER, and J. B. MOSER, Department of Biological Materials, Northwestern University, Chicago, and P. R. MEYER, JR., Department of Orthopaedic Surgery, Northwestern University, Chicago, Illinois 60611
Summary The influence of both imposed anodic potential in Ringer’s solution and surface finish on the fatigue lives of annealed 316 type stainless steel and annealed pure titanium were measured and statistically compared to fatigue data run in air. The applied potentials in simulated extracellular fluid approximated conditions existing within the body while also producing the types of surface defects actually found on removed long time implants within the time interval of the accelerated R. R. Moore fatigue tests. Differentiating tests were run a t single levels of applied cyclic stress well above the endurance limits. I n Ringer’s solution, the fatigue life of the 316 stainless steel decreased with increasing applied potential, and a t +SO0 mV was significantly shorter than when run in air. At each condition, the 316 stainless steel was independent, of initial surface finish. I n contrast, the fatigue life of titanium improved rapidly with increasingly fine surface finishes. Furthermore, compared to air, the application of +500 mV in Ringer’s solution improved the life of the rough surface finished material and markedly increased the number of cycles to failure for the electropolished specimens.
INTRODUCTION Certain metal alloys are now routinely used as surgical implant materials’ for prostheses and fixation devices. While having demonstrated acceptable strength and corrosion resistance in many biological applications, metallurgical analyses of broken orthopaedic implants indicate fatigue under cyclic loading as a major cause of s2
*Present address: University of Illinois Dental School, Chicago, Illinois. 611 @ 1975 by John Wiley & Sons, Inc.
612
BAPNA ET AL.
f a i l ~ r e . ~Moreover, whenever new designs are contemplated, particularly thinner devices requiring less removal of natural tissue, fatigue life should be a major design criterion. Unfortunately for the surgeon and the designer, with the exception of some fatigue data run in air,4-6there is a dearth of statistically significant information concerning the behavior of surgical implant metals during repeated cyclic loading in vivo or in simulated body environments. I n addition, little or no attention has been systematically focused upon those factors which are known to influence the fatigue life of metals under certain circumstance^,^ namely : compositional inhomogeneity, state of cold work, phase distribution, impurity content, nature and amplitude of applied stress, surface finish, etc. Although Wheeler and Jamess had demonstrated that crack growth rate in 316 type stainless steel was faster in Ringer’s solution than in air, preliminary studies in our laboratory revealed no significant decrease in the actual fatigue life of that material when tested in those two different environments. Also, simply running fatigue tests for several days in Ringer’s solution did not produce the type of crevice attack or surface pitting often observed in removed long time 316 type stainless steel implants>JO Anodic potentiostatic measurements” have successfully differentiated between the electrochemical stability of passivating oxide films of the common surgical implant alloys,12as well as demonstrated the capability of producing such oxide film breakdown phenomena aa pits.I3 With the harsh environment presented by the body, plus the ever present millivoltage signals and electrical potential differences existing in implants within the body,I4 it would appear that testing in air or in static simulated extracellular body fluid is not a sufficiently good approximation of the actual in vivo situation. Moreover, since the standard fatigue test is an accelerated life test, the environmental effects must also be speeded up to comply within the testing time. Therefore, we have chosen to conduct our fatigue experiments in a manner similar to that implied by Zarek,15with an external electrical potential imposed on a specimen being cyclically loaded in a simulated body fluid environment. The purpose of this investigation will be to determine the effects of different magnitudes of anodic polarization upon the fatigue life of
ELECTRICAL POTENTIAL AND SURFACE FINISH
613
two types of surgical implant metals; namely, 316 stainless steel (type A),* and pure titanium (grade 4),t when run in Ringer’s solution and possessing a number of different surface finishes. These results will be compared t o similar groupings tested in air.
METHODS The basic technique employed in this investigation was t o determine the number of cycles t o failure a t a selected applied stress for a given material of specific surface condition (transverse 60 grit, longitudinal 600 grit, or electropolished) in a particular environment (air or Ringer’s solution, plus imposed levels of anodic potential from 0 t o 500 mV). The testing apparatus (shown schematically in Fig. 1) was a n R. R. Moore type16 rotating beam fatigue machine modified with a chamber around the specimen t o permit immersion in a liquid and application of a n electrical potential. The applied weights set up a uniform bending movement on the specimen which is transformed into a continuous, completely reversing bending stress during cycling. The machine was run a t 3100 f 100 rpm (approximately 4.5 million cycles per day). Specimens were machined from single rods of purchased 12.7 mm round stock in a n annealed condition. The standard machined specimens were hour-glass shaped16 with a continuous decrease in circular cross section to the center. The Rockwell 15-N Superficial Hardness of all finished specimens was very uniform on both the surfaces and the later sectioned interiors, and found t o be 66 f 2 for the 316 stainless steel and 56 k 2 for the pure titanium. Each machined specimen was given one of the following surface finishes: (a) coarse mechanical, 60 grit transverse t o specimen long axis; (b) fine mechanical, 600 grit longitudinal; and (c) electropolished. The mechanical finishes were imparted with proper grit size emery paper being applied t o a rotating specimen. The electropolishing conditions were as follows: for the 316 stainless steel, electrolyte of 600 ml H3P04and 400 ml HzSOr a t 25°C run for 6 min a t *0.08 C max, 2 . 0 Mg max, 0.025 P max, 0.010 S max, 0.75 Si max, 17.00-20.00 Cr, 12.00-14.00 Ni, 2.00-4.00 Mo, remainder Fe; annealed condition; RN = 6 6 f 2 ; grain size 8. to .07 N max, 0.15C max, 0.0125 H max, 0.50 Fe max, 0.40 0 max, remainder Ti; annealed condition; RN = 56 f2; grain size 7.
BAPNA ET AL.
614
7 To Potentiostoi
\
\
\ \
%
Specimen (Working Electrode) Platinum (Counter Electrode) Ringer's Solution
1
L S o t u r a ted KCI Solution
Fig. 1. Schematic of R. R. Moore type fatigue machine equipped with chamber for physiological saline and application of anodic potentials.
3 V and 3 A; for the titanium, electrolyte of 943 ml glacial acetic acid and 57 ml of perchloric acid at -2OOC run for 3 min at 30 V. I n cases of applied potential, the electrochemical cell operation was similar to that described elsewhere."J7 A surrounding acrylic box (see Fig. 1) allowed the central 25 mm of the specimen to be completely immersed in Ringer's solution (9.000 g NaC1, 0.425 g KC1, 0.119 g CaC12, 0.100 g NaHC03, made up to 1 1. with deionized water and buffered to pH 7). Passing through stationary circular rubber seals in the ends of the box, the specimen could rotate freely but solution leakage was minimized. Also placed in the solution in close proximity to the rotating fatigue specimen was the open tip of a Luggin-Haber type salt bridge leading to a saturated calomel reference electrode (SCE) and a platinum wire counter electrode. The specimen became the working electrode (anode) and was electrically insulated from the rest of the fatigue machine by specially machined
ELECTRICAL POTENTIAL AND SURFACE FINISH
615
polytetrafluorethylene grips. The entire apparatus was housed in a wire mesh Faraday cage grounded a t a Wenking Potentiostat (Wenking 70TS1 Electronic Potentiostat, Brinkman Instruments, Westbury, New York). The specimen was placed in the apparatus, surrounded by Ringer’s solution and rotation begun with no weight applied. For the first hr the millivoltage generated between the specimen, and the reference cell was allowed t o come to a steady state known as the open circuit potential (OCP). A specified anodic millivoltage over and above the OCP was then chosen and set on the potentiostat which, via its feedback system, transmitted sufficient current between the counter electrode and the specimen in order t o maintain the selected anodic potential between the specimen and the reference cell. The current, measured on a potentiometric recorder (Fischer Recordall Service 5000, Fisher Scientific, Chicago, Illinois) reached a steady state within 5 min, a t which time the weight was applied creating the applied fatigue stress and the specimen was run to failure with the potentiostat continually maintaining the selected potential.
RESULTS Figure 2 shows the S-N data (S, applied reversing stress, versus N , number of cycles to failure) for some mechanically polished 316 stainless steel and titanium specimens fatigued in air. I n the early stages for both materials, as the applied stress decreases the fatigue life increases. I n the later stages, limiting stress levels (endurance limits) are reached, below which specimens are capable of withstanding greater than lo7 cycles. Superimposed on the figure is data of others indicating good agreement for specimens tested in air.4,5 It was anticipated that the variables of surface finish and applied potential might significantly alter the character of the S-N curve (e.g., shift N a t a given level of S; reduce or obliterate the endurance limit). Since in this preliminary study only about 50 machined specimens of each material were used, the present investigation was limited t o the change in N produced by a given set of variables when testing a t a single level of S well above the endurance limit. To this end, all subsequent fatigue tests were conducted a t 360 MPa for 316 stainless steel and 270 MPa for titanium. Five specimens were run at each selected condition.
616
rf
H -400In In
r
-
0
0
0
316 Stainleas 000 0 0 0 0
3002
Eidl
W
.-mc L
al
Steel
e+
~zzzzzzz,
0 0
Titanium
200-
w
N
, Number
of
Cycles to
Failure
Fig. 2. Fatigue of 2 common annealed implant materials. Open points represent number of cycles t o failure at a given stress level for specimens of 600 grit finish tested at 3100 rpm in air. Closed points indicate that specimen has not yet failed. The line through the 316 stainless steel results and the band of endurance limits shown for the titanium represents data of others. 4 . 5
Figure 3 presents the anodic potentiostatic curves of both the 316 stainless steel and titanium measured with the specimen rotating at 3100 rpm. The curves were generated by beginning from OCP and then increasing the potential in 50 mV steps. After 5 min the current per surface area of metal in electrolyte was read and the next incre-
-u W v,
-
.-0
-500
316 Stainless
0
c
al +
+500 U
.-al
g+1000 a
.
~
~ l a d
d
10' Current Density ( mo/cmL)
IO-~
IO-~
lo-'
10'
Fig. 3. Anodic polarization curves of annealed titanium and 316 stainless steel in Ringer's solution while rotating at 3100 rpm. Data gathered by 50 mV increments of applied potential over and above the open circuit potential (a).
~
~
ELECTRICAL POTENTIAL AND SURFACE FINISH
617
ment initiated. Note that the 316 stainless steel displays a transition from a passive to a transpassive state around +350 mV (versus SCE), while the titanium remains passive up to a t least +SO0 mV, the highest potential applied. From these curves it was elected to employ 0, +200 mV (passive state for 316), and +500 mV (transpassive for 316) as the applied potentials upon the specimens submerged in Ringer’s solution. These millivoltages appear to be within the ranges of potentials actually observed on metallic prostheses within the body.14 Figure 4 is a bar graph showing the average and the standard deviation of the fatigue life for the 5 specimens cycled a t 3100 rpm to failure a t 360 MPa for each of the selected conditions imposed upon 316 annealed stainless steel. Data was compared using the Student’s statistical t-test, and all subsequent statements are a t a confidence level of 95% or greater. I n any one particular environment (i.e., air or any given potential), there does not appear to be any significant difference in fatigue life resulting from the various extremes in surface finishes. However, there are significant changes in the number of cycles to failure depending upon the specific environment. 316 Stainless Steel 150,000Air
W 0:
II T1
Cycled
at
t360MPlr
Solution
3
-I
2 0
w
I1
loop00
v)
W J
0
>. 0
7
50,000
z
0
Fig. 4. Fatigue life of 316 stainless steel cycling at 3100 rpm through f 360 MPa. The bar for each particular condition represents the average N for 5 specimens while the inner line indicates f S.D.
BAPNA ET AL.
618
In Ringer's solution, the fatigue life was drastically lowered by increasing millivoltage. With respect to specimens fatigued in air, 0 mV (i.e., simply submerging the specimen in Ringer's solution) increased fatigue life, +a00 mV produced no significant change, but +500 mV was certainly of shorter life. Definite surface pitting was found in all specimens run a t +500 mV. Figure 5 indicates the results for pure annealed titanium cycled a t 3100 rpm to failure at an applied stress level of 270 MPa. In sharp contrast to the 316 stainless steel, the titanium was very sensitive to its surface finish with eleCtropolishing apparently producing the best results. Also, completely different from the 316 stainless steel results was the fact that the applied potentials (always in the passive region) enhanced the fatigue life of titanium rather than decreasing it. Particularly, beginning with an electropolished surface, the fatigue life was markedly improved by the application of +500 mV.
DISCUSSION Clearly, the most important finding of this preliminary investigation is the recognition that a harsher and/or accelerated simulated biological environment can drastically influence the fatigue life of the two common implant materials tested. Moreover, in the case of 500,000~ W
Titanium
Cycled at
K
-
~ 2 7 MW 0
T Ringer's Solution
400,000
c
0 Iv)
W
z
50,000 0
0
Fig. 5. Fatigue life of pure titanium cycling at 3100 rpm through f 270 MPa. The bar for each particular condition represents the average N for 5 specimens while the inner line indicates f S.D.
ELECTRICAL POTENTIAL AND SURFACE FINISH
619
pure titanium, it appears that there are statistically significant differences in fatigue cycles to failure depending upon the surface finish employed. For statistical comparisons, one must know the particular distribution function of the data. Fatigue data often exhibits a rather large amount of scatter about the mean and the distribution is often skewed. The observed cumulative frequency of the 5 values of N for each group in Figures 4 and 5 were plotted on Gaussian probability paper and were found in all instances to produce a straight line of correlation coefficient greater than 0.90 when subjected to linear regression analysis. Therefore, having demonstrated that the distributions of N were normal, it was justified to analyze averages, standard deviations, and t values directly in terms of N rather than log N or having to utilize other common fatigue statistical functions such as extreme-value's or Weibull di~tributions.'~ The present fatigue tests were run a t a single speed of 3100 rpm. The very nature of an accelerated mechanical test precludes that specimens remain in an environment only for a relatively short time. Since corrosion is a time dependent phenomenon, the amount of corrosion would tend to decrease with increasing speed of testing. Also, a t very high rotational speeds (the apparatus is capable of rotating a specimen at 10,000 rpm) the electrolyte solution is hurled away from the interface thereby altering surface kinetics. Yet, most fatigue machines are meant to produce accelerated tests, and one could not reasonably reduce cyclic speeds to only several per minute. Because current densities during anodic polarization stabilized rapidly in the apparatus when cyclic speeds were below 5000 rpm, a margin of safety was introduced and 3100 rpm was chosen as the standard test speed. It should be noted (see Fig. 2) that the current densities observed under these dynamic test conditons were larger than those normally observed under static conditions but the breakdown potent i a l ~ were ' ~ essentially unchanged. The application of potential in Ringer's solution is apparently capable of producing or assisting more than one mechanism. I n the case of the 316 stainless steel the pronounced shortening of fatigue life a t +500 mV is possibly associated with a fatigue corrosion phenomenon. I n such an instance the development and growth of microcracks during cycling a t high stress is enhanced by the corrosive, environment, which in the +200- +500 mV range is near or in excess
620
BAPNA ET AL.
of the breakdown potential. I n contrast, pure titanium, whose notch sensitivity is evidenced by its degradation with coarser finishes, seems to have its surface oxide layer repaired by the applied potential, which at +500 mV is still in the titanium passive region. The effect of continual surface oxide film replenishment is to counteract any polishing grooves and to heal cracks formed by stress cycling. Thus with applied +500 mV the fatigue life of all the titanium specimens was extended. Comparing the no potential applied data, there appears to be an improvement in the fatigue life of the 316 stainless steel in Ringer’s solution over and above that in air. This is thought by the authors to be due to an improved heat transfer situation with the liquid bath providing enhanced specimen cooling during testing. I n this initial study not all the necessary fatigue variables have been explored. Still to be carefully studied are the various states of cold work, grain size, impurity contents, different alloys, lowering or disappearance of endurance limits, along with changes in applied potentials, current densities, and nature (composition, pH, and aeration) of electrolyte. It would also be of great interest to prestrain specimens, to give them a trauma type loading prior to fatigue, or to change the cyclic frequency and alter the pattern from reversed bending to a closer simulation of actual gait. Also, as suggested by the present experiments, it might be extremely useful, particularly in the case of titanium, to give a prepolarizing treatment before subjecting to fatiguing in the body environment. However, be that as it may, even from the limited amount of data from the present work it is certain that the biomaterials scientists and engineers must not be content with fatigue data in air. Materials for implantation must be tested under conditons which simulate or accelerate the types of environmental influences experienced in the body. This work was supported in part by the Social and Rehabilitation Services Research Grant No. 23-P-55898.
References 1. ASTM Specification F 55, Type A: Stainless Steel Bars and Wire for Surgical Implants. American Society for Testing and Materials, Philadelphia,
Pennsylvania, 1973. 2. ASTM SpecificationF 67, Gradek: Titanium for Surgical Implants, American Society for Testing and Materials, Philadelphia, Pennsylvania, 1973.
ELECTRICAL POTENTIAL AND SURFACE FINISH
621
3. J. R. Cahoon and H. W. Paxton, J. Biomed. Muter. Res., 2, 1 (1968). 4. H. J. Grover, J. Muter., 1, 413 (1966). 5 . Metals Handbook, 8th Ed., Vol. 1, American Society for Metals, Novelty, Ohio, 1961, p. 529. 6. R. Earnshaw, Br. Dent. J., 110,341 (1961). 7. G. E. Dieter, Mechanical Metallurgy, McGraw-Hill Book Company, New York, 1961, p. 296. 8. K. R. Wheeler and J. A. James, J. Biomed. Muter. Res., 5, 267 (1971). 9. V. J. Colangelo and N. D. Greene, J. Biomed. Muter. Res., 3, 247 (1969). 10. E. J. Kaminski, R. J. Oglesby, N. D. Wood, and J. Sandrik, J . Biomed. Muter. Res., 2, 81 (1968). 11. ASTM Specification G5: Standard Reference Method for Making Potentiostatic and Potentiodynamic Anodic Polarization Measurements, American Society for Testing and Materials, Philadelphia, Pennsylvania, 1973. 12. N. K. Sarkar and E. H. Greener, Biomut., Med. Dev., Art. Org., 1,121 (1973). 13. H. J. Mueller and E. H. Greener, J. Biomed., Muter. Res., 4, 29 (1970). 14. T. P. Hoar and D. C. Mears, Proc. Roy. Soc., A , 294,486 (1966). 15. J. M. Zarek, Metals Muter., 139 (1967). 16. Metals Handbook, 1948 Ed., American Society of Metals, Cleveland, Ohio, 1948, p. 118. 17. N. D. Greene, Experimental Electrode Kinetics, Rensselaer Polytechnic Institute, Troy, New York 1965. 18. A. M. Freudenthal and E. J. Gumbell, J . Amer. Statist. Ass., 9, 575 (1954). 19. W. Weibull, J. Appl. Much., 18, 293 (1951).
Received November 27, 1974 Revised February 11, 1975
