Wear 205 (1997) 130--136
Ultrasonic in situ continuous wear measurements of orthopaedic titanium alloys M. Long, Hj. Rack * School of Chemical and Materials Engineering. Clemson University. Clemson. SC 29634, USA
Received 10 April 1996; ~cepted 17 May 1996
Abstract An ultrasonic pulse-echo technique has been demonstrated for in-situ continuous wear me~asurements.This method, based on the difference in the time of flight as recorded and converted into change of length as wear progresses, was shown to give accurate and reproducible results when compared to the conventional mass loss procedure. Additional information regarding subsurface deformation was obtained with the ultrasonic technique, strain hardening, grain refinement, crystallographic reodentation caused by sliding-induced deformation all being detectable. © 1997 Elsevier Science S.A. All rights reserved. Keywords: Ultrasonic;Wear;,Titanium;Orthopaedic
1. Introduction In-vitro wear measurements including for example pin-ondisc or pin-on-block configurations generally involve interrupted data collection of either mass loss or dimensional change [ 1,2], as recommended per ASTM G99-90. These interruptions have the potential of modifying the continuous sliding conditions since materials are removed, cleaned, measured and then replaced for continued testing. Moreover, the monitoring procedure may be long and laborious, and indeed may not be possible as may be the case with orthopaedic joint simulators. This paper demonstrates a continuous in-situ wear measurement technique based on application of ultrasonic principles [ 3,4]. Ultrasounds are defined as vibrations of a materials medium  at frequencies that are too high to be detected by an average human ear ( > 10-18 kI-lz) . Schematically, atoms in a crystal lattice may be represented as a network of elementary masses (atoms) connected to each other by elementary springs (atomic bonds). When a plane of elementary masses is displaced by an external oscillatory force, all material elements on this plane are excited collectively in step with the imposed oscillations. If all material elements were connected rigidly they would all remain in the same oscillating state or phase. In an elastic material, the elastic connections delay the transmission of motion and * Con-~7,ndh3gatathor.
introduce a time delay, i.e. a phase lag. This phenomenon is described as an elastic wave, called ultrasound when their frequency is higher than the audible range. Two types of vibration-indnced deformation resulting in compression or shear are generally possible in an infinite medium, these being consequently associated with two modes of ela:;tic wave propagation: (1) longitudinal waves (pressure or sound waves), where the particles are displaced along the direction of propagation, see Fig. i (a); (2) transverse waves (shear waves), where the particles arc displaced perpendicular to the direction of propagation, as shown in Fig. l ( b ) . Such ::': " "'=::. . ""7": " :::::::::::: .°° ° • • °°. . . . . •.••• ,.•
• ° • ° °
• • • ° , °
• ° . , ° °
° . , ~ , °
•°~°•• .... ...
° • • . • J ° , , ° ° •
° ° ° * * ° , , • ° ° °
~ . ~ , ° o ° ° • o
• • °.°°°.
/.ii i ii iii i ii'/ i i iii iiiii i :,:" : ::27:" " ::~:'." : :'"'.".".'.', O~:l~n Of ~ r tlcle
ma~t~aen Fig. I. Wavepropagationmodesof ultntsonicsignal: (a) longitudinalwave; (b) transve~ewave.
0043-1648/97/$17.00 © 1997 Publishedby ElsevierScience S.A. All rightsreserved Pll S0043-1648 (96) 07233-X
M. 1.amg.H.J. RactlWear 205 fl~7) 150-156
elastic waves may be defined by several parameters: the frequency (number of oscillations per second) and wavelength (distance between two planes in which the particles are in the same sta~c of motion) of a wave; the velocity of propagation ( speed of sound) which is a characteristic of the material; and the sound pressure (imposed by oscillations). Ultrasonic waves are generated in the material by a transmitler oscillating at the desired waveform and frequency, with collection by a detector or microphone. Both transmitting/ receiving devices rely on the piezoelectric effect, i.e. the bilateral property of selected crystals to produce electrical charges at their surface when deformed by external mechanical pressure. Adequate coupling between the piezoelectric transducer and the test material is also required to minimize loss of wave energy at the interface between transdncer and material. The "fit" at the transducer-material interface, and therefore the quality of ultrasonic signal, is reduced by rough (peak-to-valley value more than ! / 10 of wavelength) and/ or curved and/or contaminated surfaces as well as by poor acoustic impedance of the coupling medium. A variety of coupling media are used to achieve effective coupling, with glycerine, oils, grease and petroleum jelly (vaseline) being the most common. The present ultrasonic method for wear measurements is based on the pulse-echo technique [5-7], see Fig. 2, a method (ASTM-E797-87) commonly employed in sonar applications, medical diagnosis, flaw d e ~ t i o n and ultrasonic thickness gages [5,6]. The present method relies on the variation in the difference in time between signal emission and echo reception of an ultrasonic pulse, i.e. the time of flight (ToF), taken as a measure of the change in surface location with sliding distance. In-situ ultrasonic measurementsof sam8ecxwoa
I =,~ Fig. 2. Illustration of the pulse-echo ultrasonic technique .
pie thickness allow a continuous recording of dimensional change with the ultrasonic wobe being coupled to the test material. As the sample wears, the time of flight of the pulses, and therefore the sample thickness, changes as the smface is relocated.
2. Experimental details Dry unidirectional sliding and reciwocating sliding, utilizing respectively pin-on-disk and pin-on-block configurations, were evaluated. Two a/[3 orthopaedic grade titanium alloys, Ti-6AI-TNb ELI and Ti-6AI-4V ELI, were utilized as pin materials, the counterpart in each instance being hardened steel (HRc = 60). Tribological evaluation of titanium alloys consisted of the determination of friction coefficients and wear rates by both mass loss and ultrasonic methods, characterizatiou of the wear samples including examination of the wear surface of both Ti-alloy and steel counterparts by SEM and EDX. Subsurface regions of Ti-alloy pins were also examined, the pins being carefully sectioned parallel to the sliding direction, one-half being mounted, polished and etched using conventional metallographic methods. Apparent contact stresses of 0.8 and 1.1 MPa, i.e., normal loads of 16 and 70 N, and average linear velocity of 0.8 and 0.~ m s (1 Hz, 25 mm stroke), were used for unidirectional and recilYrocating sliding experiments, respectively, a minimum of three replicate tests being performed. Pricer to an experiment the steel counterparts contact surfaces were polished to a mirror finish with I p,m alumina powder, the Ti pins contact surfaces being polished in-situ with 600 grit SiC paper. Wear of the Ti pins was determined every 500 m increment by both mass loss ( +0.01 rag) and ultrasonic methods. The former were convened into length from a knowledge of the pin geometry and alloy density, densities being taken as 4.52 (Ti-6AI-TNb) and4.4 (Ti-6AI-4V) g cm -3 .LengthvalueS were obtained from centerline measurements of ToF as l=(ct×ToF)/2, per ASTM E797-87, where ct. is the longitudinal wave velocity, initially determined per ASTME494-92 by Ultran Laboratories, Boalsburg (PA), as 6010 (Ti-6AI-TNb) and 6140 (Ti-6AI-4V) m s - t. For each pin, the initial length determined by ultrasonic measurement was calibrated vs. the true length value determined by mass loss. The ultrasonic apparatus is schematically represented in Fig. 3 and included an ultrasonic pulser/receiver, a highfrequency oscilloscope with storage capability, and a transmitter/receiver ultrasonic transducer appropriate for titanium alloys [ 9 ]. The oscilloscope was interfacedto a PC for further recording of the ultrasonic signals, a typical display being shown in Fig. 4. For the incremental and in-situ thickness measurements, the pin sample was pressed against the transducer, commercial [email protected]
( 100% pure petroleum jelly) being used as the coupling medium. The back-face of the pin in contact with the transducer/coupling medium was polished on 1000 grit SiC paper prior to wear testing in order to further improve the transducer-pin interface contact, and therefore
M. Long. H.J. Rack / Wear 205 (1997) 130-136
I I ToVC.~ ""~'l DATAACQUISrrlON [
Fig. 3. Schematic o f the experimental set-up for ultrasonic measurement of pin thickness. 0.3 ¸
wear conditions appeared to be established within the first 100m of sliding travel, steady-state linear wear rates (Ixm kin-*) as determined by linear regression analysis (correlation factor R2 > 0.999 in all cases) of the slope of the length loss--sliding distance plots being given in Tables 1 and 2. No statistically significant difference (P,midi~,~o,~-~,~nS >0.96; p~.ipmcating.sndiag>0.74) between the two methods was resolved, small overall standard deviation values of wear rates being observed for both mass loss ( 0.7% ) and ultrasonic ( 2% ) methods. Excellent replication ( cr < 0.5% ) was again observed for both methods, see Tables 1 and 2. Based on all experimental measurements, a precision in the p.m range, a resolution of 10 Ixm (oscilloscope time scale vs. pin length), and a reproducibility of + 10 ttm, could be assessed. Comparison of the results from ultrasonic and mass loss measurements during unidirectional sliding showed greater
2 ~ bact-e~ho peak [
a Ma.~ l . e ~ • UlU-~..umad
0,3 , 0.2 41.1
4 T~ ( ~ )
Fig. 4. Typical pulse-echo signal display (Ti-6AI-7Nb pin).
2 3 Sliding Distance (kin)
Fig. 6. Typical wear plot for unidirectional sliding (Ti-6AI-TNb pin ). 1.5
B u r n s Rod
, • , 2 3 Sliding Distance (km)
Fig, 7. Typical wear plot for reciprocating sliding (Ti-6AI-4V pin). Table I Wear rate results (unidirectional sliding, Ti-6AI-TNb, staady-state regime)
Linear bearings I B t~hings
Wear rate (ttm km - *)
Fig. 5. Schematic of in sita ultr~onic measurements during reciprocatingsliding motion.
Incremental mass loss" Inc~mental ultrasonic ~
58 + 3 58 4- 4
0.3% (p > 0.96)
the ultrasonic signal, by minimizing surface roughness [ 10]. Finally continuous, in-situ ultrasonic measurements were also perforraed in reciprocating sliding, the pin holder being modified to hold the ultrasonic probe, see Fig. 5.
• Linear reg~ssion within staady-state regime (R 2 > 0.999).
3. Results Wear measurements are shown in unidirectional and reciprocating sliding in Figs. 6 and 7, respectively. Steady-state
Table 2 Wear rate results (reciprocating sliding, Ti-6AI-4V, steady-stata regime) Method
Wear rate (p.m k m - 1)
Incremental m a ~ loss' Incremental ultrasonic" In situ mass loss* In sita ultrasonic ~
2784-2 279 4- 5 281 + 7 282 4-9
0.4% ( p > 0 . 7 4 ) 0.4% ( p > 0.74)
• Linear regression within steady-state regime (R 2 > 0.999).
M. Long. ff,$. Rack~Wear 205 fi997) 150-136
Fig. 8. Typical SF,M micrograph of (a) t i t m i m alloy m d (b) steel surfaces ( unidirectional sliding, I0 km sliding distance). The arrows indicate sliding direction.
measured length loss values, i.e. shorter pin length, by the ultrasonic method than with conventional mass loss method (see Fig. 6), the difference remaining constant within the steady-state regime. This statistically relevant difference, ranging between 15 and 35 p,m, was consistent through :ill replicate tests, an average value for this thicknes~ deviation being 25 + 7 tim. In contrast, during reciprocating sliding, a statistically insignificant length difference of 5 pan between ultrasonic and mass loss method as shown in Fig. 7 was observed. Microscopic observation of the sample surfaces, see Fig. 8, revealed that adhesive wear was the principal wear mechanism for the materials and experimental conditions considered. Subsurface examination following unidirectional sliding of Ti-6AI-TNb (Fig. 9) showed that the microstrncture of this two-phase ( a + fl) alloy was progressively refined and reoriented parallel to the sliding direction with decreasing depth. The mechanically mixed surface region, zone HI as defined by Rice et al. [ 11 ], had an average thickness of I0 p.m. Below this zone, the plastic zone (zone H), where refinement and plastic deformation occurred, had an average thickness of 30-40 p,m, see Fig. 9(a). Following reciprocating-sliding [Fig. 9(b)], a compositionally mixed layer having an average thickness of 5-10 p.m
u ~ !
Fig. 9. gEM n,acrograph of subsurface region in (a) unidirectional sliding ( 10 km sliding dismu~ ) mul (b) reciprocating sliding mode (~ kin slklbg
d ~ ) . A typicalschematic(¢) ~ of the vmiO~ sulk.reface zone~isgiveaf~ refenmce(~lat~d fromRef. [ ! 1]). Thearrowsiadica~ slidingdin~ie~. was again observed by EDX, the plasticzone, composed of altematnd reorientedand refineda grains, extending to a 2530 pan depth from the wear surface. Finally, Fig. 10 shows an estimate of the subsurface shear swain as a function of distance below the contact surface, as determined by considering the reorientation of the a grains, the shear strain being given by 7 = tan 0/~/3 [ 12], where 0is the orientation angle as a function of depth. The maximum shear strain was lower in recil~ocating (---0.5) than in unidirectional sliding ( -- 1.7), subsurface shear strain decreasing with increasing depth, with a more gradual gradient being observed in reciprocating-sliding.
M, Long, H.J. Racl~/ Wear 205 (1997) 130--136
! ~ • Unklhec~ie~('ri.tAI-TNb) o Rcc~ ¢ri-tAl4V)
15 20 25 30 35 40 45 50
DistancefromSttrface (lan) Fig. 10. Subsurfaceshearstrainsin steady-statewearregimeas a function of subsurfacedepth.
4. Discussion No statistically relevant difference in wear rates could be resolved between the ultrasonic ToF and mass loss methods. The excellent reproducibility of wear rate values (Tables I and 2), demonstrates that the ultrasonic ToF method can be utilized for the determination of metallic alloys wear rates in both unidirectional- and reciprocating-sliding motion. Differences do, however, exist in the absolute sample length obtained from mass measurements and ultrasonic computations. These can be correlated with the subsurface deformation as the ultrasonic waves are altered by the heavily deformed near-sarface contact zone observed in the wear samples cross-sections (Fig. 9), where refinement and crystallographic reorientation of the original microstructure are observed. Observation of the subsurface deformation features induced by unidirectional and/or reciprocal sliding motion suggests that the variation of length in ultrasonic measurements is not due to the presence of cracks, but rather is due to refinement and crystallographic reorientation of the original microstructure. As sliding proceeds, a subsurface plastic zone (zone II) is formed, with its thickness increasing with sliding distance until a steady state is reached. Thus, the manifestation of a deformed subsurface zone precludes the use of a simple length computation from ultrasonic time values, i.e. 1= ( cL × ToF ) /2, the assumption of a constant ultrasonic velocity not applying within the altered subsurface region [5,6,13,14]. Limited studies of micro~ ~roctural effects on a material acoustic properties have shown that strain hardening, grain size and crystallogarphie orientation all modify the acoustic
properties of polycrystalline materials [ 14-17]. The modification of the ultrasonic velocity in the near-surface zone examined in this study can then be considered to arise from a combination of crystallographic reorientation, strain hardening and grain size refinement, all induced by sliding contact. Crystallographic reor/entatiou is commonly encountered in the near-surface region of wear samples where textures develop by rotation of crystal planes as a consequence of sliding contact [ ! 8,19]. The magnitude of these changes may be estimated by comparing longitudinal velocity values (as per ASTM-F.494-92) between the rolling and transverse directions . When considered in this investigation, an increase of 2-3% in the ultrasonic velocity in the transverse direction relative to that measured in the rolling direction was noted for both titanium alloys. The strain hardening effect can be evaluated by measurements of ultrasonic velocity in deformed Ti-6AI--4V compression samples [ 20], a decrease of 1.5 and 3% being observed following strains of 0.5 and 2, respectively. Finally, the dependency of ultrasonic attenuation upon grain size may be estimated from the prior studies of Vary [ 7] who found a velocity decrease of 2.5% associated with a 4 p.m reduction in grain size in Ti-gMo-8V-2Fe-3AI. For conditions of unidirectional sliding, a reduction of - 10 jzm in grain size can be estimated from Fig. 9(a), i.e. a 6% decrease in velocity. Considering the aforementioned factors, the modified velocity, Cs, at the surface of the pin was 1.02 X 0.97 X 0.94 × CL= 5589 m s - ' . Since the subsurface features ob~rved in the wear samples alter ultrasonic signal propagation and modify the specific ultrasonic velocity of the material, it is hypothesized that sliding-induced subsurface deformation results in an ultrasonic velocity gradient, from ct. (base material) to Cs (maximum or minimum velocity at the wear surface). The resulting procedure of computing length from the measured time must then consider the extent of the damage zone and the consequent velocity deviation. A model to evaluate the effects of subsurface deformation and reorientation on ultrasonic measurements, illustrated in Fig. 11, is proposed which distinguishes between ultrasonic wave propagation in the base material ( constant ultrasonic velocity ) and the deformed near-surface zone (velocity gradient). In the steady-state wear rate regime, the measured ToF, tu, can be written as
Fig. 1I. Descriptionof the proposedulti'asonicsul ~urfacemodel.
M. Long. H.J. Rack/Wear205 (1997) 130-136
tu=tUase mau~at"t"/subs,trace= 2
where L is the physical length of the pin, Z is the depth of the subsurface zone, and ct. is the original ultrasonic velocity measured in the base material. If the dependence of ultrasonic velocity upon subsurface deformation and reorientation is assumed to be a linear function of ckpth, then x c(x) = c L + (Cs- CL)~
where Cs is a value of the modified ultrasonic velocity at the surface of the pin and x is the distance from the bulk-subsurface interface. The ultrasonic thickness, u, of the near-surface layer is then given by . U'~ t,.b,~,~c tCs-CO~] -~
/4 (CL + "
Numerically, Z can be obtained from Fig. 9, t..~e,~,f~e calculated from Eq. ( 1 ) using t, as measured by the ultrasonic equipment, and u set equal to Z minus the difference of the absolute length values determined by the mass loss and ultrasonic methods, i.e. u = Z - ( L - L , ) . In the case of unidirectional sliding, Z = 4 0 p.m, t , , ~ , , ~ . ~ = 0 . 0 0 5 p.s, u = 15 p.m and C L = 6 0 1 0 m S - t , and Cs=5983 m s - I , a value within 7% of the estimated 5589 m s - t value. A further refinement to this approach recognizes that the ultrasonic velocity in the subsurface zone is directly related to the angle 0, defined as the sliding-induced reorientation angle between the subsurface as-deformed and the bulk microstructural features. If the angle 0 as a function o f x is represented by a fonrth-degree polynomial, then Eq. (3) becomes
U=tCl + t C s - - C L ) ~ J ]
which gives a Cs value of 5 5 0 4 m s - I , within I% to the previously estimated 5589 m s - i value.
( 1) An in-sire continuous ultrasonic method for wear monitoring has been demonstrated. (2) This method is based on the ultrasonic pulse-echo technique, where differences in the time of flight are recorded as the wear surface is relocated. (3) Additional information describing subsurface deformation may be obtained with the ultrasonic technique, strain hardening, grain refinement and crystallographic reorientafion during sliding, being potentially detected by the ultrasonic signal.
Aelmowledgements The authors would like to acknowledge the following: Dr M.C. Bhardwaj ( Ultran Laboratories, Boalsburg, PA) for his technical assistance and production of the ultrasouic transducers, Mr H. Freese (Teledyne-Allvac, Monroe, H e ) for supplying the materials, Mr P. Landreth (Clemson University, Clemson, SC) for the technical assistance with the equipment and Mr L. Shular (Shular Tool Co., Oak Ridge, TN) for sample machining.
References [!] I.e.Clarke,Wear of aaificialjointmartials,l--Frictionand wex studies:validityof weas-screeningprotocols,Engng Med., I0 (3) (1981) 115-122.  D. Dawson, Friction and wear of medical implants and prosthetic devices, ASM Hondbook, 18 (1992) 656664. [31 O. Beck, Materials clmractedzationand flaw detection by ecunstic NDE. JOM, 44 (10) (1992) 17-23. [41 R.B. Thompson, Theory and apf~ication of ultrasonicmicrostrectural clunctenzation, JOM, 44 (10) (1992) 31-35.  J. Krautkrimerand H. ~ Ultrasonic Testing of Materials, Springer, Bedin, 4th edn., 1990.  A.P. Crackr,ell, Ullrasonics,Wykeham,London, 1980.  A. Vary, Ultr'-,n,onicngasm~montof material Im3perties,in Research Techniques in Nondestructive Testing, Vol. 4. 1980, Chapter 5. pp. 159-204.  J. Black, Biological Performance of Materials-Fundamentals of Biocompatibility, Marcel Dekker,New York, 1992, 2ridedit., p. 114.  M.C. Bbexdwaj. Ultrun LMoot~ories, Bunlsbusg, PA. private communication,[995. [ 10] S. Y t Gmynn, Influenceof roughness of the bottom surface of a testpicceon the paramelev~of ultrasoundand the readingsof e l s i e thicknessgauges,Russ. J. Nondestruct. Tcrsting,30 ( 11) (1994) 851863. [ I I ] S.L. Rice, H. Nowomyand S.F. Wayne.Key Engng Mater.. 33 ( 1989) 77-100. [ 12] R. Hill, The Mathematical Theory of Plasticity. Clmendon.Oxford. 1950, p. 29. [ 13] M. Spiesand E. Schneider,Nondestructiveanalysisof texturesin rolled sheetsby ultrasonictefhniquus,TexturesMicrostruct., 12 ( 1990) 219231. [ 14] E. Schneider, Ultrasonic birefringenee effect--its application for n~l._erialscham~erization, Opt. Lasers Engng, 22 (1995) 305--323.  C.K. Jen, H. Soda, Y.S. Liu, C. Neron, A. Ohno and A. Mcl.,eaa, Acousticcharacterizationof meuzlswith colunmargtai:r:', Ultrasonic, 33(3) (1995) 181-186. [ 16] L. Shaw and D. Miracle,On the relationshipbetween ~ c t o r e and acousticemissionin Ti-6AI-4V.J. Mater. Sci., 30 (1995) 42864298. [ 17] S. Serabiun. Frequency and gnun size dependency of ultrasonic a~tenuationin polyc~stalliue materials.Br. J. Non-Destruct. Testing, 22 (2) (1980) 69-77. [ 18] V.D. Scott and H. Wilmun,Surface reoriontationcaused on metalsby abrasion--its nature,origin and relationto frictionand wear. Prec. R. Sac. Lend A, 247 (plate 5) (1958) 353-368. [ 19] D.R. Wheeler and D.H. Becldey. Texredng in metals as result of sliding, Wear.33 (1975) 65-74.  C. Robinson.A statisticalapproach to predictingthermo-meobunical behavior of Ti-6AI--4V Alloy at elevated temperatures, Master's Thesis, Masterof Science,Clem.~n University.Clemson,SC. 1996.
M. Long. H.J. Rack/Wear 205 (1997J 130-136
Marc Long is currently a Ph.D. candidate in the Bioengineering Department at Clemson University, South Carolina. He received his National Superior Engineering School degree in Mechanical Engineering from the "Ecole Nationale Superieure d'Arts & Metiers", Paris, France in 1989, and his M.Sc. degree in Materials Science and Engineering, Clemson University, in 1992. He has been working in the area of titanium alloys metallurgy and development, tribology,
fatigue and thermo-mechanical processing for the past 6 years. H.J. Rack has been conducting research in titanium alloy systems since 1970 and continues to explore the development of a detailed understanding of these materials, most recently focusing his research efforts on their tribological behavior for potential orthopaedic application. These studies include both unidirectional and reciprocating sliding conditions. In addition to his research efforts he is nationally active in the titanium community, having served as chairman of the TMS Titarium Committee.