The Concomitant Evolution of Wear and Squeaking in Dual-Severity, Lubricated Wear Testing of Ceramic-on-Ceramic Hip Prostheses

Abstract

Ceramic-on-ceramic hip bearings were tested in short-term wear tests with a systematically varied contact force. Continuous vibration and intermittent surface roughness measurements were obtained to elucidate potential causes of in vivo hip joint squeaking. The three-phase test comprised alternating cycles of edge loading and concentric articulation, always using ample serum lubricant. A 50,000-cycle wear trial in which the contact force during concentric articulation was distant from the head’s wear patch yielded no squeaking and practically no liner roughening. In 10-cycle trials of an edge-worn head coupled with a pristine liner, the contact force was varied in magnitude and point of application; immediate, recurrent squeaking occurred only when the contact force exceeded a critical threshold value and was centered upon the head’s wear patch. In a 27,000-cycle wear trial with the contact force applied near the margin of the head’s wear patch, recurrent squeaking emerged progressively as the liner’s inner surface was roughened via its articulation with the worn portion of the head. The results reveal key conditions that yield recurrent squeaking in vitro in various scenarios without resorting to implausible dry conditions. A fundamental theory explains that hip squeaking is induced by myriad stress waves emanating from asperity collisions; yet, the root cause is edge loading.

Introduction

Ceramic-on-ceramic (CoC) hip prosthesis bearings exhibit superior wear properties when the in vivo conditions are ideal, as when the bearings articulate concentrically.¹ However, some CoC bearings retrieved following in vivo use exhibited a wear pattern called stripe wear for its elongate shape on the femoral head.² This pattern is caused by edge loading, but not all such wear is stripe-shaped, and so a more apt term that relates to the root cause is edge-loading wear. Edge-loading wear was produced in a few laboratories by modifying standard hip simulator cyclic motion profiles.^3,4 The modified profiles induce small separation and edge loading between the head and liner, mimicking the in vivo behavior of a lax joint.^5,6 Studies with these conditions replicated clinically observed edge-loading wear after several million cycles. Yet, such studies did not produce the extensive wear on the bearing surface exhibited by some retrievals,^7,8 nor did they produce the squeaking noises experienced by some patients with CoC hips.⁸ This contradicts hypotheses pointing to edge-loading wear as an important cause of squeaking.^9,10 The contradiction raises questions about the relationship of edge-loading wear to liner wear and squeaking and about the efficacy of current test methods.

Given the potential for head-liner separations to occur in vivo,⁶ hip bearings have two interfaces. The intended spherical interface between the head and liner has low contact stresses¹¹ and can be protected from wear by fluid film lubrication.¹² The second interface, between the head and the liner’s edge, experiences more severe loading conditions. There, the low conformity leads to extreme contact stresses^13–15 and poor lubrication conditions. Wear from this second interface has been implicated in squeaking.^8,10,16 However, just one in vitro study to date has reproduced squeaking under lubricated, concentric articulation with an edge-worn head, and only under peak loading aligned with the wear patch.⁹ Further experiments are needed to validate hypotheses about a link between wear at the severe interface and squeaking that can originate at the mild spherical interface under a variety of in vivo loading conditions.¹⁷ Some in vitro studies described squeaking as a discrete, binary variable;^9,18 this is a limiting simplification, and continuous diagnostics are needed to uncover the problem’s root causes.

We regarded the CoC hip as a dual-interface, dual-severity bearing and examined the progressive effects of wear at both interfaces in a laboratory hip simulator. We hypothesized that wear would develop progressively, occurring first on the surfaces in the severe interface, and later emerging on the liner bearing. We also hypothesized that squeaking would emerge progressively from articulation at the mild interface, dependent on the roughening of both bearing surfaces. Tests were performed on commercial Al₂O₃ bearings with ample lubrication in a custom-built, dual-severity wear apparatus that provided both edge loading and concentric articulation. To assess wear, the bearing surfaces’ roughnesses was measured intermittently in multiple locations. To assess squeaking, component vibrations were continuously recorded. The observed progression of surface roughness and squeaking-related vibrations elucidates the fundamental causes of squeaking, and supports theories that edge-loading wear induces, or is at least an integral part of, a chain of events that causes squeaking in vivo.

Methods and Materials

Test design

A schematic of the test design is shown in Figure 1. The femoral head is mounted by repeated impaction onto a 12/14 taper on a stainless steel swing arm that can pivot on fixed axis A-A. A liner oriented at a selectable, fixed abduction angle θ can be translated vertically (y-axis) by the test machine’s actuator and rotated about the horizontal axis through its center (axis B-B) by a step motor. A spring parallel to B-B applies a force S to the lower end of the swing arm. Further, the dual-severity articulation modes are detailed as follows:

Figure 1.

Schematic of dual-severity test design. a) Concentric Articulation; top view hides swing arm and includes head section C-C to reveal liner ID wear. b) Edge Loading.

Concentric Articulation (CA, the mild condition, Fig. 1a) consists of liner reciprocation about axis B-B with the head fully reduced in the liner. A vertical actuator force on the liner and the spring force S on the swing arm induce contact force Q on the head and reaction force R on swing arm axis A-A. Q forms angle α with the vertical, and the vertical component of R (R_y) is the axial force registered by the test machine’s load cell.

Edge Loading (EL, the severe condition, Fig. 1b) consists of subluxation and reduction motion where the head slides across the liner’s edge. In subluxation, the liner is lowered distance y, and S pulls the head across the liner’s edge; simultaneously, the center-point of head-liner contact traverses angle τ on the head’s surface. In reduction, the liner is raised y, forcing the head back into the liner. Repetition of EL causes edge-loading wear on the head, spanning τ as illustrated.

The adjustability of Q (using S and R_y) and θ is a key feature. Adjustment of these variables enabled examination of the hypothesis that CA can induce liner bearing wear where the edge-worn portion of the head articulates with the inner diameter (ID) of the liner; such wear is labeled as liner ID wear in Figure 1a. As detailed below, tests that varied θ and Q (both its magnitude |Q| and orientation α) revealed that ID wear depends on both θ and Q and that squeaking can occur most readily after ID roughening surpasses a critical threshold.

Custom test apparatus

A custom, adjustable apparatus was built to implement the dual-severity test. Detailed schematics are provided in Supp. Figure S1. A few key details are labeled in Figure 2. The liner orientation is adjustable to simulate abduction angles of 45° or 60°. The liner is rotated using a NEMA-23 step motor (Powermax II, Pacific Scientific, Rockford, IL) equipped with a 10:1 reduction gearhead. The spring is connected to the swing arm 134 mm below the swing arm’s pivot axis. The spring force is measured using an in-line load cell. The entire apparatus (except swing arm and head) is attached to the vertically oriented, bottom-mounted actuator of a servohydraulic test frame (MTS, Eden Prairie, MN). The swing arm is attached via an adapter to the test frame’s top-mounted load cell. The test frame’s controller synchronizes the motions of the actuator and the step motor to effect both articulation modes as directed by a custom program. During operation, the bearings were continuously lubricated with diluted bovine serum using a gravity-fed drip system, as detailed below.

Figure 2.

Dual-severity test apparatus

Measurement protocols

Bearing surface textures were measured at intervals during the tests to quantify the progression of surface damage (Fig. 3). The liner was removed from the apparatus and held within an adjustable angle vise. The 2 μm radius tip of a stylus profilometer (SJ-400, Mitutoyo, Japan) was positioned within the liner interior such that the stylus measured along a meridional arc segment symmetrically disposed about the nadir (the arc’s lowest point). Measurements were taken at multiple ID locations; the liner was rotated to angles θ and β to position predefined measurement locations (Fig. 3a) beneath the stylus. All profilometer measurements had an evaluation length of 1.25 mm and a cutoff length of 0.25 mm. Edge-loading wear on the head was also quantified by measuring the length and width of the visually matte wear patch with a digital caliper.

Figure 3.

Liner ID surface texture measurement approach. a) Matrix of measurement locations. b) Sectioned liner and profilometer stylus.

Squeaking was quantified using vibration measurements. A laser Doppler vibrometer (CLV-3D, Polytec, Germany) was focused on a midpoint of the swing arm. The vibrometer’s voltage output was streamed via a data acquisition board to computer memory at 50 kHz. Afterwards, the data were scaled to velocity values and divided into separate segments for each half cycle of liner rotation. The power spectrum of each segment was computed using a fast-Fourier-transform (FFT) technique, and the power at particular frequencies was recorded. The frequencies were those that exhibited dramatically increased power concomitant with instances of audible squeaking. The tests were human-attended, and audible squeaking was noted as it occurred to further corroborate the signals that are interpreted as squeaking in the vibration data.

Uniform test conditions

Throughout the study, the bearings were Ø36 mm alumina implants (Biolox Forte, Ceramtec, Germany). The lubricant was bovine serum acquired with a protein concentration of 30 g/L (Hyclone, Logan, UT) and further diluted with deionized water to 17.5 g/L. The lubricant was gravity-fed to the bearing couple at ~10 drops/min. The liner rotation (about axis B-B) had a peak-to-peak amplitude of 50° at 100°/s. The liner abduction angle (θ) was 60° except in certain trials of Test DS2 described below.

Two bearing pairs were tested under varied conditions:

1. A Dual-Severity Test 1 (DS1) comprised of 2 phases. In the first phase, the first 2500 cycles alternated 1 cycle of EL with 5 cycles of CA. The spring was preloaded to S=75 N. During EL, in the subluxation half-cycle, the liner was lowered by y=0.9 mm in 0.3 s, and the headliner contact center-point slid onto the liner’s edge (Fig. 1b). In the reduction half-cycle, the liner was raised 0.9 mm in 0.5 s. During CA, the axial force was R_y=2750 N. In the second phase, the test alternated between 10 cycles of CA and 1 cycle of EL, to 50k cycles of CA.

2. A Dual-Severity Test 2 (DS2) comprised of 4 phases. In the first initial EL phase, the head was first worn over 5000 cycles of EL. The spring preload was S=75 N. For subluxation, the liner was lowered y=1.3 mm in 0.5 s. During reduction, the liner was returned y=1.3 mm in 0.9 s, while it was twice rotated ±15° as a means to lengthen the expected wear patch. Supp. Movie S1 demonstrates this test activity.

In the second varied θ and Q (VQ), twelve 10-cycle trials of CA were performed with varied θ and Q. Changing Q varied the contact pressure and the placement of the contact force with respect to the head’s wear patch; changing θ altered the area of the head covered by the liner. Three contact force angles α were used: α= 0, 7.5, and 15°; at 15°, Q was directed into the center of the head’s wear patch. Two force magnitudes, |Q|= 500 and 1000 N, were used at each α. Trials were performed at θ = 45 and 60°. Vibrations were measured.

In the third dual-severity (DS) phase, 25k cycles of CA were performed, alternating 250 cycles of CA with 10 cycles of EL. The test began with |Q|=400 N directed into the head’s wear patch (α=15°). Squeaking arose early (3000 cycles of CA); so, Q was changed to eliminate the squeaking: α was reduced to 8°, and |Q| was increased to 1000 N. Vibrations were measured during CA.

In the fourth final CA phase, 2500 cycles of CA were performed with no EL. Vibrations were measured.

Results

Dual-Severity Test 1

Test DS1 produced an elongate wear patch on the head, little roughening of the liner, and no squeaking. The wear patch dimensions are graphed in Figure 4 along with the roughness at the patch’s center. The patch was centered 14.5° from the pole (per Fig. 1b, λ=14.5°). Photos showing the wear patch’s growth are in Supp. Figure S2. The head’s roughness across the pole was essentially unchanged. The liner was measured at 9 positions on the ID, and all showed essentially no roughening – the greatest roughness was 8 nm R_a (roughness average).

Figure 4.

Test DS1: femoral head’s wear patch roughness, width, and length.

Dual-Severity Test 2

The Initial EL phase produced a 3×17 mm head wear patch, centered at λ=15° from the pole. Its roughness, measured across its narrow dimension at the center, was 87 nm R_a. A photo of the wear patch is in Supp. Figure S3.

The VQ trials produced audible squeaking only for θ =60° and only when the contact force was directed into the head’s wear patch (α=15°), at the highest load (|Q|=1000 N). Figure 5 compares the vibration signals from 3 different half-cycles of CA, with constant |Q| but varied θ and α. To quantify the squeaking, the signal power surrounding distinct squeaking-related peak frequencies was isolated. At each such frequency, the power was summed from the FFT power spectrum over a span of 12 Hz centered on the peak frequency. Summing the power in all such frequencies yielded the squeaking power, denoted as Φ_i, where i is the count of CA half-cycles. Φ_i was computed for each half-cycle, rather than each cycle, because the bearings often made different sounds in opposite directions of liner rotation. Figure 6 compares the mean value of Φ across the 12 trials in this phase. Only for θ =60°, with α= 7.5° and 15°, was the power in the squeaking-related frequencies noticeably greater than the signal’s noise level. Supp. Figure S4 graphs Φ_i for the 12 trials in this phase.

Figure 5.

Vibrations during 3 VQ trials. Top row: vibration signals; Bottom row: respective power spectra. Left: minimal vibration; Middle: increased vibration without squeak; Right: audible squeak.

Figure 6.

Mean squeaking power (Φ) from 10-cycle trials in VQ phase. Error bars are ±1 std. dev., except 60°/15° where they are ±½ std. dev.

The Dual-Severity phase produced roughening of the liner ID and intermittent bouts of squeaking. The liner surface texture was measured at 18 locations given by the combinations of θ = 60, 66, 70, 74, 78, and 82° with β = 0, 25, and −25°; Supp. Table S1 gives the complete results with 4 measures of roughness. Figure 7 compares the ID roughness at 4 locations having β =0°, along with the squeaking power (Φ_i). Φ_i rapidly increased during the first 3500 cycles with |Q|=400 N and α=15°. Squeaking that was audible from a distance >1 meter was generally associated with Φ>1 (mm/s rms)². When conditions were changed to |Q|=1000 N with α=8°, squeaking ceased, and then Φ_i trended slowly greater until squeaking became quite common after about 21k cycles. Φ_i often increased dramatically during a 250-cycle interval of CA, yielding the relatively vertical columns of data points in Figure 7. On some occasions, though, Φ_i decreased throughout a CA interval.

Figure 7.

Squeaking power (Φ_i) and liner surface roughness in Test DS2. a) DS and final CA phases; line plots show liner ID roughness at 4 values of θ , all with β =0°. b) focus on final CA phase; t highlights an interval in which squeaking did not occur after temporary bearing removal.

The final CA phase yielded audible squeaking with great regularity; Supp. Movie S2 demonstrates squeaking in this phase. An exception was during an ~500-cycle interval (t in Fig. 7b) after the bearings had been temporarily removed for surface measurement. Figure 8 shows photos of the bearings’ roughened areas after test completion.

Figure 8.

Worn areas of (a) liner and (b) head revealed by rubbing with graphite pencil. Labeled marks on the liner show the angles β where surface texture measurements were made.

Discussion

Our results demonstrate that squeaking in CoC hip bearings can emerge at different rates and in different ways, depending on the contact force Q and the liner abduction angle θ . In the short- cycle VQ trials, squeaking occurred only when |Q| was maximum and α was aligned directly with the head’s wear patch. Yet, the other VQ trials exhibited increasing vibration energy (Figs. 5 and 6) as |Q| was increased and α was aligned closer to the wear patch. The minimal vibrations seen with θ =45° can be attributed to improved fluid film lubrication when the liner covered more of the head. In Test DS2, recurrent squeaking emerged in a short interval when Q was aligned with the center of the head’s wear patch, even though |Q| was low, 400 N. The liner ID was roughened little during this interval. In the second interval of Test DS2, Q was aligned more distant from the head’s wear patch, and recurrent squeaking emerged much later, after the liner ID was substantially roughened. Q was placed most distant from the wear patch (α=1°) in Test DS1, and squeaking never emerged, even though |Q| was much greater, 2750 N.

The results from an intermediate scenario, i.e. the second phase of Test DS2 with Q indirectly aligned with the head’s wear patch, provide the best foundation for an explanation of squeaking emergence in vivo, where loading varies widely. In this theory, squeaking emerges as a chain of events inducing increasing vibration energy in a component of the prosthesis system. The evolution of a squeaking CoC hip can be described using the relative time sequence below. (Italicized times follow labels in Figure 9; they are illustrative only and are scaled from the time span of months-to-years during which squeaking typically emerges in vivo.)

Figure 9.

Diagram for the theory of the evolution of squeaking in CoC hip joints. Time labels and “Vibration E(nergy)” meters are illustrative, to provide simple relative values.

0:00: New, pristine bearing surfaces articulate concentrically, and fluid film lubrication prevents contact of asperities. Component vibration is minimal, and no noise is heard.

Interval 0:00–1:00: Small head-liner separations result in edge loading with high contact stresses that elicit edge-loading wear on the head. Patient may perceive clicking, but not squeaking.

1:00: Asperities within the head’s wear patch are large and granular; some penetrate the lubricant film. Asperity contact with the liner ID causes high magnitude micro-scale contact stress, even though macro-scale contact pressure is low. Asperity contacts radiate weak stress waves. Component vibrations increase modestly.

Interval 1:00–8:00: Over a long time, numerous cycles of CA occur with occasional EL. The head wear patch grows. Asperity contacts with liner ID create contact fatigue damage.

8:00: High-stress asperity contacts may have caused plastic changes such as twinning in surface grains, leading to intergranular cracking.¹⁹ A few grains have spalled from the liner; the voids experience energetic collisions with head asperities, sending out strong stress waves. Stress waves traverse intact grain boundaries but reflect from cracked boundaries. The liner becomes rougher. Component vibrations become stronger and more variable.

Interval 8:00–11:59: Liner ID surpasses a contact fatigue threshold, and wear rapidly becomes severe, with extensive grain pullout and rapidly increasing roughness.¹⁹ Occasional squeaks occur.

11:59: Numerous large, granular asperities from both bearing surfaces penetrate the fluid film and collide during CA, inducing strong stress waves. Intergranular cracking continues, yet many colliding asperities are peaks of firmly anchored grains. Vibration energy reaches a critical level that induces component resonance. Patient suffers persistent squeaking.

This theory of squeaking evolution is supported by test observations and scientific reports. CA between pristine surfaces occurred in the VQ trials with α=0°, and Figures 5 and 6 demonstrate the minimal vibration power. Test DS1 showed that edge-loading wear occurs much more rapidly than ID wear. Several reports show the granular microstructure of wear patches.^2,20 Figure 7 shows the trend of increasing vibration power concomitant with increasing surface roughness. Cho et al. reported that in sliding wear, alumina exhibits a sharp mild-to-severe wear rate transition, after which its bearing surface displays intergranular fracture and grain pullout.¹⁹ In Test DS2, such a transition occurred between 20k and 21.5k cycles, and it greatly widened the visibly matte, roughened area of the liner ID. At 20k cycles, the matte area spanned −15°<β <15°, and by 21.5k cycles, it spanned −30°<β <30°. Figure 7 corroborates sudden increases in the rate of roughening. Figure 7 also demonstrates that powerful squeaking can emerge first on an infrequent basis and later become routine.

An unexpected outcome in Test DS2, that squeaking became continual after cessation of EL (the final CA phase), helps explain the relationship between asperities and vibrations. When asperities collide, some of the bearings’ kinetic energy is transferred to the components in the form of minute stress waves that propagate throughout the components. The interaction of these waves with a component’s surface causes minute elastic displacements detectable with the vibrometer. The intermittent EL cycles induced severe contact stresses that created a thin, freshly damaged layer in which grains were poorly bound to the matrix. In subsequent CL, such poorly adhered grains did not effectively transmit the energy of asperity collisions into stress waves (vibrations); instead, the loosely held grains gained kinetic energy as they were pulled from the surface, becoming mobile particles. But once EL was halted, the freshly damaged layer was quickly worn away. Then, during the final CA phase, asperity collisions occurred between surface grains well-adhered to their respective substrates; so, the energy of these collisions could readily propagate across intact grain boundaries as stress waves, which consistently raised the component vibration energy to a critical, resonance-inducing level. This phenomenon also occurred without EL, around interval t during final CA (Fig. 7b).

Many authors proposed that CoC squeaking in vivo is caused in part by reduced lubrication,^18,20–22 some even proposing dry contact. Several in vitro studies tested CoC hips for squeaking with no lubrication, and most reported better success producing squeaking with dry conditions than with lubrication.^9,18,23,24 Our study used only lubricated conditions, and the results corroborate the explanation by Laurent et al that the fluid film is “disrupted” by increased surface roughness;²⁰ that is, the fluid film is penetrated by large asperities. Interpreting “disruption” in the more extreme sense of an absence of lubrication in vivo, leading to the implausible scenario of a dry CoC interface, is probably overreaching, and as our study shows, an unnecessary component of a theory of squeaking in vivo.

All known previous studies examining squeaking in vitro have used a liner with an unworn ID. One of these⁹ produced squeaking in a scenario similar to the sole VQ trial that produced squeaking; namely, the contact force was aligned directly with the head’s wear patch. Liners retrieved from squeaking CoC hips have displayed ID wear similar to that from Test DS2.^7,8

Taken together, prior observations and our results suggest that in future research of CoC hip prostheses, particularly in multi-axis simulators with flexible and programmable inputs, wear testing protocols should include joint forces directed close to or aligned with a femoral head wear patch elicited via edge loading. An in vitro model of in vivo squeaking evolution should include the following: 1) lubrication, 2) edge loading, and 3) concentric articulation wherein the contact force can cause wear asperities formed by edge loading to penetrate the fluid film. Since edge loading is the first step in a chain of events that leads to CoC wear and squeaking, future research should also be directed towards designs, materials, or techniques that will reduce its severe effects.