Wear mechanism and debris analysis of PEEK as an alternative to CoCrMo in the femoral component of total knee replacement

The polyetheretherketone (PEEK)-highly cross-linked polyethylene (XLPE), all-polymer knee prosthesis has excellent prospects for replacing the traditional metal/ceramic-polyethylene joint prosthesis, improving the service life of the joint prosthesis and the quality of patients’ life. The long-term wear mechanism of PEEK-XLPE knee joint prosthesis is comprehensively evaluated from wear amount, wear morphology, and wear debris compared to that of CoCrMo-XLPE joint prosthesis. After 5 million cycles of in vitro wear, the wear loss of XLPE in PEEK-XLPE (30.9±3.2 mg) is lower than that of XLPE in CoCrMo-XLPE (32.1±3.1 mg). Compared to the XLPE in CoCrMo-XLPE, the plastic deformation of XLPE in PEEK-XLPE is more severe in the early stage, and the adhesive peeling and adhesion are lighter in the later stage. The size distribution of XLPE wear debris in PEEK-XLPE is relatively dispersed, which in CoCrMo-XLPE is relatively concentrated. Wear debris is mainly flake and block debris, and the wear mechanism of XLPE was abrasive wear. The wear volume per unit area of PEEK femoral condyle (10.45×105 µm3/mm2) is higher than that of CoCrMo (8.32×105 µm3/mm2). The PEEK surface is mainly furrows and adhesions, while the CoCrMo surface is mainly furrows and corrosion spots. The PEEK wear debris is mainly in flakes and blocks, and the CoCrMo wear debris is mainly in the shape of rods and blocks. The wear mechanism of PEEK is abrasive wear and adhesion, and that of CoCrMo is abrasive wear and corrosion.


Introduction
Around 400 million people worldwide suffer from bone and joint diseases, and about 14 million artificial joint replacements are required yearly [1]. Artificial joint replacement, the most successful and effective surgical and rehabilitation method, is commonly used in the clinical treatment of human joint damage [2,3]. According to statistics, the 10-year survival rate of artificial knee arthroplasty is > 90% [4,5]. However, the current artificial knee joints are mostly made of cobaltchromiummolybdenum alloy (CoCrMo) or ceramic femur and ultra-high molecular weight polyethylene (UHMWPE) tibia [6,7]. CoCrMo or ceramics with high elastic modulus have poor mechanical compatibility with human bones, prone to stress shielding and stress concentration [8]. Wear debris produced by UHMWPE can lead to bone resorption and osteolysis, eventually loosening and failing prosthesis [9,10]. The artificial knee joint prostheses still have problems such as insufficient wear resistance, no long service life, and a high revision rate of implanted prostheses, which are challenging to meet the growing demand for high-reliability and long-life artificial joints [11,12].
To combat the failure caused by the wear of artificial knee joints, a high-strength, high-toughness, and wear-resistance polymer has been developed from a bionics perspective to completely replace the rigid metal or ceramic prostheses, which can form a "soft-to-soft" joint friction pair with UHMWPE [1315]. Polyetheretherketone (PEEK), a semi-crystalline thermoplastic polymer with several potential advantages, has been proposed as an alternative material for CoCrMo in femoral components of total knee arthroplasty [1619]. All-polymer artificial knee implants benefit approximately 2% of patients with metal implant sensitivities. Besides, the Young's modulus of PEEK (~3.7 GPa) is more similar to that of human bone (0.00120 GPa) compared to that of CoCrMo (~210 GPa). The lower elastic modulus of PEEK can reduce the effect of stress shielding and the risk of bone resorption and loosening failure. Moreover, compared to CoCrMo or ceramic, all-polymer implants are lighter and closer to the weight of the natural joints, resulting in greater patient comfort [17,19]. On the other hand, replacing UHMWPE with highly cross-linked polyethylene (XLPE), a new generation of acetabular prosthesis material with higher wear resistance, can reduce the wear and osteolysis of artificial knee prostheses [20].
The wear resistance and wear debris are essential to be studied because the size, quantity, and surface texture of wear debris contain rich tribological information, which can reflect the wear state and wear mechanism of artificial joints, evaluate the wear law of artificial joints, and directly affect the loosening and failure of implants [21]. The surface roughness of the PEEK femoral prosthesis was significantly higher than that of the CoCrMo prosthesis. However, this did not affect the wear rate, which remained linear during the wear test, and the wear rate of PEEKUHMWPE was comparable to those of conventional implant materials [22]. In terms of wear performance under tri-body wear/injury conditions, PEEKUHMWPE is superior to conventional knee arthroplasty materials [23]. The PEEK wear debris exhibits lower cytotoxicity [24]. When the prosthesis is PEEKXLPE, wear did occur on the surface of the materials, and the size of most particles was 15 μm, which induced an inflammatory response of the synovial membrane and release of proinflammatory cytokines and did not show any harmful influences of the prosthesis in goats [25].
Therefore, it is crucial for the optimization and production of the joint prosthesis that a novel all-polymer PEEKXLPE artificial knee joint prosthesis should be constructed and comprehensively evaluated. In order to explore the wear resistance, wear mechanism, and wear debris characteristics of the new all-polymer PEEKXLPE artificial knee prosthesis, 5 million cycles of wear tests in vitro on PEEKXLPE and CoCrMoXLPE artificial knee joints have been conducted. Based on the optimized protein degradation method and the multiple separation and extraction method of mixed wear debris with different density ratios, the wear debris in each measurement cycle was extracted and studied. Combined with the wear amount and morphology, the wear mechanism of the artificial knee joint and the generation mechanism of wear debris are explored, which provide a theoretical basis for the wear prediction and development of new artificial joint materials.

Samples
The samples of the artificial knee joint in vitro wear test were PEEK femoral condyle, injection molding, provided by Jiangsu Dejian Medical Technology Development Co., Ltd., China; CoCrMo alloy femoral condyle, casting molding, purchased from Smith & Nephew Medical (Shanghai) Ltd., China; XLPE tibial liner, provided by Jiangsu Dejian Medical Technology Development Co., Ltd., China The physical and mechanical properties of the materials are shown in Table 1. The PEEKXLPE friction pair is the experimental group, the CoCrMoXLPE friction pair is the control group, and the XLPE immersed in the lubrication is the blank control group for the exact wear loss. The artificial knee joint prosthesis samples are shown in Fig. 1. According to Ref. [26], the samples should be immersed in deionized water for at least two weeks www.Springer.com/journal/40544 | Friction before the test to achieve stable liquid absorption and exclude environmental errors. Palacios MV+G bone cement, purchased from Heraeus Medical GmbH, Co., Ltd., Germany, was used for fixing femoral and tibial specimens. The lubrication was (25±2)% fetal bovine serum (25% FBS; protein concentration = 38 g/L), purchased from Hangzhou Sijiqing Bioengineering Materials Co., Ltd., China. Amphotericin and gentamicin are added to the 25% FBS to prevent deterioration (1 L of 25% FBS solution added 100 mg gentamicin and 25 mg amphotericin).

Keen simulator tests
The wear tests were carried out on the artificial knee simulator (ProSim, Simulation Solution Ltd., UK). Before testing, femoral and tibial components were fixed onto the corresponding fixation frames according to the neutral knee mechanical alignment axis. Test loads and motions were loaded according to the total knee joint wear standard of ISO/AWI 14243-3 [26], which defines the kinematics and dynamics of the knee joint during movement under normal gait, as shown in Fig. 2. The parameters of knee simulator tests in detail are shown in Table 2. The tests in each condition have been carried out three times to ensure the stability of sample test results.
Fluid losses due to evaporation were replenished by adding deionized water at least daily during the   tests. The lubrication is wholly changed every 0.5 million cycles. At 0.5 million cycles, 1 million cycles, and every 1 million cycles thereafter, the test was stopped to measure and analyze the wear amount, wear morphology, and wear debris until the end of the test.

Wear loss
The wear loss of the tibial liner XLPE was measured by the gravimetric method according to Ref. [27]. The wear volumes of femoral condyle PEEK and CoCrMo were obtained by the laser confocal microscope (VK-X, KEYENCE, Japan). First, the wear surface of the femoral condyle is divided into four typical wear areas ( Fig. 3(a)). Then, about 30 mm 2 in each specific area is selected for scanning ( Fig. 3(b)), and divided into multiple rectangular sub-areas, scanned one by one (Fig. 3(c)). Next, the three-dimensional (3D) topography map and the wear area profile are obtained by splicing sub-regions and the correction plane ( Fig. 3(d)). The wear volume is calculated through the sinking amount of the wear region, like the scratches, spalling pits, sticking cutting grooves, and corrosion pits (Fig. 3(e)). Finally, the wear volume rate per unit area of the femoral condyle is obtained by calculating the average wear volume of several typical areas.

Wear morphology
The morphology of the wear area is observed with a Kearns high-speed camera (VW-9000, KEYENCE, Japan) to analyze its wear mechanism.

Wear debris
The protein degradation of the experimental FBS lubrication (mixed wear debris with different density ratios from artificial joints) is carried out by enzymatic degradation to remove the protein in the lubrication and wear debris surface. The mixed wear debris is separated and extracted through multiple centrifugations, as detailed in the Electronic Supplementary Material (ESM). The morphology of the wear debris in each group is observed by a field emission scanning electron microscope (FESEM; SU8220, Hitachi, Japan). Before the observation, the filter paper containing wear debris is sprayed with gold. The particle size distribution of the wear debris of XLPE is measured www.Springer.com/journal/40544 | Friction by the Nano Measurer software (Fudan University, China). In each group, the FESEM image (× 400) with more than 100 wear debris is selected for particle size distribution, and the image (× 5,000) with more than 100 wear debris is selected for the small particle size distribution (< 1 μm). The particle equivalent diameter c is based on the long diameter a and short diameter b of the particles (Eq. (1)): The compositional analysis of the CoCrMo particles was performed by an X-ray energy spectrometer (XflashFQ5060, Bruker, Germany; surface scan, working distance = 6.8 mm, and acceleration voltage = 15 kV). Figure 4 shows the wear losses of the tibial liner XLPE and the wear volume rates per unit area of the femoral condyle PEEK and CoCrMo after 5 million cycles of wear tests. In Fig. 4(a), the wear losses of XLPE in PEEKXLPE are always lower than those in CoCrMoXLPE. The difference between the two groups showed a trend of increasing first, and then decreasing. After 3 million cycles, the difference between the two groups reached the maximum, about 7.8 mg. In the running-in wear stage (01 million cycles), the wear loss of XLPE changes slowly due to insufficient matching of joint prosthesis surface. In the acceleration wear stage (1 million4 million cycles), the wear loss of XLPE changes greatly, and the increase of XLPE wear loss in CoCrMoXLPE is more significant than that in PEEKXLPE, which enters the severe wear stage earlier due to the more considerable stiffness and elastic modulus of CoCrMo. Compared with PEEK, the difference in properties such as stiffness and elastic modulus with XLPE is also greater with CoCrMo, and it is more prone to stress shielding, local plastic deformation, fatigue spalling, and wear debris, more prone to three-body wear, aggravating material damage. In the stable wear stage (4 million 5 million cycles), the increase rate of XLPE wear loss slows down. After 5 million cycles, the wear loss of XLPE in PEEKXLPE was 30.9±3.2 mg, and that in CoCrMoXLPE was 32.1±3.1 mg. The mass wear rate of XLPE in PEEKXLPE after 5 million cycles was 6.18±0.64 mg/million cycle, which is lower than that of other's tests [28], and that in CoCrMoXLPE was 6.42±0.62 mg/million cycle. Figure 4(b) shows the wear volume rates per unit area of the femoral condyle with cycles. The wear volume per unit area of CoCrMo is lower than that of PEEK, mainly caused by the difference in hardness, stiffness, and other properties between friction pairs. The hardness difference between CoCrMo and XLPE is larger than that between PEEK and XLPE. Both wear rates increase with cycles, and the changing trend is the same. The wear rate increases sharply during 01 million cycles, keeps stable during 12 million cycles, and increases significantly again during 3 million5 million cycles. After 5 million cycles, the wear volume per unit area of PEEK is about 10.45 × 10 5 μm 3 /mm 2 , and that of CoCrMoXLPE is about 8.32 × 10 5 μm 3 /mm 2 .   Figure 5 shows the wear morphologies of XLPE (in PEEKXLPE and CoCrMoXLPE) surface during 5 million cycles of wear tests. From the macro morphology, the worn scar of XLPE is mainly concentrated in the central area of the tibial liner. With the increase of cycles, the worn scar expands to all sides. The wear at the rear end of the tibial liner is more severe than that at the front end of the tibial liner, which is mainly due to the matching running-in between the femoral condyle and the tibial liner at the initial stage of wear, and then tends to be stable. From the optical morphology, the wear of XLPE is mainly abrasive wear characterized by scratches, and the wear of XLPE in CoCrMoXLPE is more severe than that in PEEKXLPE, which is consistent with the wear loss result (Fig. 4(a)).

Wear morphology
For XLPE in PEEKXLPE, after 0.5 million cycles, the XLPE surface is mainly plastic deformation and abrasive wear, characterized by many scratches, furrows, and slight ripples perpendicular to the sliding direction caused by plastic flow and accumulation ( Fig. 5(a i )). After 1 million and 2 million cycles, the worn scar of XLPE expanded laterally and longitudinally, with the longitudinal expansion tending to the front of the tibial liner. Surface roughness reduced as plastic bulges were mainly removed, and furrows and scratches were shallow and evenly distributed with a few spalling pits (Figs. 5(b i ) and 5(c i )). After 3 million www.Springer.com/journal/40544 | Friction and 4 million cycles (Figs. 5(d i ) and 5(e i )), the worn scar of XLPE no longer increased, and the depth and number of furrows increased (about 20 μm in the widest furrow). Cracks occur in the direction of the plow, and then become new furrows under the cutting of abrasive grains. Besides, there were accumulations of abrasive grains inside the furrows, and the wear surface was attached with blocky exfoliation (2040 μm in length). After 5 million cycles (Fig. 5(fi)), the XLPE worn surface showed adhesive wear with more furrows.
For XLPE in CoCrMoXLPE, after 0.5 million cycles, the XLPE worn surface is dominated by abrasive wear, with staggered furrows and scratches and no obvious plastic deformation, on which number and density of furrows are more than those in PEEKXLPE ( Fig. 5(a ii )). After 1 million cycles, the trend of worn scar is the same as that of PEEKXLPE, and the wear surface becomes relatively smooth ( Fig. 5(b ii )). Differently, after 2 million cycles, the wear of XLPE intensifies with densely distributed furrows and some peeling pits, whose wear intensification stage is earlier than that in PEEKXLPE (Fig. 5(c ii )). After 3 and 4 million cycles (Figs. 5(d ii ) and 5(e ii )), the wear of XLPE is significantly worse, and the worn scar no longer increased. Under the extrusion of abrasive grains and plastic deformation, the cracks, furrows, spalling, and adhesive accumulation appear alternately. The depth and width of furrows in CoCrMoXLPE are more extensive than those in PEEKXLPE. After 5 million cycles, the wear of XLPE is dominated by adhesion and abrasion, with fewer furrows and more adhesive particles than those in PEEKXLPE ( Fig. 5(f ii )).
The wear of XLPE in PEEKXLPE is mainly plastic deformation and abrasive wear at the early stage, and then abrasive wear gradually becomes the dominant, and adhesive wear appears at the later stage of wear. The wear of XLPE in CoCrMoXLPE is mainly abrasive wear, and then adhesive wear and abrasive wear dominate each other at the later stage of wear. Overall, the wear of XLPE in CoCrMoXLPE was more severe than that in PEEKXLPE. Figure 6 shows the wear morphologies of femoral condyle (PEEK and CoCrMo) after 5 million cycles of wear test. The wear of PEEK and CoCrMo is dominated by abrasive wear. For PEEK femoral condyle, after 0.5 million cycles, there are many scratches and shallow furrows, which have rough edges, ridge-like bumps, spalling, and horizontal stripes perpendicular to the slide direction ( Fig. 6(a i )). Subsequently, the horizontal stripes were expanded and exfoliated, forming new furrows or increasing the depth and width of the furrows (after 1 million cycles, Fig. 6(b i )). Due to extrusion and cutting, the furrows on the PEEK surface become shallower and more minor, and many discontinuous ridges and flake-like spalling appear (after 2 million cycles, Fig. 6(c i )). Then, three body wear is easily formed, which can aggravate the wear of the PEEK and lead to wider furrows and shoals of different sizes (about 1540 μm in width). Fish scale ridges and long ridges are distributed on the edge of the shoals, and transverse lines are perpendicular to the furrow appearing around (Figs. 6(d i ) and 6(e i )). After 5 million cycles, the furrows on the PEEK surface are dense and evenly distributed, and there are adhesive deposits at the bottom of deeper furrows, resulting in adhesive wear (Fig. 6(f i )).
For CoCrMo femoral condyle, after 0.5 million cycles, the CoCrMo wear surface has fewer furrows with lower roughness and smooth edges ( Fig. 6(a ii )). After 1 million cycles, the number and depth of furrows increase, and some corroded areas appear (Fig. 6(b ii )). After 2 million cycles, the furrows are dense and evenly distributed with shallower depths. There is no obvious spalling and adhesion but still a few spalling pits and corrosion areas (Fig. 6(c ii )). Then the wear of PEEK becomes more severe with more and deeper furrows. After 5 million cycles, the corrosion area increased significantly (Figs. 6(d ii )6(f ii )). The wear of PEEK femoral condyle is mainly abrasive wear and adhesive wear occuring at a later stage. The wear of CoCrMo femoral condyle is also dominated by abrasive wear, but at the middle and later stages of wear, corrosive wear occurs alternately.   For the XLPE wear debris in PEEKXLPE, after 0.5 million cycles, the particle size is normally distributed, mainly between 15 μm (60%70%), while, that larger than 15 μm only accounts for 0.4%. During 0.5 million3 million cycles, the particle size distribution gradually shows 2  distribution, the wear debris is mainly concentrated between 01 μm (65%70%), and that between 15 μm is significantly reduced. The particle size is distributed normally again after 4 million cycles. The wear debris in size of 01 μm accounts for only 5.4%, and that between 15 μm accounts for a maximum of 70%, and the medium-sized (520 μm) wear debris begins to increase. After 5 million cycles, large-size wear debris (> 20 μm) increased significantly. For the XLPE wear debris in CoCrMoXLPE, during 04 million cycles, the particle size is normally distributed, and the size of wear debris is concentrated in 15 μm (50%60%); large-size wear debris (> 20 μm) decreases gradually with the increase of cycles, of which the size of more than 30 μm exists only during 00.5 million cycles. During 4 million-5 million cycles, the particle size showed 2  distribution, the proportion of wear debris in size of 01 μm increased to 53.42%, that of 15 μm decreased to 30.57%, the medium-sized (520 μm) wear debris is significantly reduced, and the large-size (> 20 μm) wear debris only accounts for 0.2%. Compared with the size distribution of XLPE wear debris in PEEKXLPE, at the early stage of wear, there were fewer small-sized (0-10 μm) wear debris and more large-size (> 10 μm) wear debris (Fig. 7(e)). At the later stage of wear, the proportion of small size (01 μm) wear debris is large, that of medium size (120 μm) wear debris is little, and the largesize (> 20 μm) wear debris is almost absent. The size distribution of XLPE wear debris in PEEKXLPE is relatively dispersed, and that in CoCrMoXLPE is relatively concentrated. Figure 8 shows the morphologies of XLPE wear debris in PEEKXLPE. At the early stage of wear (0.5 million cycles), there are mainly spherical debris of different sizes and tearing debris with a robust spatial hierarchy, irregular contour, and complex texture. Spherical wear debris is formed by the exfoliation of asperities on the XLPE surface due to the repeated rubbing under the alternating compound motion. Then the large-size wear debris is broken under the shearing force, forming spherical wear debris in different sizes. Tear debris is usually associated with crack propagation. During 0.5 million-3 million cycles, XLPE wear debris is most flake wear debris with a smooth surface. This is because plastic flow occurs in the local area with high stress to form a smooth surface by rolling, and then dislocation lines are generated on the surface and sub-surface of XLPE by repeated extrusion. Microcracks are formed, expanded, and peeled off to form flake wear debris under cyclic stress. For wear debris with a large area and small thickness, the boundary contour is complex and curly, mainly caused by plastic deformation and abrasive wear. After 4 million5 million cycles, XLPE wear debris is mainly spindle and flat block wear debris, with complex boundary contour and apparent concave-convex undulating texture on the surface, which is mainly formed by the interaction of adhesion and abrasive wear. When the material is torn off from the surface, it will form spindle wear debris, and flat wear debris will be formed under the action of fatigue stress.

Wear debris
Therefore, combined with the analysis of wear morphology and wear loss, the wear mechanism of XLPE in PEEKXLPE is mainly plastic deformation and abrasive wear at the early stage and adhesion and adhesive wear at the later stage. Figure 9 shows the morphologies of PEEK wear debris in PEEKXLPE. At the early stage of wear (0.5 million cycles), PEEK wear debris is mainly spherical wear debris with little quantity and size (12 μm), formed by rubbing and peeling, and rod-shaped wear debris (15 μm) with tortuous boundary contours and large surface texture fluctuations. After 1 million 2 million cycles, there are flake debris and tear debris formed by alternating contact stress and abrasive wear. The flake debris is about 20 μm in size with a little edge curl and smooth surface, and the tear debris is about 2530 μm in length with a rough surface and spatial hierarchy. After 3 million cycles, the wear of PEEK becomes serious, and the number of wear debris increases. There are mainly flake and block debris with  www.Springer.com/journal/40544 | Friction a smooth surface and relatively small size (10 μm). After 4 million cycles, the PEEK debris is mainly spindle-shaped block debris and fewer needle-shaped debris (2 μm) generated by the friction pair, rubbing the wear debris under alternating loads. After 5 million cycles, the PEEK wear debris increased significantly, and some fine debris was adsorbed on the massive debris. The amount of PEEK debris is obviously less than that of XLPE debris, which is sporadically distributed. At the early stage of wear, PEEK wear debris is mainly flake debris and tear debris, and block debris are dominated generated by abrasive wear in the later stage. Figure 10 shows the morphologies of XLPE wear debris in CoCrMoXLPE. At the early stage of wear (0.5 million cycles), XLPE debris is mainly flake debris (20 μm) with complex boundary contour and curly shape and strip debris (70 μm) with apparent texture, which are mainly caused by plastic deformation and furrow cutting. After 1 million cycles, XLPE debris is mainly flake debris with curly contour edge and smooth surface and spindle block debris with apparent concave-convex texture. After 2 million 3 million cycles, the number of XLPE debris increases significantly, the size and shape of wear debris have no noticeable change, and it is still dominated by flake and block debris. After 4 million cycles, the XLPE debris is mainly block debris with a reduced quantity and size. There are many small spherical or quasi-spherical debris (< 3 μm), which is caused by repeated rolling, rubbing, and shearing of large-size block debris under a high cyclic stress. After 5 million cycles, the XLPE wear debris is dominated by block debris (< 20 μm) with neat edges and irregular shapes and spherical debris. Figure 11 shows the morphologies of CoCrMo wear debris in CoCrMoXLPE. During 5 million cycles, CoCrMo wear debris is little in number and size, and a large number of nano-scale wear debris are scattered on the filter paper with aggregation. After 0.5 million cycles, CoCrMo debris is mainly long-shaped block debris with a length of about 2030 μm, whose surface is complex, with layered and petal-like texture, irregular contour edges, and  | https://mc03.manuscriptcentral.com/friction fine particles attached. After 1 million cycles, it is mainly spindle-shaped block debris with uneven texture and some small-sized wear debris gathers. After 2 million cycles, there are many strip debris (~20 μm) with layered and honeycomb structures and block debris (~5 μm). After 3 million cycles, there is mainly 5-10 μm-size honeycomb block debris with complex textures and spatial hierarchy. After 4 million cycles, the size distribution range of wear debris increases. The large-size wear debris is mainly spindle-shaped block debris, and the small-size block debris is formed by the broken of some honeycomb block debris with thin thickness undercutting. After 5 million cycles, the CoCrMo wear debris is mainly spindle-shaped block debris (25 μm) with a concaveconvex surface, and flake debris (2 μm) with a smooth surface and irregular boundary contour.

Wear mechanism
The analysis of wear amount, wear morphology, and wear debris shows that the friction pair material has an essential influence on the wear test results of the joint prosthesis, and the wear mechanism of different friction pairs is quite different.
In the wear-in period, the friction pair does not fit the grinding surface well due to the asperities on the surfaces. Under loading, the surface of the friction pair will be cut to form furrows and peel off, resulting in scratches, furrows, and spalling. The wear debris is mainly large-size strip and tear debris. However, the number and depth of scratches and furrows differ on different friction pairs. On the one hand, the roughness and hardness of PEEK and CoCrMo are different, and the number of micro protrusions on the femoral condyle surface and the number and depth pressed into the soft XLPE surface are different. On the other hand, PEEKXLPE is in soft contact with a larger actual contact surface, and contact stress between PEEK and XLPE is less than that between CoCrMo and XLPE under the same axial load. Therefore, the plowing effect of PEEK on XLPE is smaller than that of CoCrMo on XLPE. At the same www.Springer.com/journal/40544 | Friction time, the friction pair undergoes plastic deformation along the slide direction. When reaching the limit, cracks are formed at defects (inclusions, pores, dislocation accumulation areas, pitting pits, etc.). The source expands from the sub-layer to the wear surface, forming new furrows or local peeling, and part of the peeling is rubbed or torn to flakes, blocks, or tear debris. Part of the exfoliated particles are discharged from the friction contact area under the shear and lubricating flow to be free wear debris, and the other part is still in the friction contact area, acting as the abrasive for three-body wear, which indicates that the wear enters the accelerated fatigue wear stage. On the wear morphology, the furrows are deepened, the corrosion layer appears, and the wear debris is mainly flake and block debris. Under the high cyclic stress and rolling, the larger-sized exfoliated particles are fragmented into smaller-sized particles, which are further sheared and broken into many spherical or spherical abrasive particles with smooth edge contours and slight texture fluctuations. Part of it is discharged from the contact area to be wear debris, part of it is rolled on the wear surface to form an adhesion, and part of it is dispersed in the lubrication as microspheres to retard wear, which makes the wear to be a stable wear stage. The wear morphology is furrow, adhesion, etc., and the wear debris is main block and spherical wear debris. Figure 12 shows the wear mechanism of the XLPE tibial liner in PEEKXLPE and CoCrMoXLPE. XLPE, a thermosetting polymer with a network and semi-crystalline structure, has good mechanical properties and chemical stability. Its wear performance is mainly related to the hardness difference and surface roughness of the grinding pair. The hardness difference of PEEKXLPE is significantly less than that of CoCrMoXLPE, and the number of micro peaks on the PEEK surface is higher than that of CoCrMo. As a result, in the wear-in stage (0.5 million1 million cycles), the wear of XLPE in PEEKXLPE is more severe than that of CoCrMoXLPE, with more and shallower scratches and furrows. The XLPE wear debris in PEEKXLPE is mainly tearing debris due to extensive plastic deformation and cyclic alternating stress, and that in CoCrMoXLPE is tearing debris and curled flake debris, resulting in strong plowing. Part of the peeled particles are discharged from the contact area, and another part is stored in the contact area with a strengthened edge contour, which will damage the friction pair and accelerate the plastic deformation and crack source propagation of XLPE. The greater the particle hardness is, the faster the crack propagation speed is, the more peeling pits and peeling deposits are, the more serious the three-body wear is, and the more and deeper the furrows are formed, as shown in Figs. 5(c)5(e). In the accelerated fatigue wear stage (2 million4 million cycles), the furrows and peeling accumulation on the XLPE surface in CoCrMoXLPE were significantly more severe than those in PEEKXLPE. At this time, the XLPE wear debris in PEEKXLPE is mainly flake debris with a smooth surface, while that of CoCrMoXLPE is block debris and flake debris. With the cycles increasing, the more friction heat and higher temperature of lubricating fluid in PEEKXLPE friction pair are due to the poor thermal conductivity of PEEK compared  with that of CoCrMo. The protein of FBS lubricating fluid in PEEKXLPE is degraded, resulting in poor lubrication and an increased rate of XLPE wear loss, as shown in Fig. 4(a). During 4 million5 million cycles, the increase rate of XLPE wear loss slows down, and the wear enters a stable period with adhesion on the XLPE worn surface. Under the high cyclic stress, rolling, and shearing, large-scale peeling is broken into small spherical or spherical particles with smooth edges, some form free wear debris outside the contact area, and some are microspheres in lubrication to reduce wear. However, due to the slight hardness difference of PEEKXLPE friction pair with small fragmentation, the size of wear debris is more significant than that of CoCrMoXLPE, which is mainly massive wear debris and spherical wear debris, while CoCrMoXLPE is mostly spherical wear debris with a smaller size. Figure 13 shows the wear mechanism of PEEK and CoCrMo femoral condyles. The greater the hardness difference, the easier to form a transfer film on the material's surface with high hardness during the wear tests. There is a transfer film on the CoCrMo surface as a protection film to reduce wear. In addition, PEEK, a thermoplastic polymer, is prone to plastic deformation, crack propagation, falling off, and aggravating abrasive wear. Under the same conditions, the wear volume of PEEK is always higher than that of CoCrMo and increases rapidly. However, the corrosion resistance of CoCrMo is weaker than that of PEEK, which will lead to corrosion in the wear process. At the initial stage, the wear of the femoral condyles is mainly scratches and furrows due to the micro protrusions between friction pairs. The scratches on the PEEK surface are more than those on CoCrMo, accompanied by plastic ridges, and the PEEK wear debris is mainly rod debris. Under cyclic alternating loads, carbide micro bulges on CoCrMo surface were broken and strengthened into hard particles, which will damage the local surface with deeper furrows [29]. The wear debris is mainly flake and strip debris. The good thermal conductivity of CoCrMo is beneficial to dispersing frictional heat and maintaining FBS lubrication in the frictional contact area, which is more likely to form a stable protein film and reduce wear. Due to high-temperature corrosion resistance, corrosion areas appear on the CoCrMo surface. Under three body wear, the material is ploughed, and the corrosion spots are removed, but the protective film and corrosion areas will also be reformed and appear alternately, as shown in Figs. 6(b ii )6(f ii ). The increase of plastic deformation and crack propagation leads to material peeling and entering the accelerated fatigue wear stage. The PEEK wear debris is mainly flake and block debris, and the CoCrMo wear debris is mainly small block debris. At the later stage, adhesive wear occurs on the PEEK surface due to the relatively poor thermal conductivity and the small hardness difference between the friction pairs. The wear mechanism of PEEK is changed from abrasive wear to abrasive wear and adhesive wear, while the wear mechanism of CoCrMo is abrasive wear and local corrosion.

Conclusions
The wear loss of XLPE in PEEKXLPE (30.9±3.2 mg) was lower than that in CoCrMoXLPE (32.1±3.1 mg) after 5 million cycles of wear tests. The wear of XLPE in PEEKXLPE is mainly plastic deformation, abrasive wear, and adhesive wear. The wear of XLPE Fig. 13 Wear mechanism of (a) PEEK femoral condyle and (b) CoCrMo femoral condyle.
www.Springer.com/journal/40544 | Friction in CoCrMoXLPE is dominated by abrasive wear and adhesive wear. Generally, the wear of XLPE in CoCrMoXLPE is more serious than that in PEEKXLPE. The wear of the PEEK femoral condyle is mainly abrasive wear, and adhesive wear occurs at the later stage. The wear of CoCrMo femoral condyle is also dominated by abrasive wear, but at the middle and later stages, corrosive wear occurs alternately.
The size distribution of XLPE wear debris in PEEKXLPE is relatively dispersed, which in CoCrMoXLPE is relatively concentrated. The wear debris of both is mostly flake debris and block debris, but the XLPE wear debris in PEEKXLPE has a smoother surface and more regular edge contour than that in CoCrMoXLPE. The PEEK wear debris is mainly in flakes and blocks, and the CoCrMo wear debris is mainly in the shape of rods and blocks. The surface of the CoCrMo wear debris is rough and has nano-grinding particles attached.