Introduction

Silicon nitride (Si3N4) is a technical ceramic that has been widely used in many industrial applications, such as turbine blades, bearings, cutting tools, and spark plugs [1, 2]. In addition to excellent mechanical properties, Si3N4 ceramics have also been reported to be biocompatible and bacteriostatic [3, 4]. Owing to the above advantages of Si3N4, it has been utilized as spinal fusion implants [5]. In addition, Si3N4-based materials have been considered to be used as total hip replacements (THR) and total knee replacements (TKR) both in bulk form as well as coatings, since Si3N4-based materials typically show high wear resistance [3, 5, 6].

Owing to their good wear properties, zirconia (ZrO2) and alumina (Al2O3) have been the main ceramic bulk materials considered for joint bearings, but both of them still have some disadvantages [7]. Yttria-stabilized zirconia (YSZ) exhibits tetragonal to monoclinic phase transformation when exposed to the in vivo environment, sometimes leading to implant failure [8, 9]. Zirconia Toughened Alumina (ZTA) ceramics have shown improvements compared to pure Al2O3 ceramics, e.g., higher fracture strength. However, neither ZrO2-based ceramics nor Al2O3 ceramics have shown bacteriostatic properties as Si3N4 has demonstrated [10,11,12]. Another advantage of Si3N4 over ZrO2 or Al2O3 in terms of use as an implant is that wear debris from Si3N4 can dissolve slowly in aqueous solutions [13], forming biocompatible ions, which potentially reduces any negative effect the wear debris may have on the surrounding tissues.

Sintering additives are needed to assist the densification of Si3N4. Before sintering, Si3N4 raw powder is mixed with a small amount of sintering additives, typically Y2O3 and Al2O3, which aid sintering and densification via formation of a liquid phase [4, 14]. Upon solidification, the additives form an intergranular amorphous phase or slightly crystallized phase at the grain boundaries [15, 16]. An alternative way of achieving dense Si3N4 may be through the use of more biocompatible and bioactive sintering additives, such as SrO, MgO, and SiO2 [4, 17]. These sintering additives have the potential to improve the bioactivity of the Si3N4. For example, it has been found that Sr could trigger osteoblast formation and inhibit osteoclasts, potentially leading to increased bone formation [18]. Researchers have also reported that Sr and Mg could improve osteogenesis and vasculogenesis [17, 19] and that Si may increase bone mineralization [20]. In an earlier study by the authors, a new Si3N4-based bioceramic was developed by adding 10% and 18% of SrO, MgO, and SiO2 as sintering additives [19, 21]. Bioactivity over time was confirmed through exposure to a phosphate buffer solution, resulting in a nanotopographic surface which could be potentially beneficial for osseointegration and cell adhesion [22,23,24,25,26]. Meanwhile, indirect cell studies with MC3T3-E1 (osteoblast) cells showed no toxicity [19, 21].

It is well known that wear properties play a significant role when Si3N4-based bioceramics are to be used as implants for total hip replacements (THR) and total knee replacements (TKR). However, the wear properties of the newly-developed Si3N4 bioceramics have not been evaluated. Hence, the aim of this work was to evaluate the wear performance of this new Si3N4 bioceramic. The Si3N4 of this study was densified using a spark plasma sintering (SPS) instrument with SrO, MgO, and SiO2 as sintering additives. Surface roughness and morphology were characterized by optical profilometry and scanning electron microscopy (SEM), respectively. The phase composition was analyzed by X-ray diffraction (XRD) measurements. A multidirectional wear test was used for wear resistance tests.

Materials and Methods

Spark Plasma Sintering

The starting raw materials, including Si3N4 powder (β phase, with an average particle size of approximately 20 μm), SrO, MgO, and SiO2 (as sintering aids), were mixed for 3 h in a polyethylene jar using Si3N4 balls and ethanol (> 99%) as a mixing medium. Two compositions were manufactured, i.e., 2%SiO2–4%MgO–4%SrO–90%Si3N4 (wt%) and 2%SiO2–8%MgO–8%SrO–82%Si3N4 (wt%). These materials are hereinafter denoted as SN10 and SN18, respectively. According to our previous studies, Si3N4 sintered with 10-wt% and 18-wt% glass phase showed significant differences in terms of densification behavior, mechanical properties, phase composition, and ion release. Hence, in this study, we choose to move forward with both 10-wt% and 18-wt% glass phase to explore their effects on the wear behavior of the Si3N4. Si3N4 discs were sintered using spark plasma sintering (SPS-825, Fuji Electronic Industrial Co., Ltd, Japan). A traditional Si3N4 ceramic with a composition of 4%Al2O3–6%Y2O3–90%Si3N4 (wt%) was also manufactured as a control and was denoted as SNC. The detailed sintering process has been described elsewhere [19, 21], but it is reproduced herein for clarity. The SN10, SN18, and SNC samples were sintered with slightly different sintering parameters, as follows: a heating rate of 100 °C/min from room temperature to the sintering temperature, an applied pressure of 60 MPa for the three groups, and a dwelling time of 3 min at the sintering temperature, then a sintering temperature of 1680 °C for the SN10 group and SN18 group, and a sintering temperature of 1700 °C for SNC group. All samples were sintered in a N2 atmosphere. All the sintered samples were prepared as discs (Ø20mm x 2 mm). The sintered samples were wet ground and polished on an automatic polishing machine to achieve smooth surfaces. The average relative density of the SN10 group and the SN18 group was approximately 92.2% and 91.4% [19], respectively, while the relative density of the control group was 93.7%.

Multidirectional Wear Tests (MWT)

The wear behavior of the prepared Si3N4 against ultra-high molecular weight polyethylene (UHMWPE) was assessed using cylindrical UHMWPE pins with a nominal length of 19.1 mm and a diameter of 9.5 mm. The pins were made of UHMWPE GUR1020 (Peter Brehm GmbH, Weisendorf, Germany). Multidirectional wear tests (MWTs) were performed in 0.2-μm filtered fetal bovine serum solution (25%) at 37 °C. Prior to the test, the pins were pre-soaked in serum and cleaned according to the standard [27]. According to ASTM F732-17 [27], wear testing was carried out with a nominal load of 213 N, resulting in an estimated contact pressure of 3 MPa. A frequency of 2 Hz and a sliding velocity 40 mm/s were used for 2,000,000 cycles (2.0 MC), using a 5-mm square path for a sliding distance of 20 mm/cycle. To evaluate the wear of the UHMWPE pins, a gravimetric method was used [28]. The volumetric wear was calculated to obtain the wear factor (k), which is the volumetric wear divided by the applied load and multiplied by the sliding distance [28].

Material Characterizations

Phase identification was carried out by XRD (Diffractometer D5000, Siemens, Germany). XRD patterns were recorded in the range of 5°–90° with a scan step of 0.2 s/step and a step size of 0.012° using a Ni-filtered Cu Kα X-ray resource (40 kV and 40 mA). The average surface roughness was measured before and after wear testing using a Nexview™ NX2 3D Optical Surface Profiler, at 10X and a field of view (FOV) of 2.0, corresponding to an area of 500 × 500 μm2. Five measurements were performed on each sample to obtain Sa (arithmetical mean height). Morphological characterization was carried out by scanning electron microscopy (SEM, Zeiss Merlin with AZtec EDS/EBSD, Oberkochen, Germany) before and after wear tests using secondary electrons (SE) at 20 kV. The studied samples were sputtered with Ag/Pd coating for 40 s to avoid charging effects. Before observation, the samples were etched with 5% HF solution for 60 min for grain exposure. After every 0.5 MC, the pH of the lubricant solution was measured at room temperature (around 21.5 °C) using a pH meter (METTLER TOLEDO, Ohio, USA).

Statistical Analysis

IBM SPSS Statistics v 22 was used for all statistical analyses. A repeated measures ANOVA was used to determine statistical differences in roughness, coefficient of friction, and wear factor as a function of material and time point. A critical level of α = 0.05 was used to determine significance.

Results and Discussion

Phase Identification

The phase composition of the materials is shown in Fig. 1. It was observed that all three groups were composed of α-Si3N4 and β-Si3N4. For the samples sintered with SrO, MgO, and SiO2, no other crystalline phases containing Sr, Mg, or Si were detected, indicating that these additives (SrO, MgO, and SiO2) were amorphous after sintering. Even for the SN18 sample where the content of sintering additives was 18 wt%, the additives (SrO, MgO, and SiO2) did not crystallize, neither did they react with each other to form other compound. For the reference sample (SNC), no crystalline phases containing Al or Y were detected either (Fig. 1). Hence, these additives (Al2O3 and SiO2) were also amorphous after sintering. The sintering additives did not alter the crystallization behavior of the Si3N4.

Fig. 1
figure 1

XRD patterns of the studied Si3N4 showing the presence of both alpha and beta phases

Full size image

Coefficient of Friction

The coefficient of friction of the studied Si3N4 is presented in Fig. 2, from which it can be seen that in all tests it ranged between 0.06 and 0.10 (Fig. 2), which is comparable to previous studies on Si3N4 bulk material and coatings [29, 30]. Previous work has found that a tribofilm forms on the surface of Si3N4 in aqueous environments during wear tests [31,32,33]. This tribofilm is a thin layer consisting of SiO2 and Si(OH)2, and this layer can improve the wear resistance of Si3N4 by acting as a self-lubricating layer and reducing the coefficient of friction [5, 34]. No statistically significant difference could be found between the three Si3N4 at any of the time points, nor was there any significant effect of time detected. In other words, both the SN10 and the SN18 samples showed friction coefficients as low as that of the SNC sample. Having a low coefficient of friction is one of the most important prerequisites for biomaterials to be used as total hip replacements (THR) and total knee replacements (TKR) [16]. Hence, the new Si3N4 sintered by SrO, MgO, and SiO2 fulfilled this prerequisite.

Fig. 2
figure 2

Coefficient of friction of the Si3N4 with wear cycles ranging from 0.5 MC to 2 MC. The bar charts are showing mean values ± standard deviations

Full size image

Wear Factor

The wear factors of the UHMWPE pins sliding against the studied Si3N4 are displayed in Fig. 3. There was no statistically significant difference between the wear of the UHMWPE pins against the different material groups. While the wear factors measured in these tests were higher for the first 1.5-M cycles when compared to previously reported values, they were lower for the last 500-k cycles—by one order of magnitude. For instance, the wear factor of UHMWPE pins sliding against a CoCrMo alloy (Ra = 15 nm) was reported to be 2.2 × 10–6 mm3/N.m [35], and UHMWPE pins sliding against Charnley implants (Ra = 10 nm) after clinical use was 2.1 × 10–6 mm3/N·m [36]. While the contact pressures were lower in those studies, another reason for the lower wear factors could be due to differences in surface roughness of the mating materials. Indeed, on the pins’ surfaces (Figs. 7–9), the formation of scratches and plastic deformation (highlighted by arrows) was observed. This was likely the result of abrasion and adhesion of the polyurethane with microscopic asperities on the ceramics’ surfaces [37]. On the other hand, the repetitive movement during the wear cycles led to a preferred orientation of the UHMWPE molecules in the main sliding direction, resulting in strain hardening [38,39,40], which may explain the lower wear factor toward the end of the test. A multidirectional wear path is commonly chosen to better mimic clinical wear [41] because path shear and tensile forces occur in several directions [42]. Despite the multidirectional movement, the same pattern was followed over time, and the increased hardness of the polymer may be a reason for the lower wear rate in the last part of the test. Another contributing reason may be the formation of a tribofilm, as previously discussed.

Fig. 3
figure 3

a Wear factor of the UHMWPE pins sliding against the different silicon nitride materials from 0 to 2MC. b The wear factor at 1.5–2 MC is presented at a different scale for better visualization. The bars are showing mean values ± standard deviations

Full size image

Surface Roughness

The surface roughness of the studied Si3N4’s before and after wear tests are presented in Fig. 4. As can be seen, both the SN10 and the SN18 samples showed surface roughness values lower than 50 nm before the wear tests, which was in accordance with the standard [43]. However, that of the SNC sample was slightly (around 5%) higher than the recommended value. It can also be observed that the surface roughness of the SNC sample was, in general, the highest during the wear test, an effect that was statistically significant for all time points, except vs SN10 at 0–0.5MC and at 1.5–2.0MC. In fact, the surface roughness of the SN10 sample increased progressively during the test, reaching around 50 nm at the end of the test. Whereas, the SN18 sample maintained a surface roughness value in agreement with the standard [43] during the whole test, which was significantly lower than the SNC sample at each time point. The increase in surface roughness of the SN10 sample indicated that some dissolution of the surface could have taken place during the test. However, the actual dissolution rate of the new Si3N4 needs further investigation. It is a complex matter depending on many factors, such as the type of liquid medium, temperature, and pH. A comprehensive study on the dissolution rate of the materials was not carried out herein because it was out of the scope of the present study. The focus herein was on an initial evaluation of the wear behavior of the two new Si3N4 ceramics.

Fig. 4
figure 4

The surface roughness of the studied Si3N4 before wear tests and at each time point until 2 MC. The bars are showing mean values ± standard deviations

Full size image

Surface roughness (Ra) of UHMWPE pins before and after wear tests is shown in Fig. 5. The pins had similar surface roughness (1.5 μm) before the wear test. No statistically significant difference in roughness was found between the pins ran against the control material or against SN10, except at the 1–1.5MC time point. Also, there was no significant difference between the pins against SNC and SN18 except at 0 cycles and at 1–1.5MC, where the pin ran against SN18 had a significantly lower roughness. Therefore, while a generally lower roughness may have been expected for the pins running against the new materials, which had a lower roughness than the control throughout the tests, this effect was not evident. One limiting factor here may be the relatively low number of samples tested. Another reason may be the high scatter of the data, possibly due to an abrasive process in the beginning of the tests—with the highest scatter—and the polymer strain-hardening effect and/or tribofilm resulting in a lower scatter in the remainder of the tests. This is also in accordance with the lower wear factors toward the later parts of the test. Indeed, the effect of the tribofilm may have canceled any effect of the difference in roughness of the ceramic materials.

Fig. 5
figure 5

Surface roughness (Ra) of UHMWPE pins before wear tests and at each time point until 2 MC during wear tests. The bars are showing mean values ± standard deviations

Full size image

The surface morphology evolution of the pins can be seen in Figs. 6, 7, and 8. The pins showed a smooth surface before the wear tests (Figs. 6a, 7a, and 8a). For the pins slid against the SNC sample, plastic deformation occurred on the surface of the pin after 0.5 MC (Fig. 6). On the other hand, for the pins which slid against the SN10 sample, a few machining marks were visible after 0.5 MC, and plastic deformation occurred after 1.0 MC. After 2.0 MC, deformities can be observed on the pin surface (Fig. 7). As for pins slid against the SN18 sample, plastic deformation only occurred after 1.0 MC and after 2.0 MC (Fig. 8). Hence, the type and relative amount of the glass phase in the Si3N4 showed effects on the surface morphology of the pins after wearing, suggesting it may also have had an effect on the tribofilm.

Fig.6
figure 6

Surface evolution of the UHMWPE pins which slid against the SNC sample. a Before wear test; b after 0.5 MC; c after 1.0 MC; and d after 2.0 MC. Regions where plastic deformation/wear occurred on the pin surface are marked by arrows

Full size image
Fig. 7
figure 7

Surface evolution the UHMWPE pins which slid against the SN10 sample. a Before wear test; b after 0.5 MC with few machining marks; c after 1.0 MC with plastic deformation; and d after 2.0 MC showing deformities on the pin surface. Plastic deformation/wear occurring during the wear cycles are highlighted by arrows

Full size image
Fig. 8
figure 8

Surface evolution the UHMWPE pins which slid against the SN18 sample. a Before wear test; b after 0.5 MC; c after 1.0 MC; and d after 2.0 MC. Plastic deformation/wear occurring during the wear cycles were highlighted by arrows

Full size image

Surface Morphology

The surface morphologies of the studied Si3N4 before and after wear tests were observed with SEM and the results are displayed in Fig. 9. The as-sintered Si3N4 showed relatively smooth surfaces after polishing, except for a small number of shallow voids (Fig. 9a–c). The shallow voids could be pores formed because of incomplete densification or surface defects caused by the grinding and polishing. To more clearly visualize the Si3N4 grains, the surfaces were etched with HF acid and the corresponding SEM images are shown in Fig. 9g–I, from which long rod-like β-Si3N4 grains were observed [44]. During the wear tests, slight dissolution occurred on the surface of the ceramic discs (Fig. 9d–f), which was in accordance with earlier studies [19, 21]. The glass phase slowly dissolved exposing the Si3N4 grains, resulting in a rougher surface, which could therefore increase the wear of the counter surface. However, the pin wear rates decreased over time, which could be due to polishing and hardening effects of the polymer surface, as mentioned earlier. No particles could be found on the surface of the materials, nor any visible debris of the ceramics. This could be indicative of good wear resistance, since the presence of particles or debris on the interfaces of the ceramics would not only result in the formation of more particles but also aggravate friction and wear due to a plowing effect [45].

Fig. 9
figure 9

Surface morphology of the studied Si3N4 characterized by SEM: a–c polished surface before wear testing; d–f after wear testing for 2 MC; g–i polished and etched surface before wear testing. The surfaces were etched with HF to visualize the Si3N4 grains

Full size image

pH Measurements

Figure 10 shows the pH of the wear test solutions. The pH value of the fetal bovine serum solution was approximately 7.7. The wear test solutions showed a pH of around 8.5. This means that basic ions, probably ammonium ions (NH4+) and their modifications, were released from the Si3N4 surfaces to the wear test solutions during the test. Si3N4 has an oxidized surface layer of a silicon oxynitride (Si–O–N). Friction occurring at the surface of Si3N4 might have resulted in the formation and release of ammonium ions. It has been reported that this might induce bacteriolysis [46], which renders Si3N4 bacteriostatic properties. Both the SN10 sample and the SN18 sample had a statistically significant higher pH value than that of the SNC sample at all time points except at 1.5–2.0MC. It has previously been shown that not only does the pH interfere with the dissolution of the material, but it also affects the coefficient of friction and wear [47]. When the system has a higher pH and a higher surface charge, the tribofilm seems to be thinner, resulting in a higher electrostatic repulsive force between the debris generated during wear and the tested surface [48]. Further investigation into the effect of pH on the current system would be of interest, especially considering that inflammation may give rise to a local pH decrease [49, 50]. In vivo, the synovial fluid pH is around 7.3—7.4 for normal conditions [49, 50], whereas pH values ranging between 6.6 and 7.4 have been found in fluids of inflamed joints [51].

Fig. 10
figure 10

pH of the wear test solutions at time points from 0.5 to 1.0 MC. All measurements were performed at room temperature (T = 21.7 ± 0.7 ℃). The bars are showing mean values and the error bars standard deviations

Full size image

Conclusions

In this work, the wear properties of two new Si3N4 bioceramics were evaluated. The new Si3N4 bioceramics were sintered using SrO, MgO, and SiO2 as sintering additives, rather than traditional aids, i.e., Al2O3 and Y2O3. Multidirectional wear tests showed that the new Si3N4 bioceramics had promising wear resistance, including low coefficients of friction and a low wear factor of UHMWPE pins at longer cycle times. The formation of a lubricating tribolayer appeared important to the wear factor of the mating UHMWPE surface. There was no statistically significant difference in pin wear rates between the new Si3N4 compositions and the reference Si3N4 sintered using Al2O3 and Y2O3 as additives. Although a slight dissolution occurred on the surface of the Si3N4 during the test, no particles or debris were found at the friction interface. It can be concluded the new Si3N4 bioceramics showed promise in terms of wear performance but further tests simulating more realistic in vivo conditions are necessary.