Pure PEEK (99%) was purchased from Junhua Technology (Changzhou, China). GO liquid dispersion with a concentration of 1 mg/mL was provided by Dazhan NAMI (Guangdong, China). Concentrated sulfuric acid, acetone, and ethanol were obtained from Kolon Chemical (Chengdu, China).

Pure PEEK was crushed into a 10 mm × 10 mm × 3 mm sheet. The sheets were successively polished with 200-, 300-, 500-, 800-, 1,200-, 2000-, 3,000-, 5,000-, and 7000-grit silicon carbide sandpaper and then sequentially put into acetone, anhydrous ethanol, and ultra-pure water for ultrasonic cleaning for 15 min, respectively, to remove impurities from the surfaces.

The cleaned PEEK sheets were immersed into a solution of sulfuric acid at a concentration of 95 wt% for 5 min at 25°C and then washed in a 1 mol/L NaOH solution to remove the residual sulfuric acid. The above sheets were then dried in an oven at 60°C. Afterward, the sulfonated PEEK pieces were designated as SPEEK. The SPEEK specimens further modified with GO (designated as GO-SPEEK) were prepared as follows: first, the SPEEK specimens were immersed in a GO dispersion at a concentration of 1 mg/mL for 5 min, and then the specimens were removed and dried in an oven at 60°C for 30 min. The schematic diagram of the GO-SPEEK sample preparation procedure is shown in Figure 1.

Scanning electron microscopy (SEM, Zeiss Gemini 300, Germany) was used to observe the surface morphology of PEEK, SPEEK, and GO-SPEEK. All samples were sprayed with gold for 2 min before the SEM observation with a scanning wavelength range of 3,000–500 cm⁻¹. Fourier transform infrared spectroscopy (FT-IR, Hermo Scientific Nicolet iS20, America) was used to analyze the surface chemical composition of the samples by using attenuated total reflection. The crystal state of samples was tested by X-ray diffraction (XRD, Rigaku Smartlab SE, Japan) with a scanning range of 2θ from 10° to 80° and a step length of 5°/min. The Raman spectra of the different samples were recorded using the LabRam HR Evolution (SPM-960, Japan) laser Raman spectrometer, with a laser wavelength of 633 nm and a wavenumber range of 2,500–1,000 cm⁻¹. In addition, X-ray photoelectron spectroscopy (XPS, Thermo Scientific ESCALAB, United States) with Al Kα radiation was utilized to detect the chemical composition and elemental state of the elements.

The hydrophilicity of the samples was assessed by the static droplet method using a contact angle tester. The volume of water droplets was 4 μL. After the droplet touched the surface of the samples for 5 s, the water contact angle was recorded. Three parallel samples were selected for each group, and three different locations on the surface of each sample were tested.

The antimicrobial capacity of the samples was assessed by using the bacterial strains of Escherichia coli (E. coli, ATCC25922) and Staphylococcus aureus (S. aureus, ATCC25923). The bacterial inoculum was prepared with 10 g/L peptone, 5 g/L NaCl, and 3 g/L beef paste. E. coli and S. aureus were then added to PBS at a concentration of 10⁶ cells/mL (calibrated according to colony forming units (CFU) using the diffusion plate method). The tested samples were sterilized using medical alcohol immersion as well as UV irradiation and then placed in individual wells of a 24-well culture plate. A 1-mL aliquot of the above bacterial suspension was added to each sample in the wells and incubated at 37°C for 24 h. To examine the bacterial morphology, the samples were washed three times with PBS to remove the unadhered bacteria and immersed in 2.5% glutaraldehyde for 1 h at 4°C. The samples were subsequently dehydrated in ethanol and then vacuum-dried. After gold spraying, the samples were observed under FESEM. To quantitatively explore the live bacteria colonies on different surfaces, the above-mentioned 24-h incubation samples were rinsed using PBS and placed separately in 3 mL of PBS. After being ultrasonically shaken to detach the adhered bacteria from the samples, the resulting bacterial suspension was diluted 10,000-fold and incubated at 37°C for 24 h in a standard agar medium (Huang et al., 2021), after which the active bacteria were counted by the plate method. Three samples for each group were tested, and each test was repeated three times (n = 3).

The biomineralization capacity of different samples was assessed by exploring the hydroxyapatite (HA)-precipitation activity on the surfaces by immersing the samples in a simulated body fluid (SBF) solution. The SBF solution was formulated as follows: NaCl (8.035 g), NaHCO₃ (0.355 g), KCl (0.255 g), K₂HPO₄ (0.176 g), MgCl₂ (0.146 g), CaCl₂ (0.292 g), and Na₂SO₄ (0.072 g). During the preparation, the ingredients were dissolved in deionized water and kept at a temperature of 36.5°C, and the solution was adjusted to pH = 7.35 by using 1 mol/L hydrochloric acid and Tris (hydroxymethyl)aminomethane (Tris) (Barooghi et al., 2018). The samples were placed in centrifuge tubes after the SBF solution was added, and the tubes were transferred into a constant temperature incubator at 37°C. At the target times of 14 days and 28 days, the samples were removed, dried, and sprayed with gold to observe the surface morphology by SEM. In addition, to reveal the surface elemental distribution of the SBF-immersed samples, EDS-mapping analysis was conducted on the SEM equipment using an SEM-EDS system. The substances formed on the different surfaces were also investigated by means of XRD.

To display the effect of surface modification on the mechanical properties of PEEK, a tensile test was conducted on the samples using an electronic universal testing machine (TY8000-A) in accordance with the ASTM standard D638 (Laureto and Pearce, 2018; Zhao et al., 2023). The shape of the test sample is shown in Figure 2. Three samples for each group were used to obtain the average.

Figure 3 shows the PEEK, SPEEK, and GO-SPEEK surface morphology. As shown in Figures 3A, B, after being ground and polished, the PEEK surface is smooth and flat, Figures 3C, D show the SPEEK surface morphology; a three-dimensional (3D) porous structure with an average pore size of about 8 μm is formed on the surface. After further coating with graphene oxide, the porous surface varies compared to the sulfonated surface. Part of the 3D porous structure is covered by the GO layer, leading to a decrease in the average pore size (Figures 3E, F). To further disclose the morphology of the grafted GO, a magnified image (the inset in the upper right corner of Figure 3F) is taken at the location indicated with the letter A; it demonstrates that the GO deposited on the SPEEK sample exhibits a wrinkled-like sheet appearance, which is similar to the results reported by Kumar et al. (2023).

Figure 4A depicts the XRD crystal phase of the different samples. As can be seen, PEEK is a semi-crystalline polymer exhibiting relatively sharp crystalline peaks, with higher-intensity peaks at 2θ about 19°, 21°, 24°, and 29°, corresponding to the diffraction of (110), (113), (220), and (213) crystalline planes respectively, which is similar to the previously reported results (Cao et al., 1994). SPEEK and GO-SPEEK also have characteristic peaks at the same positions with weakened intensity, indicating that the surfaces of the samples have been modified. The FTIR spectra of different samples are shown in Figure 4B. SPEEK and GO-SPEEK show new absorption peaks at 1,050 cm⁻¹ and 1,251 cm⁻¹. The 1,050 cm⁻¹ peak corresponds to the S=O symmetric stretching absorption peak, and the 1,251 cm⁻¹ peak corresponds to the O=S=O asymmetric stretching absorption peak (Zhao et al., 2013; Chen et al., 2018). This indicates that the -SO₃H functional group has been introduced into the PEEK polymer chain by the sulfonation reaction (Zhao et al., 2013). Because of the adsorption effect of the micropores generated by the sulfonation, GO is confirmed as being successfully loaded on the SPEEK surface. As shown in Figure 4C, GO-SPEEK has two characteristic peaks that appear around 1,350 cm⁻¹ and 1,607 cm⁻¹, corresponding to the D-band and G-peaks (Ferrari and Basko, 2013). The peaks represent defects in the C-atom lattice and in-plane stretching vibrations of the sp2-hybridized C-atoms, respectively.

It was reported that the biological properties of the biomaterials are closely related to their surface hydrophilicity (Wang et al., 2019; Kopac, 2021). Therefore, a water contact angle tester was used to analyze the hydrophilicity of different samples. The results in Figure 4D reveal that the water contact angle of the original PEEK surface is 82.7° ± 3.8°, and the value decreases to 55.2° ± 2.6° for SPEEK due to the introduction of the more hydrophilic -SO₃H by the electrophilic substitution reaction (Alimohammadi and Ramazani, 2023). Introducing the GO on the SPEEK surface further reduces the water contact angle to 10.9° ± 1.1°, which might be due to the hydrophilic functional groups, such as hydroxyl and carboxyl groups, on the modified surface (Dikin et al., 2007), causing further improvement of the GO-SPEEK hydrophilicity.

XPS was adopted to explore the chemical composition before and after surface treatment. The wide-scan spectra show that only C and O are detected for the PEEK surface. A minor amount of S could be detected for the SPEEK surface, and the relative S percentages of PEEK, SPEEK, and GO-SPEEK are 0%, 2.57%, and 0.68%, respectively (Figure 5A), suggesting a significant decrease of S after being grafted with GO. The high-resolution spectra of C 1s of different samples are shown in Figures 5B–D. The characteristic peaks appearing at 284.8 eV and 286.5 eV correspond to the C-C/C-H and C-O groups, respectively (Ouyang et al., 2018b; Al-Gaashani et al., 2019). C=O bond and π-π* bonds at 287.7 eV and 291.4 eV are observed for PEEK (Figure 5B, which change little after sulfonation (Figure 5C), similar to previous reports in the literature (Ouyang et al., 2018b). However, an O-C=O bond appears for GO-SPEEK (Figure 5D), suggesting that GO is successfully loaded on the SPEEK surface (Torrisi et al., 2020). This can also be evidenced by the increased C content of GO-SPEEK compared to SPEEK (Figure 5A) (Al-Gaashani et al., 2019).

It has been previously observed that there is a significant competitive surface adhesion relationship between the invading bacteria and the host cells, which is described as surface competition (Gristina, 1987). Free bacteria adhere to the implant surface and subsequently form a stubborn biofilm that disrupts host-cell adhesion, leading to a failure of implant-cell integration. It is essential to prevent biofilm formation to avoid bacterial infections (Hemmati et al., 2021). Figure 6A shows the antibacterial effect of different samples on E. coli. As can be seen from the SEM result, after 24 h of co-culture, the PEEK surface shows the highest number of E. coli, and the bacteria exhibit a typically rod-like morphology. The profile of the bacteria is clear, and the cell membrane is relatively smooth. In contrast, the E. coli on the SPEEK surface reveals a deformed and elongated shape, suggesting that the SPEEK sample creates an uncomfortable environment for bacteria to live. Note that few E. coli could survive on the GO-SPEEK surface, and the attached E. coli displays a destructed appearance. The agar dilution method proves the above SEM result; the number of E. coli on the samples follows the order of PEEK > SPEEK > GO-SPEEK. The plate count result is shown in the histogram and indicates that the PEEK surface has the highest number of E. coli colonies at 3.45×10⁶, higher than the 2.72×10⁶ colonies formed on the SPEEK surface, and much higher than the 4.6×10⁵ colonies formed on the GO-SPEEK surface. It is evident that the GO-grafted PEEK surface has a significant bactericidal effect on E. coli; the number of E. coli is reduced by approximately 86% after the PEEK surface modifications.

S. aureus is one of the leading causes of hospital- and community-associated infections (Huang et al., 2011) and is a large proportion of the bacteria that infect orthopedic implants (Arciola et al., 2018). Therefore, it is even more important to examine the resistance of orthopedic implants to S. aureus. Figure 6B shows the antibacterial properties of different samples against S. aureus. The SEM results show that the S. aureus grows well on the PEEK surface and exhibits the typical morphology of grape bunches. Fewer S. aureus are observed on the SPEEK surface, but they still maintain the shape of intact balls. However, the S. aureus on the GO-SPEEK surface seems lysed and barely visible. The SEM result is highly consistent with the agar plate culture result: fewer S. aureus colonies are on the SPEEK surface, and many fewer colonies are on the GO-SPEEK surface than the PEEK surface. The quantitatively counted result reveals that the PEEK surface has the largest number of S. aureus colonies, 3.98×10⁶. The number of colonies decreases to 1.95×10⁶ for the SPEEK surface (a decline of approximately 51%), while the GO-SPEEK surface is observed to be more resistant to S. aureus with an approximately 94% reduction of bacteria colonies to about 2.3×10⁵. The grafted GO plays an important role in endowing GO-SPEEK with high efficiency in killing both Gram-negative and Gram-positive bacteria.

The grafted GO might greatly improve the antimicrobial effect of PEEK in two ways: first, the wrinkled GO nanosheet could provide many sharp edges (Ouyang et al., 2018b; Kumar et al., 2023), which can cause physical damage to the bacterial cytosol and destroy the bacterial membrane, thus causing cytoplasmic loss and ultimately leading to death of bacteria (Akhavan and Ghaderi, 2010). Second, the bacterial membrane is mainly composed of phospholipid macromolecules, and GO has a unique two-dimensional structure with sp² carbon, resulting in strong dispersion between GO and phospholipids, which allows GO to continuously capture the phospholipids of the bacterial membrane and leads to the eventual destruction of the bacterial membrane (Tu et al., 2013). Some scholars have noted that the hydrophobic interactions between GO and phospholipids would contribute to bacterial phospholipid translocation and overturn of the cytosolic membrane and, therefore, exacerbate damage to the bacterial membrane (Dallavalle et al., 2015; Guo et al., 2021b). In this work, the surface-modified PEEK materials exhibit higher bactericidal capacity against the Gram-positive S. aureus than the Gram-negative E. coli. The reason may be that the two kinds of bacteria have different acid resistance. E. coli has the ability to transfer Gln to Glu, which produces gaseous ammonia. The free ammonia could neutralize protons, leading to an increase in intracellular pH and improving their resistance to acidic environments (Lu et al., 2013). In contrast, S. aureus is less resistant to acids. In our article, the sulfonation treatment introduced -SO₃H groups, which are highly acidic, on PEEK (Sutherland et al., 1994; Ouyang et al., 2016). Therefore, the surface-modified PEEK shows much stronger antibacterial activity for the S. aureus strain.

The bioactivity of the biomaterials could be assessed by evaluating the ability to deposit HA on the surface after a certain time of in vitro SBF immersion. PEEK, SPEEK, and GO-SPEEK were placed in an SBF solution and incubated at 37°C in a constant temperature incubator. After 14 days and 28 days of immersion, the samples were dried and observed by SEM. As shown in Figures 7A, D, there is no morphological change on the PEEK surface after soaking for 14 days and 28 days, and no deposited material is observed on the PEEK surface. SPEEK, after 14 days of immersion, displays distinct ball-like deposits on the surface (Figure 7B), and the deposits become more dense after 28 days (Figure 7E). As for GO-SPEEK, the surface is almost totally covered by the ball-like deposits, even after 14 days (Figure 7C). The GO-SPEEK surface is completely covered by densely white precipitations after 28 days of immersion in the SBF solution (Figure 7F). The EDS results from a typical ball-like deposit (indicated with a letter B in Figure 7F) on the surface of GO-SPEEK after 28 days of immersion in SBF suggests that these balls are abundant in calcium (Ca) and phosphorus (P), and the Ca/P ratio is about 1.60 (Figure 7G), which is very close to the Ca/P ratio in HA (1.67). Figure 7H shows the EDS two-dimensional mapping result of Figure 7F. It can be seen that C, O, Ca, and P are generally evenly distributed throughout the surface. The Ca and P distributions are highly correlated, further indicating that Ca combines with P to form Ca/P precipitations on the GO-SPEEK surface. The XRD profiles of the samples before and after SBF immersion for 28 days are depicted in Figure 8. As can be seen, apart from the featured peaks belonging to PEEK, the peaks located at 2θ ≈ 32.4° appear on the surface-modified samples, which fit well to the (211) crystalline plane of HA (Almasi et al., 2014). These results indicate that sulfonation could enhance the biomineralization capacity of PEEK, while the introduction of GO could further boost the formation of HA on the SPEEK surface. GO-SPEEK exhibits a much stronger biomineralization capacity than the PEEK or SPEEK samples.

The sulfonation could introduce -SO₃H groups on the samples, which was reported to enable ion exchange and electrostatic interaction of ions with the SBF solution, thus favoring surface mimetic mineralization (Hamai et al., 2018). In addition, it has been reported that the negatively charged GO surface could adsorb a significant amount of positively charged calcium ions from the SBF solution, which facilitates the combination of the PO₄ ³⁻ ions and subsequently accelerates the precipitation of HA in the mineralization process (Gao et al., 2015). Accordingly, the biomineralization capacity of the investigated samples in this work is GO-PEEK > SPEEK > PEEK.

In general, the adhesion of osteoblasts on the surface of a biomaterial means that osseointegration is initiated (Gongadze et al., 2011). The morphology of the MC3T3 cells after 5 h of incubation on different sample surfaces was imaged by SEM, as shown in Figure 9A. The cell is weakly adherent to the PEEK surface with an overall spherical shape, and the elongation of the cell pseudopods is not observed. In contrast, the cell on SPEEK is irregularly shaped and stretched much wider with distinct pseudopods. The cell on GO-SPEEK is observed to be fully spread and tightly bound to the surface, implying that the GO-SPEEK surface provides the optimal circumstances for the osteoblasts to attach. The CCK-8 assay was used to further disclose the adhesion and proliferation conditions of cells on the samples. It can be seen from Figure 9B that, at the initial incubation times of 1 h and 5 h, no significant difference in the viable cells adhered to the samples was observed. At 24 h, the numbers of adherent cells on the different samples all increased compared to 5 h. As time progressed, the cells on the three kinds of samples all proliferated. These results prove that the PEEK, SPEEK, and GO-PEEK samples in this study all exhibit good cytocompatibility, and they boost cell growth on the surfaces. Notably, the cell proliferation is observed to be accelerated on the SPEEK sample, and the acceleration is more pronounced on the GO-PEEK sample than on the PEEK sample.

The enhanced cell adhesion and proliferation will influence subsequent differentiation and ECM mineralization (Luo et al., 2015; Huang et al., 2021). The osteogenic differentiation of MC3T3 cells on different samples was evaluated by the determination of mRNA expression levels of several osteogenesis-related gene markers. As shown in Figure 10, the RUNX2, OCN, and Col-I mRNA expression levels of the cells on each sample sustainably increase with an incubation time of up to 14 days. At each time point, the expressions of mRNA of each marker in the cells on the SPEEK sample are higher than that on the PEEK sample. However, the GO-SPEEK sample exhibits the strongest positive effect of boosting the osteogenic differentiation of cells.

RUNX2 is a transcription factor necessary for early osteoblast differentiation. OCN, responsible for calcium ion binding, is a late-stage maker of osteoblast differentiation, and Col-I is the most abundant bone matrix protein (Huang et al., 2021). In this work, GO-SPEEK remarkably enhances levels of RUNX2 mRNA compared to SPEEK and PEEK (Figure 10A), indicating that GO-SPEEK can accelerate the differentiation by promoting a mature phenotype at earlier time points (Kaur et al., 2024). The upregulation of RUNX2 mRNA expression was reported to facilitate the mRNA expression of OCN and Col-I (Huang et al., 2019), thus leading to more pronounced OCN and Col-I proteins to secrete into the ECM and thereby accelerating ECM mineralization. The ECM mineralization results in our investigation confirmed the above point. As seen in Figure 11A, it is obvious that much denser ECM mineralization is deposited on the GO-SPEEK surface than on the PEEK and SPEEK surfaces (especially the PEEK surface) at each time point. In addition, at 14 days, much denser stained deposits are observed on the SPEEK and GO-SPEEK samples than at 7 days. The quantitatively determined ECM mineralization results are in accordance with the above staining pictures, as shown in Figure 11B. It is indicated that both SPEEK and GO-SPEEK reveal promote ECM mineralization compared to PEEK, and the mineralization level was much higher on the GO-SPEEK surface than on the SPEEK surface at each incubation time.

In this study, we built a 3D porous structure on PEEK through sulfonation and found the modified surface promotes the adhesion and osteogenic functions of MC3T3 cells. Additionally, GO grafted on the sulfonated PEEK surface was proven to further improve the cell–material interactions. The generated microporous structure of the implant surface has been reported to be favorable for cell adhesion and osteogenic functions (Lee et al., 2004; Liu et al., 2021). Cai et al. (2020) used a femtosecond laser to construct a porous surface on a PEEK surface, which was also proved to significantly promote the adhesion and proliferation of rat bone marrow mesenchymal stem cells. Apart from the microporous structure, GO could also provide an affirmative effect to regulate cell behavior. Su et al. (2020) combined GO with polydopamine and loaded it on the surface of a titanium alloy, which greatly enhanced the osteogenic mineralization of hMSCs. The previous study disclosed that the combination of π–π stacking, electrostatic forces, and hydrogen bonding of the GO surface allows it to take up osteoinductive factors from the medium via non-covalent binding, thus enhancing osteogenesis (Lee et al., 2011). Some studies have proposed that the GO can activate osteogenesis-related signaling pathways and promote bone formation (Wu et al., 2015; Zhang et al., 2016; Wu et al., 2018). Moreover, it is notable that, in this work, both sulfonation and GO grafting are beneficial for the enhancement of PEEK hydrophilicity (as shown in Figure 4D). The hydrophilic surfaces were confirmed to be conducive to regulating the osteoblastic response (Sun et al., 2017; Huang et al., 2019; Kopac, 2021; Tseng et al., 2021). Hence, the improvement of the hydrophilicity of PEEK after surface modification is also responsible for the ameliorated cell-material interactions (including cell adhesion, proliferation, osteodifferentiation, and ECM mineralization) in our investigation.

The mechanical properties of the samples evaluated by using the tensile experiment and the typical tensile curve, as well as the acquired data, are shown in Figure 12. PEEK, SPEEK, and GO-SPEEK all have tensile yield zones, indicating that the samples are generally ductile materials. The tensile stress versus tensile strain curves of different samples substantially look analogous (Figure 12A), suggesting that little variation of the mechanical properties of PEEK occurs after the surface modification. Figure 12B compares the tensile strength and the elastic modulus of different samples, and the data are summarized in Table 2. SPEEK and GO-SPEEK have similar values of yield/break tensile strength and elastic modulus compared to the original PEEK because sulfonation and the subsequent GO-grafting treatment were only applied to the surface of the samples; the mechanical properties of the bulk material were maintained. Although the measured elastic modulus of GO-SPEEK (1.55 GPa) is slightly less than PEEK (1.69 GPa), which might be due to the formed micropores on the surface, the value is still much larger than that of the 1.33 GPa of the cancellous bone (Clausing et al., 2023). With the maintained mechanical properties, the enhanced surface biomineralization capacity, and cell-material interactions, as well as the highly efficient bacterial killing activity, GO-SPEEK is expected to be extensively used in the bone replacement and orthopedic field.