**4. Discussion**

RFMS was successfully utilised to modify the surfaces of PEEK polymer at low temperatures with a CaP thin film. This was done with the objective of potentially enhancing the PEEK material's surface bioactivity in preparation for exposure to physiological conditions. Through the manipulation of the sputtering parameters, namely deposition time, this work has shown that the desired CaP thin films can be achieved successfully, in a one step process, without the need for subsequent processing (e.g. thermal annealing), albeit at a CaP ratio below that expected for pure HA. To ascertain the potential of these surfaces, several different analytical techniques were used to characterise the resultant thin films chemically and physically, including ToFSIMS, XPS, and SEM. The XPS data obtained for PEEK was typical of this material. Each peak assignment for the curve fitted elements were as expected [24–26,28,29,31]. The O/C value obtained for PEEK, 14/1, was lower than that which was anticipated (16/1) [2,23,25], small deviations such as this have been attributed to the presence of hydrocarbon contamination on the PEEK surface [28]. With regard to the ToFSIMS analysis of PEEK, according to Pawson et al. [28] and Lub et al. [30], specific peaks, within the survey, at *m*/*z* of 39, 51, 77, 91, 115, 152, and 165 have been considered as representative of aromaticity, such as the structure of PEEK. The peak at *m*/*z* 165 was originally found in a polycarbonate, which is known as structurally very similar to that of PEEK, this particular peak has been recorded as being useful in determining the polymer structure as it contained no O, due to the polymer being analysed after rearrangemen<sup>t</sup> [30]. Peaks outlined by Pawson et al. [28] which were thought to be indicative of PEEK included *m*/*z* 104, 105, 139, 163, and 195–197. The peak at an *m*/*z* ratio of 39 has been treated with care as, according to literature, some of the intensity has been known to be attributed to potassium as a result of remnants left from the

polymerisation process. It was not detected via XPS here. According to Pawson et al. [28], peaks below *m*/*z* 60 within the negative survey have not been found to be useful diagnostically and intensities are low, which, from looking at other literature, was expected. Work by Henneuse-Boxus et al. [27] has detailed a number of peaks for the *m*/*z* ratios of PEEK fragments which have been found within the negative survey presented here, these included 25, 41, 49, 73, 108, 121, and 196, which were recorded as corresponding to C2H<sup>−</sup>, C2OH<sup>−</sup>, C4H<sup>−</sup>, C6H<sup>−</sup>, C6H4O2 −, C7H5O2 − and C13 H8O2 −, respectively. The peak at the *m*/*z* ratio of 197 has been assigned to the PEEK monomer. Positive ion intensity surface maps for the PEEK surfaces have included peaks present at *m*/*z* 40 and 57, which represent Ca+ and CaOH<sup>+</sup>, respectively. Low levels of both ion fragments, in comparison to a modified surface, were noted as expected due to this surface being essentially pure PEEK (PEEK Optima LT1 grade). This intensity was likely contributed to by a fragment other than Ca+ or CaOH<sup>+</sup>. Fragments present at *m*/*z* 104, 139 and 163 are known to be indicative and diagnostic of PEEK, these were present in higher intensities than those of the CaP modified surface, this was expected. These ions have been commonly found in low intensities [28]. The ToFSIMS negative ion intensity surface maps for PEEK surfaces at the selected masses exhibited low levels of PO3 − and PO2 −, in comparison to a CaP-modified surface, as expected due to the fact that this particular surface has had no exposure to any modification procedure. The surface maps for *m*/*z* 108, 121, and 196 have been associated with aromatic polymers or fragments of PEEK [27], it was expected that these would be present in a higher intensity than they were found, in comparison to a modified surface. C*n*H*m* clusters were observed in much higher intensities than the peaks chosen ( *m*/*z* 108, 121, and 196) and are known as attributable to carbonaceous species, these have been thought of as generally indicative of a polymeric material; however, these were considered too generic to utilise on this occasion as, according to literature, they have not been considered diagnostic of PEEK [28]. SEM micrographs of PEEK modified with CaP via RFMS for various deposition times have been shown in Figure 1. The morphology of the surface of the polymer was characterised for time periods up to 600 min, during which time the polymer surface morphology developed, from the early time point of 10 min pitting of the surface could be clearly observed, whereas a more intricate lattice-like or network-type microstructure was exhibited at later time points. It was thought that this development was a consequence of the eventual coalescence of the pits to establish porosity via the nucleation and growth of the surface structures. The size of the pits, according to the area and Feret's diameter calculations, has been shown to rise as time increased, until 450 min, when the dimensions started to decrease. This, along with the knowledge that the number of pores declined as time deposition increased, indicated that an infilling of the surface morphology has occurred after 450 min. The nature of the porous microstructure exhibited at latter deposition times was similar in nature to previous work [32–35] on polymers. As such, the surface of the PEEK substrates obviously undergo etching at the start of the sputtering process, with Ca and P species then becoming embedded into the PEEK and the resultant CaP coating nucleating from these embedded species, resulting in a hybrid intermixed zone between the coating and the substrate. Evidence for this hybrid intermixed CaP coating and PEEK substrate zone is shown by the cross-sectional depth profile illustrated in Figure 16, where there is clear overlap between the PO3 − and C4H− ions species (representative of the CaP coating and the PEEK substrate, respectively). It is suggested that, as the Ca and P species do not embed homogeneously over the entire surface during sputtering, this results in the formation of the lattice-like network as opposed to a normal dense, non-porous and continuous coating as normally observed with such coatings on metallic substrates [20].

In order to investigate the nature of the CaP thin films once they had been deposited onto the PEEK substrate, for depositions times 10, 60, 300, and 600 min, ToFSIMS was employed including the survey, surface mapping and depth profiling functionalities. XPS was used to investigate the chemistry at the time points (10, 60, 300, and 600 min). The positive ToFSIMS surveys were displayed in Figures 11a and 12a for the samples HA10 and HA600, respectively. Each of the spectra displayed the characteristic peaks expected for a ToFSIMS analysis of a CaP modified surface [9,19,36]. In each spectrum, Ca+ species ( *m*/*z* 40 and 57) as well as the isotopes ( *m*/*z* 42 and 44) were present. Various peaks within the positive spectrum considered to be representative of polymers with aromatic groups were present (*m*/*z* 39, 51, 77, and 91) along with peaks thought indicative of PEEK (*m*/*z* 104 and 105). The negative ToFSIMS spectra for both the HA10 AND HA600 samples (Figures 11b and 12b respectively), revealed the dominance of the key ion, P (*m*/*z* 63 and 79) for both time points. ToFSIMS results revealed that, as the length of time that the CaP material was sputter deposited onto the surface of the PEEK substrate increased, the relative intensity of the PEEK fragments diminished in both the positive and negative spectra, whilst the relative Ca and P intensities increased as shown in Figure 14. These results were consistent with the XPS results. The WESS and high-resolution spectra for XPS for the HA10 and HA600 samples (shown in Figures 5–8) exhibited peaks which were characteristic of CaP materials and were as expected [9,12,37]. It was noted from these results that the Ca/P was very low for all the samples and well below the expected 1.67. Previous work by Surmenev et al., showed that the Ca/P ratio was much higher when HA was sputter deposited onto polymeric materials (PTFE), and that it was difficult to form CaP coatings on such substrates [38]. The results here are contradictory to those previous findings and are most likely due the build-up of negative charge on the surface of the insulating PEEK substrate. This prevents re-sputtering of P by negatively charged O, resulting in the low Ca/P observed here [39]. ToFSIMS surface mapping for both the PO2− and PO3− and Ca ions has shown a similar trend. Incidentally, it was also reported by various authors that both Ca and P travel from the target to the substrate as neutral species [38] depositing to form CaP thin films [40–43]. When analysing the ToFSIMS spectra, it was noted that PO2− and PO3− ions were dominant ions in the negative ToFSIMS survey, and with increasing deposition time their presence increased, correlating well with the increase in both O and P highlighted in the corresponding XPS. This information, along with the decrease in the C % AC, with time, caused the O/C ratio to increase. High levels of C present on the CaP–modified PEEK surfaces have been reported in literature [9]. It was considered that this level of C may have been due to 'adventitious C contamination' because of the adsorption of impurity hydrocarbons, which corroborated the ToFSIMS results. ToFSIMS analysis indicated that, in both the positive and negative surveys, the peaks thought to be representative of hydrocarbon clusters (C*n*H*m*) had high intensities, (*m*/*z* 29, 49, and 73); however, as deposition time increased, the intensity of these peaks depreciated. It was also considered that some of the intensity of the C 1*s* high resolution XPS scan may have been attributed to C–C/C–H bonds. It was postulated from analysing XPS and ToFSIMS data that the decrease in C intensity was attributable to the sputter deposited thin film growing with time and, therefore, fewer PEEK C–C/C–H bonds were being detected, as the detection depth limits of the XPS system was limited to ~5–10 nm. After 600 min of sputter deposition it was realised that a number of the peaks defined within literature as being specific and diagnostic of PEEK, *m*/*z* 104 and 105, were still being detected by ToFSIMS (detection limit of 1–2 nm); however, only the 105 peak was considered significant (>0.4% of largest peak) [9]. It was suggested that this peak may have been due to the presence of another organic species, as corroborated by XPS, and the porous nature of the surfaces produced; however, this does merit further investigation. The shakeup peak often affiliated with aromatic ring structures present in PEEK have been exhibited in the C1s high resolution survey scans for 10 and 60 min but have disappeared for the 300 and 600 min modifications. The XPS no longer detected the underlying polymer substrate from 300 min, which suggested that the thin film deposition had grown so that the polymer was out of the detection limits of the XPS instrument. ToFSIMS surface mapping showed a similar trend, whereby all six of the fragments for PEEK, in the positive and negative spectra, were reduced in comparison to the neat PEEK sample, while the Ca, PO2− and PO3− ions had increased intensities. The ratio of each bond within the C 1*s* scan to the total % AC of C was similar to that expected for PEEK for the 10 and 60 min deposition sample with the addition of the CO3<sup>2</sup>− bonds having possibly stemmed from the modification of the substrate with amorphous HA as outlined by Surmenev et al. [44]. There was strong evidence that suggested that there was a formation of a hybrid interaction layer between the PEEK and CaP material. In order to investigate the chemical nature of the interaction, ToFSIMS depth analysis was carried out, revealing that the P-related ions are readily detected on the surface of the modified PEEK substrate to a depth of ~ 0.65 μm, whereas the Ca ion signal was detected for much longer and appeared to tail o ff. This data indicated that sputtered species were possibly embedded into the polymer material up to a depth of ~1.21 μm, as determined by Zeta analysis. It was suggested that, once the Ca embedded into the polymer, P proceeds to grow on this Ca rich layer, with all species known to be eligible for re-sputtering in the dynamic RFMS environment. Within the known literature [30,45] several potential hypotheses are provided as to the reason for intensity spikes of each ion of interest. It was thought that, in the case of this work, the most likely cause was matrix ionisation e ffects that have been associated with surface pollutants, such as an oxidation layer. The evidence to substantiate this event comes from the fact that oxidation was thought to coincide with the small spike in intensity for signals at the beginning of each depth profile [46]. Another potential hypothesis that has been recorded was that the full depth profile regime has been met after a few seconds of sputtering time. The interface between the CaP film and the PEEK substrate has been found to be denoted by a slow decay of both the Ca2+ and PO3 − signals and the gradual appearance of the PEEK signals, which indicated the presence of a hybrid layer where the Ca ions were embedded into the PEEK material. The literature [47] has outlined the sequence of events leading to the formation of a CaP thin film via RFMS. At the initial stages of deposition an enrichment of Ca at the substrate surface is known to take place, thought to be due to the re-sputtering of P ions, leading to a Ca/P higher than stoichiometric HA. This work has not highlighted this trend as the surface at 10 min (the HA10 sample) has been shown by XPS analysis to have already become P rich. It was thought to be the case that the investigation would have needed to have been carried out much earlier to replicate the Ca enrichment trend, as within the work completed by <sup>L</sup>ópez et al. [48]. Evidence to conclude that Ca was indeed embedded into the PEEK material intensified as it was realised that there was less % AC of Ca-related material on the polymer surface in comparison to the thin films sputtered onto Ti, under the same conditions in previous work by the authors [3,7,20]. It is suggested that this may have been due to the heavier nature of Ca, in comparison to the P ion during bombardment, or the softer nature of the polymer substrate. This would account for the P rich surface, the lower than expected Ca/P ratio, as well as the lower than expected Ca intensity in the ToFSIMS survey scans at all time points.

Ideally, here the aim was to produce CaP coatings onto PEEK to provide a means to enhance their bioactivity. Key to this was to produce coatings that have properties commensurate with those properties outlined in the ISO 13,779 2 (2018) and ASTM F1609 standards, ideally mimicking the properties of HA, with a Ca/P ratio of around 1.67. It is clear from the results produced here that this has not been achieved as the reported Ca/P ratios are well below 1.0. The next phase of any work here would be to optimise the sputtering process to achieve enhanced coating properties to align with the ISO and ASTM standards. Aspects of substrate biasing, the process gas pressure, and the process gas used would all be important in order to achieve this. It would also be important to undertake mechanical testing of the coatings, understanding the coating thickness better, and importantly understanding the dissolution behaviour of these surfaces before progressing to more involved in vitro testing of these surfaces. If the appropriate chemical, physical, mechanical, and in vitro properties were achieved, these coatings could provide a basis for enhancing the bioactivity of PEEK materials for use in orthopaedics in order to improve bone apposition, both in terms of how quickly this can be achieved and the enhancement of the bond with bone when compared to pure PEEK devices.
