1. Introduction
There is increasing interest in the use of Polyetheretherketone (PEEK) in orthopaedic implant devices due to its excellent biocompatibility, its radiolucency, chemical resistance, sterilizability, ability to be easily processed, and its favourable mechanical properties (in comparison to human cortical bone) [
1,
2]. To date it has found applications in spinal fusions cages, dental implants, and maxillofacial reconstruction [
1,
3,
4]. However, despite its obvious potential in load bearing orthopaedics and reconstructive surgery, a major clinical concern is that PEEK is bioinert and that it will not provide a suitable interface for driving successful osseointegration, in vivo [
2,
5].
Various approaches have been suggested to enhance the surface bioactivity of PEEK, including surface modification via chemical or plasma treatment [
6], coating of the surface via plasma spraying [
7], sputtering [
8,
9], or by the direct fabrication of a composite material containing bioactive agents via injection moulding [
10] or direct 3D printing [
11,
12]. A range of different bioactive agents have been studied as additives for PEEK based composites, namely the likes of hydroxyapatite [HA—Ca
10(PO
4)
6(OH)
2], [
13] strontium (Sr) substituted apatites [
14], fluoro-hydroxyapatite [
15], β-tricalcium phosphate [β-TCP Ca
3(PO
4)
2] [
16], calcium silicate [
17,
18], bioglass [
19], and titanium dioxide [TiO
2] [
20]. Although these different additives all offer obvious advantages when added to the PEEK matrix to create a composite material, the use of these materials could potentially alter bone homeostasis and depending on their concentration and scale (micro versus nanoparticles) they could prove toxic to osteoblasts and this needs to be considered when designing the composite material of choice [
13,
14,
15,
16,
17,
18,
19,
20].
The development of bioactive PEEK-based composite materials provides several obvious advantages over other approaches, most notably that composite materials can have tunable mechanical properties for specific applications along with the added benefit of enhanced bioactivity on its surface, enhancing its osseointegration. A range of direct processing techniques have been utilized to deliver such composite materials, which includes compounding and injection moulding [
18], extrusion free forming in combination with compression moulding [
21], selective laser sintering [
22], cold press sintering [
23], hot pressing [
24], and electrostatic bonding [
25]. Of these different approaches, injection moulding is still one of the most used techniques [
2]. The main shortcoming of injection moulding, and the other techniques highlighted here for manufacturing PEEK and PEEK based composites with bioactive agents is the lack of flexibility for making parts with complicated geometry and the significant waste that can be generated [
10]. One way to get around these inherent issues is to develop an additive manufacturing approach to creating composite materials. To date there are a significant number of reports in the literature highlighting how 3D printing can be used to develop pure PEEK and PEEK-based composites implant devices using either Selective Laser Sintering (SLS) [
22,
26,
27] or Fused Filament Fabrication (FFF) [
5,
11,
12] However, the SLS approach comes with a high cost, requires significant safety measures and results in significant waste of the feedstock powder, which cannot be re-used for implant preparation thereafter due to the potential for contamination. In comparison, FFF utilises a continuous filament during 3D printing which produces minimal waste and is accessible due to its relatively low cost. Furthermore, FFF printers can be easily and cheaply modified and upgraded to develop printers better suited for an intended application area. FFF does however have significant shortcomings, such as poor mechanical properties in the resultant 3D printed structures in the Z direction. PEEK has a very high melting temperature (343 °C) when compared to other biomedical polymers that can be 3D printed such as Polylactic Acid (typically between 130–180 °C). This means that when either 3D printing PEEK or a PEEK-based composite the chamber, the print bed and hot-end temperatures need to be high, typically well beyond those available in normal commercial FFF printers [
28,
29,
30,
31]. As such, obtaining the necessary properties in the 3D printed PEEK or PEEK-based composites will depend upon delivering the appropriate processing parameters during printing. Previous novel work, believed to be one of the first study of its kind, presented the properties of PEEK and hydroxyapatite (HA) composites manufactured by directly 3D printing (using FFF) from a composite PEEK/HA composite (between 0–30 wt% HA) with respect to their thermal properties, mechanical properties, surface morphology and crystallinity, and highlighted how the properties of the PEEK/HA composites could be produced to be in line with human cortical bone [
11]. Several other reports have also been published showing how bioactive PEEK/apatite composites can be manufactured using FFF 3D, albeit using significantly different processing parameters and 3D printers [
5,
12]. Despite this, the published works is still very limited in this area.
One deficiency in the previous studies was the lack of surface analyses undertaken on the 3D printed PEEK/HA composites, which is a critical factor when considering such materials for use in implant devices in vivo. In this novel paper, we would like to address this issue and highlight further the significant developments in the direct 3D printing of PEEK/HA composites using an FFF approach. The key aim of the work was to prove that FFF 3D printing could deliver PEEK/HA composites with controllable concentrations of HA on the surface of the 3D printed structures without the need for any further processing steps to expose the bioactive HA materials (in essence a one-step process to producing bioactive PEEK/HA composites). To the best of our knowledge, this is the first time this has been reported in the literature using advanced surface characterisation techniques as detailed below. In this work we report on the direct 3D printing of the extruded PEEK/HA composite filaments via an FFF approach using a custom modified commercial printer Ultimaker 2+ (UM2+). The 3D printer was modified to operate at higher temperatures (as detailed in the Materials and Methods Section), allowing the properties of the printed PEEK bodies to be customised accordingly, namely the uppermost surface in this case. The 3D printed specimens were then subject to extensive surface characterisation regime via physical, and chemical techniques (as detailed in the Materials and Methods Section), along with an in vitro study (using U-2 OS osteoblast-like bone cells), to ascertain the potential for FFF to be a go-to manufacturing technique to produce PEEK/HA composites for orthopaedic implant devices where direct bone apposition between the implant surface and human bone is crucial for their long-term success.
4. Discussion
The core aim of the work undertaken in this study was to provide a way to regulate concentrations of bioactive HA materials directly on the surface of FFF 3D printed PEEK/HA composite structures in a one step process and to investigate their different chemical, physical, and in vitro properties. Our previous work studied the crystallinity, morphology, bulk properties, and mechanical properties of the same materials and found that the approach taken here could be used to successfully manufacture PEEK/HA composites up to 30 wt% HA [
11]. This worked highlighted that the materials were printable, and importantly could be delivered with mechanical properties that matched those of human cortical bone [
11]. Other studies have utilised FFF 3D printing of PEEK/HA composites; however, they only investigated the use of up to 10 wt% PEEK/HA and did not achieve the same mechanical properties as those presented here [
5,
12]. As such, the results presented in this next study deliver the next logical step required in this work, which, to our knowledge, demonstrates the first study of the surface properties of novel FFF 3D printed PEEK/HA composites using advanced surface characterisation techniques such as XPS and ToFSIMS in combination with standard techniques such as XRD, FTIR, and SEM/EDX.
The FTIR results (taken using ATR), shown in
Figure 2a highlight that the 3D printed PEEK sample is pure PEEK, with no peaks present from HA and all the peaks with the expected relative intensities for a pure PEEK sample [
18]. With the addition of HA into the PEEK matrix, as shown in
Figure 2a–d, for the 5HA–30HA samples, respectively, the presence of the PO
43− (υ
3) band (between 1200–900 cm
−1) is evidence that HA is present in the top 1–2 µm of the surface of the PEEK/HA 3D printed composites (in line with the analysis depth of the ATR techniques utilised here) [
32]. As the concentration of HA increases in the composites, the relative intensity of this PO
43− band increases significantly with respect to the normalised intensities of the PEEK, indicating a greater prevalence of HA at the surface of these samples as the HA content is increased. This is further corroborated by the XRD results, shown in
Figure 3. The pure PEEK sample (0HA) in
Figure 3a shows the PEEK to contain no additional phases with all the major reported peaks (highlighted by dashed lines for the (110), (111), (200), and (211) diffraction planes) present at the correct 2θ positions and at the correct relative intensities, as would be expected for pure PEEK [
11,
18]. After the addition of HA into the PEEK matrix, as shown in
Figure 3b–e, for the 5HA–30HA samples, respectively, clear diffractions peaks can be observed for HA (as highlighted in
Figure 3), with the peak positions and relative intensities all in line with those expected for HA as outlined in the ICDD file# 09-0432. In addition, the relative peak intensities for the HA material are seen to increase relative to the PEEK as the concentration of HA increases in the samples as shown in
Figure 3b–e for the 5HA–30HA samples, respectively. It is also apparent that the diffraction peaks for PEEK or HA show no significant change in their peak width as the concentration of HA increases in the composite materials. The XRD and FTIR results highlight that no new additional phases are present in these 3D printed PEEK/HA composite samples under the conditions employed here [
10,
25].
The combination here of the high ambient temperature within the print chamber (230 °C), the nozzle temperature (400 °C), and the print-bed temperature (280 °C) are all important to deliver semi-crystalline PEEK/HA composites with high levels of crystallinity as reported previously in our work (44.59–49.91%) [
11]. These results are higher than the current reported values in the literature [
39]. Previous work by Wu et al. and Zanjanijam et al. showed that as the ambient temperature of the printing chamber is increased, the crystallinity of the polymer matrix increases [
40,
41]. In such conditions, there are no issues relating to rapid cooling of the printed sample and the polymer chains have adequate time to crystallise, enhancing the overall crystallinity of the samples, as observed here. The nozzle temperature is also very important here for enhancing crystallinity, and that in combination with the higher ambient chamber temperature prevents rapid cooling of the printed specimens, which in turn prevents warping. None of the samples in this study were seen to warp and this is a direct consequence of the high temperatures employed. The print bed temperature is also elevated in this work, well beyond the values reported in previously related experiments by others, although this parameter has received a lot less attention in the literature than the nozzle temperature and the ambient temperature [
29,
40]. This also helps to prevent warping as it slows down the cooling rate of the polymer when printed. In combination, the temperature of the print nozzle, the print bed, and the print chamber are all critical parameters that collectively ensure the best quality prints without warping and enhanced crystallinity across all the samples. They also act to ensure minimal voids in the printed sample and consistency in the printed layers, as shown in the SEM cross-sections in
Supplementary Figure S1. This is a consequence of the elevated temperatures of the print bed and the print chamber, allowing more time for molecular diffusion to occur which results in less surface voids being created. [
40,
42] However, some small voids are observed in the composite 3D printed samples and suggests that adhesion between the layers is affected by the addition of HA into the PEEK matrix. Consequently, the enhanced crystallinity also delivers enhanced mechanical performance in such 3D printed samples, although that is not the focus of the work in this study. Other parameters that can influence the quality of the print are the nozzle diameter and the feed rate of the filament [
11,
40]. In this study both the nozzle diameter (1.0 mm) and the feed-rate (40 mm/s) are much higher than previously observed in a range of different studies utilising FFF to 3D print PEEK. Typically, a nozzle diameter of 0.4 mm has been utilised [
40,
41], although several studies have utilised wider diameter nozzles [
40]. For this work, a higher diameter nozzle was employed to prevent clogging of the nozzle with the PEEK/HA composites and to minimise void space in the printed samples. In addition, the printing speed was on the higher end of those seen for FFF 3D printing of PEEK in other studies, although it is comparable to that utilised by Opaldo et al. to 3D print PEEK/HA composites of up to 10%
wt/
wt HA in PEEK [
5]. However, it should be noted here that each of the different studies utilise different 3D printers, often custom built, each with unique printing capabilities, which sometimes limits the processing conditions that can be employed.
The associated SEM results here also point to increasing levels of HA on the surface of the composite samples, as can be seen from the images in
Figure 3. No HA particles can be observed for the pure PEEK (0HA) sample, as shown in
Figure 3a(i,ii), but the 5HA through to the 30HA samples all show the presence of HA particles of up to 5µm in diameter. The particles are evenly distributed across the surface of the samples and seem to be more prevalent as the concentration of HA increases in the samples. The FTIR, XRD, and SEM results all show that the PEEK/HA samples have HA homogeneously distributed across the surface of the 3D printed parts and as such present bioactive HA materials that should enhance osseointegration if implanted, in vivo. To confirm that the HA particles are on the uppermost surface of the 3D printed samples, and not just on the sub-surface where they could be detected by the likes of XRD and FTIR, surface sensitive techniques such as XPS and ToFSIMS analyses were also employed in this study. XPS analysis of the different samples clearly showed her that Ca and P species were certainly detectable in the top 10nm of the PEEK/HA composite samples, as highlighted in
Figure 5 and
Figure 7, with an obvious enhancement of the Ca2p and P2p peak intensities (illustrated in
Figure 7c,d, respectively), indicating a greater availability of HA on the uppermost surface of the samples as the HA content in the PEEK/HA composites increases from 0–30 wt%. It is also interesting to note that the Ca/P ratio detected for these samples appears to decrease significantly as the HA content in the composites increases. The reported Ca/P ratios (as highlighted in
Table 3) decrease from 2.78 ± 0.87 (5HA) to 1.59 ± 0.09 (30HA). The O/C ratio also increases here (from 0.11 ± 0.00 (5HA) to 0.17 ± 0.00 (30HA)) at the same time. The increase in the O/C ratio is likely to be a consequence of the increasing number of associated PO
43− groups, which correlates with the corresponding decreasing Ca/P of the same samples. This can be confirmed as the relative intensity of the lower binding energy O1s peak (around 531.6–531.9 eV) is seen to increase relative to the higher binding energy O1s peak (around 533.6–533.8 eV) in all the PEEK/HA 3D printed samples, as can be observed in
Figure 6b and
Figure 7b for the 0HA and 30HA samples, respectively. No Ca or P contamination were detected on the surface of the 0HA sample (pure PEEK), and no other impurities were detected using XPS (such as Na), at least within the detection limits of this technique (~0.01 atomic concentration %). It is important to note that ToFSIMS analysis of the same surfaces showed small amounts of contamination across all the 3D printed PEEK and PEEK/HA composites, with the likes of Na, F and K observed, along with a range of organic species that would not be associated with PEEK, namely C
2H
3+, C
2H
4+, and (C
2H
5+) as typical examples. These can clearly be observed across all the positive ion ToFSIMS spectra in
Figure 8a 0HA and
Figure 8b 30HA, Specific peaks with a high intensity included
m/z 39 and 51, are indicative of either aromaticity or ionically diagnostic of PEEK/PEEK fragments by Pawson et al. [
37] and are clearly shown across all the 3D printed samples. Ca
+, CaH
+ and CaOH
+ peaks can be observed at
m/z of 40, 41, and 57 on all samples. However, when the normalised peak intensities for all the samples are calculated, as shown in
Figure 9, the concentrations of these Ca species are negligible in the 0HA (pure PEEK sample) when compared to the PEEK/HA 3D printed composites. Higher concentrations of Ca
+, CaH
+ species are observed in all the PEEK/HA composites when compared to CaOH+, with the 20HA sample reporting the highest concentrations of Ca
+ and CaH
+ species. It would have been expected here that these concentrations would have been higher in the 30HA sample; however, small variations in the concentration of the HA in the extruded filament or slight variations in the 3D printing environment could play a role in this anomaly and need to be further investigated. Despite this, the distribution of Ca species (Ca
+, CaH
+ and CaOH
+) are all observed to be homogenous across the entire area of the samples analysed. This indicates that the Ca species in HA is available in the uppermost surface regions of these samples (1–2 nm), which is the desired outcome. Previous studies on PEEK/HA composites manufactured using extrusion free forming [
1] and compounding and injection moulding [
13] have shown that such materials are more favourable for implantation than pure PEEK materials, due to the addition of bioactive HA.
In vitro testing of the different 3D printed composites revealed that all the samples support the attachment, growth, and viability of U-2 OS osteoblast-like cells for at least 7 days. The 30HA 3D printed sample had the highest cellular metabolism at 7 days when compared to the pure PEEK samples (0HA), which may be a consequence of the higher HA content in this sample, as highlighted in
Figure 12. However, it should be noted that MTT records cellular metabolism and not cellular viability or proliferation directly, therefore, it cannot be ruled out that the cellular metabolism observed for the 5HA–20HA samples here is a consequence of touch contact inhibition [
43]. The cell shape (morphology) is highly indicative of the osteoblast-like cell behaviour in relation to both adhesion and viability, as it should be observed that a cell that has a positive interaction with suitable surface properties will show signs of cell spreading on the surface. The corresponding SEM images of the U-2 OS cells at 7 days, as shown in
Figure 14, highlights that cell are adhering and spreading on all the samples; however, the 0HA, 5HA, and 30HA samples have more visibly rounded cells than the 10HA and 20HA samples. This suggests that the cells have not fully interacted with their surroundings, and full cell attachment has not occurred yet at this time-point. Filopodia are seen to be protruding from the cells on all the 3D printed samples (as highlighted in
Figure 14), indicating that the cells have begun to probe the underlying substrate for topographical features to further guide their attachment. For the 10HA and 20HA samples, the U-2 OS cell exhibit a definite flattened morphology when compared to the 0HA, 5HA, and 30HA samples and appeared to adhere securely to the substrate surface with spread out elongated filipodia. However, for the 30HA sample there did appear to be a high proportion of the U-2 OS cells with a rounded cell morphology when compared to the 10HA and 20HA samples. The 30HA sample does show a significant proportion of flattened cells as well. Cell viability and proliferation testing using the PicoGreen™ assay, shown in
Figure 13 highlighted that cells adhered well to all the 3D printed samples at 7 days and corroborates the results observed in the MTT and cell morphology tests at the same timepoint. In particular, the 5HA and 10HA samples showing higher DNA concentrations than pure peek (0HA) and are therefore exhibiting enhanced U-2 OS viability and proliferative capacity. It has been suggested that with an increasing HA content in the 20 HA and 30HA samples that there may elevated levels of free Ca
2+ in the surrounding cell culture medium due to dissolution of the HA, therefore this could inhibit cellular proliferation [
10]. This may explain the slightly lower DNA concentrations detected here via the PicoGreen™ assay. The lower proliferative capacity may also indicate the onset of osteoblast differentiation in the 20HA and 30HA at this early timepoint. There appears to be no correlation here between the surface roughness, surface morphology or contact angle with respect to the MTT, PicoGreen™ assays, or cell morphology results. This could have been a consequence of the temperatures of the colling composite material, with small differences perhaps introducing subtle changes to the surface chemistry and morphology and influencing the results. It is obvious that the processing conditions here, namely the control of the temperature may need studies more accurately and possibly in real-time to provide a means of altering the printing conditions in situ and removing the possibility of inconsistency in the surface properties of the samples produced. It has been reported that the surface roughness does decrease with increasing nozzle temperatures in 3D printing [
42]. The contact angle for the 0HA is low at 43.56 ± 4.11°, lower than might be expected for pure PEEK as determined in other studies [
4,
10,
44]. Further to this the water contact angle is seen to increase significantly for all the 3D printed PEEK/HA composites, with values in the range 77.40 ± 10.03° − 93.08 ± 5.44°, highlighting a tendency towards less hydrophilic surfaces, with the 30HA being borderline hydrophobic [
30]. It was shown that the contact angle can reduce significantly when agents are added to the PEEK matrix to manufacture composites [
10,
13,
44,
45]. However, there are a range of conflicting results reported in the literature with respect to the water contact angle of PEEK/HA bio-composites. Some studies have highlighted a slight increase in the contact angle after the addition of HA to PEEK/HA biocomposites, whilst others report a decrease in the water contact angle with the addition of HA particles to PEEK [
10,
44,
45,
46,
47]. Whereas it would be expected that the water contact would increase with the addition of HA particles on the surface of the sample, localised surface properties and the surface topography can heavily influence this [
46]. Typically, the adhesion and proliferation of osteoblasts has been associated with good wettability (hydrophilicity), with the osteoblast cells exhibiting strong preference for hydrophilic surfaces [
44]. All the samples here were hydrophilic in nature, with only the 30HA sample showing borderline hydrophobicity. Notwithstanding this in the results reported here, the sample surfaces are quite rough, and this may have had an influence on the measured contact angles and merits more detailed examination. Porosity is also known to influence the osteoblast response, more so than any associated surface roughness, as highlighted by Spece et al. [
48]. Despite these findings, it could be suggested that the HA content on the surface of the samples has the most significant influence on the in vitro behaviour observed in this study, which has been previously suggested for PEEK/HA biocomposites [
1,
2,
3,
5,
13]. For all samples, the results do indicate healthy cell functioning and a positive cell-surface response between the U-2 OS cells and the 3D printed samples up to 7 days, and further in vitro testing is required to discriminate between the potential bioactivity of these different FFF 3D printed PEEK/HA composites. These will include the consideration of the initial cellular adhesion events and in-depth consideration of cellular differentiation to determine their suitability going forward for orthopaedics. These results highlight that 3D printing is a useful tool for printing PEEK and HA composites, and that the printing technology is developing to enable the manufacture of customised medical devices in the future [
49,
50,
51].