Developing alkaline titanate surfaces for medical applications

Improving the surface of medical implants by plasma spraying of a hydroxyapatite coating can be of critical importance to their longevity and the patient’s convalescence. However, residual stresses, cracking, undesired crystallisation and delamination of the coating compromise the implants lifetime. A promising alternative surface application is an alkali-chemical treatment to generate bioactive surfaces, such as sodium and calcium titanate and their derivatives. Such surfaces obviate the need for high temperatures and resulting micro-crack formation and potentially improve the bioactive and bone integration properties through their nanoporous structures. Also, metallic ions such as silver, gallium and copper can be substituted into the titanate structure with the potential to reduce or eliminate the infections. This review examines the formation and mechanisms of bioactive/antibacterial alkaline titanate surfaces, their successes and limitations, and explores the future development of implant interfaces via multifunctional titanate surfaces on Ti-based alloys and on alternative substrate materials.


Introduction
The number of people in the world aged 65 years or over is projected to more than double in 2050 from its 2015 value. The global population of this age range currently is approaching 1.5 billion [1]. According to the National Joint Registry (NJR), the prevalence of hip replacements in the United Kingdom has continually increased, with 101,384 primary hip replacements out of over 109,000 total hip replacement procedures (including revisions) being conducted in 2019 [2]. Currently, failure of hip implants occurs predominantly through aseptic loosening, accounting for 40% of single-stage revisions in 2018, with infection affecting 6% [2]. Despite current outlooks showing greater hip survivorship than previously thought; 57.9% of hip replacements last 25 years according to Evans et al. [3], further advances are still required to ensure improved quality of life for an ever-aging population while mitigating the need for subsequent revision surgeries.
Implant surfaces are critical to in vivo devices. Controlling the properties of these surfaces to achieve appropriate extracellular environments, therefore enhancing the implant-tissue bond, especially in the case of total joint replacements to create strong and long-lasting natural fixation, is a demanding and necessary paradigm within orthopaedic research.
This extends further when considering additional bone-related research, such as with dental implants. Due to the complexity of the implanted environment, each implant must be specifically designed for the intended implant site. Regarding dental implants [4,5], a trans-mucosal component is required to penetrate the soft tissue, such as abutment devices, which lies between the anchoring implant in the bone (usually a screw), and the functional component of the dental prosthesis (i.e. the crown) [4]. Therefore, carefully designing how these different materials and structures interact physically and chemically is essential. Ultimately, the critical factors in the success of these components are the chemical, structural and morphological properties of the implanted surface.
A key research area, therefore, for such implants is the modification of surfaces to enable greater adhesion to surrounding bone tissue [6,7]. Durability and in vivo success of an implant are dependent upon osteoconductive growth around an implant through osteoblastic recruitment. Presently, the only FDA-approved method for providing hydroxyapatite (HA) structure on the surface of implants is via high-temperature (>10,000°C for the flame centre; > 1500°C for particle temperature [8]) plasma spraying delivering a thick coating of hundreds of microns. HA is a form of calcium phosphate, which mimics the main mineral component, crystal and chemical structure of cortical bone: Ca 10 (PO 4 ) 6 (OH) 2 [8]. However, despite these coatings being ideal for improving implant biocompatibility, they suffer from inherent issues such as residual stresses, for example, 20-40 MPa tensile, and cracking resulting from the manufacturing temperatures used [9,10]. This ultimately results in coating spallation, and in turn, generation of aseptic loosening (sustained osteolysis and failure between the implant and bone in the absence of infection [11]) in vivo through macrophage activation and subsequent inflammatory responses [9][10][11][12][13].
To negate the limitations of implant coatings, surface modifications have been considered, such as the production of sodium titanate layers. By directly modifying the surface, the issue of coating spallation can be minimised. Kokubo et al. demonstrated that sodium titanate, generated apatite in vitro and could be synthesised at 60°C; much lower than conventional plasma spraying methods. The sodium modifier in the Ti-O framework, allowed for ion-exchange reactions to take place with Ca 2+ ions in the extracellular environment/simulated body fluid (SBF), which over time could produce the necessary apatite that can lead to bone maturation [14]. Generation of such surface layers is achieved by the immersion of titanium, typically (Cp-Ti) in 5 M NaOH solution at 60°C for 24 h. This process is then followed by heat treatments at 600°C for 1 h (ramp rate of 5°C min −1 ), in order to densify and further crystallise the resulting titanate structure [15][16][17]. Successful studies both in vitro and in vivo resulted in the implantation of NaOH-treated Ti-6Al-2Nb-Ta alloy femoral stems and acetabular cups into 10,000 patients in Japan [18]. Further work by Kizuki et al., Yamaguchi et al. and many others has demonstrated the ability for incorporating various ions into the titanate structure through solution-based ion-exchange reactions [19][20][21]. Other elements such as calcium (Ca), magnesium (Mg), silver (Ag) and strontium (Sr) have been successfully incorporated into the titanate structure, searching for improved bioactivity and antibacterial properties, as well as improved tailoring for specific applications [20][21][22]. This review will explore the impact of medical titanate structures, since the initial work by Kim et al. [23] and Kokubo et al. [24] in 1996. Older reviews on the subject [25][26][27][28], have only detailed the alkaline titanate film formation through wetchemical conversion. Here we explore the impact of alternative titanate morphologies (such as nanorods [29], nanotubes [30] and nanobelt sheets [31]), material structures (titanate glasses [32]) and material properties (piezoelectric biomedical titanates [33]) and provide a prospective on their biomedical potential. Furthermore, the antibacterial potential of these surfaces has only partially been described within a broader review of antibacterial surfaces by Spriano et al. [34]. In this review, all literature on doped titanate structures with antibacterial properties have been collated and their impact critically discussed. Therefore, an additional perspective on the efficacy of the titanate surfaces as antibacterial materials will also be presented, regarding the trade-off between bacterial death and the recovery of cellular response in the same environment. This is increasingly important as the focus on antibacterial surfaces is an ever-prevalent issue, considering the current rise in antibiotic resistance, widely considered to be the next pandemic [35], and therefore the need for alternative solutions to antibiotics [36]. In the final section, we present future directions and the latest research which explores the need to provide titanate structures in alternative forms and to expand the applicability of alkaline titanate surfaces. The latter is key to maximising the impact by delivery on alternative substrate materials (i.e. ceramics, polymers, composites), thus imparting the biomedical properties of alkaline titanates onto candidate bioinert materials [37].

Ti and its alloys
Advantages of Ti as a biomedical material Titanium (Ti) is a lustrous transition metal found in group 4, period 4 of the periodic table [38]. Despite finding applications in many fields, including aerospace and automotive industries, its prevalence in biomaterials has increased significantly over the past few decades [39][40][41]. A key feature of Ti, due to the amorphous passivated surface layer, is its bio-inertness; bioinert materials are defined as a material, by which, 'no chemical reactions occur between the implant and the tissue … [and] no direct contact with the adjacent bone tissue [is observed]' [42]. The proposal of Ti and its alloys as an alternative to the widely used 316L stainless steel (SS) and Cobalt-Chromium (Co-Cr) alloys, was due to its excellent corrosion resistance, and minimal allergenic and immunogenic potential when implanted; Ni, Co and Cr have all been quantified as being harmful to the human body [43]. Furthermore, its Young's modulus (105)(106)(107)(108)(109)(110)(111)(112)(113)(114)(115)(116)(117) GPa, however, still higher than cortical bone: 3-20 GPa [44]) results in lower amounts of stress shielding compared to 316L SS and Co-Cr-Mo alloys (Table 1) [43].
Within the Ti alloy family, Ti-6Al-4V has been particularly used in medical applications, most notably hard tissue replacements, such as orthopaedic joint arthroplasties. Despite the preferable mechanical properties compared to 316L SS and Co-Cr alloys, α (Cp-Ti) and α+β (Ti-6Al-4V) alloys still have elastic moduli greater than human bone, which can result in stress shielding effects. Interestingly, Ti-β alloys, such as Ti-Nb-Ta-Zr alloys, which have a body-centred cubic (BCC) structure, do present Young's modulii that are more favourable for orthopaedic applications [66] (Ti-Nb-Ta-Zr alloys exhibited values ca. 48-55 GPa, with 40 GPa being seen in the Ti-35Nb-4Sn system; roughly half the value of Ti-6Al-4V), due to its BCC structure [67]. However, a number of issues do persist with β alloys, particularly difficulty in homogenously melting Ti, as well as the inclusions of Nb, Ta (β-stabilisers) and Zr (Neutral elements), which can cause biocompatibility issues due to chemical macro-segregation (difficulty in forming a single equiaxed β phase), limiting their commercial usage in medical settings [66].
NiTi shape memory alloys (SMA) have also been of interest to many surgeons as a useful material for medical devices requiring in vivo movement. Not only does it possess low elastic modulus (ca. 30 and 80 GPa for its martensitic and austenitic forms, respectively), but it also exhibits superelasticity and shape memory effect (SME). Despite extensive use in various medical applications, such as orthodontic wires and stents, there are concerns of the dissolution of Ni ions, which have the potential to induce allergic, toxic or carcinogenic effects, as described by Shabalovskaya [68]. Due to failures of devices in vivo, in vitro performances being inconsistent, as well as a report of 33 failed stents/grafts which had been retrieved from patients (5-43 months post-implantation), exemplify the lack of understanding of the surface chemistry of this material, in particular, its susceptibility to intergranular corrosion. Critically, alloys containing Ni should be avoided in the first instance; however, if their properties cannot be replicated, such as SME, surface modification is a logical step to improve the corrosion resistance and minimise any adverse biological effects.

Inherent limitations
Despite titanium's excellent properties, fundamental issues remain regarding its deployment as a biomaterial. First, its lower hardness (ca. 200 H v [69,70]) and wear resistance result in an inability to be used for articulating surfaces, such as the femoral head [71]. Second, prior to any surface modification, pure Ti cannot confer a bond to living bone (passivated TiO 2 layers are known to be bioactive; however, the thin (ca. 5 nm) nature of such a coating is insufficient protection to corrosion [72]) and therefore, over time, its fixation in vivo is not stable. Cementation of the implant using poly (methyl methacrylate), or PMMA, to the surrounding tissue has been employed since 1953; however, this process also possesses limitations, since the exothermic curing reaction can initiate tissue necrosis [73].
In addition, chemical issues persist, especially when considering the complexity of the live environment in which these materials will be implanted. No matter the metal/alloy being used, corrosion will occur due to the extremely harsh body environment, in which body fluid contains chloride ions (Cl − ), proteins, as well as the variation in pH levels depending upon the area of implantation (3.5-9) [74]. Depending on the alloy being implanted, and its particular alloying elements, the corrosion resistance will vary considerably due to the passivating film formed from the alloy inclusions. For example, Nakagawa et al. [75] showed a Ti-0.2Pd alloy exhibited greater corrosion resistance (the amount of Ti dissolved (0.1% NaF/24 h) at pH 4 was 800 and 22 μgcm −2 for pure Ti and Ti-0.2Pd, respectively) over a wide pH range (3-7 at 37 ± 0.1°C, through dilution of 0.05-2% NaF with H 3 PO 4 ) due to the surface concentration of Pd. In addition, the ability of alloys to re-passivate their surfaces following corrosion, the chemical makeup of this re-passivated film compared to the native oxide, plus the ease of dissolution and reprecipitation, are significant factors in the corrosion resistance of the alloy [76]. In addition, the presence of proteins can either have a positive or detrimental effect on the corrosion resistance of the alloy being implanted.
Additionally, osseointegration, and more simply biocompatibility, of an implant material heavily relies on its topography and chemical nature [74]. Release of metallic ions, whether through leaching, wear or otherwise, can cause inflammation, irritation, and/ or sensitisation of the surrounding tissue. If not carefully balanced, inflammatory responses through pro-inflammatory cytokines, chemokines and matrix metalloproteases, may result in osteoclast activation, ultimately causing bone deterioration, osteolysis and implant loosening [77]. Ti-6Al-4V suffers from the addition of Al and V, both considered cytotoxic elements, which have been known to be associated with longer-term health conditions, such as Alzheimer's, neuropathy and ostemomalacia [78][79][80][81]. The topic of corrosion and limitations of Ti alloys in medical settings has been extensively reviewed by Geetha et al. [74].

Current processes employed for surface modification of Ti
To address the issue of bone bonding, many researchers have investigated the ability to confer bone bonding to previously bioinert implant materials, specifically metals and alloys. This has been succinctly reviewed by Jäger et al., taking into account mechanical, chemical and physical surface modifications of titanium materials [82]. However, despite all these different modification processes, one method, in particular, has received a great deal of attention and has been commercially exploited: plasma spraying of hydroxyapatite (HA).
Plasma spraying of HA: the current 'gold standard' Since the work performed by Getter et al. in 1972, HA has been investigated and used widely as a coating material for biomedical implants [83][84][85]. Its mimicry of the mineral component of bone, making up 70-90% of its dry mass, makes it an ideal surface modification, since it is inherently bioactive [86][87][88]. Furthermore, the primary method, which remains the only FDA-approved method for conveying a HA coating to implants, is through plasma spraying; their mechanical adhesion of 55-62 MPa is above the minimum requirement of 50.8 MPa [88,89]. Plasma spraying ( Figure 1) utilises a carrier gas (usually argon, or a mixture with other gasses) to carry HA particles through a low voltage and high current electrically discharged plasma, as described by Herman [90]. This melts the particles sufficiently, that upon impingement on the substrate surface, they solidify into a coating.
Plasma spraying offers rapid deposition rates (1 × 10 6 particles m −2 s −1 ) and sufficiently low costs. However, despite these advantages, plasma spraying, as well as other thermal spraying techniques, such as vacuum arc coating, has inherent disadvantages including poor adhesion [91]; non-uniform coating density [92]; and excessive temperatures (ca. 1500 K [8]) leading to deleterious phase transformations and residual surface stresses (20-40 MPa tensile) [93][94][95], which can result in the formation of micro-cracks [96]. Their brittle nature can result in spalled particles in vivo. The production of these spalled particles may induce phagocytic pathways through macrophage activation and may eventually lead to aseptic loosening of the implant, although direct determination of the cause of this multifactorial aetiology is still fiercely debated within the literature [97][98][99].

Alternative methodologies for HA generation
In recent years, different approaches have been considered as alternative methodologies for the current FDA standard. Alternative coating methods for generating HA on implants have been reviewed by Yang et al. and demonstrate the advantages and disadvantages of each coating method [100]. However, substantial issues remain within all techniques mentioned, especially regarding biomedical coatings, resulting in alternative lines of enquiry. For example, ion-implantation, which is the process of bombarding the sample surface with an ion beam of sufficient energy to implant the bombarding ion into the surface (Figure 2), as well as electrochemical methods, have shown promise in bioactive scenarios, through the production of apatite in SBF. Studies by Armitage et al. [101], Nayab et al. [102] and Rautray et al. [103], demonstrated successful implantation of calcium in titanium (Ti) and titanium alloys, with further work on the implantation of oxygen (O) [104], sodium (Na) [105], magnesium (Mg) [106] and even silver (Ag) [107] into Ti structures being detailed within the literature. Despite their advantages, the above techniques require complex, and costly, equipment in order to generate such surfaces/coatings. Furthermore, these processes cannot successfully coat complex geometries, such as surface plasma-sprayed, porous implants (e.g. 300+ μm thick, 30% non-interconnected porosity, mean pore size 250 μm, R a = 21-28 μm [108,109]), resulting in further alternative methods being sought [103].
In addition, a number of other technologies, although not FDA approved, are utilised to deposit CaP coatings onto implants (Table 2). Plasma electrolytic oxidation for example, is used by Nobel Biocare to produce calcium phosphate coatings on their dental implants [110]. The complexity of the regulatory process to successfully demonstrate long-term efficacy, coupled with the multiple requirements of the produced CaP coating (appropriate chemical phases, optimal surface roughness/porosity, long-term stability, some degree of sub-stoichiometry to induce bioactivity, etc. [111]) has resulted in limited FDA approval, and hence clinical usage. Other techniques are reviewed by Rahman [112]. It is, therefore, advantageous to develop coatings and surface modifications, which can generate apatite in vivo, since this in situ formation is more likely to form apatite with appropriate properties for subsequent bone formation; hence the investigation into alkali-titanate structures.

Medical alkaline titanates
Ever since the work of Li et al. in 1994, on the induction of bone-like HA on titanium substrates through the generation of gel-like titania on its surface, surface modification through chemical routes has been considered of great practical importance, with ever-increasing numbers of papers ( Figure 3) [113]. Their study demonstrated that the hydroxyl groups formed on the surface were bioactive in nature, in addition to the negative surface potential. Therefore, appropriate generation of such groups on titanium, while ensuring a negative surface charge, would hypothetically induce apatite formation in vitro in SBF, and in extension in vivo [113].
The formation of osteoconductive HA, through submersion in SBF, is a well-known technique utilised in in vitro assessment of a substrate's bioactivity (ISO 23317:2014) [114]. In comparison to HA found within the physiological environment, the SBF solution is regarded as supersaturated, as demonstrated by Lu et al. [115]. The higher the supersaturation of a solution, the greater the probability of molecular collision, which ultimately leads to the formation of stable nuclei and, therefore, HA crystal growth [116]. As SBF is supersaturated with respect to Ca 2+ and HPO 2− 4 (S°( HA) = 1.42; S°is defined as the thermodynamic saturation level, where S°> 0 indicates the compound will precipitate [117]), and metastable thus spontaneous apatite growth has been reported to occur from nucleation points within a vessel, such as cracks/scratches. However, a significant number of studies have contraindicated the use of SBF as an in vitro predictor of in vivo bioactivity, for example, β-TCP has extensively been shown to bond to bone in vivo, yet will not demonstrate a bioactive response in SBF [118]. This test, therefore, should be corroborated with additional in vitro assessments such as cytocompatibility assays (Neutral Red, AlamarBlue), osteogenic potential assays (Osteocalcin, Alkaline phosphatase), prior to in vivo testing. For example, a recent review by Kokubo et al. [25] and additional studies by Bohner et al. [117] and Kokubo et al. [119,120], described the efficacy of SBF as a technique for assessing bone-bonding capabilities through apatite formation. The review draws upon an additional review by Zadpoor [121], in which of 33 studies investigated, only 76% (25/33) predicted in vivo bioactivity through the in vitro SBF method, with 5 of the 8 that did not predict bioactivity, demonstrating no apatite formation on any of the biomaterials tested. Other predictive in vitro environments exist to understand specific medical material/implant applications. For example, artificial saliva since its local environment is significantly different to most orthopaedic scenarios.
Both classical (direct nucleation) and non-classical (amorphous calcium phosphate as the precursor) pathways have been studied extensively in the literature, with the study by He et al. [122] demonstrating via in situ liquid cell transmission electron microscopy a more in-depth perspective. It was found that mineralisation initiates due to ion-rich and ion-poor (Ca 2+ and PO 3− 4 ) solutions, with both classical and non-classical nucleation pathways observed. The ion-rich and ion-poor solutions phase separate, with the driving force for HA crystal growth being the reduction in surface energy, followed by agglomeration and coalescence. A schematic, taken from the study by He et al. is provided below (Figure 4).

The inception of bioactive titanates and their apatite-forming potential
The first studies that introduced alkali and heat treatment processes in order to improve Ti biocompatibility were by Kokubo et al. and Kim et al. in 1996 [23,24]. Their work was followed on from corrosion studies by Revie et al. [123], Hurlen et al. [124], Arsov et al. [125], and Prusi et al. [126], who suggested that hydrated TiO 2 would be produced in alkaline solutions such as KOH. This work was then carried further by Li et al., who demonstrated TiO 2 gels (produced via solgel method) successfully induced bone-like apatite in  SBF [113]. Kim et al. hypothesised that if such layers could be subsequently generated in vivo, bone-like apatite may be generated, enhancing the bioactivity of such a surface [23]. The initial alkali-treatment outlined by Kim et al. consisted of a 5-10 M NaOH or KOH (5 mL at 60°C) treatment of Ti or Ti alloy substrates for 24 h, followed by rinsing in distilled water, ultrasonic cleaning for 5 min and air drying at 40°C. Subsequently, heat treatments were also incorporated to increase the stability of the produced titanate structure (the methodology of which is outlined in Figure 5) [127].
The mechanism of sodium titanate formation Initially, the passivated surface layer, TiO 2 , is partially dissolved by the alkali solution: . Stage V can proceed by two different scenarios: One is HA growth only by ACP dissolution-HA reprecipitation followed by HA self-assembly (stage V-a), and the other one is the ACP dissolution-HA reprecipitation followed by HA growth via agglomeration and coalescence and then delamination from ACP (stage V-b). For the classical nucleation pathway, in stage III, the HA crystals directly nucleate from the ion-rich liquid phase. The HA crystals grow by agglomeration and coalescence (stage IV, bottom). Reproduced from He et al. [122], with permission from Science. As demonstrated by numerous studies [124,125,[128][129][130][131], the above reaction occurs concurrently with the hydration of the Ti metal; Equations (2-4), causing oxygen penetration into the top 1 μm of the surface. Ti The negatively charged HTiO − 3 generated through the alkali attack combine with the alkali ions present within solution, most notably Na + , which results in an alkali-titanate layer, with the chemical formula: M x H 2−x Ti y O 2y+1 ; 0 , x , 2 and y = 2, 3, or 4. The depth of penetration for both sodium and oxygen is ca. 1 μm with ca. 8 at.% Na at the surface [132]. Heat treatment of the samples was then conducted between 400 and 800°C for 1 h to assess the effects on the formed layer [23]. The original amorphous sodium titanate that was formed during the NaOH treatment, partially converted into crystalline sodium titanate at temperatures above 600°C, with small quantities of rutile (TiO 2 ) also being formed (XRD; Figure 6). By 800°C, fully crystalline rutile and sodium titanate had formed, with no amorphous layers present. The sodium titanate formed following heat treatment is isomorphic to the layer formed initially [132]. Not only can this process occur on pure Ti, but its applicability to medically relevant Ti alloys, including Ti-6Al-4V, Ti-6AL-2Nb-Ta, and Ti-15Mo-5Zr-3Al [23,24,133,134], is also seen.
Successful formation of sodium titanate structures has also been seen on equiatomic NiTi SMA [135][136][137][138][139][140]. As described previously (Section Advantages of Ti as a biomedical material), NiTi has a varied history, due in part to its corrosion resistance, and leaching of the potentially immunogenic, toxic and carcinogenic Ni ions [68]. It was, therefore, a natural step to attempt to modify the surface of NiTi devices, to not only improve corrosion resistance, but to also enhance the biological properties. However, when attempting to utilise the same sodium titanate modification process, it was found that this alone is detrimental to its corrosion resistance. This is likely due to alkali attack of the surface enhancing Ni leaching, as well as the formed sodium titanate layer exhibiting less corrosion resistance than inherent passivated titanium oxide layers (i corr = ca. 1 × 10 −4 v. 3.9 × 10 −5 A cm −2 , respectively) [135,138]. Therefore, most studies on NiTi require additional coating (e.g. hexadecyltrimethoxysilane [135,138]) of the sodium titanatemodified surface to allow its utilisation, for example in cardiovascular applications.
A more recent study by Conforto et al. [141] illustrates that these early interpretations may not be fully correct. Here, diffraction rings were clearly present in TEM diffraction patterns, thus it is likely to be nanocrystalline in nature rather than amorphous, with a monoclinic structure of Na 2 Ti 6 O 13 . Indeed in the original XRD ( Figure 6) in the pre-heat-treated sample, broad peaks are present, which is indicative of nanocrystallinity [142]. Kim et al. [23] interpreted the crystalline titanate structure as Na 2 Ti 5 O 11 ; however, a study by Bamberger and Begun [143] highlighted that this structure is unlikely to exist, with the monoclinic Na 2 Ti 6 O 13 structure being more likely, in the  general form A 2 Ti n O 2n + 1 . The structure, illustrated in Figure 7, has unit cell dimensions of a = 15.131 ± 0.002 Å, b = 3.745 ± 0.002 Å, c = 9.159 ± 0.002 Å, β = 99.30 ± 0.05°, with a crystal space group of C2/m, C 3 2h (Z = 2) [143,144].
Apatite formation mechanism of alkali-titanate surfaces Apatite formation (pictorially described in Figure 8) occurs on alkaline titanate surfaces through complex ion-exchange reactions either in vitro in SBF or in vivo. The level of electrostatic attraction of Na + within the sodium titanate layer is not enough to hold it in place when H 3 O + ions are in solution. Therefore, the exchange of H 3 O + with Na + results in Ti-OH bonds forming on the surface. Furthermore, this reaction causes an increase in the pH of the environment surrounding the surface, which in turn causes the surface to become negatively charged, as detailed by Gold et al. [145]. The mechanism of apatite formation is a result of research data conducted by various research groups [14,146,147].
Takadama et al. used X-ray Photoelectron Spectroscopy (XPS) and X-ray Diffraction (XRD) to quantify the composition of the top surface layers of the substrate [146]. It was found ( Figure 9) that HA formation began at around day 2, with complete conversion occurring within 3 days. Furthermore, highresolution scans of Ca 2p, O 1s, Na 1s and P 2p all corroborated the initial findings. From the Na 1s spectra, Na + began eluting within 30 min, while Ca 2p peaks were found as early as 30 min (deconvoluted as calcium titanate: Ca 3 Ti 2 O 7 ). The Ca 2p peaks exhibited a slight shift in binding energy at 2 days, which was deconvoluted as hydroxyapatite. P 2p did not exhibit a peak until 48 h submersion in SBF, indicating Ca had ion exchanged into the surface prior to the attraction of phosphate ions onto the surface. However, the most significant piece of data was the formation of Ti-OH bonds within 30 min of SBF immersion. This partially confirmed the hypothesis that these bonds were essential for indirect apatite formation through calcium titanate formation [146].
XPS results were corroborated by Takadama et al. whereby transition electron microscopy (TEM) combined with energy-dispersive X-ray (EDX) analysis was employed to understand the structural alteration during apatite formation [14]. Initially, a fine network structure of ca. 500 nm was observed on the alkaliand heat-treated samples. Upon immersion in SBF, Ca inclusion was noted within 0.5 h, with an amorphous calcium titanate and calcium phosphate (Ca:P = 1.4) forming within 24 and 36 h, respectively. By 72 h, the Ca:P ratio was approximately 1.65, close to stoichiometric HA (1.67) [14]. Yamaguchi et al. further demonstrated the morphological changes that occurred during the submersion of sodium titanate layers within SBF, producing apatite as seen in Figure 10 [148].
Li et al. explored vacuum pressures during the heat treatment process (ca. 10 5 -10 −3 Pa) [149]. Their results showed that at higher vacuum pressures, the structure exhibited larger pores sizes, while improving the HA formability in SBF [149]. Takedama et al. postulated that the formation of apatite on alkali-treated titanium occurred through electrostatic interactions between the surface, and specific ions within the aqueous solution [14]. Kim et al. investigated Zeta (ζ) potential, in regard to soaking time in SBF [147]. The ζ potential (V) was quantified using the Smoluchowski equation [150], seen below: where u e is the electrophoretic particle mobility (m 2 V −1 s −1 ), h is the solution viscosity (Pa s (or N s m −2 )), 1 r is the relative permittivity/dielectric constant and 1 0 is the permittivity of a vacuum (8.8 × 10 −12 Fm −1 (or NV −2 )). The work corroborated that ionexchange reactions between Na + and H 3 O + occurred first, generating negative Ti-OH bonds (negative ζ potential), which then attracted positive Ca 2+ (increasing the ζ potential due to localised Ca 2+ concentrations), followed by negative phosphate ions, forming calcium titanate (within 0.5 h) and calcium phosphate (within 42 h), respectively ( Figure 11) [147]. Within 72 h, apatite had been formed on the surface, due to the much lower solubility of HA with respect to calcium phosphate in the body environment. Furthermore, it is well known that HA has a negative charge in the body environment due to the presence of hydroxyl and phosphate groups on its surface, as demonstrated by the potential at 72 h ( Figure 11) [151]. Interestingly, a trend that has been seen in a few studies [147,152], is the incorporation of carbonate, sodium and magnesium into the apatite layer formed in SBF, which is more akin to that of bone-like apatite. This is due to the SBF being supersaturated with respect to apatite even under normal conditions [153]. In order to completely assess the dependence of pH of the treatment medium on apatite formation, systematic alteration of NaOH and HCl solution pH from 0 to 14 was conducted at 60°C for 24 h, followed by heat treatment at 600°C. The study by Pattanayak et al. indicated a pH of 1.1 ≤ x ≤ 13.6 inhibited the occurrence of apatite formation, while pH values below 1.1 and above 13.6 generated apatite within 3 days ( Figure 12(A)) [154]. It was also demonstrated that apatite formation depends on the surface charge, rather than surface roughness and specific crystalline phases. When Ti metal is subjected to acid treatment, as outlined previously, followed by subsequent heat treatment, the surface ζ potential is positive ( Figure 12 (B)). Conversely, for alkali-treated substrates, the ζ potential is negative. The theory that ζ potential affects apatite formation is further substantiated through the fact that natural Ti, despite the same heat treatment process, does not generate apatite on its surface, since the surface ζ potential is neutral ( Figure 12(C)) [154].
Further reactions occur within the aqueous SBF solution, whereby, the negative surface attracts Ca 2+ , forming an amorphous calcium titanate surface. The accumulation of Ca 2+ ions result in an overall positive surface charge, which subsequently attracts PO 3− 4 (phosphate) ions, forming an amorphous calcium phosphate. This surface is metastable and subsequently matures into apatite, which was found to occur within 3 days in SBF [148].

Development of novel titanate structures for biomedical applications
Novel titanate structures have been explored, typically as composite coatings, porous scaffolds, orthopaedic microspheres, and protein/drug delivery carriers. These have been derived to achieve more cost-effective manufacturing routes, with targeted, location-specific biological effects. For example, Mozafari et al. formulated 50-125 nm ZrTiO 4 /Bioactive glass thin films via sol-gel followed by spin coating to form consecutive multi layers [155]. In a methods article by Triviño-Bolaños et al. [156], sodium titanate was moulded, pressed and sintered to form porous rutile TiO 2 /Na 0.8 Ti 4 O 8 / Na 2 Ti 6 O 13 scaffolds to eliminate the use of toxic solvents often used in nanoparticle synthesis and/or to reduce costs associated with methods such as hydrothermal synthesis. Further examples of porous scaffolds (ca. 75% porosity via MicroCT, ca. 100-500 μm pore sizes, with ca. 100 μm wide struts) were developed by a slurry coating of polymer foams using TiNbZr powders, which were sintered and followed by hydrothermal titanate formation to mimic geometries of cancellous bone ( Figure  13(A)) [157]. In addition, electrochemical routes by anodic oxidation have been used within the literature to form 'nanoflower-like' titanate coatings, driven by applied voltages/currents or 350 V and 70 mA cm −2 in a custom electrolyte of β-glycerophosphate, calcium acetate and NaOH. The high surface area nanoflowerlike constructs were found to be bioactive in SBF, in good agreement with previous alkaline titanate studies ( Figure 13(B)) [158].
Bio-silks, which are readily developed for medical composite matrices and are generated out of silk fibroin, have been altered to increase mechanical performance by integration of 2D titanate nanosheets followed by ion-exchange reactions with silver solutions for antibacterial dental applications [159]. By incorporating nanoparticles within the silk matrix, a change in the mechanical properties can be achieved through interaction between the particles and the fibroin chains. An alternative application by Colusso et al. showed silk titanate multilayers fabricated by consecutive spin coating with an ability to change colour based on atmospheric humidity changes applicable in sensing applications, which may be utilisable in a biomedical setting to provide minimally invasive detection and/or diagnosis. Modifying the relative humidity range from 10 to 80% induced repeatable modification in the transmittance wavelength, attributed to the change in refractive index of the multilayer due to swelling via water uptake and bonding. The titanate-containing silk generated a greater response compared to silk alone, due to the negatively charged titanate nanosheets interacting more with water [160].
Nano-grid titanate geometries have been produced by anodising Ti foil, such as by Zhang et al., which additionally introduced zinc through hydrothermal incorporation and showed in vitro potential for antitumour osteosarcoma applications ( Figure 14) [161].
ZnO nanorods or nano particles have been used for anti-tumour applications in the past; however, their ease of detachment from the implant results in lower efficacy, hence, loading of Zn into these nano-grid structures. Nano-grids with a Zn content of 0.15 M were found to be optimal, had affirmatory abilities to significantly inhibit UMR-106 tumour growth in vivo and had no impairment to the body. The geometry of these structures is of importance to biomaterial  applications, due to not only enhanced release of beneficial ions, but also enhanced adhesion, migration and proliferation of normal osteoblasts.
Hsu et al. developed titanate microspheres using a water-in-oil emulsion method and a non-toxic camphene porogen for titanium. Orthopaedic microspheres are often formed of Ca/P ceramic and glasses for injecting into bone, cartilage or muscle as a regenerative filler with enhanced cell infiltration [162]. Here titanate spheres contained up to 74% porosity with pore sizes up to 200 μm ( Figure 15(A)) [163]. Silica-and Nb-substitution into the titanate structure was shown by Milosevic et al. to improve absorption of Ovalbumin, gentamicin and methyl blue with 1 g of titanate absorbing between 9 and 90 mg at pH 5.0-7.0 and showing desorption characteristics at pH 7.0 indicating the potential of substituted-titanate structure for protein/drug delivery [164]. Li et al. produced magnetic yolk-shell titanate microspheres of ca. 560 nm in diameter, which were produced using an Fe 2 O 3 core coated with SiO 2 and TiO 2, followed by titanate formation for catalysis applications; however, their potential could be extended to MRI imaging ( Figure 15(B)) [165].
Nanostructures: from nanotubes to nanograins Nano-geometries of alkaline titanate structures are of interest since they can increase effective surface area and therefore cellular interaction rates for biomedical applications and nanotubes have been developed as a carrier material for cellular stimuli, such as proteins or delivery of therapeutic ions owing to the ease of functionalisation. In 1998, Kasuga et al. first reported synthesised nanotubes of ca. 8 and 100 nm in diameter and length, respectively, with surface areas of ca. 400 m 2 g −1 from sol-gel produced TiO 2 powders via a hydrothermal reaction at 120°C [166]. Nanotubes and nanosheets of Na 2 TiO 3 O 7 have been formed using a commercially available P25 TiO 2 nanoparticle precursor. TiO 2 powders (21 nm diameter) treated in NaOH solutions at temperatures up to 150°C exhibited increased surface areas from 66 to 337 m 2 g −1 due to the resultant formation of 50 nm diameter tubes, that were hundreds of nm in length [167]. Similar tubes were electrochemically deposited onto Si substrates for use as optical/semiconductor films by Kim et al. [168] while other manufacturing modifications have successfully produced elongated tubes by rotating particles at up to 20 rev.min -1 during synthesis [169]. The P25 TiO 2 hydrothermally-derived titanate nanotubes were tested with rat cardiomyocytes and were shown to be cytocompatible over 24 h, despite reporting endocytosis and diffusion of particles into the cell, suggesting potential future cardiovascular applications ( Figure 16) [170]. The authors went a step further by exploiting the presence of negatively charged surface hydroxyls to bond polyethylene glycol (PEG) and polyethylene imine (PEI) for therapeutic and gene therapy applications also showing no cytotoxic effects on cardiomyocytes in concentrations up to 10 μg mL −1 [171].
More recently, Rodrigues et al. highlighted the potential for rare earth (La 3+ , Ce 4+ , Pr 3+ , Nd 3+ , Er 3+ and Yb 3+ ) doping of sodium titanate nanotubes generated via a microwave assisted hydrothermal method [172]. The synthesised nanotubes were then thermally treated (200,400, 600 and 800°C, 2 h) in order to produce structural and morphological changes which ultimately can modify their optical properties. However, in the context of implantable medical devices, these materials will be significantly limited due to the toxicity of the RE elements.
Alkaline titanate nano-whiskers, specifically K 2 TiO 3 and K 2 Ti 4 O 9 , were found to be bioactive in SBF over a 12-d period, precipitating calcium titanate, Ca/P, and hydroxyapatite on their surface [173]. Zhao et al. formed nanowire scaffolds by hydrothermal treatment of Ti substrates, which were post coated with electrochemically deposited hydroxyapatite. They showed encouraging proliferation and differentiation of MG63 osteoblast cells after 7 days [174].
Titanate, titanium-based metallic and titaniumincorporated glasses Another important class of titanate materials are titanate glasses. In a biomedical context, glasses have been widely used from biodegradable, bioactive coating and scaffold materials to enhance osseointegration [175], to cancer detection through fibreoptic cables [176]. The glass structure is highly modifiable through the incorporation of elements. For example, the bulk glass composition can be modified to illicit controllable degradation profiles [177], which can be utilised in bone repair scenarios [178]. Titanium-based glasses pose a novel alternative to presently used glasses, such that their mechanical properties are significantly higher than conventional glasses in tensile loading (1.4-2 GPa v. ca. 7 MPa for a typical glass [179]), as well as the ability to dope such structures with functional ions. This will enable wider use of glass structures in medical applications, such as load-bearing scenarios, which they have been limited previously. In this review, we compare titanate glasses with titanium-based metallic and titanium-incorporated glasses.
Titanate glasses were first reported by Jijian et al. in 1986 and contained up to 60 wt-% TiO 2 [180]. High-density (up to 4.5 g cm −3 ) barium titanate glass microspheres have shown promise for high-resolution imaging of biological samples [181]. In contrast to titanate structures, titanium-based metallic glasses within the biomedical field are interesting candidates for non-corrosive, load-bearing applications. Titanium metallic glass alloys exhibit mechanical properties, such as tensile strength (ca. 1.4-2 GPa [182]) and elastic strain limit (up to 2% [183]), which supersede that of implant materials, such as Ti6Al4V (0.85-1.1 GPa [182]). Reported manufacturing routes are so far limited to melt casting of small (up to 6mm diameter) rods and spin casting of 30 μm ribbons, Qin et al. [184] and Jiang et al. [185], respectively. Titanium glasses have a relatively high glass-forming window, in compositions from Ti 10 Zr 30 Cu 60 [185] to Ti 40 Zr 10 Cu 36 Pd 14 [184]; however, high-resolution TEM revealed nanocrystals of up to 15 nm embedded within the amorphous matrix for Ti 10 Zr 30 Cu 60 composition suggesting a glass-forming limit. Ti 40 Zr 10 Cu 36 Pd 14 glasses can be treated in a NaOH solution to form sodium titanate prior to submerging in Hanks Balanced Salt Solution (HBSS) for 30 days to observe formation of an apatite layer, suggesting limited in vitro reactivity [184].
Another approach is to use TiO 2 inclusions in phosphate-based glasses (PBGs). Examples include 25 mol-% TiO 2 to improve network connectivity by increasing ionic cross-linking between phosphate units and/or positioning itself within the structural backbone to reduce the prevalence of P-O-P chains, which 'depolymerise' by hydrolysis [186]. Alternatively, Das et al. melt quenched anti-wear strontium bismuth titanate borosilicates glasses, using post-heat treatment for crystallisation of wear resistance ceramic phases, attributing TiO 3 units as effective nucleation sites for crystallisation [187].

Piezoelectric biomedical titanates
The interest in biomaterials that are piezoelectric stems from the piezoelectric properties of human bone (coefficient of ca. 0.7 pC/N) [188], which have been known to stimulate osteogenic cells to facilitate bone remodelling [189,190]. In particular, barium titanates (BaTiO 3 ) [191,192] have potential uses in piezoelectric medical composites, in particular, in medical imaging, due to its high refractive index (ca. 2.1) increasing imaging resolution by a factor of three [193]. To achieve sufficient electrical polarisation, 80 vol% BaTiO 3 has been desirable in orthopaedic structures [194]; below 70%, no measurable piezoelectric properties where found [195]. Bowen et al. [195] hypothesised the reason for the reduction in the piezoelectric properties was due to 'mechanical clamping' of the BaTiO 3 materials due to the stiffer HA matrix preventing adequate strain to be generated during poling.  cytotoxic composition after 72 h incubation [190]. Ions of Ba 2+ were released, ca. 582-826 ppm by 72 h, due to the reactivity of less stable ceramic phases produced during the sintering process. However, less soluble composites, such as the 80% BaTiO 3 -HA composite (244 ppm Ba 2+ ) formed non-soluble secondary phases (CaTiO 3 and Ca 3 (PO 4 ) 2 ) and remained cytocompatible [190].
Generating porosity in such scaffolds adds to the potential medical application of these structures, since enhanced bone ingrowth combined with piezoelectric properties draws on the natural properties of bone. HA-coated BaTiO 3 porous scaffolds were generated by Etherami et al. and produced by the ceramic slurry coating of a polyurethane template (burned out in a post process) and dip coated with a Gelatine/HA layer. Gelatine was used to enhance the mechanical stability of the scaffolds, since it has been known to exhibit microscale crack-bridging properties, which leads to potential enhancement of the toughness of the scaffold, similar to collagen fibres in bone. The composite exhibited cytocompatibility with seeded MG63 osteoblast cells for up to 7 days, successfully adhering to and incorporating through the porous ceramic network, while the crystalline structures exhibited piezoelectric properties of 4.5 pC/N [189].
Other approaches to generate and improve medical piezoelectric materials include modifying BaTiO 3 structures with elements such as Ca, Sr, Zr and C, to improve electrical characteristics, mechanical properties and cellular cytocompatibility. Sr 2+ ions have been reported to improve and suppress osteoblast and osteoclast proliferation, respectively, while increasing the dielectric constant in piezoelectric titanates; the increase is due to Sr 2+ decreasing the oscillation space of Ti and distorts the structure of the ferroelectric domain [196]. Phromoyoo et al. added zirconia to increase the piezoelectric coefficient in egg shell synthesised β-TCP/BaTiO 3 /Zr (β-TCP/ BZT) composites showing enhanced mechanical hardness, due to the shift from intergranular to predominantly intragranular fracture, with >80 vol% needed to produce piezoelectric properties [188]. BaTiO 3 was electrodeposited by Rahmati et al. onto Ti6Al4V surfaces for their use as osteoinduction/piezoelectric coatings in load-bearing orthopaedic implants, such as hip stems or dental implants. In vitro bioactivity was confirmed by apatite formation 7 d post-submersion in simulated body fluids (SBF) [197].
Composite approaches have also been used, such as Beta-Tri Calcium Phosphate (β-TCP)/Sr-doped BaTiO 3 composites were 3D printed using polyvinyl chloride (PVC) as a binder to form Ca/P leaching osteogenic scaffolds. Other applications include fixation bone cements. Tang et al. mixed graphene and BaTiO 3 (30-80 vol%) into polymethyl methacrylate PMMA suggesting that 0.5 vol% graphene increased electrical conductivity and reduced the relative required proportion of BaTiO 3 to 60 vol%. Without graphene, they obtained a compromise between piezoelectric performance (achieving 0.33 pC/N at 60 vol% BaTiO 3 ) and compressive strength (ca. 85.3 MPa) showing similar strength to natural human bone. Graphene inclusion increased piezo performance to 0.8 pC/N attributed to improved charge transfer and remaining cytocompatible [198]. A comparative look at the various piezoelectric, material and chemical properties of alkaline titanate materials described within the literature, in comparison with human bone, has been compiled in Table 3.
Pre-modification: the potential for multifunctional surfaces with peroxide pretreatment A study by Janson et al. [208] has developed dualfunctionality surfaces that incorporate titanate structures. This work, following on from previous work by Tengvall et al. [209], looked into the generation of a titanium-peroxy gel layer; suggested by Tengvall et al., to be: where a hydrogen peroxide (H 2 O 2 ) treatment of titanium has potential bactericidal properties. By combining a hydrogen peroxide treatment (30 wt-%, 80°C, 1 h), with subsequent sodium (5 M, 15 min) and calcium hydroxide (0.1 mM, 15 min) treatments, these surfaces demonstrated bactericidal effects (reduced viability by 93%) on Staphylococcus epidermidis through direct and biofilm inhibition tests, while ensuring cytocompatible surfaces to MC3T3 human dermal fibroblast cells [208]. Reduced soaking times in alkali media were utilised to lessen reactive oxygen species (ROS), which are toxic to both bacteria and potentially human cells. This approach's objective reduced the release of ROS species to prevent cytotoxicity, while maintaining bactericidal properties; this compromise is prevalent in many elution-based bactericidal materials, since human cells tend to be more susceptible compared to bacteria. H 2 O 2 pre-treatment is also effective for the formation of apatite on sodium titanate-treated NiTi alloys surfaces by Cheng-lin et al. [139]. It was found that H 2 O 2 pre-treatment led to the creation of more Ti-OH groups, as well as reducing the amount of Ni 2 O 3 , Na 2 TiO 3 and remnant NiTi phases. As a result, the induction period of apatite formation is shortened from >24 h to 12 h by the slow kinetic formation process of Ti-OH groups via the exchange of Na + ions from Na 2 TiO 3 with H 3 O + ions in SBF.  [203] (Continued) The increase of solid loading led to a reduction of porosity, smaller lamellar pore width and thicker lamellar ceramic wall resulting in the increase of the active piezoelectric ceramic phase in the HA/BT composite. Despite lower mechanical properties, the porous scaffolds with larger pore sizes (>100 μm) are more suitable for tissue ingrowth, while those with small pores sizes (≤100 μm) can provide good mechanical properties and facilitate the nutrition and fluid transport through pore channels in the aligned graded porous structures.

Limitations of first-generation sodium titanates
Before discussing some of the exciting flexible titanate structures in Section Modifiable titanate structures based on ion-exchangeability in vitro, it is worth exploring some of the limitations of the first-generation sodium titanate structures.

Ca reagent contamination, its effects on apatite formation and shelf-life assessment
Kizuki et al. [210] demonstrated that even 0.0005% of Ca ions present within the initial NaOH solution used, will preferentially enter into the sodium titanate layers formed, inhibiting apatite formation ( Figure 17; above 1.5 ppm Ca concentration apatite nucleation is inhibited), since the sodium ions diffusion is reduced due to partial replacement of Na + with Ca 2+ in the surface structure. Generally, the reduction in apatite formation was a direct correlation to greater NaOH volumes used, i.e. a larger availability of Ca 2+ ions. It was found that just 1.5 ppm of Ca 2+ was enough to reduce Na + release from 1.21 to 0.87 ppm ( Figure  18) [210]. Fundamentally, there is a need for very careful planning of the type of reagent used, since contamination on this scale can be considered negligible depending on the application. The lack of stability of sodium titanate structures in humid environments is also a concern [211]. Humid environments (Kawai et al. assessed for 1 week at 80°C and 95% humidity [211]) inherently contain H 3 O + within the water vapour [212], ion-exchange reactions can still persist, causing a decrease in the sodium content and hence reduced apatite formation in implantation. This is a significant issue for medical implants, since there is a substantial wait between manufacture, shelf-life and implantation, so careful inert packaging is required. This effect must also be considered during all stages of production from preparation and handling of these materials, characterisation and validation and storage.

Inability to produce titanate surfaces on certain Ti alloys
Alkali-treatments are not so effective to produce apatite on Ti-Zr-Nb-Ta alloys [213]. Specifically, the Ti-15Zr-4Nb-4Ta alloy utilised had the advantage of having high ultimate tensile strength (ca. 453 MPa) [214], but the inclusion of Nb and Zr inhibit apatite formation, as illustrated by Cho et al. [215] and Niinomi [216]. If the mass% of Nb and Zr in the alloy is less than 10%, it is postulated that apatite formation is possible on NaOH-treated alloys when the number of sodium titanate molecules on the surface is sufficient to nucleate apatite. Sodium niobate can also form and there is a delay in the ion-exchange of Na + with H 3 O + , likely due to stronger interaction between Nb-O and Na + . Also, Zr has higher corrosion durability in NaOH solution compared to Ti, which forms a thin Na-free zirconia hydrogel, where OH − formation is rare; further inhibiting sodium titanate and subsequent apatite formation. Therefore, additional or modified steps are required in order to confer apatite-  forming titanate structures not just non-Ti-containing materials, but also many Ti alloys [216].

Modifiable titanate structures based on ionexchangeability in vitro
The limitations outlined in Section Pre-modification: the potential for multifunctional surfaces with peroxide pre-treatment, can be overcome by the flexibility of the titanate structure, enabling facile ion-exchange. Here we consider approaches to improve biocompatibility and impart bacterial resistance.

Bioactive titanates
The majority of studies regarding titanate surfaces for biomedical applications focus on their bioactive potential both in vitro and in vivo. The development on the original sodium titanate surfaces, to incorporate ions such as calcium, magnesium, strontium, through ion-exchange reactions, has seen many alternative produced surfaces for biomedical implants.

Calcium titanate
Calcium as an element has been incorporated in a number of coatings and materials designed to be bioactive and is essential in biomineralisation of bone; 99% of the calcium in the body is found within hydroxyapatite [217] and has been incorporated into various materials in order to aid bone growth and osteoblast proliferation [218,219]. Studies of calcium titanate (CaTiO 3 ) coatings have demonstrated increased osteoblast adhesion [220], apatite-forming ability [221] and improved osseointegration in a rabbit model [222] compared to untreated or roughened Ti. By incorporating an additional step into the hydrothermal treatment, which soaks the sodium titanateconverted samples in a calcium-rich (saturated Ca (OH) 2 ) solution (36.5°C, 24 h), Rakngarm et al. demonstrated the ability of sodium titanate surfaces to incorporate alternative ions (Ca 2+ ) into the structure on Cp-Ti and Ti-6Al-4V alloys, enabling faster apatite generation; HA deposition within 24 h immersion in SBF, and complete coverage within 7 days, compared to only sodium titanate treatment (up to 1 month) or acid treatment (10 days) [223]. Kizuki et al. [19,224] improved the apatite-forming ability (no apatite nucleation pre-water treatment, with total apatite-surface coverage following water treatment) of calcium titanate surfaces, as well as scratch resistance by incorporating a heat treatment at 600°C , followed by submersion in H 2 O at 80°C for 24 h (<10 mN for pre-heat treatment; 48 mN following heat treatment; 54 mN following subsequent water treatment). The increase in scratch resistance due to the conversion of Ca x H 2-2x Ti 3 O 7 to CaTiO 3 postheat treatment also stopped apatite formation in SBF, no apatite nucleation observed by SEM, due to the suppression in Ca 2+ leaching from the calcium titanate layer. Therefore, an additional hot water treatment stage (80°C for 24 h) is required to generate a calcium-deficient titanate layer on the surface due to ion-exchange with H 3 O + ions, facilitating calcium release. Due to this initial release (0.15 ppm), subsequent apatite formation through ion-exchange reactions detailed previously can occur [19,224]. This identical hot water treatment is effective at generating apatite formation on Ti-Zr-Nb-Ta alloys, which had previously not been achieved without this additional treatment [225,226].
Bone bonding of NaOH (5M, 60°C, 24 h), CaCl 2 (100 mM, 40°C, 24 h), heat-(600 or 700°C) and water-treated (80°C, 24 h) Ti metal and Ti alloys, specifically from the Ti-Zr-Nb-Ta alloys, namely Ti-15Zr-4Nb-4Ta, Ti-29Nb-13Ta-4.6Zr alloys and gum metal (Ti-36Nb-2Ta-3Zr-0.3O) were investigated by Fukuda et al. and Tanaka et al. [227,228]. Both studies demonstrated successful bone bonding without fibrous encapsulation on rabbit tibial implants ( Figure 19). During detachment testing, facture occurred within the main bone portion, as opposed to the interface, meaning the interfacial bond strength is sufficient to prevent delamination; with all failure loads of the 700°C and water-treated samples exhibited failure loads > 50 N. One postulate is that by developing better bone-bonding capabilities (minimising fibrous encapsulation and enabling mature bonding) on alloys that are free of cytotoxic elements, it will result in a new generation of implant materials that can potentially replace the current generation of Ti alloys, through potentially negating any negative biological effects. In 2019, the first clinical trial of CaTiO 3 Schanz screws demonstrated marked improvement (median values of 25.36 Nm v. 16.68 Nm, respectively; p = 0.043) in fixation strength over stainless steel (SS) and HA, confirming its potential efficacy [229].

Magnesium titanate
Typically, calcium titanate structures are substituted with an additional cation (Mg, Zn or Sr), although multiple cations are also possible. These multi-cationic titanates not only produce titanate structures with equivalent bioactivity but also possess their own unique functionalities, which can be optimised through varying the compositional ratio [230].
Magnesium is a highly abundant element within the body, with 60% of its in vivo distribution being contained within bone [231]. It's essential participation as a cofactor to >300 enzymes, as well as reducing HA crystal size, without which would ultimately result in brittle bones, makes it an ideal element to incorporate within the titanate structure for biomedical applications [82]. Magnesium titanate was first produced via ion-exchange through submersion of pure Ti in an MgCl 2 aqueous solution, by Shi et al. [232]. The resulting Mg-containing surface showed improvements in cell attachment (38% v. 25%) and proliferation (10 × 10 4 v. 8.5 × 10 4 cells/well), linked to the increased protein (16 v. 8 μg cm −2 ) adsorption observed on the treated surface compared to untreated Ti. It is widely accepted that electrostatic interactions are important for protein adsorption, with Mg 2+ acting as a cation bridge between the solution proteins and the underlying Ti surface. Higher protein adsorption aids in cell proliferation due to the increasing expression of integrins that mediate cellular adhesion. Yamaguchi et al. demonstrated an Mg-containing CaTiO 3 using a CaCl 2 -MgCl 2 mixture following an initial NaOH solution treatment. Ca/Mg ratios between 1:0 (pure calcium titanate) and 0:1 (pure magnesium titanate) were investigated [21]. The apatite-forming ability of the  Figure 20 for corresponding values). Furthermore, Ca-containing MgTiO 3 showed greater failure loads and a higher bone-implant contact through a 24-week rabbit tibial in vivo test, supporting the biological markers found in vitro [233].
Another approach is to fully substitute MgTiO 3 by excluding CaCl 2 . MgTiO 3 exhibited apatite formation after heat treatment without needing additional water or aqueous solution treatments, and that the apatiteforming ability was greater than NaTiO 3 , as well as increased albumin adsorption, higher MC3T3-E1 cell attachment, spreading and faster proliferation over 7 days. This was due to the greater electrostatic attraction between the divalent Mg 2+ ions and negative albumin over Na + or Ti-O − [234].

Strontium titanate
Strontium has been shown to regulate bone regrowth, by activating osteoblast activity and reducing osteoclastic activity and hence has been a popular choice as a bioactive agent [235]. Initial work on ion exchange of Sr by Yamaguchi et al. [236] demonstrated Sr-containing CaTiO 3 did not exhibit apatiteforming ability until a further H 2 O or SrCl 2 treatment was employed, which showed higher Ca 2+ release (0.05 ppm) and lower Sr 2+ release (0.06 ppm), compared with SrCl 2 treatment alone (0.01 and 0.92 ppm, respectively over 7 days) [236]. Okuzu et al. also found higher cell viability, greater expression of β catenin, integrin 1β, cyclin D1, ALP and Opn, as well as higher ECM mineralisation and Ocn expression over 7 days for Sr-containing MgTiO 3 (Figure 20), compared to Ca-doped titanate alone. Greater failure loads (32 v. 22 N at 24 weeks) and higher bone-implant contact (40 v. 36%) was also achieved in the Sr-containing MgTiO 3 compared to Ca-doped titanate alone, through the same rabbit tibia model (Figure 21) [233].
The SrCl 2 ion-exchange treatment has been used on selective laser sintered (SLS) Ti-6Al-4V scaffolds by Shimizu et al. [237]. The chemical process Figure 19. SEM images of the surface of an AcaHW-GM plate (defined below) after detaching tests at (A) 4 weeks and (b) 26 weeks. AcaHW-GM plates were prepared as follows: plates of gum metal were first soaked in 10 mL of 1 M aqueous NaOH solution at 60°C for 24 h. After removal they were gently rinsed with ultrapure water for 30 s and dried at 40°C. The plates were subsequently soaked in 20 mL of 100 mM CaCl 2 solution at 40°C for 24 h, washed and dried in a similar manner. Next, they were heated to 700°C at a rate of 5°C min −1 in an electric furnace in air and kept at that temperature for 1 h, followed by natural cooling, followed by soaking in 20 mL of ultrapure water at 80°C for 24 h, and then washed and dried. (A) Some bone residue (Bone) was observed on the AcaHW-GM plate (P) at 4 weeks. (B) Much more bone residue (Bone) was observed on the AcaHW-GM plate (P) at 26 weeks. Reprinted from Tanaka et al. [228] with permission from Springer Nature.
successfully imparted the nanostructure of titanate on the microscale roughness of the scaffold. XTT cell viability showed no significant difference between titanate and untreated scaffolds, but the CaTiO 3 -and SrTiO 3 -modified scaffolds showed higher ALP and integrin 1β gene expression in vitro. In vivo testing in a rabbit model further detailed enhanced failure loads and bone-implant contact (ca. 35% v. 27% for SrTiO 3 and CaTiO 3 , respectively) at 2 and 4 weeks, with SrTiO 3 (51.3 and 103.6 N for 2 and 4 weeks, respectively) outperforming CaTiO 3 (51.3 and 103.6 N for 2 and 4 weeks, respectively) [237]. The SrTiO 3 -modified scaffolds also contained Ca ions, which may have led to a synergistic behaviour observed when both ions are delivered, increasing osteogenesis; however the underlying molecular mechanism and the various factors affecting this is still unclear [238].

Barium titanate
Zhou et al. [239] described the formation of barium titanate films, which were subsequently doped with strontium (Sr) ions, for improved bioactivity for osseointegration enhancement; Sr has been shown to activate various signalling pathways in osteoblasts and osteoclasts, such as the calcium-sensing receptor, inositol 1,4,5 triphosphate, MAPK-ERK1/2 and Wnt/ NFATc, enabling bone stem cell osteogenic differentiation and inhibiting osteoclast activity. Due to Sr and Ba having similar ionic radii (1.12 and 1.34 Å, respectively [240]) and electronic structures, substituting Sr into the barium titanate structure is a logical step to not only modify the piezoelectric properties of barium titanate, but to also improve the bioactivity. With Sr substitution, the morphology becomes rougher with a micropore structure (ca. 30-300 μm), and the water contact angle (WCA) (BaTiO 3 = ca. 50°v. Sr-BaTiO 3 = ca. 40°), elastic modulus and hardness (BaTiO 3 = ca. 1 GPa v. Sr-BaTiO 3 = ca. 0.75 GPa) decreased. The film exhibited a sustained release of Sr ions (3 ppm day 1, 4 ppm day 14) and proved beneficial for in vitro cell adhesion, osteogenic differentiation, in vivo osteogenesis and osteointegration. The benefits of lower mechanical properties (benefit cellular proliferation and osteogenesis) and WCA between 40°and 60°(enhance cell adhesion), combined with the properties highlighted above for Sr inclusion, demonstrate the improved properties for Sr-doped structures over BaTiO 3 alone.
The link between bioactivity and piezoelectric/ferroelectric material properties is complex since bone growth is heavily dependent on multifactor processes. It has been suggested that localised stresses are detectable by osteocytes, which results in enhanced bone remodelling at areas of higher stress. Furthermore, electrical polarisation may induce specific cellular pathways that can lead to enhanced bone growth. Further studies on barium titanate ceramics have demonstrated apatite-forming ability being dependent on the polarity or partial charge of the surface, with positively charged surfaces showing no bioactivity while negatively charged surfaces were able to bond cations such as Ca 2+ , subsequently allowing the typical process of apatite formation of chemically derived titanates to occur [33,[241][242][243]. Furthermore, although not covered in this review, the use of barium titanate materials for medical sensing applications is becoming an increasingly attractive research area [244][245][246].

Zinc titanate
Zinc is another essential element within the body, being the second most abundant metal after iron. Its functions in the body are numerous, but can be divided between structural roles, catalytic functions and regulatory functions, and are essential to growth and repair mechanisms in vivo, such as promotion of bone formation [247,248]. Initial work by Yamaguchi et al. used a mixture of Ca(CH 3 COO) 2 and Zn (CH 3 COO) 2 at 40°C for 24 h to exchange Na + ions in NaTiO 3 . The initial Zn-containing CaTiO 3 (sodium hydrogen titanate was isomorphously transformed into zinc-incorporated calcium hydrogen titanate, Zn x Ca y H 2-2(x + y) Ti 3 O 7 ) exhibited apatite nucleation after SBF submersion (3 days) but had low scratch resistance (0.9 mN). Again, heat treatment (600°C; forming Zn x Ca 1-x Ti 4 O 9 ) improved scratch resistance (41.7 mN; likely due to the dehydration and further crystallisation), but reduced apatite formation in SBF. Acetic acid treatment returned apatite formation, via the partial exchange of Ca 2+ , which was not possible with water, due to the presence of Zn. This acetic acid-treated Zn-containing CaTiO 3 released both Ca 2+ and Zn 2+ ions, though Zn release was very low at 0.03 ppm after 12 days while Ca release was higher at 0.24 ppm, and restored apatite nucleation in SBF. It was suggested that the Zn ions present inhibited Ca release from the titanate surface, and hence inhibited apatite formation [248].

Antibacterial titanates
In addition to bioactive ion incorporation into alkalititanate structures, antibacterial ions have also been considered to negate issues regarding implant infection rates, while avoiding antibiotics overuse. Additionally, it allows scope for generating multifunctional interfaces, whereby bioactive and antibacterial properties are combined in one surface [249].

Silver titanate
One of the main antibacterial ions, which is widely prevalent in the literature, is silver (Ag) [250,251]. Studies by Inoue et al., Lee et al. and Kizuki et al. investigated the ion-exchangeability of Ag + with Na + in the titanate structure [224,252,253]. The initial study by Inoue et al. described the formation of titanate nanotubes through 10 M NaOH treatment at 160°C for 3 h, followed by heat treatment at 300°C for 1 h. The nanotubes were submersed in 12 mL, 0.05 M silver acetate, generating nanostructured silver titanates with loaded silver nanoparticles (metallic colloids; see Figure 22).
The surfaces produced also generated an antibacterial effect that was shown to be due to Ag-ion elution which, due to the faster elution speed of Ag + (82 ppm for Ag + v. 7.9 ppm for Ag(0)), could possibly be more potent against Multidrug Resistant S. aureus (MRSA) than metallic silver alone [252]. Kizuki et al. expanded the study by Inoue et al. to produce a silver titanate layer, on Ti and a Ti-15Zr-4Nb-4Ta alloy, without forming metal colloids on the surface, as well as investigating the effect on apatite-forming and bone-bonding abilities of the resultant products [224]. All samples experienced an AgNO 3 solution step at varying molarities (0.01-25 mM) following previous NaOH, CaCl 2 (optional), and heat-and water treatments. Despite excellent in vitro antibacterial activity against a wide range of bacterial types (S. aureus, E. coli, etc.), in vivo activity was insufficient due to the very low incorporation of Ag into the Ti surface as a result of its cytotoxicity [254][255][256][257].
A different approach by Shuai et al. [258] involved the functionalisation of polydopamine-coated BaTiO 3 nanoparticles with a silver ammonia solution. The cobenefits of the piezoelectric properties of BaTiO 2 , as well as the antibacterial potential of Ag, make this an attractive future research area ( Figure 23).

Gallium titanate
Gallium (Ga) has been previously reported to be an ideal substitute for Ag in the antibacterial setting [259,260]. Its history in the field of medicine is expansive, having been notably used in chemotherapeutic drugs [261]. Its similarity to Fe(III) in ionic radius and charge, allow replacement within target molecules, which has resulted in an ideal antimicrobial agent, whose presence can cause Ga(III)-induced bacterial metabolic distress through a 'Trojan horse' mechanism ( Figure 24) [259,262]. Furthermore, inhibition of bone resorption through reduced Ca 2+ release from bone makes it an ideal element for incorporation in orthopaedic devices [263]. Cochis et al. incorporated gallium into the surface of titanium through electrodeposition [260]. These surfaces are antibacterial against Acinetobacter Baumannii (MRAB12); a multi-drug resistant (MDR) nosocomial pathogen, which is rapidly emerging in implant-associated infections, with a higher efficacy compared to silver [264].
Only three biomedical studies have focussed on the incorporation of gallium into the titanate structure for antibacterial applications. Yamaguchi et al. [249] demonstrated the successful incorporation of gallium into the calcium titanate structure through a mixture of CaCl 2 and GaCl 3 , as well as purely GaCl 3 treatments (with additional water treatments post-heat treatment), generating Ga-doped calcium titanate and gallium titanate surfaces, respectively. Both Gacontaining calcium titanate and gallium titanate surfaces killed A. baumannii, with Ga 3+ release rates of 0.35 and 3.75 ppm, respectively, while also producing apatite in SBF. A. baumannii, however, is susceptible to Ga ions, requiring only 2-100 μM to produce an inhibitory affect compared to S. aureus (0.32-5.12 mM) [259,265].
Wadge et al. [266] assessed the potential viability of Ga-ion-exchanged titanate, produced using a 4 mM Ga(NO 3 ) 3 ion-exchange solution (more biocompatible compared to GaCl 3 ; FDA approved for cancerassociated hypercalcemia treatment), against S. aureus, a less susceptible pathogenic species. The Ga titanate structures tested would not leach enough Ga 3+ ions (4-40 μM) into solution to have an antibacterial effect on S. aureus (Figure 25). To demonstrate both a bioactive and broad-spectrum antibacterial effect requires further work. It was also found that without subsequent heat treatment, gallium titanate surfaces present a cytotoxic effect on MG63 cells (cell viability 24.2%), due to the higher release rate of Ga 3+ ions into the surrounding solution; heat treatment was necessary to enhance the coating stability.
Similar to the study by Wadge et al., whereby Ga (NO 3 ) 3 was used, Rodríguez-Contreras et al. [267] highlighted the further potential of using non-toxic Ga(NO 3 ) 3 at higher concentrations; 100 mM, to produce gallium-doped calcium titanate coatings on porous (macroporosity of 347 ± 1 μm and a microporosity of 8.6 ± 0.2 μm) additively manufactured Ti structures ( Figure 26). It was found that the higher gallium nitrate concentration resulted in the formation of a thicker apatite layer, indicating accelerated nucleation. Furthermore, no cytotoxicity (cell viability >70%) was seen for both the 5 and 100 mM treated samples. The gallium-treated samples, irrespective of Ga concentration used, resulted in an inhibition halo for P. aeruginosa. However, no inhibition halo was noted for S. aureus or S. epidermidis, while E. coli required the highest concentration (100 mM Ga(NO 3 ) 3 ) to exhibit an inhibition halo. Therefore, it is still not clear whether Reprinted from Shuai et al. [258] with permission from Elsevier. Figure 24. Diagram demonstrating the similarity between Ga and Fe ions and how the 'Trojan Horse' mechanism functions. Due to the similarities shown, and the process by which both ions are transported via transferrin, results in bacteria being unable to distinguish between Ga 3+ and Fe 3+ . Once the ions have been transported, the inability for bacteria to reduce Ga 3+ , which is the normal process for Fe 3+ in order to generate energy via metabolic processes, results in Ga-induce metabolic distress, and eventually bacterial death.
Ga-doped titanates are capable of broad-spectrum antibacterial effects, compared to Cu or Ag.
Zinc and other heavy metals Ag and Ga are heavy metals with high atomic weights and are believed to elicit an antimicrobial effect due to their interaction with proteins. There is evidence that heavy metals bind to sulphur atoms in cysteine molecules (an amino acid commonly present in a wide variety of proteins) or to amine groups which causes an 'oligodynamic effect' [268]. Due to the large size of the binding heavy metal ion, the shape of the protein becomes distorted and cannot perform the same biological functions [269]. Zinc (Zn) and copper (Cu) are both heavy metal ions that have also been investigated as potential antimicrobial ions in biomaterial contexts. ZnO-TiO 2 systems are prevalent in the chemical industries in gas sensors, catalysts and anode materials [270]. The delivery of zinc ions has been investigated in attempts to demonstrate antimicrobial effects; either as zinc nanoparticles, zinc oxide or zinc salts [271,272]. Zinc oxide is used extensively in bone cements and periodontal dressings. ZnTiO 3 has been proven to have bactericidal effects against Escherichia coli by Stoyanova et al. when prepared as sub-micron sized particles [273].
Copper is also widely known as both an antimicrobial and antiviral (especially pertinent with the current SARS-CoV-2 (COVID-19) pandemic [274]) ion which has successfully been incorporated into titanate nanotubes using ion-exchange methods. Compared to Agdoped titanate nanotubes, the antimicrobial activity of the Cu-doped titanates was effective against a smaller range of pathogens but the samples did successfully release Cu ions which were able to elicit antibacterial effects [275].

Iodine-doped titanate
Iodine's broad antibacterial spectrum makes it ideal for generalised disinfectants and is regularly seen in surgical settings as a topical disinfectant under the name Povidone-iodine (PVP-I) [276]. Up to 10.5% could be loaded into a calcium-doped titanate surface, with sustained release of 5.6 ppm over 90 days, >99% reduction over the same time period (reduced to 97.3% after 6 months demonstrating some bacterial regrowth) and produced no cytotoxicity while simultaneously generating apatite with 3 days in SBF [277].
Ikeda et al. [278] also demonstrated iodine-doping into a calcium titanate surface (5M NaOH, 60°C, 24 h; 100 mM CaCl 2 , 40°C, 24 h; 600°C heat-treat, 1 h; 10 mM ICl 3 , 80°C, 24 h). From the XRD, a shift in the calcium titanate spectral peaks indicates the incorporation of I into the crystal structure. It was also shown that ALP, Ocn, Opn, Integrin β1 and Col1a1 expression, as well as MC3T3-E1 cell proliferation, viability and morphology was not significantly affected due to the presence of iodine; however, iodine-doping significantly reduced the number of S. aureus (methicillin-susceptible: MSSA) compared to CaTiO 3 alone in vitro: ca. 0.13 ± 0.18 v. 6.1 ± 4.1 (CFU x10 4 ), respectively, at 24 h. After 16 weeks of implantation, both bone-implant contact and failure load were significantly better for the iodine-doped titanate (ca. 20% and 20 N) compared to pure Ti (ca. 5.5% and 2 N), but had no significant difference compared to CaTiO 3 alone (ca. 26% and 24 N), indicating no adverse effects. An in vivo bacterial assessment demonstrated pus and thick granulation tissue for the pure Ti (5.7 ± 0.35 (CFU × 10 4 /mL)), which was not observed in the iodine-doped group (0.09 ± 0.06 (CFU × 10 4 /mL)). The iodine-doped sample also did not induce any renal dysfunction, nor affect thyroid hormone levels. The initial preliminary trials for iodine-doped calcium titanates are extremely promising and demonstrate a step forward in multifunctional titanate structures.

Limitations on clinical deployment of antibacterial titanates
Despite many successful studies demonstrating the antimicrobial efficacy of loaded-titanate structures, limitations still exist, jeopardising their commercial exploitation. To date, there is no commercially available antimicrobial surface for orthopaedic and dental applications. Translation from 'bench-to-bedside' is complex and requires the navigation of significant hurdles (cost, regulatory, ethics) [279]. Presently, getting a novel drug from first testing through to FDA approval and finally to market, takes more than 13 years, and 95% of drugs entering human trials fail. The usual issues are due to misleading, non-reproducible claims of efficacy of the initial pre-clinical trials. Values showing significant differences (p < 0.05) are indicated with an asterisk. Reprinted from Rodríguez-Contreras et al. [267] with permission from Elsevier.
Particularly in in vitro trials, where failure to replicate the implanted environment, as well as understanding not only the effect of the material but also unexpected side effects, is a significant issue that inhibits further clinical efficacy. In the case of antibacterial titanate structures, there are several possible limitations that can prevent its further translation. The antibacterial properties of the material rely on elution of cations, so there is a finite concentration that can be achieved pre-implantation, and therefore can only generate short-term antibacterial effects often with a burst release of the active cation upon implantation. There is also potential for encapsulation due to the deposition of a protein conditioning layer or the antibacterial release may not prevent biofilm formation. Such layers may inhibit the effective release of the active agent, reducing its efficacy further.
The authors postulate that the formation of transient degradable coatings and/or protective coatings, in combination with titanate surfaces, can address these issues. If a degradable coating is applied on top of the titanate surface, not only can this be modified to release additional antibacterial agents, but also the transient nature of the coating will inhibit bacterial attachment in the short-term. Once degraded, the titanate surface will be exposed, enabling longer-term multifunctional cation release, which can sustain the antibacterial effects, while improving bioactive properties. This is further discussed in the future outlook section (Combinatorial material approaches for a wide-range of applications). Another approach is to directly load antibiotics into the sodium titanate structure, Yilmaz and Türk [280]. They demonstrated that following the NaOH treatment (10 M, 60°C, 24 h) of pure Ti plates, a gentamicin sulphate solution (50 mg mL −1 ; a broad-spectrum aminoglycoside antibiotic, which targets both Gram-positive and -negative bacteria) could be used to load the titanate surface. Cell viability remained high (85%) for the antibiotic-loaded sample, with apatite nucleation measured in as little as 3 days, as well as ca. 90 wt.% of the drug being released within 125 min (57% after 15 min). The rapid release inhibited bacterial invasion and prevented early contamination of the sample. For long-term usage, not only is burst release of the antibiotic critical to inhibiting biofilm formation but also retention of some of the drug over longer periods is necessary to inhibit subsequent infections. Therefore, antibiotic-loaded-titanate structures are a promising area for trialling long-term antibacterial studies.

Critical comment
It is clear that a number of possibilities exist regarding the use of substituted-titanate structures for orthopaedic implant applications. However, not only is there a complex interplay between various ions in the extracellular environment, which are continuously being replenished/controlled, but also the presence of various proteins, cells, bacteria and micromotion during movement. It is also essential to understand how various ions affect cells and bacteria, depending on their concentration, speed of elution and potential for reacting with other ions in solution to form compounds and complexes, and in extension, their effects once formed. Usually, there is a trade-off between killing bacteria and having cytotoxicity towards cells; bacteria tend to be more resistant than cells, and hence to attain bactericidal levels, some cells will die, so finding an acceptable window, if that exists, is key.
Co-release or multi-release coatings, where each doped element possesses a different mechanism or mode of action, offer a significant advantage over single-release coatings due to reduced bacterial resistance, and potentially synergistic effects of the doped elements. Yamaguchi et al. [21,236,248] have highlighted the potential for such co-doped titanate structures, such as Mg, Zn or Sr being modified into the CaTiO 3 structure to allow dual-bioactive ion potential. Similarly, doped CaTiO 3 with Ag and Ga enable the potential combinatorial approach of bioactive and antibacterial properties [249], opening up scope for multifunctional, tailorable biomedical surfaces. The authors suggest that multi-layered, multi-component structures offer a broader, multifunctional, and more promising option to the present structures, as suggested in Section Limitations on clinical deployment of antibacterial titanates and discussed further in Section Combinatorial material approaches for a wide-range of applications. Silver has long been the most successful and widely used antibacterial element in modified coatings; however, gallium or iodine show promise to provide broad-spectrum antibacterial properties, with combined bioactivity. By combing these elements in titanate structures, which offer enhanced surface area due to their nanoporosity, with exfoliating layers such as Mg or similar, multiple doped coatings can be utilised to meet the challenge of perturbing bacterial growth, increased capacity and longer-term antibacterial action and improved bone bonding.
If antibacterial titanate structures are to be successful, they must address the issues raised in Section Limitations on clinical deployment of antibacterial titanates of long-term antibacterial efficacy, while ensuring limited burst release of the active agent. This is critical to attaining clinical success. To achieve the understanding of how titanate structures, with additional materials/drugs, behave in vivo requires in vitro use of co-cultures of bacteria and cells and the presence of appropriate proteins. Additionally, understanding how combinations of various bioactive and antibacterial cations affect the chemical, structural and biological properties of the titanate surface is key to finding optimal modes of action against various bacterial types. Figure 27 has been included as it summarises the bioactive, antibacterial and multifunctional cations, which can be utilised in combination to generate effective multi-ion surfaces for a broad range of medical applications.

In vivo trials & subsequent clinical deployment
There have been a small number of in vivo studies of implanted alkali-titanate surfaces of biomaterials, demonstrating their clinical potential.

Hip stems
Yan et al. studied the effect of NaOH (4 M, 60°C, 24 h) and heat-treated (600°C, 1 h) Ti implants ( Figure 28) through implantation into rabbit tibiae, which were harvested 4, 8 and 16 weeks post-implantation [281]. At 16 weeks, the failure load of the treated implants was ca. 45 N, compared to 14 N for untreated Ti; a marked increase in bone bonding, with no adverse tissue response. There was no significant difference between alkali-treated samples and SBF apatiteformed Ti samples, indicating that the titanate samples can generate apatite in vivo and are equivalent to in vitro grown apatite surfaces. A similar study by Fujibayashi et al. also exhibited corroboratory results [282].
Regarding total hip replacements, the alkali treatment (5 M NaOH, 60°C, 24 h; followed by 600°C, 1 h heat treatment) has been employed on a Ti-6Al-2Nb-Ta acetabular shell and femoral stem. Implantation of 70 prostheses in 58 patients was performed at two different university hospitals between 2000 and 2002. Follow-up of these uncemented hip replacements showed that none of the implants required revision during an average follow-up period of 57.5 months (ca. 4.8 years) [288]. The average JOA score (Japanese Orthopaedic Association score: calculated from pain (/40), range of motion (/20), ability to walk (/20), ability to carry out daily activities (/20), which is quantified by an orthopaedic surgeon; however, it is unclear whether an independent surgeon was used [289]) improved from 46.9 for preoperative assessments, to 91.0 at the final follow-up; a significant increase. No observed osteolysis occurred in the 70 implanted hips, with all gaps closing within 12 months radiologically. These results demonstrated the efficacy of this surface modification in a clinical setting during a relatively short observation period [288]. The devices were made commercially available in 2007 [16].
Longer-term survival rates (min. 8 years; average 10 years) were reported by So et al. [290] and Kawanabe et al. [288]. The overall revision rate was 98% at 10 years, with no radiographic signs of loosening in any of the retrieved implants, as well as no radiographical gaps in any patient 12 months post-operation; survival rate was defined as revision occurring for any reason, with 2 patients requiring revision at 2 weeks and 8 years post-operation due to a periprosthetic femoral fracture and infection of the femoral component, respectively. Bone was found histologically in the pores of the implant within the first 2 weeks (retrieval of a fractured implant), with deep bone bonding in the pores by 8 years from the other implant. Limitations do exist, however, as alternative studies on total hip arthroplasties have shown, as highlighted in Table 4, that the samples size and study length should be increased to confirm alkali-treated implant superiority [290].

Spinal fusion cages
Spinal fusion cages ( Figure 29) have also been considered; however, the level of osteoinductivity from alkali-treated surfaces alone was lower than necessary for such a device; indeed, most spinal devices require an autograft in order to induce bone growth. Therefore, a combinatory surface treatment was employed in order to improve the osteoconduction and osteoinduction of such surfaces for spinal fusion devices. NaOH treatments (5 M NaOH, 60°C, 24 h) subjected to water-(ultrapure, at 40 or 80°C for various periods up to 48 h) and heat treatments (600°C, 1 h) formed a Na-free titanium oxide onto pure Ti (>99.5%), which was also found to bond to living bone [301]. The premise for such a treatment is that the conversion from the sodium titanate gel into anatase should confer a more effective apatite nucleating surface, based on compositional and structural studies conducted by Uchida et al. and Wei et al. [302][303][304]. For example, apatite formation requires appropriate crystallographic matching between the apatite crystal and the material surface; the apatite (0001) plane matches better to the (110) anatase plane compared to the (101) rutile plane. Similarly, apatite formation is inhibited if the titania gel is amorphous. By conducting water treatments, a higher quantity of Ti-OH groups are found on the surface, which enhances apatite nucleation. Fujibayashi et al. discovered that such surfaces formed on porous sintered pure Ti (5 × 5 × 7 mm 3 ,  [305][306][307] and osteoconductive [308] properties. To improve the osteoinductivity further, the water treatment can be replaced with 0.5 mM HCl, in particular for porous structures, where water treatments could not remove all the sodium within complex 3D structures [306]. A comparison between water and acid post-treated samples was conducted through implantation into canine muscles, with the new bone area (BA) fraction (bone growth in total pore area) and bone incidence (number of bone-induced samples/number of implanted samples) measured [306]. BA and incidence after 3 months for water-treated samples were 0.5 ± 0.6% and 1/4, respectively, with acid-treated samples exhibiting area and incidence values of 8.3 ± 2.5% and 4/4, respectively. The acidtreated samples exhibited significantly (p < 0.05) superior osteoinductivity at 3, 6 and 12 months [306]. Alternative HCl concentrations have been suggested by Pattanayak et al. to induce apatite formation, as well as differing heat treatment and/or Cl − ions adsorbed, without morphological change. These surfaces formed apatite in SBF within 1 d via a different mechanism ( Figure 30) to the described sodium titanate method (see Figure 5). Initially, the adsorbed ions (chloride and sulphates) dissociated from the surface, generating an acidic environment, which caused the Ti-O bonds to become positively charged on the surface (affirmed via zeta potential and XRD analyses). This attracted the negative phosphate ions, which following their accumulation, caused the surface to become negatively charged. This subsequently attracted the positively charged calcium ions from solution and formed an amorphous calcium phosphate. The timeline for these exchanges is suggested due to the presence of Ca and P in XPS as a function of SBF soaking time (1 and 30 min, 1, 6 and 12 h). P is seen in as little as 30 min, while Ca develops after 1 h for the heat-treated samples. Over time, this amorphous surface layer matured into crystalline apatite. Furthermore, as stated before, the acid-treated surface is not affected by humid environments; a significant issue of alkali-treated surfaces. An example of this was conducted by Kawai et al., whereby a porous Ti metal subjected to acid-and heat treatments was implanted into a rabbit femur [314]. The surrounding bone deeply penetrated into the porous structure within 3 weeks, in contrast to that of purely-acid and purely-heat-treated surfaces [314].
A canine model was set up by Takemoto et al. to establish the in vivo efficacy [316]. Aimed as a lumbar interbody fusion device, the structure consisted of a porous titanium construct (50% porosity; average pore size ca. 300 μm). Five devices were treated with 5 M NaOH (60°C, 24 h), followed by 0.5 mM HCl (40°C, 24 h), ultrapure water-(40°C, 24 h) and heat treatments (600°C, 1 h); another five constructs remained untreated. After 3 months, the radiological evaluation showed all treated devices achieved interbody fusion, while only 3/5 of the untreated did. Furthermore, histomorphometric evidence demonstrated the treated samples exhibited greater BA ingrowth percentage: ca. 16.7% v. 13.4%; as well as increased bone contact: ca. 34.9 v. 10.5. Following the canine model, five spinal fusion devices were implanted into patients in Japan between 2008 and 2009 [317]. Not only did bone union occur in all of the patients, but there was no need for autologous iliac crest bone grafting, an ideal advantage for such a device. All clinical results improved from pre-to post-operatively, with no adverse effects occurring during all follow-ups tested. However, a further, larger, longer-term study is required to fully evaluate its efficacy [317]. For example, the leaching of ions in a confined volume may rapidly alter the local pH, which may result in adverse effects on living cell activity in small pores of porous structures, which should also be assessed.

Bone screws/dental root-shaped implants
Despite their success in clinical trials, the initial consensus for wet-chemically derived titanate structures was that their application should be limited due to the lower torsional shear stress (9.5 MPa [318]) they can withstand. The torsional moment in these applications can exceed 50 lb.-in (>5.6 Nm) [319]. However, more recently, modifications have been made to the original treatment methods (5 M NaOH, 60°C, 24 h) in order to confer titanate structures onto bone screws, with the intention of withstanding the forces during insertion.
For example, Zhu et al. [320], used a pre-treatment of 2M H 2 O 2 /0.1M HNO 3 on Ti-6Al-4V screws, followed by a relatively low concentration of NaOH treatment (1 M, 383 K, 4 h). Furthermore, the subsequent calcium treatment utilised calcium acetate (0.04 M, combined with NaOH), rather than CaCl 2 or CaO. The screws were then implanted into rabbit femoral condyles, with the CaTiO 3 -coated screws exhibiting greater ALP activity and cell proliferation than the untreated screws at 7 days. After 12 weeks of implantation, the untreated screw showed no differences at the bone-screw interface, with no physical bonding. However, the CaTiO 3 -coated screw demonstrated a combination with bone tissue in as little as 2 weeks. These observations are likely due to the CaTiO 3 promoting cell adhesion and proliferation, chemical bonding between bone and the Ca-rich coating, as well as high surface roughness, increasing the bone contact area.
Wang et al. [321] modified the temperature of synthesis conditions with subsequent 5M NaOH and 0.05 mM CaCl 2 solutions both held at 160°C with dwell times of 10 and 24 h, respectively. The CaTiO 3--coated screws were compared against HA plasma sprayed screws. 12 weeks post-implantation the histomorphometric analysis showed no significant difference between the two treated screws in terms of BA percentage (78.33 ± 0.71 v. 79.11 ± 0.41% for CaTiO 3 and HA, respectively; uncoated = 42.28 ± 1.04%). However, greater strength was seen for the HA-coated screws compared to the CaTiO 3 coating at 12 weeks (ca. 40 v. 26 fixation index). The CaTiO 3 coating had nano-apertures (ca. 1-4 nm), while HA coating showed apertures with pore size between 100 and 200 μm. These larger apertures may explain the greater mechanical bonding.
Pure CaTiO 3 screws have also been investigated, such as the study by Gathen et al. [322]. In this study, four paediatric patients (2-14 years) and four cadaver tibias were used, with the comparison between CaTiO 3 , HA and SS screws. CaTiO 3 screws generated a significantly better fixation index (ca. 0.81) than both the HA (ca. 0.66) and SS (ca. 0.51) screws.
Dental root-shaped implants with a sodium titanate coating have also been investigated. Dental implant treatment requires rapid healing times for clinical loading and high rates of osseointegration, which is more specifically required in elderly patients with completely edentulous jaws. Camargo et al. [5] highlighted the potential use of alkali-titanate-treated (5 M NaOH, 60°C for 1, 2, 3, 5 and 7 d), grit-blasted Ti implants for dental repair. Results showed the formation of sub-micron-structured alkali-titanate layer.
In vitro tests showed, alkali-treated Ti surfaces had the ability to stimulate mineralisation upon soaking in SBF, while in vivo histomorphometrical analyses showed similar values for BA (BA%: 35.8 ± 6.9 v. 35.3 ± 9.0, respectively) and bone-to-implant contact (BIC%: 50.6 ± 12.8 v. 62.0 ± 15.0, respectively) for both commercially available, grit-blasted, acid-etched Ti implants and the alkali-treated counterpart, demonstrating their clinical potential. Overall, the collected studies highlight the potential for alkali titanates as a coating for orthopaedic screws and dental rootshaped implants, broadening the applicability of titanate materials.
Improving applicability: generating titanate structures onto non-Ti substrates Presently, nanoporous titanate structures generated through wet-chemical conversion have been limited in their applicability to only titanium-containing materials and alloys. In order to convey the beneficial properties of titanate structures, including nanoporosity and bioactive and/or antibacterial properties, onto or embedded in alternative materials, such as polymers, ceramics and other metals/alloys, novel methodologies are needed to be considered.
One methodology which can be utilised is coating alternative materials with a layer of Ti, which is the subsequently modified using the conventional NaOH treatment (5M, 60°C, 24 h). Wadge et al. [37] investigated this, for the first time, utilising a combinatorial approach of DC magnetron sputtering Ti coatings, which are then converted ( Figure 31). It was found that more equiaxed sputtered films generated thicker titanate structures due to the optimisation of surface area and the angle of growth of the titanate nanocrystals (see Figure 32). This methodology can potentially be applied to biodegradable fracture fixation devices.
Wadge et al. [323] demonstrated the potential application of sodium and calcium titanate thin films on biodegradable Mg substrates and measured the potential benefit to their corrosion resistance for biodegradable medical applications, such as fracture fixation devices, for example, bone plates. This study demonstrated that the conversion of Ti coatings into a calcium titanate structure increased the E corr and decreased the i corr values by 0.16 V and 0.25 mA cm −2 , respectively; exemplifying their potential and raising the need for further investigation (Figure 33).
A different approach is to use a sol-gel methodology generating sodium titanate structures on 316L SS (Devi et al. [324]). This also demonstrated a nanoporous topography, albeit with a different morphology and no chemical bond to the metal surface. The ability to generate different titanate morphologies on alternative substrates broadens the applicability of these surfaces.

Future outlook
Despite the advancements made within the medical literature to date to not only generate multifunctional  Relative correlation between the surface area of the sputtered Ti coating produced, and the relative effect on the produced titanate thickness in comparison to pure Ti metal conversion, as described by Wadge et al. [37]. The graphs at the bottom are purely a relative representation of the surface area and titanate thickness dependence. titanate structures, with antibacterial, bioactive, and/ or piezoelectric/ferroelectric properties but to also generate novel material structures/types, such as nanotubes and titanate glasses, there are still several questions that need novel solutions, as well as interesting areas which should be explored further. These are highlighted below.
Combinatorial material approaches for a wide range of applications By utilising the advantageous properties of specific titanate structures (porosity, bioactivity, ion-exchangeability, antibacterial) and combining this with all types of materials, either through coating, embedding or multilayering, with further advantageous properties (antithrombogenic, improved radiopacity, drug eluting), novel and effective combinations can be made. For example, the combination of a titanate-converted Ti material, which is doped with a drug (for example Aspirin which is used in stent treatments, or alendronate and minodronate which are used in osteoporosis treatment), with an additional coating on top (such as PEG or gelatin), has been used in studies by Mohanta et al. [325] and Yamaguchi et al. [325,326], respectively. The potential impact of titanate materials often benefits from the generation of -OH groups on their surface, which enables connection to drugs and polymeric layers. Combining this with the porosity on the surface, drug loading and mechanical interlocking between additional coatings makes titanate structures for novel applications ideal. They key issue for future studies will be to investigate which drugs can be loaded into the structure, while providing additional multilayers that will facilitate the elution post-insertion. Ensuring appropriate bonding with the titanate structure is also critical, and potentially biodegradable multilayers may provide the added benefit of disrupting bacterial attachment and subsequent biofilm formation. These properties should then be assessed in relation to bactericidal and cellular efficacy, in both in vitro and in vivo Figure 33. Schematic representation of the desired effect of titanate structures being applied to degradable Mg substrates, through the combination of DC magnetron sputtered Ti thin films and subsequent wet-chemical conversion [323]. models, as well as understanding the elution mechanism, release profile and whether environmental factors affect these.
The path to broad-spectrum titanates, standardised testing, and 'smart' coatings Most studies focus on small sets of bacterial types (e.g. E. coli, Staphylococcus aureus), and if the materials being tested can reduce/kill the bacteria, then these surfaces are branded as antibacterial, or even antimicrobial; the distinction being antibacterial is solely bacteria, while antimicrobial includes other microbes, such as fungi. This should not be sufficient to quantify a material as antibacterial and antimicrobial, since only a small subset of bacterial types are tested in each study; due to time constraints or availability of different strains. The present cut-off for most studies is sub-optimal, as well as clear differences in the type of assessment (direct v. indirect), bacteria used, additional factors (omission of co-culture, proteins, etc.) making clear comparison difficult. Expanding this further is key for appropriate standardisation of antibacterial assessment.
A review by Cloutier et al. succinctly sums up the current issues on antibacterial coatings, and the suboptimal testing regimes [327]. Multipronged approaches are key to the development of next-generation antibacterial coatings, with multiple modes of action, since every bacteria has different levels of susceptibility to each mode of action (contact killing, metabolic or membrane disruption, etc.). Most in vitro testing does not mimic in vivo conditions; co-culture with cells, cultures with multiple strains, relevant proteins to specific environments, host immune responses, and biofouling, and combining this with intended applications/environments, is similarly overlooked.
A good example of the targets and goals that should be adopted by researchers in terms of the range of assessments required to provide a more complete assessment was a recent study by Ikeda et al. [278]. This study investigated the doping of iodine in the calcium titanate structure and provided the following assessments: SEM, XPS, XRD, Raman; in vitro cell proliferation, adhesion (SEM), fluorescent immunostaining; ALP activity, gene expression (ALP, Ocn, Opn, integrin β1, Col1a1); in vitro antibacterial assessment (CFU, LIVE/DEAD, crystal violet assay, SEM); and in vivo animal testing (biomechanical detachment, histological examination, XPS post-retrieval, blood test to test renal and thyroid function, antibacterial assessment (implant site observation, CFU, histological assessment)). Despite only one bacteria type being tested, the level of investigation has the necessary detail, and if incorporated with multiple bacteria types, and potentially co-culture in vitro, would be a significant step towards a standardised process. Limitations were also presented within the study which is helpful in determining future targets. For example, larger implants and larger experimental animals should be used to suitably compare to real in vivo scenarios. Long-term antibacterial effects and toxicity need evaluation for iodine-containing titanate surfaces. Finally, antimicrobial activity against refractory infections such as osteomyelitis (a difficult-to-treat disease causing bone destruction due to repeated inflammation) was not investigated, and hence validation using a refractory infection model is required.
Other targets include broad-spectrum and application-specific testing protocols are needed to ensure that if a surface can be claimed as antibacterial, it should satisfy a wide-range of bacterial types, with specific arrays being used for medical implants that are nosocomial, i.e. specific to hospitals, and designing appropriate protocols [328]. Standardisation of these methodologies is also paramount to avoid wastage of research effort. Longer-term stability of coatings is also a severely overlook property, and possibly one of the main reasons for the absence of studies at the clinical stage.
Specific stimuli is also an exciting target area. This includes 'On-demand' release of antimicrobial agents, such as delayed, controlled and/or sustained release, which can be triggered through pH changes, increasing/decreasing temperature, contact with specific environments (e.g. blood), exogenous stimuli (magnetic, ultrasound, etc.) and/or levels of O 2 [327]. Typically, titanate structures release cations upon suspension in aqueous environments. Therefore, to impart ondemand release, combination with additional surfaces or materials to inhibit/modify the ability for cationic release, should be targeted. For example, utilising the piezoelectric properties of some titanate materials, may allow doping of the structure, which in a particular stressed orientation could inhibit release. However, once an appropriate electrical stimulus is applied, it can distort to release the required agent. This would likely be useful in sensing applications, such as for diabetic patients, which could enable localised, controllable release of a drug or specific cation.

Interlinked titanates for scaffold generation
A key direction in the field of wet-chemical titanates will be cross-linking of titanate structures to be used as bridges or interfaces between materials, particularly in the production of scaffolds. The potential for crosslinking and bonding between titanate struts has been developed by Wadge et al. [329]. Through modification (5 M NaOH, 60°C, 24 h) of Ti-6Al-4V microspheres, interlinking between microspheres was seen, with clear densification of the titanate film ( Figure 34). However, fully utilising this property in specific applications has yet to be realised. Potential future studies could investigate the production of porous scaffolds, which can be used as an implantable filler for bone defects. Further than this, if the reaction where the titanate struts interlock can occur in situ, this would allow injectable pastes which can selfassemble into defects, particularly if the titanate materials can be developed onto degradable materials (either polymeric or Mg).

Remarks and conclusion
The field of nanoporous alkaline titanate structures for medical applications has developed extensively in the past two decades, with highly promising clinical data to suggest their potential for orthopaedic and dental applications. We have highlighted the key processes that have been used to generate such structures, the development of alternative structures (nano-whiskers, titanate glasses, etc.) and materials properties (piezoelectric/ferroelectric), as well as the movement into multifunctional doped titanate coatings to address the significant burden of antibiotic-resistant infections.
Despite the numerous advances, some critical limitations, as well as potential further areas to explore, still exist. It is, therefore, critical that researchers consider these areas, particularly the formalisation of standardised, broad-spectrum antibacterial testing, longterm stability antibacterial titanate coatings, as well as combinatorial surfaces which incorporate titanate structures, in order to bring such advances into clinical practice. This can only be achieved if extensive collaboration, between clinicians, scientists, patients and regulatory bodies is realised. It is critical that translation is considered at all aspects during research, with the long-term goal being clinical deployment.
To conclude, in this review, we have demonstrated that alkaline titanate materials offer a flexible platform to produce a diverse range of multifunctional nanostructures that can be utilised to impart useful properties, such as bioactivity, piezoelectric behaviour, drug delivery and antibacterial protection. The potential to be applied to any surface through suitable interlayer and their unique ability for facile multi-element substitutions make them of significant interest. We have demonstrated that the understanding of the property, processing and performance of the material combinations is still in its infancy in terms of complex in vivo interactions. Alkaline titanate multifunctional properties allow for future innovations in addition to those listed above from corrosion and degradable materials, antibacterial surface, controlled release drug delivery, scaffold generation for biomaterials and sensors and are of interest to the biomedical industry and beyond.