The phase, morphology and surface characterization of Ti–Mo alloy films prepared by magnetron sputtering

Magnetron sputtering is a physical vapor deposition process which allows the formation of metastable materials because of the particle bombardment process. Ti–Mo alloys are typical biomedical b-Ti alloys because of the remarkable b-stabilizing effect and the favorable safety of the Mo element. In this paper, Ti–15 wt% Mo and Ti–30 wt% Mo alloy films prepared by magnetron sputtering are studied. The substrate temperatures are 50 C, 100 C, 200 C and 300 C, respectively, which are far below the b-transus temperatures of Ti–Mo alloys. The two films both show metastable b-Ti growth at 50 C and 100 C, but have different phase evolutions as the substrate temperature further rises. The diversity is attributed to the different stabilities of metastable b growth in the two films. The different morphology evolutions with the increasing substrate temperature between the two films are discussed. The natural surface oxidation layers of the two films are analyzed by XPS.


Introduction
Titanium (Ti) and its alloys offer a wide range of adaptable structures for biomedical applications. 1 Interest in b-Ti alloys as biomaterials was derived from the excellent combination of good mechanical properties with low elastic modulus (E) close to that of human bone (E ¼ 10-40 GPa (ref. 2 and 3)), leading to better bone remodeling and osseointegration processes compared with aand a + b-Ti alloys. 4 It is known that the properties of b-Ti alloys are greatly affected by alloying elements. On one hand, solid-solution strengthening is the most important strengthening mechanisms involved in these alloys. On the other hand, by the addition of enough b stabilizing elements (Mo, Nb, V, Fe, Ta, etc.), metastable b-Ti alloys can be produced by the fast cooling process following the heating treatment above T b (b transus temperature). A series of b-Ti alloys have been developed, such as Ti-Mo, [5][6][7][8][9][10][11][12] Ti-Nb, 13 Ti-Mo-Nb, 14 and Ti-Nb-Zr-Fe (Ta). 15 Among these alloys, metastable Ti-Mo alloys receive much attention because of the remarkable b-stabilizing effect and the favorable safety of the Mo element. According to Bania et al.,7 only 10 wt% of Mo content is needed to fully stabilize the b phase at room temperature in a quenched Ti-Mo alloy. Ho et al. 5 found that the cast binary Ti- (10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20) wt% Mo alloys show metastable b-Ti growth, while those with less than 10 wt% Mo content exhibit a martensitic transformation from the b phase to orthorhombic a 00 or hexagonal a 0 phase. Magnetron sputtering of thin lms has developed into a mature eld of physical vapor deposition (PVD) technologies. The relations between deposition parameters and lm structures are of fundamental importance because understanding these relations helps to control the lm growth. [16][17][18][19][20][21][22][23] During the sputtering process, the atom congurations on the substrate are essentially depend on two parameters: (i) the ratio T s /T m which controls the conventional heating process, in which T s and T m are the substrate temperature and the melting temperature of deposited materials, respectively, and (ii) the energy delivered to growing surface of the lm by particle bombardment (atomic scale heating), which is related to the deposition conditions, such as working pressure, substrate bias or sputtering power. [24][25][26] It is necessary to note that the atomic scale heating stimulates highly non-equilibrium cooling processes which results in the formation of metastable and unstable materials. This is one of the main reasons for the formation of special structures of the lms under particle bombardment.
According to conventional heating processes, it is reasonable to affirm that b-Ti phase would be formed in Ti alloy lms when they are deposited above T b . However, in practical cases, the deposition above T b is seldom used because of two issues. One is the interface reaction between the lm and the substrate at such high T s , the other is a subsequent stabilization of b-Ti phase to low temperatures when cooled down. To the best of our knowledge, Musil et al. 27 rstly reported the formation of metastable b-Ti alloy lms formed by magnetron sputtering. Because of atomic scale heating from particle bombardment, btype Ti-10 wt% Cr and Ti-10 wt% Fe lms can be formed when T s is below 100 C (T b,Ti-Cr $ 667 C, 27 T b,Ti-Fe $ 595 C (ref. 28)). Regent et al. 29 reported the same result for Ti-10 at% Mo alloy lms on a unheated substrate. These results show a feasible way to form metastable b-Ti alloy lms at a low T s (T s < T b ), which avoids the two issues mentioned above. However, it gives rise to another two problems. The rst is the formation of b-Ti determined by particle bombardment on the growing surface of the lms during deposition. Enough bombardments are desired to provide a sufficient energy for the adatoms to form high temperature b phase at the low T s range, while too much bombardments cause interface local heating, implantation, vacancy or disruption of crystallizations in the lms. For instance, from XRD results, Ti-10 wt% Cr alloy lms have a a-Ti growth without substrate bias, but change into an amorphous growth at À150 V. 23 Also, they could have a b-Ti growth when under proper bombardment conditions. 27 The second problem is to keep the metastable b-Ti phase in the lms during the whole deposition, which depends on alloy composition of the lms and the T s . Substrate heating could partly decrease defects, and improve lm/substrate interface bonding, crystallinity or structural density of the lms, but it may cause the decomposing of metastable b-Ti phase during the deposition. 30 Therefore, the stability of metastable b-Ti phase in the sputterdeposited lms needs to be investigated. This work is mainly devoted to the second problem.
In the biomedical eld, however, magnetron sputtering is still a novel technique. Nanostructural surfaces of biomedical Ti alloys are interesting when it comes to the bone/Ti alloy implant interfaces. The high surface energy of nanomaterials leads to desirable cellular responses and high osseointegration. 31 Ivanova et al. 32 studied the bioactivity of Ti lms with different thickness (3, 12, or 150 nm), and revealed a direct link between the nanoscale surface roughness and cell attachment of the lms. More recently, b-Ti alloy lms have attracted more attentions because of enhanced biomedical properties of b-Ti alloys compared with aor a + b-Ti alloys. Coating of less biocompatible materials with b-Ti alloy lms has became an alternative way to improve the biomedical properties. 33,34 For instance, Tallarico et al. 34 prepared biomedical b-type Ti-Nb-Zr alloy coatings on stainless steels by magnetron sputtering, and investigated the growth and surface chemistry of the coatings. Gonzalez et al. 33 studied the morphology, microstructure and compositions of biomedical b-type Ti-Nb alloy lms deposited at 200 C, with 15 at%, 20 at%, 30 at% and 40 at% Nb content, respectively. Besides these biomedical applications, b-Ti alloy lms can also serve as hydrogen storage materials because hydrogen has a high solubility in b-Ti alloys. Shi et al. 35,36 showed that Ti-Mo alloy lms have an improvement in hydrogen storage capacities compared to pure Ti lms, and the resistance against embrittlement of the lms was enhanced because of the ductility of b-Ti phase and the reduction of the hydrogenise formation. Magnetron sputtering allows the production of metastable b-Ti alloy lms, therefore, it is now prompting new researches for those applications. Controlling the metastable b-Ti growth in the lms is primarily important, however, very complex, because it is highly related to several conditions, such as T s , particle bombardment and the compositions of the lms. Understanding these is necessary for determining preparation parameters, lm compositions, and service temperatures of the lms in practical applications.
Ti-15 wt% Mo and Ti-30 wt% Mo alloys are both popular biomedical Ti-Mo alloys, 6,37 however, their as-sputtered structures have yet not been studied. In the present work, Ti-15 wt% Mo and Ti-30 wt% Mo alloy lms formed by magnetron sputtering technique are investigated, respectively, with main concerns on phase, morphology and surface characterization of the lms. T b , of pure Ti is 882 C and the approximate amount of b transus reduction per wt% of Mo addition is about 8.3 C. Therefore, T b of Ti-15Mo and Ti-30Mo can be estimated as 758 C and 633 C, respectively. 38 The T s during deposition is 50 C, 100 C, 200 C and 300 C, respectively. The metastable b-Ti growth of the two lms are obtained at 50 C and 100 C, but they show different phase evolutions as the T s further rises, which is attributed to the different stability of metastable b-Ti phase. Accordingly, the different morphology evolutions in the T s range between the two lms are shown and discussed, respectively. Besides, surface characterization is an important topic of biomedical materials because it plays a key role in the living tissue response to the metal presence. Considering that Ti and Mo elements are both chemically reactive, nature surface oxidation layers of the lms are analyzed using X-ray photoelectron spectroscopy (XPS).

Experimental
We deposited the Ti-Mo alloy lms on silicon (Si) (100) substrates by direct current magnetron sputtering. The alloy targets of two compositions were used: Ti-15 wt% Mo and Ti-30 wt% Mo, which were prepared by melting appropriate amounts of Ti and Mo powders. Hereaer, Ti-15 wt% Mo and Ti-30 wt% Mo will be simplied as Ti-15Mo and Ti-30Mo. We used high-purity (99.999%) argon as the sputtering gas. The base pressure was 5.0 Â 10 À4 Pa and the sputtering pressure was 0.8 Pa. Before starting the actual deposition, we presputtered the target for 20 min with a shutter covering the substrate. The actual deposition time of the lms was 120 min. The target-substrate separation was kept at 40 mm. The sputtering power was 180 W. The substrate temperatures were 50 C, 100 C, 200 C, and 300 C, respectively. Three samples were prepared for each condition. Aer each deposition, we kept the T s for 20 min and then we cooled the lms to room temperature in the vacuum.
We analyzed phase, texture, morphologies of the lms using X-ray diffraction (XRD) operated on PANAlytical X'Pert PRo MPD instrument (Almelo, Netherlands), Tescan MIRA3 XMU scanning electron microscope (SEM; Brno, Czech Republic), and FEI Tecnai™ F30 G2 transmission electron microscope (TEM; Hillsboro, OR) with accelerating voltage of 300 kV, respectively. Surface roughness of the lms were measured using Shimadzu SPM-9600 atomic force microscope (AFM; Kyoto, Japan). The instrument was operated in the contact mode, and the measuring range was 5 Â 5 mm 2 . Bulk compositions of the lms were detected by a SEM-based INCA energy dispersive spectrometer (EDX) with an X-MAX detector (Oxford Instruments, UK). The calibration was performed using a Ti standard, and the quantization was processed with a built-in XPP correction procedure. Surface chemical compositions of the lms were analyzed using Thermo Fisher X-ray photoelectron spectroscopy (XPS; America) with a Al Ka (hn ¼ 1486.6 eV) monochromatized focused X-ray source. The spectrometer was calibrated against the reference binding energies (BEs) of clean Cu (Cu 2p 3/2 at 932.6 eV), Au (Au 4f 7/2 at 84.0 eV) and Ag (Ag 3d 5/2 at 368.2 eV). The analyzed area is 400 Â 400 mm 2 . In addition to the survey spectrum (pass energy of 200 eV, step energy of 1 eV), the following core levels were systematically recorded at higher energy resolution (pass energy of 50 eV, step energy of 0.1 eV): C 1s, O 1s, Ti 2p, Mo 3d. The binding energy positions were determined through curve tting with the mixed 30% Lorentzian and 70% Gaussian curves and Shirley baselines using the Avantage soware. Core levels were calibrated by the standard C 1s peak at 284.8 eV. Fig. 1(a) shows typical XRD patterns of the Ti-15Mo alloy lms at different T s . It can be seen that (i) the Ti-15Mo alloy lms have a metastable b growth at 50 C and 100 C, and (ii) the orthorhombic a 00 phase was formed at a higher T s . Fig. 1(b) shows close up of the reections from 36 to 41 , in which a single b(110) peak is observed below 100 C, and a composition of b(110) and a 00 (111) peaks appears at higher T s . The (110) plane is the densest plane of b-Ti phase which has a bodycentred cubic structure. Increasing the T s is supposed to help the (110) growth of the lm because of the reduced surface energy of the lm. 39 However, the b(110) peak is decreased as the T s rises in this work, which is also found in our previous work for Ti-5Al-4Mo-4Cr-2Zr-2Sn alloy lms. 30 This is because, with the occurrence of the phase evolution, the growth of the lm is determined by a combined effect of the reduced surface energy of the lm and the interfacial energy between different phases. The a 00 phase is a well-known transition phase during the transformation from b-Ti to a-Ti. The formation of a 00 phase indicates that the metastable b-Ti phase has changed because of the activation energy provided by heating substrate above 100 C. Therefore, in Ti-15Mo alloy lms, a sufficiently low T s (no more than 100 C) is necessary to stabilize the metastable b-Ti phase during the whole deposition. We can also infer that metastable b-type Ti-15Mo alloy lms can not be subjected to an excessive service temperature, such as the hydrogenation temperature in hydro-treating, 36 to prevent the decomposing of b-Ti phase. Fig. 2(a) shows the bright eld (BF) image and the selected area electron diffraction (SAED) pattern of the Ti-15Mo lm at 100 C. The pattern further conrms the b-Ti growth of the lm at this temperature because the four rings correspond to (110), (200), (211) and (220) reections of b-Ti. The polycrystalline orientation of the crystallites shows the adatom diffusion is negative at 100 C, so the crystallization follows the random orientations of nuclei. 40 The formation of b phase suggests that the particle bombardment conditions provide a sufficient energy for the adatoms to form the hightemperature b phase. The fast dissipation of this energy in close proximity of every incident atom results in extremely fast cooling and freezing of the b-Ti phase on the growing surface of the lms during the deposition. Fig. 1(c) shows typical XRD patterns of the Ti-30Mo alloy lms at different T s . The lms could keep a b-type growth in the whole T s range. This result shows that (i) the stability of The Mo elements could not only incorporate in the b-Ti lattices, but also exist at the grain boundary in a state of amorphousness or very tiny crystallites. In the latter case, their qualities are so small that the diffraction intensity is below the detection limit.

Results and discussion
The atomic radius of Mo and Ti elements are 1.36Å and 1.47Å, respectively. As the T s rises, the enhanced solid solution of those free Mo elements into the Ti matrix tends to decrease the lattice parameter of Ti matrix and right shi the peaks. The slight peak broadening with the T s ( Fig. 1(d)) is caused by the decreasing of the crystalline size which is attributed to impurity segregations. The impurity incorporations are hardly avoidable during real depositions. At higher T s , the impurity segregations to surfaces and grain boundaries could suppress crystalline growth of the fundamental structure. 40 Fig. 2(b) shows the BF image and corresponding SAED pattern of the b-Ti growth of the lm at 300 C, suggesting certain orientation growth at this temperature.
Besides the b phase growth, another important issue of these lms is their morphology evolutions which particularly arise in the PVD process. Fig. 3 and 4 show the SEM results of surface and fracture morphologies of the Ti-15Mo and the Ti-30Mo alloy lms, respectively, suggesting that the morphology evolution of the two lms have some similarities with the increasing T s . When the T s ranges from 50 C to 200 C, they both present a relatively smooth surface morphology with small surface particles, and a columnar fracture morphology with distinct boundaries. When the T s rises to 300 C, the surface particles of two lms both grow up greatly, leading to a coarse and roughness surface structure, and the surface and fracture morphologies both become dense at this temperature. Fig. 5 shows the AFM average surface roughness (R a ) of the two lms both sharply rises at 300 C, which is in accordance with the SEM observations.
One problem is that the Ti-15Mo alloy lm can not keep the b-type growth above 100 C. In spite of its b-type growth at 50 C, Ti-15Mo alloy lm shows a poor structure with voids and open boundaries, see Fig. 3(a), which is the microstructure of zone 1 in the low T s range of the Movchan and Demchishin model. 41 This poor structure is rather friable and not suitable for practical application, and the voids absorb impurities in the lm interior which degrades the corrosion performance of the metallic lm. Thus, it is necessary to improve the growth of Ti-15Mo alloy lm deposited at 50 C. According to the structure Fig. 2 The BF image and the SAED pattern of (a) the Ti-15 wt% Mo film at 100 C and (b) the Ti-30 wt% Mo film at 300 C. zone model (SZM) from Thornton,42,43 the zone 1 growth of the lm can change into a dense and smooth non-columnar zone T growth at the low T s range in assistance of the effect of particle bombardment. The principle of this low temperature sputtering of lms with zone T growth is based on replacement of the equilibrium substrate heating by non-equilibrium atomic scale heating. Therefore, the poor structure of the Ti-15Mo lm deposited at 50 C can be improved without compromising its metastable b growth by enhancing the particle bombardment, such as decreasing the working pressure, rising the sputtering power, or using the substrate bias. Fig. 6 shows the fracture image of the Ti-15Mo alloy lm deposited at 50 C when the low working pressure 0.2 Pa and the substrate bias À100 V were used. The lm structure is signicantly improved because of the disappearance of voids and columnar growth mode, which can be dened as the microstructure of zone T in the SZM from Thornton. 42,43 When the T s rises to 300 C, the morphologies of two lms become completely different, as shown in Fig. 3(d) and 4(d). The rst difference is that the surface particles of the two lms have completely different shape features at 300 C. This can be related to the different phases of the two lms at 300 C. As a comparison, Zhang et al. 44 showed that the surface particles of b-type Ti-Cr alloy lms have similar spherical shape features under their experimental conditions, with 10 at%, 25 at%, 40 at% Cr contents, respectively. The second difference is that the columnar fracture morphology of Ti-15Mo alloy lm disappears, which is replaced by a smooth and noncolumnar one, while that of the Ti-30Mo alloy lm keeps at 300 C. The SZMs from Movchan and Demchishin 41 and Thornton 42,43 qualitatively represent the microstructure of single-phase lms with the increasing T s /T m . These models, however, strongly change when impurities or additives are incorporated in the lm. 39,40 Fig. 4 The SEM surface and fracture images and the EDX results of the Ti-30 wt% Mo alloy films at different substrate temperature.   6 The dense structure of Ti-15 wt% Mo alloy films deposited at 50 C when the low working pressure (0.2 Pa) and the substrate bias (À100 V) were used, which is accordance with the zone T growth in the SZM. This is because the impurities or additives may lead to the formation of a segregation layer 40,45 or a new phase 39,46 on the phase boundaries of the forming majority phase, which would change the growth mode of the lms. In the present work, Ti-30Mo alloy lms have a single phase in the whole T s range, and the transition from a poor columnar growth to a dense columnar growth is consistent with the transition from zone 1 to zone 2 in the SZMs. However, it must be noted that the impurity incorporation or the co-deposited Mo elements have inevitably inuenced the growth of the lm during deposition, although the columnar mode of the lm is sustained. For Ti-15Mo alloy lms, the formation of new phase disturbs the columnar growth of the lms, and leads to a structure with no distinct columnar growth. The similar results were found in Zr-Cu-N nanocomposite coatings. 47 At a low Cu content of 1-2 at%, the lm exhibits a columnar growth, while at a high Cu content of 20 at%, Cu as an additive phase, which is detected by the XRD, can block the growth of all grains already in the coalescence stage and results in ne globular structure of the lm. Barna et al. 46 investigated the microstructures of Al 1Àx Pt x alloy lms with 0 # x # 0.4 using TEM, and found the formation of Al 5 Pt and Al 6 Pt intermetallic phases on the boundaries of Al islands. Equiaxial grains of the minority phase decorate grain boundaries and triple points, thus decrease migration of grain boundaries and grain coarsening. 48 These results all show that the formation of new phase can strongly affect the development of the lm structure. Therefore, Ti-15Mo and Ti-30Mo alloy lms have different structural features at 300 C.
The EDX and XPS techniques are used to acquire bulk and surface compositions of the two lms, respectively. It is known that the EDX detection depth is up to several micrometers, while the XPS detection depth is only nanoscared from the lm surface. Typical EDX spectra and corresponding composition results of the Ti-15Mo and the Ti-30Mo alloy lms at different substrate temperatures are inserted in Fig. 3 and 4, respectively. Besides major Ti and Mo peaks, C peaks are also detected by EDX. Fig. 7 provides the XPS survey spectra of the two lms deposited at 100 C and 300 C, respectively, where the C and O peaks are both detected. These impurities in the lms mainly come from two sources. The rst is the adsorption of the residual gas of the vacuum system during deposition, which affects the crystalline growth and the texture evolution of the lms, 39 or leads to the intrinsic compressive stress in the lms. 49 The other is the adsorption of impurities when the lms are exposed in the air aer deposition. The C component is commonly from the adsorption of hydrocarbon molecules or Ccontaining gas molecules, and the O component is from the formation of metallic oxides and the adsorption of oxygen or oxygen-containing gas molecules. 34 Fig. 8 shows the typical XPS Ti 2p and Mo 3d spectra of the Ti-15Mo lms deposited at 300 C, respectively. Ti 2p spectra ( Fig. 8(a)) are tted by 4 components associated to metallic Ti, TiO, Ti 2 O 3 , and TiO 2 , while only the TiO 2 component is pronounced for all lms. Mo 3d spectra (Fig. 8(b)) are tted by 3 components associated to Mo, MoO 2 , Mo 2 O 5 . The MoO 2 component is more prominent for all lms, although metallic component and Mo 2 O 5 are also noticeable. These results show that the surfaces of the lms are naturally oxidized in the air. These partially oxidized surface layers are desirable because they could provide great corrosion protection for implant materials.
The EDS and XPS alloy compositions for the two lms at 100 C and 300 C are shown in Table 1, respectively, where the C and O components are excluded. The comparison between EDS and XPS results suggests an enrichment of Ti on the lm surface. This type of composition diversity can be caused by two reasons. The rst one is related to the different sputtering yields between Ti and Mo atoms. Ti-Mo alloy targets are used in this work. Ti and Mo atoms alternately denote the main species  which are reected from the surface of the target during deposition. Therefore, the composition of Ti and Mo atoms arriving on the growing surface of the lm alternately changes during deposition, leading to a difference between the surface composition and the bulk composition of the lms. The second reason is connected with the formation of surface oxidation layer of the lms. The Ti enrichment in the surface oxidation layers has been found for Ti-13Nb-13Zr and Ti-50Zr alloys. 50 Recently, this result was also found for co-deposited Ti-Nb alloy lms. 33 In this co-deposition process, pure Ti and pure Nb targets are operated at constant conditions, respectively. We can infer that, during the lm thickening, the composition of the Ti and Nb atoms arriving on the growing surface of the lm is constant, and the lm have a uniform alloy composition from its growing surface to the internal. As a result, the Ti enrichment on the lm surfaces can be attributed to the surface oxidation behaviours of the lms aer deposition.

Conclusions
Biomedical b-Ti alloy lms prepared by magnetron sputtering received much attentions for the use as biomedical surface modications layers. Controlling the b-Ti growth in the lms is a focused issue, which is related to the deposition parameters and the alloy compositions of the lms. Ti-15 wt% Mo and Ti-30 wt% Mo alloys are both popular b-Ti alloys, but their lms formed by magnetron sputtering have not yet been studied. This paper investigates the sputter-deposited Ti-15 wt% Mo and Ti-30 wt% Mo alloy lms when the substrate temperature is 50 C, 100 C, 200 C and 300 C, respectively. XRD results show that Ti-15 wt% Mo alloy lms could keep a metastable b growth only at 50 C and 100 C, and the b-Ti growth at 100 C is further conrmed by the SAED pattern. As a comparison, Ti-30 wt% Mo alloy lms could keep a metastable b growth at a higher substrate temperature up to 300 C, which is also conrmed by the SAED pattern. Besides the different phase formations, the two lms exhibits different morphology evolutions as the substrate temperature rises. SEM observations show that the two lms both have a zone 1 growth at the low substrate temperature range. When the substrate temperature rises to 300 C, Ti-30 wt% Mo alloy lms change into a dense and columnar zone 2 growth which is accordance with the SZMs, while Ti-15 wt% Mo alloy lms change into a noncolumnar growth, which is considered to be connected with the formation of new phase in the lms. AFM measurements indicate that surface roughness of the two lms both sharply increases at 300 C. XPS analysis show a predominance of TiO 2 and MoO 2 in the surface oxide layers of the lms. The comparison between the EDS and the XPS results shows an enrichment of Ti on the oxide surface layer for all lms.

Conflicts of interest
There are no conicts to declare.