3.1. Structure, Morphology and Chemical Composition
Figure 1,
Figure 2 and
Figure 3 represent the XRD patterns of the bulk, and melt-spun alloys, separated by composition (Ti
60Zr
10Si
15Nb
15, Ti
64Zr
10Si
15Nb
11, Ti
56Zr
10Si
15Nb
19) and by the melt-spinning conditions, in the case of the ribbons (rotation speed, MS1 (28 m/s) vs. MS2 (36 m/s)). The structure of the bulk alloys consists of the following phases: The predominant bcc β–(Ti,Nb) phase and the silicide compounds: Nb
5Si
3 and Si
3Ti
2Zr
3. A great deal of overlap is noticed on the diffraction patterns for the bulk alloys. Hence, the exact attribution of the diffraction peaks would be problematic, also considering the similar angular position of the peaks, exhibited by these compounds. The structure of the melt-spun alloys shows a dependence on the alloy composition. A second dependence on the structural evolution is related to the variable parameters of the melt-spinning processes, i.e., rotation speed/cooling rate. A predominant (high-intensity) peak located at 2θ = 38°, attributed to the bcc β–(Ti,Nb) phase can be observed in both MS1 and MS2 melt-spinning processes. This β phase is noticed at other angular positions, suggesting that this phase is predominant. Consequently, these alloys should be considered β-alloys. Amorphous ribbons were obtained for Ti
56Zr
10Si
15Nb
19 and Ti
64Zr
10Si
15Nb
11 alloys (
Figure 3) in the MS2 process, due to the higher cooling rate. The broad bands associated with these two compositions, obtained in the MS2 process, are characteristic of amorphous materials. Considering that the germination/crystallization phenomenon does not take place, there are no planes on which the X-ray beam would diffract on. A crystal is composed of periodically arranged atoms in a 3D space. On the other hand, amorphous materials do not possess that periodicity and atoms are randomly distributed in the 3D space. The scattering of X-rays by atoms is the main cause of the appearance of these broad bands.
It is evident that the peripheral speed of the copper wheel significantly affects the cooling speed of the alloy, and consequently, of the final structural arrangement. Even if most of the variants of these alloys could not be obtained in amorphous form, the crystalline structure is evidently fine, as observed in the crystallite sizes (shown in
Table 2) estimated with the Scherrer equation for the predominant β phase. One can notice that the crystallite size for the predominant β phase is close to or smaller than 20 nm. This phenomenon of structural refinement has an observable influence on structural stability during heating, as it will be later shown. Other phases found on the diffractograms were predominantly intermetallic silicon compounds: (Ti,Nb)
5Si
3 (hexagonal lattice—03-065-3599) and Si
3Ti
2Zr
3 (hexagonal lattice—00-047-1339).
A possible explanation related to the difference in structural evolution for the samples in the MS2 batch, the fact that only two compositions could be obtained in amorphous form, it could be related to the mixing enthalpy (ΔHmix (kJ/mol)) and the relative difference between the diameters of the alloying elements atoms (δ parameter). Concerning the mixing enthalpy, the calculated values are the following: Ti60Zr10Si15Nb15 ΔHmix = −28.4 kJ/mol, Ti64Zr10Si15Nb11 ΔHmix = −33.3 kJ/mol, Ti56Zr10Si15Nb19 ΔHmix = −32.4 kJ/mol. These values are characteristic for alloys which can be amorphized relatively easily. However, if we consider the δ parameter, there are clear differences between the three variants of Ti-based alloys: Ti60Zr10Si15Nb15 δ = 8.02, Ti64Zr10Si15Nb11 δ = 8.63, Ti56Zr10Si15Nb19 δ = 8.61. Considering these calculated values, only the Ti64Zr10Si15Nb11 and Ti56Zr10Si15Nb19 alloys are in the potentially amorphous materials’ domain, while the Ti60Zr10Si15Nb15 is located outside of this region, hence the nanocrystalline structure for the latter composition would be expected, and not the amorphous one.
The melt-spun ribbons from the MS2 batch were subjected to a heat treatment in argon atmosphere, past the phase transformation temperatures noticed on the DTA curves, up to T
max = 800 °C (bellow the allotropic phase transformation between α-Ti and β-Ti, which occurs at 890 °C, when no alpha or beta-stabilizer elements are present), presented in
Section 3.4. The diffraction patterns for these samples are presented in
Figure 4. The Ti
60Zr
10Si
15Nb
15 and Ti
64Zr
10Si
15Nb
11 alloy compositions exhibit almost identical structures, compared to the ones exhibited by the melt-spun ribbons (MS1 or MS2): The intermetallic silicon compounds: (Ti,Nb)
5Si
3 (hexagonal lattice—03-065-3599) and Si
3Ti
2Zr
3 (hexagonal lattice—00-047-1339), and the bcc β-(Ti,Nb) eutectic. In the case of the Ti
56Zr
10Si
15Nb
19 composition, a different behavior is observed. At the 34.94, 37.70, 39.84, 52.31, and 75.58° angular positions, the diffraction peaks can be attributed to the hcp α-Ti phase, namely the (100), (002), (101), (102), and (112) directions. This phenomenon could be attributed to the higher Nb content, compared to the other alloy compositions. The presence of beta-stabilizers (in this case Si and Nb) lowers the phase transformation temperature between the bcc β-(Ti /Nb) and the hcp α-Ti phases.
The morphology and the chemical composition of the bulk alloys can be observed in
Figure 5,
Figure 6 and
Figure 7. The bulk Ti
60Zr
10Si
15Nb
15 alloy is characterized by a light matrix-like silicon-rich region, with embedded darker titanium-rich compounds. In the case of this alloy composition, the Ti-rich regions are Zr-depleted, as shown in
Figure 5. The bulk Ti
64Zr
10Si
15Nb
11 alloy exhibits polyhedral crystals, embedded in a lighter matrix. The polyhedral crystals are Si and Zr-rich, depleted in titanium, while the lighter phase represents the β–(Ti,Nb) compound. Similar characteristics are shown by the bulk Ti
56Zr
10Si
15Nb
19 alloy, with darker silicon-rich regions and lighter titanium-rich regions. The chemical composition for each region is reported in
Table 3.
There are clear differences in terms of morphology between the alloy compositions. However, due to the complexity of the quaternary system, it would be hazardous to give definitive conclusions.
The discussion related to the alloy structural development relies on the thermodynamics of the system. Considering the relative novelty of this quaternary composition of Ti-based alloy, the quaternary phase diagram could not be found in the literature. However, if one looks at the binary phase diagrams, some predictions could be made. From the Ti–Zr [
29], Ti–Nb [
30], and Zr–Nb [
31] binary phase diagrams, one can notice that mostly solid solutions are formed between these elements, regardless of the composition and temperature. The presence of solid solutions was evidenced by the XRD analysis, from the bcc–β-Ti,Nb- and the bcc–ZrTiNb-attributed diffraction peaks. The addition of Si to the ternary Ti–Zr–Nb system makes the system significantly complicated. Several silicide compounds can be formed, according to the Ti–Si [
32], Zr–Si [
33], and Nb–Si [
34] binary phase diagrams, namely: Si
2Ti, SiTi, Si
4Ti
5, Si
5Ti
3, and SiTi
3 from the Ti–Si phase diagram; Si
2Zr, α-SiZr, α-Si
4Zr
5, β-SiZr, β-Si
4Zr
5, Si
2Zr
3, Si
3Zr
5, SiZr
2, and SiZr
3 from the Zr–Si phase diagram; and Nb
3Si, α-Nb
5Si
3, β-Nb
5Si
3, and NbSi
2 from the Nb–Si phase diagram. The formation of these silicide compounds depends to a great extent on the enthalpy of formation. The diffraction patterns contain peaks for Me
5Si
3-type silicide compounds, where Me = Ti, Nb, Zr. The enthalpy of formation at 298K for these silicide compounds is: Ti
5Si
3 ΔH
S = −573 kJ/mol [
35]; Zr
5Si
3 ΔH
S = −625.4 kJ/mol [
35]; Nb
5Si
3 ΔH
S = −62.2 kJ/mol [
36]. Thus, the formation of the Ti
5Si
3 and Zr
5Si
3 phases should be more favorable. It could be said that the formation of these silicide compounds relies on the competition between the Me elements (Me = Ti, Zr, Nb). Considering the relatively close value for the enthalpy of formation between the Ti
5Si
3 and Zr
5Si
3 phases, the presence of the Si
3Ti
2Zr
3 compound can be explained, as seen on the X-ray diffraction patterns presented in
Figure 1,
Figure 2 and
Figure 4.
Another aspect that might influence the change in morphology as a function of the chemical composition of the alloys could be related to the particularity of the binary Me–Si phase diagrams, in respect of the very narrow domains associated to the formation of silicide compounds, especially the ones evidenced by the X-ray diffraction structural analysis. The titanium content in the Ti–Zr–Si–Nb alloys, correlated to the titanium content in the Ti–Si binary phase diagram, shows that the three proposed compositions, namely Ti
60Zr
10Si
15Nb
15, Ti
64Zr
10Si
15Nb
11, and Ti
56Zr
10Si
15Nb
19, could be associated to three distinct domains. The Ti
5Si
3 silicide compound appears for titanium concentrations between 60.5 and 64.5 at. % [
37], from the liquid state, at 2393.9 K [
38], while the Ti
5Si
4 peritectic appears at a concentration of titanium equal to 55.6 at. % [
37], from the liquid+Ti
5Si
3 phase, at 2192.6 K [
38]. Thus, the Ti
56Zr
10Si
15Nb
19 composition would be located closer to the peritectic transformation; however, the formation of the Ti
5Si
4 would not occur due to the lower Si content. Moreover, the Ti
60Zr
10Si
15Nb
15 composition would be close to the formation of the Ti
5Si
3 silicide, while the Ti
64Zr
10Si
15Nb
11 composition would be positioned inside but towards the end of the Ti
5Si
3 domain.
Consequently, considering the chemical composition presented in
Table 3, as well as the hypotheses presented previously, the morphological regions associated to the EDS spectra could be summarized as follows: For the Ti
60Zr
10Si
15Nb
15 alloy composition, spectrum 1 represents the solid solution composed of Ti+Zr+Nb, while spectrum 2 represents a mechanical mixture of silicide-type phases; for the Ti
64Zr
10Si
15Nb
11 alloy composition, spectrum 1 represents a mixture of silicide phases, while spectrum 2 represents a Ti+Nb solid solution mixed with silicide phases; and for the Ti
56Zr
10Si
15Nb
19 alloy composition, spectrum 1 represents silicide phases, while spectrum 2 represents a mixture of solid solution and silicide phases.
The morphology and chemical mapping for silicon, for the melt-spun ribbons obtained with 28 m/s peripheral speed, for all three compositions, is shown in
Figure 8. The polyhedral crystalline structures, noticed in the case of the Ti
64Zr
10Si
15Nb
11 and the Ti
56Zr
10Si
15Nb
19 bulk alloys inside the eutectic matrix, are no longer visible in the case of the melt-spun ribbons, however silicon-rich needle-like grains are visible. This difference in structural development is a direct cause of the increased cooling rate, compared to the cooled copper crucible cast bulk alloys. Due to this increased cooling rate, the nucleation and crystal growth is hindered. Further increasing the cooling rate, as is the case for the MS2 batch of alloys, leads to either nanostructured melt-spun ribbons, where some grains can be observed after mechanical polishing (
Figure 9) (Ti
60Zr
10Si
15Nb
15 alloy), to featureless/amorphous melt spun ribbons, where no discernable chemical composition variation could be observed, which is to be expected, considering that the nucleation is entirely suppressed (Ti
64Zr
10Si
15Nb
11 and the Ti
56Zr
10Si
15Nb
19 alloys).
3.2. Mechanical Properties
Considering that these alloys would be intended for biomedical applications, where the mechanical characteristics are just as important as the biocompatibility and the corrosion resistance, some of the mechanical properties (hardness, Young’s modulus) were assessed by instrumented indentation. The variation of these two characteristics can be observed in
Figure 10 and
Table 4, as a function of the chemical composition and the processing parameters.
One particular requirement for alloys which should be used in orthopedic applications is that the material should exhibit a significantly reduced Young’s modulus, as close as possible to the one of the adjacent bone tissue. Depending on the type of bone structure, Young’s modulus can vary significantly: For trabecular bone, between 10–14 GPa, while for cortical bone, between 18–20 GPa [
39]. Moreover, the bone modulus varies in magnitude depending on the direction of measurement [
1]. The observation of other authors, that with increased Nb content, the elastic modulus is lower, applies to our findings, as well. In terms of hardness, a growth of the hardness values compared to the bulk material can be observed for both MS1 and MS2 ribbons of Ti
64Zr
10Si
15Nb
11 and Ti
60Zr
10Si
15Nb
15 alloy, while the alloy with a lower titanium concentration rather shows a decrease of the hardness values (
Figure 10). However, one must consider the difference in measuring error, especially for the bulk alloys, which is the direct result of the inhomogeneity of the samples, in terms of structural constituents. The mechanical properties of these constituents were further measured by instrumented indentation on the bulk samples and the results are the following: the silicide constituent exhibits H
it 14.09 ± 1.8 GPa, E
it = 210.88 ± 13.45 GPa, while the β-Ti,Nb component exhibits H
it = 6.04 ± 0.93 GPa, E
it = 113.28 ± 12.51 GPa. The measuring error decreases with the refinement of the structure, the MS2 melt-spun ribbons, either as-cast or thermally treated, exhibiting the lowest values. A heat treatment process consisting of heating over the phase transformation temperature followed by in-furnace cooling was also applied to the melt-spun ribbons from the MS2 batch. Only the Ti
64Zr
10Si
15Nb
11 alloy presented significantly higher hardness compared to bulk material’s hardness value. Further analysis of the Ti
64Zr
10Si
15Nb
11 bulk alloy sample is presented in
Figure 11 where the loading-unloading curves for the structurally different regions are presented. The characteristics of the remaining bulk alloys are similar, hence the loading-unloading curves for the Ti
56Zr
10Si
15Nb
19 and Ti
60Zr
10Si
15Nb
15 alloys are not shown, in order to better observe the curve features.
The graph from
Figure 11 represents the variation of the applied load as a function of the penetration depth, on two distinct regions, the softer eutectic matrix (black curve) and the harder hexagonal silicide intermetallic compound (red curve). Considering the significant difference in penetration depth for the same applied load, it is obvious that the eutectic matrix (the β-phase) is significantly softer. One can notice certain deflections from the regular path on the loading curve for the intermetallic compound region, which signify “pop-in” events. These steps are an indication that one of several phenomena can occur: Micro-cracks, phase transformations, dislocations nucleation, strain transfer across grain boundaries, all due to the applied load.
If we consider the Hall–Petch (H–P) relation, which helps to conclude that with decreasing crystallite size, the strength/hardness of the questioned material increases, it seems that, concerning the MS1-series samples, the threshold is located above 21 nm crystallite size. This critical threshold concerning the crystallite size, past where the H–P relation is no longer valid, signifies that the samples from batch MS1 can be characteristic of the reverse H–P relation. Below this critical crystallite size, a further refinement of the structure leads to a decrease of the mechanical properties, as can be noticed from the variation of the crystallite size as a function of the hardness (lower crystallite size is related to lower hardness,
Table 2 and
Table 4). The reverse H–P relation suggests a shift in the dominating deformation mechanisms from dislocation-mediated plasticity to crystallite-boundary-associated plasticity, such as crystallite–boundary sliding, crystallite–boundary diffusion, and crystallite rotation.
Figure 12 represents the surface morphology of the thermally treated melt-spun ribbons, and the Berkovich indenter imprints performed in the material. One can observe that the material is homogeneous, due to the fact that the imprints exhibit almost identical shapes and sizes. Oxide traces are visible, mostly due to the polishing process. Crystalline grains cannot be observed for this magnification. Thus, the ribbons are retaining their structural refinement obtained due to the melt-spinning process, even after the heat treatment past the phase transformation temperature.
If we consider the measuring error related to the results from the bulk samples, it was shown that there is a direct correlation between the inhomogeneity of the material and the distribution of the values. Moreover, if we discuss the hardness of a material, several other factors can affect this characteristic. It might be assumed that the specimen material is originally stress-free prior to indentation. However, in many materials, stresses, tensile or compressive, may be present within the specimen as a result of processing, (temperature induced, as is the case of fast-cooling processes, thus related to the as-cast melt-spun ribbons, or thermal treatments, thus related to the heat treated ribbons) or sample surface preparation (cold working from mechanical polishing). The presence of residual stress can influence the results of instrumented indentation experiments, considering that the material recoils with different degrees (as a function of the type and level of internal stress), and therefore, it interacts differently with the diamond indenter.
In spite of the fact that the newly developed titanium alloys presented in the literature exhibit Young’s moduli relatively closer to that of the bone (the lowest value found in the literature was 14 GPa for a Ti–19 Nb–14 Zr (at. %) shape memory alloy [
40]), and consist of highly compatible alloying elements, their wear resistance under loading conditions is reported to be still very poor.
The H/E ratio, (where H is the indentation hardness and E is the indentation elastic modulus), called the elastic strain to failure, gives information on the wear resistance of the material in question. Higher values for this ratio, meaning a combination of high hardness and low elastic modulus, would confer the material’s increased fracture toughness. Furthermore, the H2/E2 ratio gives information about the elastic resilience of materials (i.e., their ability to elastically absorb energy without yielding). Moreover, the H3/E2 ratio is an indicator regarding the material’s resistance against plastic deformation. Lower values of this ratio signify a poor resistance to plastic deformation.
Observing the results from
Table 4, one can extract correlations which indicate certain predictions. It seems that the alloy variant with better wear resistance and better resistance to plastic deformation is the Ti
64Zr
10Si
15Nb
11 alloy, after melt-spinning in a vacuum and with a peripheral speed of 36 m/s, followed by thermal treatment past the phase transformation temperature and free cooling with the furnace. However, the metastable aspect of the proposed alloys would probably benefit primarily the manufacturing aspect, and to a lesser degree the final part properties, considering that the structure of the final material is affected on the one hand by the laser sintering followed by cooling (during additive manufacturing), and on the other hand by various treatments the part might be subjected to, for example, HIP (hot isostatic pressing).
3.4. Thermal Analysis
The thermal stability and crystallization temperatures were measured by differential thermal analysis. The analyses, performed in argon atmosphere, up to 1530 °C for each set of melt-spun ribbons and also for the bulk material, demonstrate that exothermal processes appear in all cases of melt-spun ribbons, in contrast with the bulk alloys, where no exothermal reaction was observed. With the increase in the percentage of the amorphous phase, the tendency to produce exothermic processes grows. Moreover, even if some of the melt-spun ribbons are not entirely amorphous, but nanostructured, they exhibit exothermal reactions, to a lesser degree, all in the 550–800 °C region. Above 1300 °C, all samples, regardless of processing characteristics, start to partially melt, as can be observed in
Figure 18,
Figure 19 and
Figure 20, for the Ti
56Zr
10Si
15Nb
19 alloy, the Ti
60Zr
10Si
15Nb
15 alloy, and the Ti
64Zr
10Si
15Nb
11 alloy variants, respectively.
The difference in behavior during heating, as a function of the cooling rate, which was obtained during the melt spinning processes, can be observed in
Figure 18 for the Ti
56Zr
10Si
15Nb
19 alloy. The MS1 melt-spun ribbon exhibits one broad peak with the maximum at 656 °C. The presence of this broad peak certifies that a certain degree of amorphization or structural refinement occurs during melt spinning, even if the cooling rate is not as high as the one obtained during the MS2 castings. In the case of the MS2 melt-spun ribbon, two exothermal peaks, located at 566 °C and 668.5 °C can be observed, the latter with a significantly higher intensity. The curve representing the variation of DTA as a function of temperature for the bulk Ti
56Zr
10Si
15Nb
19 alloy does not exhibit any exothermal peaks. This observation should be an indication regarding the stable (bulk) or metastable (melt-spun ribbons) characteristic of the alloy, as a function of the processing parameters. Over 1300 °C, the eutectic is no longer stable, thus the melting process of this phase can be inferred from the endothermal peaks, visible on the DTA curves. Similar behavior is exhibited by the samples from the Ti
64Zr
10Si
15Nb
11 alloy: the MS1 melt-spun ribbon exhibits a broad band starting with 600 °C, the MS2 ribbon exhibits a sharp exothermal peak with the maximum at 643.3 °C, the bulk alloy has a stable structure, starting to melt after 1300 °C. Regarding the Ti
60Zr
10Si
15Nb
15 alloy, the variation of the DTA as a function of the temperature is significantly different, compared to the remaining compositions. In the case of the MS1 ribbon, instead of a sharp exothermal peak, which would have been expected after 600 °C, a broad band is visible. This observation leads to the conclusion that the cooling rate during melt-spinning is not sufficient for complete amorphization, however, a certain degree of the amorphous phase is still present. This behavior is in agreement with the structural analysis.