3.1. Composition and Microstructure
The XRD analysis results of the treated samples at different boriding temperatures are shown in
Figure 5. In all processing temperatures, the dominant diffraction peaks were assigned to the TiB
2 phase and showed relatively higher intensity than those of the TiB phase. It resulted that the surface of the base sample formed a thick and dense layer of TiB
2. Similar results have been reported in previous studies [
14]. TiB
2 was the dominant phase for the samples treated at all boriding temperatures. At lower boriding temperatures, a very thin boride layer composed of TiB
2 and TiB was formed on the Ti surface. However, borided layer composition changes with increasing boriding temperature; Aich et al. [
26] noted that under the same XRD analysis conditions, the thickness of the TiB
2 layer increased with the processing temperature increasing. In addition, the TiB phase became undetectable. Here, a similar trend can be observed.
Figure 6 shows the Gibbs free energy (Δ
rG) profile of the boriding reaction in the temperature range from 300 to 2000 °C (see Equations (2)–(4) below), as calculated using the NIST-JANAF thermochemical numeric data [
27]; the values corresponding to specific temperatures are listed in
Table 1.
As is well-known, the reaction can only occur when the Δ
rG is less than zero, and the smaller the Δ
rG, the easier the reaction occurs [
28]. From
Figure 6, the Gibbs free energy values of the boriding reactions were negative, which indicates that the borides’ relative energy was lower than that of titanium and boron in the energy level distribution, and the borides could be formed with sufficient activation energy. Under the same conditions, when the value of free energy was larger, namely less negative, the reaction proceeded with more difficulty. Due to the lower Gibbs free energy required to form TiB
2 compared to TiB, TiB
2 was preferentially produced at the same boriding temperature with sufficient Ti and B elements. As shown in
Table 3, at higher temperatures, a lower Gibbs free energy for both TiB and TiB
2 formation was required, and hence, they formed more easily. Therefore, for the same products, higher boron contents are expected for samples borided at higher temperatures. From the perspective of material analysis, it is necessary to calculate the thermodynamic parameters. Thermodynamic calculations were performed on pure titanium, and the simulation results showed the types of borides that may be generated during the boriding of pure titanium in this study. During the boriding process, boron atoms diffuse from the molten salt into the titanium lattice from the surface to the substrate and react with titanium. With increasing boriding time, the number of boron atoms on the surface is higher than that in the bulk of the sample. Hence, most of the reactions between boron atoms and Ti occur on the surface. In this case, the reaction shown in Equation (2) was more likely to occur than those shown in Equations (3) and (4), resulting in TiB
2, rather than TiB, being formed on the outermost surface of the Ti sample. This is consistent with the preferential formation of TiB
2 detected by XRD at 1150 °C. Beneath the continuous TiB
2 layer, boron content was relatively low. At this condition, boron content dominates the boriding reaction. Under this condition, the Gibbs free energy of the reaction shown in Equations (3) and (4) was more negative than that of 1/2Ti + B → 1/2TiB
2 (see
Table 3) with the same boron content, which meant that boron affinity potential with titanium of TiB was larger than that of TiB
2, indicating that TiB was more easily formed than TiB
2 in the region with lower boron content in the bulk of the samples.
Figure 7 shows SEM micrographs and EDX line analysis results of cross-sections of the treated layer of the titanium substrates borided at different boriding temperatures. A light gray double-borided layer composed of monolithic TiB
2 and needle-like TiB whiskers was observed under the surface. No lack of discontinuity and adhesion between the breaker layer and the substrate was observed. This shows that the whisker-like TiB firmly fixed the boride layer on the substrate. This dual layer with higher boron content at the surface was consistent with the thermodynamic results. Previous studies [
29,
30,
31,
32,
33] showed that TiB has a B27 structure, which makes the growth rate of TiB along the [010] direction faster than along the other crystal directions. In the B27 structure, the zigzag B chain and the Ti channel without B are parallel to the [010] direction. Therefore, based on the periodic bond chain theory, TiB grows faster in the [010] direction than the other directions, such as [101] and [100], making TiB morphologically needle-like or whisker-like. TiB
2 has a simple hexagonal structure with a graphite-like B layer perpendicular to the [001] direction, which indicates that the growth rate of TiB
2 along the [001] direction is slower than that along the [100] or [010] direction. Therefore, the morphology of TiB
2 is layered perpendicular to the [001] direction. From
Figure 7, the thickness of the total borided layer (TiB + TiB
2) produced at 1150 °C in this study was about 30 µm, which consisted of a 7.5 µm-thick continuous and smooth monolithic TiB
2 layer, with the remainder consisting of TiB whiskers.
As the boriding temperature increased, the apparent improvement in the boriding of Ti was attributed to the composition and the content of the boriding mixture. The reactions producing B atoms from sodium tetraborate and aluminum in the crucible are shown below.
The TiB layer contained high-aspect-ratio whiskers inclined with respect to the substrate surface, some with a length of a few micrometers. Beneath this layer, we observed some discontinuous TiB particles, suggesting that some whiskers grew almost perpendicular to the surface (
Figure 7). Additionally, comparing the SEM images of samples produced at 1050 °C and 1150 °C (
Figure 7b,c, respectively), it was found that the thickness of the TiB
2 layers was similar. This is consistent with previous findings, where the thickness of the monolithic TiB
2 layer increased while the thickness of the TiB whisker layer decreased with increasing boriding temperature above the phase transition temperature [
34].
In addition, it can be seen from
Figure 7 that the cross-sectional morphology of processing temperatures 950 °C and 1050 °C were different. Especially TiB had no obvious needle-like structure at this processing temperature. It may be due to the different diffusion rates of B in TiB
2 and TiB. The diffusion rate of B at the processing temperature of 950 °C was much lower than that at the processing temperature of 1050 °C. The specific calculation results are described later.
3.2. Modeling of Layer Growth Kinetics
Figure 8 shows the modeling of layer growth kinetics and the relative unit cells during the B atoms’ diffusion into titanium. When B atoms diffuse into titanium, a TiB
2 layer is formed on the surface first. TiB
2 is a hexagonal crystal system, C32 structure, space group is P6/mmm, the number of molecules contained in the Bravais lattice unit cell is M = 1. The origin is at the center of symmetry, and the lattice constant at room temperature is a = 0.304 nm, b = 0.304 nm, and c =0.322 nm. The relative unit cell structure model of TiB
2 is shown in
Figure 8. As boriding time increases, B atoms diffuse to the inside to form a TiB layer. TiB is an orthorhombic crystal system, B27 structure, space group Pnma, the lattice constant at room temperature is a = 0.305 nm, b = 0.456 nm, and c = 0.612 nm. The relative unit cell structure model of TiB is shown in
Figure 6. The unit cell structure model of Ti has two types. When the boriding temperature is below phase transition temperature, Ti is α-Ti, space group is P63/mmc, and the lattice constant at room temperature is a = 0.293 nm, b = 0.293 nm, and c = 0.466 nm. When the boriding temperature is above phase transition temperature, Ti is β-Ti, space group is Im3-m, and the lattice constant is a = 0.282 nm, b = 0.282 nm, c = 0.282 nm. Based on the results of the experiment, a uniform continuous TiB
2 layer was formed on the surface, and active B atoms in the molten salt bath further diffused into the substrate through the TiB
2 layer to form a needle-like TiB phase, which formed a two-phase composite sublayer structure with the remaining Ti substrate. The structure formation of the composite boride layer was determined by the thermodynamic and kinetic processes of the reaction of Ti and B. The outermost layer was uniform and continuous, and the two sub-layers (TiB whisker layer + Ti substrate) maintained the borided layer structure in the crystallographic orientation relationship.
Diffusivities of B in TiB
2 and TiB phases were determined from the research of B. Sarma et al. [
34]. In that study, the diffusivities of B in TiB
2 and TiB were expressed with the following formula:
R = 8.314 J·mol
−1·K
−1, where
Q and
D0 values for B diffusion in TiB
2 are 187.1 kJ mol
−1 and 6.8 × 10
−8 m
2 s
−1, respectively. The corresponding values in the TiB phase are 190.4 kJ mol
−1 and 437.6 × 10
−8 m
2 s
−1, respectively. These values were determined from the research of B. Sarma et al. [
34].
Table 4 presents the B diffusivities in the boride phases with different boriding temperatures used in the present calculations. As shown in
Table 4, the diffusion rate of B became faster with the boriding temperature increasing. In addition, when the processing temperature was higher than 950 °C (such as 1050 °C and 1150 °C), the diffusion rate of B in TiB and TiB
2 was much higher than when the processing temperature was 950 °C. It is the reason why TiB had no obvious needle-like structure at this processing temperature. Comparing the diffusion rate of B in TiB and TiB
2, it was found that the diffusion rate in TiB was higher than that in TiB
2 at the same boriding temperature. Z. Fan et al. also found that the diffusion rate of B in TiB is higher than that in TiB
2 [
35]. That is the reason why a needle-like TiB layer is formed beneath the layer.
In order to study the acceleration kinetics of the growth of the boride layer near the transition temperature, in addition to knowing the diffusion rate of B in the boride, the study of the diffusion rate of titanium in B is also necessary. However, due to the limited solid solubility of B in Ti, and Ti and B easily reacting to form compounds, it is very difficult to measure the diffusion coefficient of B in the Ti substrate. This may be the reason why B diffusion data does not exist in the literature. Another question is whether B diffusion in the Ti phase occurs through interstitial or substitution mechanisms near the transition temperature. Assume that there is no interstitial diffusion of B in Ti. In addition, the higher concentration of substitution vacancies in the Ti substrate near the phase transition temperature can promote the increase of B diffusion through the substitution mechanism. In fact, previous studies have also shown [
36] that the ω phase formed near the transition temperature may be a vacancy supersaturated phase, which tends to promote substitution diffusion near the phase transition temperature. Based on these, it can be reasonably assumed that B atoms diffuse through the substitution sites in Ti at a speed comparable to Ti itself.
As we all know, the growth of the interface reaction layer is a thermally activated process, and growth rate (
k) is closely related to temperature. This temperature dependence can be described by the following equation. The specific values are shown in
Table 5.
where
k0 is a constant (a frequency factor),
Qk is activation energy for layer growth, both
k0 and
Qk are materials constants,
T is absolute temperature, and
R is the gas constant. These values were determined from the research of Z. Fan et al. [
35]. It can be seen from
Table 5 that the growth rate of TiB was higher than that of TiB
2, which is consistent with the results observed in
Figure 7 (the thickness of TiB was higher than that of TiB
2). In addition, it was also found that the diffusion rate of B atoms in the TiB phase along the tip direction was more than 45 times that in the TiB
2 phase. This result is the same as in the research of M. Keddam et al. [
37].
According to the results in
Table 5, the diffusion distance was calculated from the diffusion coefficient using Equation (10). Note that
d is diffusion distance,
D is diffusion coefficient, and
t is processing time (it was 7200 s in this research).
According to the results in
Table 6, the thickness of TiB was much greater than that of TiB
2. This is consistent with the results observed in
Figure 7. It can be seen from
Table 6 that when the processing temperature is 1150 °C, the thickness of TiB
2 is about 8 µm, which is the same as the results actually observed in this research (
Figure 7). However, at this processing temperature, the calculated thickness of TiB was much larger than the actual measured result (the thickness of TiB was about 22.5 µm in this research). This was due to insufficient boron content. As we all know, TiB is formed by the reaction of B atoms obtained from TiB
2. In this research, there were not enough boron atoms to react with titanium atoms, which led to the actual thickness of TiB being small. Namini A. S. et al. also found that the B atoms in TiB are obtained from TiB
2, which makes TiB grow and TiB
2 gradually become smaller [
38].
3.3. Surface Vickers hardness
Figure 9 shows the surface Vickers hardness of untreated and treated samples with different applied loads. The surface Vickers hardness of the untreated sample with applied loads of 0.49 N and 0.98 N was 335 HV and 253 HV, respectively. With increasing boriding temperature, surface Vickers hardness increased to maximum values of 2767 HV and 2340 HV at 1150 °C for loads of 0.49 N and 0.98 N, respectively. A previous study [
39] showed that boriding of a Ti-6Al-4V alloy formed a layered structure mainly composed of TiB and TiB
2, where the surface Vickers hardness of TiB was 2351–2757 HV, and that of TiB
2 was about 3248 HV. The coexistence of TiB and TiB
2 phases on the surface of the sample may lead to a lower surface Vickers hardness. Moreover, the surface Vickers hardness of our samples was lower than that of the hardness of the boride layers measured in the previous study. This was the result of the relatively thin boride layer on our samples, where the lower hardness of the base material also contributed to the measured surface Vickers hardness. This was evident from the surface Vickers hardness decreasing with increasing applied load; smaller loads resulted in smaller penetration depths and less contribution of the softer substrate material.
The hardness of the surface layer is determined by a series of factors, including grain size distribution, phase composition, the thickness of the layer, and the crystallographic orientation between the strengthening phase and the substrate. Thicker hardened layers result in relatively higher measured hardness, as observed for the sample borided at 1150 °C. At this processing temperature, the highest content and layer thickness of the hardest compound (TiB2) were observed, resulting in the highest measured surface Vickers hardness for this sample. It is worth noting that the surface Vickers hardness testing results here show comprehensive performance of the hardened layer on the surface, not just the hardness of the borided layer, because the borided layer generated at lower processing temperatures was too thin for surface Vickers hardness testing. The test result was inevitably affected by the Ti substrate.
3.4. Tribological Properties
Figure 10 shows the trend of borided layer thickness and wear depth with different boriding temperatures after wear testing. As shown in
Figure 10, with increasing boriding temperature, wear depth decreased, and thickness of the borided layer increased. It can be seen from
Figure 10 that the wear depth of the treated sample was lower than that of the untreated sample. In addition, it was observed that the wear depth tended to decrease as the boriding temperature increased. It was due to the thickness of the hardened layer reducing the wear as the boriding temperature increased. The wear depth decreased gradually with the boriding temperature increasing. During the designed temperature range, when the boriding temperature increased to 1150 °C, the wear depth reached the minimum value of 3.02 μm.
Figure 11 shows the corresponding SEM micrographs, EDX analysis, and 2D profilometric view of the worn surfaces of the untreated and borided samples after wear testing, from which the wear depth and width values were determined.
Figure 11a shows evidence of severe plastic deformation with heavy smearing on the worn surface of the untreated sample due to repeated force applied to the sample surface during wear testing. The observed scratches and grooves indicate abrasion effects. From 2D profiles showing the wear depths, it is clearly seen that boriding of Ti substantially improved its wear resistance under a 4.9 N loading force. A wide and deep wear track formed on the untreated sample during wear testing. A continuous wear track with an average depth of 58 μm and a width of 1240 μm was formed on the untreated sample. The rough 2D wear track seen in the untreated sample was larger and deeper compared to those in the treated samples. It was due to severe plastic deformation with heavy smearing and scratches occuring on the surface of the untreated sample during wear testing.
The wear depth can be calculated with Equation (1) by supposing that an abrasion partner ball was not worn. The results are shown in
Table 7. As shown in
Table 7, it was found that the wear depths of the borided samples were smaller than that of the untreated sample. It was due to the formation of a boride layer on the surface. The boride layer has high hardness and acts as a protective Ti substrate during wearing with the grinding ball. In addition, Duan Yu. et al. also found that the wear of the treated sample by boriding is reduced [
40]. The wear depth decreased gradually with increasing boriding temperature. Hence, the boriding of the pure Ti substrate substantially increased its wear resistance, consistent with the hardness measurements. The wear track of the sample borided at 950 °C was much deeper than those of the other borided samples, but slightly shallower than that of the untreated sample due to some increased hardening from the very thin boride layer. In this processing temperature, the binding force between the hardened layer and base material was low, and microparticles were removed from the surface during wear tests, producing the linear scratches observed in the wear scars (
Figure 11b). At the boriding temperature of 1050 °C, the worn surface showed limited deformation and smearing effects, and the wear depth was significantly smaller than in the untreated and 950 °C samples, attributed to the formation of a sufficiently thick hardened layer. Comparing the measured wear depth value of treated samples and the untreated sample, when the boriding temperature was 950 °C, the wear track of the treated sample was relatively deeper, but lower than that of the untreated sample. This is attributed to the increased surface hardness enhancing wear resistance. The borided layer was very thin, the binding force at this processing temperature was low, and microparticles would be generated to increase wear loss. This consideration is supported by the fact that linear scratches were observed in the observation result of wear scars.
When the boriding temperature increased to 1150 °C, the worn surface of the treated Ti had no large-scale deformation and abrasion powder, and the wear marks were very narrow. It is estimated that the formation of the borided layer reduces the tendency of the adhesive to wear, resulting in very shallow and smooth wear marks. Samples have not been peeled or peeled off. The wear depth clearly decreased with increasing thickness of the hardened layers. The wear resistance of the treated samples was significantly improved due to the hard and adhesive layers formed during the boriding process (
Figure 7). A hardened layer was achieved after boriding, containing a surface layer with ultra-high hardness with a slightly softer layer beneath. Such a surface layer is beneficial, especially for wear-resistant materials, as it improves the adhesion and wear resistance of the soft titanium substrate. The softer TiB layer beneath the surface layer acts as damping and prevents delamination during wear, especially under heavy force. For the conditions studied in this research, the maximum boriding temperature of 1150 °C resulted in minimum wear width and depth values of 400 μm and 3.02 μm, respectively. As shown in
Table 7, the calculated wear depth values exceeded the measured ones. This suggests that the abrasion of the ball surface occurred in addition to the abrasion of the hardened layer, resulting in a transition from mechanical wear to abrasive wear. When the boriding temperature was 1050 °C, the worn surface of the treated sample showed very limited deformation and smearing effects. As a result, the wear depth decreased significantly, which may be related to the sufficiently hardened layer being formed. Referring to
Figure 7 and
Figure 10, it can be clearly seen that the wear depth was shallower when the hardened layer was thicker. This extraordinary improvement in wear resistance of the treated sample was attributed to the hard and adhesive subsurface layer formed during boriding (
Figure 7). As pointed out above, a surface layer with ultra-high hardness was achieved on the outermost layer, with the substrate having gradually decreasing hardness. The hard surface layer increased both adhesive and abrasive wear resistance of the soft Ti substrate. The layer just below the outermost layer with relatively low hardness may have a good damping effect which prevents delamination during rubbing, especially under heavy loading conditions.
Although not clearly observed in the micrographs, it is well-known that oxidation readily occurs due to friction heating during wear testing because of the high affinity of Ti for oxygen. Therefore, we assume oxidative wear occurred in the samples at all boriding temperatures. The hardened layer can prevent the physical reaction of friction, thus reducing oxidation wear. The examination of the worn surface of the treated sample showed that there was no obvious plastic deformation, which was consistent with the extremely low wear rate. In addition, it can be seen from
Figure 11 that some oxidative wear had occurred. This was due to the friction heat generated in wear testing, causing oxidation to occur. Stott F. et al. also got the same result [
41].
The surface pressure calculated from the wear width after wear testing of the untreated sample and those borided at 950 °C, 1050 °C, and 1150 °C were 0.41, 0.44, 1.52, and 3.90 N/mm2, respectively. The surface pressure of the untreated sample was lower than that of the borided samples, and the surface pressure increased with increasing boriding temperature. This was the reason why wear width decreased with increasing boriding temperature (corresponding to higher hardness values) for the same applied force. The surface pressure values of the 950 °C treating sample and the untreated sample were of the same order of magnitude, indicating similar wear severity. This was attributed to the very thin hardened layer formed at this low boriding temperature.
The different surface pressures observed for the samples prepared using different boriding temperatures are expected to affect the fricative values measured during the wear testing. Therefore,
Figure 12 shows the fricative values as a function of time for all samples. With increasing wear testing time, the fricative value first increased and then decreased to a stable value. Over time, the actual contact area between the sample surface and the ZrO
2 ball increase under fixed positive pressure, and the elastic deformation of the contact surface increases, resulting in an increase in the fricative value. When contact time is extended beyond a certain critical time, the elastic and plastic deformation of the contact surface no longer increases. With increasing contact time, the wear and plastic deformation of the contact point changes the surface morphology and the pressure state of the sample. After a certain time, the maximum dynamic friction force saturates and the wear enters a stable stage, resulting in a stable fricative value.
The average fricative values for the untreated sample and those borided at 950 °C, 1050 °C, and 1150 °C were 99.1, 94.1, 88.9, and 93.2 dB, respectively. The fricative value of the untreated sample was higher than those of the borided samples, indicating that boriding reduced the fricative value. The maximum fricative value was observed for the sample borided at 950 °C, attributed to the thinner hardened layer that allowed the formation of particles via flaking, which wear with the sample and the grinding ball. The fricative value of the sample treated using a boriding temperature of 1050 °C was lower than that of the 1150 °C sample. The higher concentration of hard TiB
2 in the 1150 °C sample (as confirmed by XRD) resulted in lower elastic deformability during wear testing. Due to the high surface hardness of TiB
2, the hardened layer was brittle. It was due to the surface hardness reaching the maximum value at this processing temperature. The higher the surface hardness, the easier the brittle fracture occurs. It was due to the weakening of the elastic deformation ability of the sample. Jahandari S. et al. also found that when the brittleness index is higher, elastic deformability is lower [
42]. During wear testing, the sample deforms due to the absorption of mechanical energy. Brittle materials absorb less energy than soft ones during wear testing, resulting in larger vibrations and higher fricative values. The lowest fricative value was observed for the sample treated at a boriding temperature of 1050 °C due to the optimal ratio of TiB
2 to TiB. Hence, this material is considered suitable for practical use, as it will induce the least vibration under wear conditions.