Dry wear characteristics of TC21 titanium alloy at elevated temperatures

TC21 alloy is a new type of high damage tolerance titanium-based alloy, but its elevated-temperature wear characteristics such as wear mechanism and wear transition are still unknown. In present study, dry wear tests of TC21 alloy were carried out at experimental temperatures of 20 °C–300 °C under various applied loads. Volumetric wear rate was plotted against experimental temperature under each applied load to exhibit its variation trend and mild-severe wear transition. Scanning electron microscopy was used to examine worn surface morphologies. Confocal scanning laser microcopy and Vickers microhardness tester were utilized for characterizing the friction-affected microstructure and mechanical property in the subsurfaces. Four wear mechanisms, namely abrasion, adhesion, mild surface deformation and severe surface deformation, were observed. Severe surface deformation was found responsible for severe wear behavior, and it was aroused by the near-surface softening originating from dynamic recrystallization (DRX). The severe wear transition temperature was found to be decreased linearly with increasing applied load. By linearly fitting the relation between applied load and transition temperature, a critical surface temperature of 399.4 °C for severe wear transition is estimated, and it is further deduced to be the critical temperature for DRX realization of surface material during sliding.


Introduction
Titanium-based alloys are extensively used in aerospace, automobile, biomedicine, chemistry and food processing industries owing to their outstanding properties such as low density, good room and elevatedtemperature strength, excellent biocompatibility and corrosion resistance [1][2][3][4]. TC21 alloy (i.e.Ti-6Al-2Sn-2Zr-1Cr-3Mo-2Nb-0.1Si alloy) is a new type of high damage tolerance titanium-based alloy with high or moderate strength. The fracture toughness and crack propagation rate are two important requirements for the design of TC21 alloy, they are designed to be comparable to high-toughness Ti-62222S alloy [5,6].
As a α + β titanium alloy, TC21 alloy is considered for structural parts in aircraft and automobile applications, thus many aspects such as the microstructure transformations, heat treatments and mechanical properties have been studied in details [7][8][9][10]. The alloy is also considered for further engineering applications such as gear, shaft parts, engine valve and piston pin, in which friction and wear performance are also needed to be taken into account. Similar to other titanium alloys, TC21 alloy has drawbacks such as big friction coefficient, low shear strength and small strain-hardening coefficient, etc, which greatly limit its applications in tribological fields. Several surface modification technologies have been tried to address the poor wear performance problem facing TC21 alloy, such as pack boronising, carburizing and plasma chromizing [11][12][13][14]. Compared with the most widely used TC4 (i.e.Ti-6Al-4V) alloy, TC21 alloy is rarely studied in the tribological fields besides a few of investigations regarding fretting wear [15,16]. A lot of aspects of wear characteristics, including dry sliding wear properties at room and elevated temperatures, surface damage mechanisms, mild-severe wear transition Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
(hereafter referred as to 'severe wear transition') and near-surface microstructure evolution are still unknown. Among them, severe wear transition is extremely important for titanium-based alloys, because mild wear behavior is often considered to be safe, but severe wear behavior is unacceptable in the viewpoint of engineering applications [17][18][19][20]. Up to now, the formation of tribo-oxide layer and its effect on wear transition of titanium alloys are the mostly discussed. It has been often reported that tribo-oxide layers formed on the worn surfaces of TC4 and TC11 (Ti-6.5Al-3.5Mo-1.5Zr-0.3Si) alloys can significantly prevent severe wear transition at room and elevated temperatures [17][18][19]. However, the fundamental mechanism for severe wear transition without the formation of tribo-oxide layers on the surfaces of titanium-based alloys has not been revealed yet. Therefore, a comprehensive investigation on these aspects is needed, and it is helpful to evaluate tribological application and take directional measures to improve wear properties for TC21 alloy.
In this work, the wear rate variations of TC21 alloy with experimental temperature were studied, from which severe wear transition under various applied loads was identified. The wear mechanisms were confirmed by observation of worn surface morphologies. Microstructures and hardness in the near-surface regions were analyzed for several selected specimens before and after severe wear transition. An apparently linear correlation between applied load and severe wear transition temperature was also discussed, from which an important finding is concluded that severe wear transition is dependent on a critical surface dynamic recrystallization (DRX) temperature of 399.4°C.

Experimental procedures
A hot-rolled TC21 alloy bar with dimensions of j100 × 150 mm was received from Xian Chaojing Technology Co., Ltd. The chemical composition of TC 21 alloy is listed in table 1. Phase composition and microstructure were examined using an X-ray diffractometer (XRD) and a confocal scanning laser microscope (CSLM), respectively. The essential mechanical properties including compression properties and hardness were measured using a material test system and a hardness tester, respectively.
Dry wear tests were carried out on a pin-on-disc type wear test machine under various applied loads and experimental temperatures. The pin specimens with dimensions of j6 × 13 mm were machined from TC21 alloy bar. The conterface disks had the dimensions of j70 × 10 mm. AISI 5100 steel was used to make disks through the processes of machining, quenching, tempering and grinding. The average hardness of the disks was as high as HRC54 (i.e. corresponding roughly to 572HV). Before each wear test, the contact surfaces of pins and discs were polished with diamond polishing slurry to 0.4-0.5 μm Ra surface roughness and then cleaned in an acetone solution. During wear testing, pin sample was hold static against disk and applied to a constant load while disk rotated at a constant speed of 318 rotations per minute. The corresponding sliding speed on the wear track of disk was 1.0 m s −1 . Each Pin sample slid against disk through a distance of 753 m (i.e. 4000 rotations). The experimental parameters were as follow: load ranging from 10 to 50 N, experimental temperature varying from 20°C to 300°C. The chosen temperature range could cover a part of working temperature range of wear components made of TC21 alloy, such as engine valve and piston pin. A resistance split furnace was attached on the wear test machine. The friction pairs were right in the center of chamber when two parts of the furnace were closed together. Wear testing and heating can be carried out simultaneously in the furnace chamber. A thermocouple probe was inserted into the furnace chamber through a hole, which guaranteed the desired experimental temperature with an accuracy of ±5°C. Wear performance was characterized by volumetric wear rate. Volumetric wear rate could be obtained by dividing the volumetric wear loss by the sliding distance. By examining the reduction of pin length before and after wear test using a high measurement accuracy micrometer (0.0001 mm), the volumetric wear loss was calculated. Each wear rate value was an average obtained from three or four wear specimens.
Worn surfaces were observed using a scanning electron microscope (SEM), the chemical compositions on the worn surfaces were detected using an energy dispersive X-ray spectrometer (EDS) attached to SEM. Subsurface microstructures of worn specimens were examined using CSLM. Subsurface hardness of several selected worn pin specimens was checked using a Vickers microhardness tester with an applied load of 100 g and a dwell time of 10 s.

Results and discussion
3.1. Microstructure and mechanical properties Figure 1 shows XRD pattern of as-received TC21 alloy. TC21 alloy consists of two constituent phases, i.e. α-Ti and β-Ti, and their diffraction peaks were indexed. By means of integrating peak areas of (101) and (110) peaks for α-Ti and β-Ti phases, the volumetric fractions of the two phases are reckoned. α-Ti phase takes up 76.7% volume fraction, while β-Ti phase accounts for 23.3% volume fraction. Figure 2 illustrates the optical microstructure of TC21 alloy. TC21 alloy is composed of light-colored equiaxed grains of primary α-Ti phase, and surround them are the dark-colored mixture of lamellar α-Ti phase embedded in β-Ti phase matrix. The average grain size of primary α-Ti phase is around 5.5 μm, while the average width of lamellar α-Ti in the mixture is about 0.08 μm. The compressive mechanical properties and hardness of as-received TC21 alloy are summarized in table 2. Figure 3 shows the variations of wear rate with experimental temperature under different loads. Apparently, applied load demonstrates a considerable effect on wear rate, increasing applied leads to wear rate rising. It is noted there are two types of variations of wear rate with experimental temperature. The first one is that wear rate  increases slowly with increasing experimental temperature within a certain load range, then goes up continuously with a big slope as experimental temperature exceeds a critical value. The first type includes five wear rate curves under the loads of 10, 15, 20, 25 and 30 N. The second one is that wear rate is at a high level even at room temperature, e.g. larger than 35 × 10 −12 m 3 m −1 , and it rises fast in a big slope with increasing experimental temperature. The second type embodies three wear rate curves under the loads of 35, 40 and 50 N.

Wear rate variations with experimental temperature
The first type of wear rate curves presents the typical feature of severe wear transition (e.g. a mild wear stage with low wear rate and a severe wear stage with high wear rate), while the second type only exhibits the characteristic of severe wear. In addition, under applied loads of 10, 15, 20, 25 and 30 N, there is a certain critical experimental temperature for severe wear transition that can be determined from each wear rate curve, i.e. the temperature at the turning point. The critical temperature is found to be decreased from 275°C to 20°C as applied load is augmented from 10 N to 30 N. Furthermore, the wear rates at various critical experimental temperatures are found to be at low level, less than 35 × 10 −12 m 3 m −1 . These values are comparable to those wear rates also at severe-wear transition state for other titanium-based alloys such as TC4 and TC11 alloys reported by others [21,22].

Wear mechanisms
For the first type of wear with distinctive severe wear transition, for example, the wear under 10 N proceeded in such a way: at 20°C, the surface underwent abrasion and adhesion wear, which resulted in a number of grooves and scaly swelling patterns arranged along the sliding direction ( figure 4(a)). The scaly swelling patterns were formed by plastic deformation, adhesion and subsequent separation between asperities on contact surfaces, and they were not peeled off the pin surface due to low temperature and insufficient adhesion force. At 100°C and 200°C, the surface presented typical mild plastic deformation and adhesion wear, which manifested itself by the flat surface with a regular edge and scaly swelling patterns as well as spalled small plates (figures 4(b) and (c)). At these high temperatures, the plastic deformability of the studied alloy increases, but the yield strength decreases. The surface material is apt to be mildly deformed under the combination of applied load and friction force, which typically brings about a flat surface. At experimental temperature above the critical severe-wear transition temperature, namely 275°C, the surface demonstrated typical severe plastic deformation and adhesion, which yielded a smooth surface with an extruded edge and irregular adhesion scars ( figure 4(d)). The severe plastic deformation is actually a certain extent of plastic flow of surface material, in which the surface material flows along the sliding direction and finally causes a smooth surface with an extruded edge. At other loads like 15, 20, 25 and 30 N, mild surface deformation and adhesion wear also dominated the wear process at experimental temperatures in mild wear (figure 4(e)), while severe surface deformation and adhesion wear worked at experimental temperatures in severe wear ( figure 4(f)). For the second type of wear with only severe wear, the   surface was seriously damaged, which led to a smooth surface, an evident extruded edge as well as adhesion scars of different sizes ( figure 4(g)). The details of adhesion scars can be found in the high magnified image in figure 4(h). During chemical analysis of the worn surfaces using EDS, a phenomenon was found, that is, the worn surfaces were not essentially oxidzed, i.e. the content of oxygen content was detected to be negative by EDS mapping. The major elements of worn surfaces are list in table 3. For example, even at high experimental temperature of 200°C under 25 N, oxygen element reached the maximum content, only 0.55wt.%, indicating no tribo-oxide layers formed during sliding. It is well accepted that a compacted and intact tribo-oxide layer can protect the surface of titanium alloys from wear and delay severe wear transition [17][18][19]. The contents of oxygen element in protective tribo-oxide layers formed on Ti and alloys in mild wear usually are higher than 14.0% [18,23,24]. Furthermore, XRD analysis was also performed on worn surfaces at 25 N and 30 N. No tribo-oxides were found on the surfaces tested at 20°C and 200°C, as shown in figure 5. In the present study, absence of tribo-oxide layer means that wear behavior transition of TC21 alloy is different from other titanium-based alloys such as TC4 and TC11, the severe wear transition is not influenced by tribo-layer at all. According to SEM observation results, severe surface deformation wear mechanism accompanies the appearance and development of severe wear. Consequently, it actually dominates the severe wear transition.

Hardening and softening on the surfaces
According to Archard wear law, surface hardness is an important factor influencing wear loss [25]. Considering that the surfaces were mostly flat after wear testing, the hardness could be measured easily on the worn surfaces. The surface hardness was plotted against experimental temperature, as shown in figure 6. For the first type of wear with mild-severe wear transition under loads of 10-30 N, the surface hardness increased slightly before the critical experimental temperature, kept higher than 400HV, but suddenly dropped to a low level of 370-393HV.  The surface hardness level in mild wear regime is substantially higher than 348HV i.e. the hardness of asreceived TC21 alloy, indicating occurrence of surface hardening in mild wear regime, while the sudden drop of hardness and subsequent low level of hardness imply a transformation to surface softening in severe wear regime. Apparently, the softening promotes the flow of surface material and formation of an extruded edge, and finally gives rise to high wear rate. It is thus the real reason for severe surface deformation wear mechanism, and it leads to the severe wear transition. Both EDS and XRD analyses reveal the absence of O element and oxides on the worn surfaces in mild wear regime, namely oxidation is almost negligible on the surfaces. The surface hardening is thus caused by the plastic deformation, i.e. it is actually a type of strain hardening. However, the cause for surface softening is more completed than that for surface hardening. It could be unveiled by further detailed analysis on surface microstructure evolution throughout severe wear transition. Figure 7 shows the cross-sectional microstructures of several selected specimens worn in mild and severe wear regimes. In the case of 25 N loading, the severe-wear transition temperature was 75°C. At 20°C, the nearsurface material was plastically deformed towards the sliding direction, which led to a plastic deformation zone (PDZ) formed within 27.5 μm underneath the surface ( figure 7(a)). At 50°C, the PDZ extended to 33.1 μm depth underneath the surface, and a thin layer with about 1.76 μm thickness was observed in the topmost part of the PDZ, in which the plastic deformation flow lines run almost parallel with the surface ( figure 7(b)). This suggests that the thin layer suffers a large plastic strain. At 150°C, the PDZ further extend to a larger depth of about 57.8 μm, and the original equiaxed α-Ti phase grains and dark lamellar mixture of α-Ti and β-Ti phases were totally transformed into a type of fibrous structure within the depth of 0-4.4 μm ( figure 7(c)). At 200°C, the plastic deformation zone was a little thinner than that at 150°C, only had a thickness of about 55.2 μm, but the fibrous structure layer grew thicker in the topmost part of the PDZ, about 9.7 μm thickness ( figure 7(d)), in which almost no trace of the original dark lamellar mixture of α-Ti and β-Ti phases at boundaries of α-Ti phase grains could be found. This suggests that plastic deformation mostly happened in the region occupied by the mixture of α-Ti and β-Ti phases. In the case of 30 N loading, the severe-wear transition temperature was 20°C. The corresponding microstructural evolution was found similar to that under 25 N. At 20°C, a plastic deformation zone was also formed within 70.6 μm beneath the surface ( figure 7(e)). At 50°C, the plastic deformation also contained a fibrous structure layer in the topmast part. The white fibrous structure layer was much thicker than that under 25 N, about 24.3 μm thickness ( figure 7(f)). The above results show that the most significant microstructure change after severe wear transition is a white fibrous structure layer formed underneath the surface. However, the details of white fibrous structure layer have not been further confirmed due to the limited resolution of CSLM used. In order to characterize the mechanical properties in PDZ underneath surface, microhardness was measured on the cross sections of several selected specimens worn at 25 N and 30 N. Microhardness was measured along five different parallel lines on the cross sections. Each measurement of microhardness started from the position at the depth of 5 μm underneath the surface with a spacing of 5 μm or 10 μm towards the substrate until the original microhardness of the studied material was reached. Each microhardness value was actually an average obtained from five tests. Figure 8 shows variations of microhardness with depth from surface for specimens  8(a)), at 50°C, the hardness decreased continuously with depth increasing in the PDZ, and finally dropped to an almost constant hardness of about 351HV, showing a typical strain-hardening effect in the PDZ underneath the surface. This follows the strain change in the subsurface, that is, with increasing depth, plastic strain decreases and consequently the strain-hardening effect weakens. Nevertheless, at 150°C, a softening effect replaced the strain-hardening within the depth of 2-10 μm, namely a low hardness valley occurred within that depth range for the specimen tested in severe wear regime. Beyond that range, the hardness went up and decreased again with depth increasing. In the case of 30 N ( figure 8(b)), the variations of hardness in subsurfaces were similar to that under 25 N. For the specimen tested at 50°C in severe wear regime, a low hardness valley also occurred within 25 μm underneath the surface. The softened depth ranges under the two applied loads were basically in agreement with those of fibrous structure layers observed in figures7(c) and (f). This indicates that the near-surface softening leads to the formation of fibrous structure layer in subsurface and brings about the surface morphological features such as severely plastic-deformed surface and an extruded edge.

Change of microstructural and hardness in subsurfaces
Since the near-surface softening was the most dominant factor for severe wear transition, the reason for it was explored again by examining the resulting microstructure change for specimens tested at 30 N using optical microscope analysis. After careful polishing the worn surfaces for removing of a very thin layer (about 2-4 μm) from the surfaces, optical images of the microstructures immediately underneath the surface in mild and severe wear were taken. Figure 9 shows the near-surface microstructures at 20°C and 50°C. At 20°C in mild wear regime, the primary equiaxed α-Ti phase grains were found to be deformed towards sliding direction, which gave rise to light-colored α-Ti phase grains in strip or in egg shape. Meanwhile the dark-colored microstructure between α-Ti phase grains i.e. the part comprised of lamellar mixture of α-Ti and β-Ti phases apparently experienced a much larger plastic deformation, which led to a series of intensive flow lines extending along the sliding direction ( figure 9(a)). At 50°C in severe wear regime, light-colored α-Ti phase grains can be still recognized in a plastic deformation state, while a great number of fine grains also occurred in β-Ti phase matrix (figure9(b)). According to the softening and specific microstructure in the near-surface region, it is deduced that a part of β-Ti phase matrix was DRXed during severe wear. The large plastic deformation together with enough frictional heating in the near surface region activates DRX and subsequent softening of β-Ti phase matrix. Friction-induced DRX and wear transition have been previously reported in many nonferrous alloys such as copper alloys, aluminum alloys and magnesium alloys [26][27][28]. For two-phase titanium alloys, β-Ti phase frequently has a lower thermal stability than α-Ti phase, and is thus DRXed easier than α-Ti phase. For example, other previous investigations on TC4 alloy reported that β-Ti phase was DRXed earlier than α-Ti phase during hot deformation at 650°C and 700°C, and β-Ti phase had a larger volume fraction than α-Ti phase in DRXed microstructure [29,30].

Correlation between applied load and critical experimental temperature
Applied load is plotted against critical experimental temperature for severe wear transition in figure 10. Apparently, applied load decreased in a linear relationship with critical experimental temperature. The slope and intercepts of vertical and horizontal axes for the linearly fitted line are −0.078, 31.15 N and 399.4°C, respectively. The correlation between applied load F and severe wear transition temperature T can be thus  represented as equation (1) using the slope −0.07 and the intercept of horizontal axis T 0 = 399.4°C.
The intercept of vertical axis physically means that at applied load of 31.1 N, the critical experimental temperature will be 0°C, while the intercept of vertical axis implies that at an applied load approaching 0 N, the critical experimental temperature will be 339.4°C, if the surface wear process keeps unchanged without interference with protective tribolayer. Because a little applied load close to zero can only give rise to an almost ignorable temperature rising, the critical experimental temperature of 339.4°C is actually the surface temperature during wear testing. Therefore, the physical meaning of 339.4°C can be regarded as a critical surface temperature for TC21 alloy sliding at 1.0 m s −1 , that is, when the surface temperature is increased to 339.4°C under the combination of frictional and environmental heating, the severe wear transition is activated. This deduction can be further derived from the bulk surface temperature rising relationship [31], as expressed by equation (2).
where T b and T are the surface and environmental temperatures, respectively. α is the friction-heat partition coefficient for pin specimen, μ is the friction coefficient. F and v are the applied load and sliding speed, respectively. l b and A n are the average heat diffusion distance and nominal contact area for pin specimen, respectively. K mp is thermal conductivity of pin specimen. Equation (2) can be altered into another form, as expressed by equation (3).
At the severe wear transition temperature, wear rate is low and no extruded edge is formed, so pin length dreduces a little, and normal contact area is enlarged only a little. The parameters l b and A n could be regarded as approximate constants. Similarly, K mp can also be considered as an approximate constant for TC21 alloy, since the thermal conductivity of two-phase titanium alloy like TC4 alloy lies in a thin range of 6.9-8.6 W mK −1 at temperatures between room temperature and 250°C [29]. The coefficient of friction μ of TC21 alloy fluctuated around 0.40, it also can be thought as a constant. Therefore, an approximate constant K can be used to replace the parameters in the right of equation (3), from which a simplified equation (4) is given as follows: It is noted that equation (1) resambles equation (4) in structure. At the severe wear transition state, they acturally reperesent the identical physical meaning about the relation between applied load and experimental temperature. Therefore, the constant of 0.078 and T 0 of 339.4°C in equation (1) correspond the constant K and friction heating-induced surface temperature T b , respectively. Base on the findings such as the near-surface softening effect and corresponding fine microstructure generating from DRX in severe wear regim, the surface