Influence of Aluminum on the Wear Properties of High‐Carbon Metastable Austenitic Steels

The two‐body abrasive wear properties of metastable austenitic steels (MAS) against SiC abrasive paper are investigated at different wear loads. To ensure a metastable austenitic microstructure, the alloying compositions are chosen such that the martensite start temperature of the MAS is approximately at room temperature, while the proportions of carbon, manganese, and aluminum change. The abrasion test results are compared to martensitic (40MnB5) and austenitic steel (Hadfield steel). An up to four times lower weight loss is found for the MAS compared to the Hadfield steel and up to 6.7 times lower weight loss compared to the martensitic steel. It is found that the wear resistance of the MAS increases significantly with wear load. Wear resistance of over 1300 Nm mm−3 is achieved at the highest wear load of 32 N. The wear properties of the MAS are associated with an increase in the surface hardness resulting from a mechanically induced austenite to martensite phase transformation. It is shown that the addition of aluminum to the MAS reduces the wear resistance. This is explained by an increase in stacking fault energy and the associated restriction of the mechanically induced transformation to martensite.


Introduction
Extending the service life of components subject to wear is one way of reducing the consumption of materials and resources and, thus, significantly improving the sustainability of products and reducing maintenance costs. Commonly, martensitic steels are used for components under abrasive stress. Their wear resistance increases with the initial hardness, which depends mainly on the carbon content of the steel. [1,2] Hadfield steels form another group of abrasion-resistant steels. Their austenitic microstructure is particularly resistant to wear due to their high work hardening ability. However, martensitic steels-especially those having a high-carbon content-feature brittle behavior, and austenitic Hadfield steels demand high manganese concentrations and their fabrication is intricate and costly. It was shown that decreasing the manganese content in Hadfield steels improves their resistance to abrasive wear. [3,4] Lower manganese contents cause a reduction in the thermodynamic stability of the austenite, so a mechanically induced martensite formation may additionally support the hardening in the near-surface region. The improvement of wear properties in steels when austenite is retained in the microstructure is observed in numerous steels. [5][6][7][8][9] Gong et al. [6] showed in rubber wheel tests that both impact toughness and wear resistance of martensitic high-carbon steel are improved by the introduction of retained austenite in the microstructure. In their case, the retained austenite-introduced by a partitioning treatment at 250°C-transforms completely into martensite during wear. Similar results are obtained for carbide-free bainitic steels under abrasive tests. [7][8][9] Efremenko et al. [10] performed wear tests on steels containing 1.2 wt% carbon, 2.6 wt% manganese, and 1.6 wt% silicon in carbide-free bainitic and fully metastable austenitic states. The carbide-free microstructures showed good wear resistance. However, the metastable austenitic state outperformed them by showing half the wear rate. It was shown that the high resistance against abrasive wear for these metastable austenitic steel (MAS) is due to a mechanically induced transformation of the metastable austenite to martensite. The dependence of wear resistance on the wear load of such MAS has not yet been investigated. Additionally, the wear resistance under low abrasive load conditions of conventional Hadfield steels is low and is increased by the addition of aluminum. [4,11] Thus, the addition of aluminum in MAS and its effect on the wear properties should be investigated as well. Motivated by the results of Efremenko et al., [10] a calculation model for the estimation of the martensite start temperature (M s ) was proposed and further MAS compositions, which should exhibit similar wear properties, have been developed. [12] Some of them are the subject of this study.
The underlying thesis of this study is that high wear resistance is achieved regardless of the steel composition, provided that the metastability of the austenite is given. Although austenite stability depends on many factors-chemical composition, grain size, morphology, stress state, the strength of neighboring phases, and grain orientation-Wong [13] stated that the chemical composition DOI: 10.1002/srin.202200545 The two-body abrasive wear properties of metastable austenitic steels (MAS) against SiC abrasive paper are investigated at different wear loads. To ensure a metastable austenitic microstructure, the alloying compositions are chosen such that the martensite start temperature of the MAS is approximately at room temperature, while the proportions of carbon, manganese, and aluminum change. The abrasion test results are compared to martensitic (40MnB5) and austenitic steel (Hadfield steel). An up to four times lower weight loss is found for the MAS compared to the Hadfield steel and up to 6.7 times lower weight loss compared to the martensitic steel. It is found that the wear resistance of the MAS increases significantly with wear load. Wear resistance of over 1300 Nm mm À3 is achieved at the highest wear load of 32 N. The wear properties of the MAS are associated with an increase in the surface hardness resulting from a mechanically induced austenite to martensite phase transformation. It is shown that the addition of aluminum to the MAS reduces the wear resistance. This is explained by an increase in stacking fault energy and the associated restriction of the mechanically induced transformation to martensite.
is the primary factor affecting the austenite stability and that M s can be used in the first approximation to describe it. The composition of the steels was chosen so that their M s is approximately the same, while the manganese, carbon, and aluminum contents vary.
This work aims to characterize the two-body abrasive wear properties of several MAS and to compare them to conventionally used abrasive-resistant steels. Further to investigate whether these properties are coupled to certain chemical compositions, or rather the general stability of austenite.

Microstructure
The as-quenched microstructures of the MAS samples are shown in Figure 1. All MAS show similar microstructures: thermally induced martensite α 0 th (brown or dark) embedded into austenite (beige to orange). Additionally, surface martensite α 0 sur (white or light) is revealed. Differences in the microstructures of the different MAS can be seen in the grain sizes and the fractions of thermally induced martensite.

Abrasion Tests
The weight loss in dependence on wear distance at different applied wear loads is shown in Figure 2. It is evident that the overall weight loss, regardless of the wear load, is lower for the MAS compared to the martensitic 40MnB5 and the austenitic Hadfield steel. It is further shown that the addition of aluminum increases the weight loss of the MAS. Regardless of the wear load, the MAS 100 shows the lowest weight loss, followed by the MAS 120. These steels are followed by the aluminum-containing MAS in alternating order. It can be further deduced that the change in mass loss with increasing wear load is much lower for the MAS compared to the reference steels.
The wear resistance (WR) is calculated according to Equation (1) (F -wear load, d -wear distance, and V -wear volume). A density of 7.85 g cm À3 was applied for all steels to calculate the volume losses from the measured weight losses.
The dependence of the wear resistance on the wear distance is shown in Figure 3. In general, all steels tested exhibit an increase in wear resistance with increasing wear distance. However, the rate of increase varies considerably depending on the steel. The aluminum-containing MAS show nearly constant wear resistance over wear distance, which correspond to the abovementioned linear increase in weight loss. For the 40MnB5 and the Hadfield steel, the wear resistance only slightly increases with ongoing wear progress. A significantly greater increase in wear resistance is shown by the MAS 100 and 120. Figure 4 shows the wear resistance after 96 m of the tested steels at different wear loads. It is apparent that the wear resistances of all steels increase with increasing wear load. The rate of increase in wear resistance after 96 m is particularly pronounced for the MAS and is highest for the MAS without aluminum. Assuming a linear increase of the wear resistance with increasing wear load (Figure 4), the wear resistance rates, i.e., the rate of increase of wear resistance, can be determined as the slope of a linear fit. For the MAS 100, 120, 100Al, 120Al, and the reference steels 40MnB5 and Hadfield steel, rates of 30.8 AE 8, 29 AE 6, 13.8 AE 4, 12.4 AE 1, 5.1 AE 2, and 5.6 AE 3, respectively, are obtained. These values are given in Nm mm À3 per Newton of wear load. The MAS 100 and 120 show more than twice the wear resistance rate compared to the aluminum-containing MAS and approximately five times that of the reference steels. The measured wear resistance and the calculated wear resistance rate of the Hadfield steel are in good agreement with the literature. [14]

Surface Characterization
The pre-and postwear roughness parameters R a and R z of the specimens are shown in Table 1. The aluminum-free MAS and the reference steels show similar prewear roughness, while the aluminum-containing MAS exhibit a slightly lower prewear roughness. The postwear roughness after a wear distance of 96 m of all steels is in the range of 0.4-0.8 and 2.5-6 μm for R a and R z , respectively. Interference microscopy images and roughness profiles are shown in Figure 5 for pre-and postwear conditions at wear loads of 8 and 24 N and a wear distance of 96 m. In the prewear condition of the MAS (Figure 5b), distinct scoring can be seen on the surface due to the removal of the decarburized layer after hot rolling. These scores are still visible after the wear tests at 8 N wear load.
In addition to the interference microscopy images, the postwear surfaces were investigated by the means of scanning electron microscopy (SEM) imaging. These micrographs- Figure 6 shows several examples-display the same characteristics independent of the composition and wear load. After the abrasion tests, grooves of different widths and depths and build-up ridges can be seen. Furthermore, microcracking and chip delamination are observed at the ridges.

Hardness Measurement
The results of the Vickers hardness measurements are shown in Table 1 and shown in Figure 7. Figure 7 shows the postwear  hardness values measured directly on the worn surface in dependence on the applied hardness measuring load. In the as-polished condition, the hardness is not dependent on measuring load and penetration depth for any sample. The as-polished hardness equals mainly the postwear hardness at 10 kgf. It appears that the hardness of the martensitic 40MnB5 is independent of both the hardness measuring load and the wear load. However, the remaining steels, namely, the MAS 100, 100Al, 120, 120Al, and the reference Hadfield steel, show substantial alteration in hardness as a function of hardness measuring load. The extent of the hardness increases with decreasing hardness measuring load is not related to the wear load. Generally, the surface hardness of the MAS exceeds the hardness of both reference steels. Further, the near-surface hardness of the aluminum-containing MAS is lower than their counterparts without aluminum addition. Table 1 shows the as-polished and postwear surface hardness of the MAS. The hardness increase is calculated as the ratio of the difference in hardness (postwear minus as-polished hardness) and the as-polished hardness times 100%. The MAS show an as-polished hardness in the range of 230-430 HV10. The postwear surface hardness is not related to the applied wear load, but rather to the alloying composition. Note that the aluminumcontaining MAS show a higher as-polished hardness compared to the MAS without aluminum addition. The postwear hardness and the hardness increase of the aluminum-containing steels are, however, lower. The MAS 100 exhibits a hardness increase of 349 AE 26% on average, whereas the  Figure 6, the wear resistance of the tested steels in dependence on the hardness increase is shown. In general, increased wear resistance is associated with a higher hardness increase.
The martensitic reference steel 40MnB5 shows both the lowest wear resistance and the lowest hardness increase. A significant hardness increase of up to 150% is observed for the austenitic Hadfield steel, which is consistent with prior research. [3] The aluminum-containing MAS show similar hardness increases to the Hadfield steel, ranging from 100 to 200%. The most profound hardness increase is found for the MAS without aluminum addition. Their hardness increases between 200% and 400%. The linear dependence between wear resistance and hardness increase is found to be wear load dependent, with the same hardness increase leading to higher wear resistances at higher wear loads. Figure 8 shows the measured X-ray diffraction (XRD) patterns of the MAS in the as-polished condition and after abrasion tests at different wear loads, as well as the crystallographic planes of the corresponding phases (austenite γ; martensite α). Note that highcarbon martensite exhibits tetragonal distortion of its unit cell and, hence, a splitting of the specified α planes occurs. Therefore, the labeled crystallographic α-planes are a simplification. The intensity scale is displayed in logarithmic increments since otherwise, the presence of the martensitic phase would hardly be discernible in the as-polished state. It is evident that the intensity of the martensitic phase in all MAS is substantially increased after the wear tests and a distinct peak broadening is observed for the austenitic and the martensitic phase. No dependence on the wear load can be identified. Detailed consideration of the reference steels XRD results is not included, since they do not undergo phase transformation during wear. Table 1 shows the calculated phase fraction of the MAS in the as-polished and postwear conditions. As stated before, the as-polished condition is characterized by up to 96% austenite and just a minor fraction of thermally induced martensite. Note that there is one exception: The MAS 120Al exhibits a greater fraction (16%) of thermally induced martensite in the as-polished state. During wear, in all MAS the martensite fraction increases substantially, reaching up to 72% (Table 1). It is observed that the aluminum-containing MAS show lower martensite fractions in the postwear condition compared to the aluminum-free MAS.

Discussion
The microstructures of the MAS investigated in this paper consist of austenite, thermally induced martensite, and surface  Figure 1). Surface martensite is formed either spontaneously as a result of the presence of a free surface [15,16] or mechanically induced during polishing. However, as surface martensite is only a few hundred nanometer thick, macroscopic measurements (hardness and XRD) are hardly affected. Accordingly, the bulk microstructure components are austenite and thermally induced martensite, which are confirmed by the evaluation of the as-polished XRD measurements (Table 1). Differences in the as-polished hardness of the MAS result from different amounts of thermally induced martensite (Table 1) as well as from different extents of solid solution strengthening of the austenite due to different chemical compositions ( Table 2). Specifically, the MAS 100, 100Al, and 120 have similar fractions of thermally induced martensite. Therefore, the higher as-polished hardness of the MAS 100Al compared to the MAS 100 can be attributed mainly to solid solution strengthening due to the aluminum addition. In case of the MAS 120, the higher as-polished hardness is due to solid solution strengthening as a result of the higher carbon content. The MAS 120Al shows the highest amount of thermally induced martensite and, thus, the highest as-polished hardness of the MAS.
Often, a greater wear resistance is caused by a higher hardness of the material. [17][18][19] In this study, however, the wear resistance does not correlate with the as-polished hardness of the MAS. Rather, the greater wear resistance is found to correlate with a higher hardness increase of the MAS in the near-surface region during wear ( Figure 6). A linear dependence between wear resistance and hardness increase during wear is also found in previous studies. [2,3,10,20] However, no load dependence was investigated before. For the wear loads investigated in this study, the dependence of wear resistance and hardness increase varies with wear load: the higher the wear load, the more pronounced is the increase of wear resistance with increasing hardness. Within the group of the MAS, the postwear hardness changes with alloying composition and is not related to the applied wear load. The measured postwear hardness of the MAS of up to 1100 HV is in good agreement with the measured hardness of high-carbon thermal martensite. [21] The MAS 100 and 120 tend to have a slightly higher postwear surface hardness than their aluminum-containing counterparts. Furthermore, the increase in hardness is higher for the MAS without aluminum addition. Combining these findings with those of the weight loss measurements, it can be concluded that aluminum addition leads to higher weight losses as a result of the lower work hardening (hardness increase) of the abraded surface. The material parameter that is significantly influenced by the addition of aluminum and, in contrast, has a strong effect on the workhardening behavior is the stacking fault energy (SFE). In austenites with SFE below 20 m Jm À2 , the formation of martensite nuclei by dislocation motion is favored, whereas SFE over 20 m Jm À2 is associated with the preferred form of austenite twin nuclei. [22][23][24] This results in the well-known transformation-induced and twinning-induced plasticity mechanisms (TRIP and TWIP), which present an additional hardening mechanism in steels.
It is known that the addition of aluminum shifts the SFE to higher values. [25] Therefore, the addition of aluminum may change the deformation mechanism in austenite from TRIP to TWIP. The SFEs (at 300 K) given in Table 2 were calculated by applying the thermodynamic parameters given in Table 3 to the calculation model proposed by Allain et al. [26] . Note that the calculated SFE is intended only for illustrating the effect of aluminum addition and should not be considered absolute. However, the calculated SFE values illustrate that the addition of about 1.5 wt% aluminum shifts the SFE from values below 20 m Jm À2 (TRIP) to values above 20 m Jm À2 (TWIP) and thus triggers a change in the deformation mechanism from TRIP to TWIP. That in turn inevitably leads to less mechanically induced martensite during wear and thus to lower hardness values, as observed in all MAS with aluminum addition when compared to those without aluminum addition. XRD results (Table 1) show that the martensite fraction in all MAS increases substantially during wear. However, the increase in martensite fraction, again, is in general lower for the aluminum-containing MAS (up to %60%) compared with the aluminum-free MAS (up to 70%).
It can be concluded that aluminum addition suppresses the mechanically induced martensite formation during wear, which is in agreement with the results obtained by SFE calculations. Therefore, the postwear hardness and hardness increase of the aluminum-containing steels are lower as well. As mentioned before, the austenite stability can be approximated by M s . In some empirical calculation formulae, [12,[27][28][29] aluminum addition increases the M s , thus it stabilizes the martensitic phase. This is what can be seen in the greater amounts of thermally induced martensite (Table 1) of aluminum-containing MAS compared to aluminum-free MAS. However, compared to the effect of carbon and manganese on M s , the aluminum effect is low. The effect of aluminum on the change in austenite deformation behavior through its effect on the SFE is more pronounced in MAS than the martensite stabilizing effect. Therefore, the  martensite stabilizing effect can be balanced by the greater influence on SFE. Despite the differences in chemical composition, microstructure, hardness, and work-hardening behavior, no significant changes in the wear mechanism are observed ( Figure 9) of any tested steel or with changing the wear load. Microploughing and microcutting can be identified as wear mechanisms. Microcutting leads to a direct detachment of material, whereas microploughing primarily leads to the formation of ridges. Repeated deformation of the ridges by passing adjacent abrasive grains continues work hardening. Material is finally removed by low-cycle microfatigue, starting with crack initiation, and crack propagation, which finally lead to the delamination of wear debris. Therefore, it is reasonable that the change in weight loss as a function of wear load is mainly due to the deeper penetration of the abrasive grains as a result of higher wear load and, thus, a greater volume is detached by one stroke. This is in good agreement with the linear dependency between weight loss and the wear load. Accordingly, the weight loss is a function of the surface hardness since hardness determines the penetration depth at a given wear load.

Conclusion
The two-body abrasion behavior of MAS with and without aluminum addition was analyzed and compared to an austenitic Hadfield steel and martensitic steel. It was found that higher wear resistances can be achieved in MAS compared to a martensitic 40MnB5 and the austenitic Hadfield steel for wear loads between 8 and 32 N. The high wear resistance of MAS is achieved by a mechanically induced phase transformation from metastable austenite to martensite during the wear tests.
The following conclusions can be drawn from the obtained results: 1) A linear correlation between the wear resistance and the hardness increase is found; 2) The increase in wear resistance with increasing hardness increase depends on the wear load: the higher the wear load, the more pronounced the increase in wear resistance; 3) Compared to 40MnB5 and the Hadfield steel, the wear resistance at 32 N of the MAS 100 is 6.75 and 4.5 times higher, respectively; 4) The high wear resistances of the MAS are attributed to an increase in hardness during wear due to a mechanically induced transformation from metastable austenite to martensite. A hardness increase of up to 390% was observed; 5) The addition of 1.5 wt% aluminum increases the SFE and triggers a change in the deformation Figure 9. Wear load-dependent wear resistance of the tested steels in dependence on the hardness increase. Table 3. Parameters used to calculate the SFE at 300 K according to ref. [26]. x i and χ i are, respectively, the fractions of element i in at% and the mole fraction. T γ Neel 0.00001x 3 Mn À 0.08984x 2 Mn þ 11.76x Mn À19.92x C À 12.72x Si À 6.61x Al þ 152.4½K [34] T ε Neel 580χ Mn ½K [26] β γ =μ B 0.7χ Fe þ 0.62χ Mn À 0.64χ Fe χ Mn À 4χ C [35] β ε =μ B 0.62χ Mn À 4χ C [35] www.advancedsciencenews.com l www.steel-research.de steel research int. 2023, 94, 2200545 mechanism from TRIP to TWIP. It, therefore, impairs the superior wear properties of the MAS by suppressing the mechanically induced austenite to martensite phase transformation; and 6) The SFE-along with the M s -must be considered in the development of wear-resistant MAS. Based on the results and the conclusions drawn, we suggest that there are a large number of MAS with different chemical compositions but with similarly excellent wear properties. Their M s may vary to the extent that a partially or fully austenitic microstructure is present at the service temperature. Additionally, mechanically induced transformation of austenite to martensite has to be feasible by adjusting the SFE. Sufficiently high carbon content is also required, to ensure a high hardness of the martensite. Since, for example, carbon and manganese reduce the M s while increasing the SFE a calculation model has to be developed that predicts ranges of compositions, where M s and the SFE fulfill the mentioned conditions.

Experimental Section
Materials: The designations and the chemical compositions of the steels tested are shown in Table 2. The alloy design of the MAS was chosen such that their M s was approximately the same and in the range of room temperature. [12] This was accomplished by balancing carbon and manganese-a lower carbon content demands a greater manganese content to keep the martensite start temperature low. Carbon exhibited a greater impact on the martensite start temperature compared to manganese. Therefore, slight changes in carbon content required relatively large manganese content changes. Thus, the MAS 100 and 120 have approximately the same martensite start temperature, while their composition differed not insignificantly.
Hot-rolled sheets of the MAS were fabricated from cast and subsequently forged slabs at COMTES FHT (Czech Republic). Afterward, 2 mm was milled off of each side of the sheets to remove any decarburized layers. 10 Â 10 Â 100 mm specimens of the MAS steels were cut out of these sheets. The MAS specimens were heat treated in a muffle furnace continuously purged with inert gas and quenched in water to room temperature. The steel 100 was austenitized at 950°C for 600 s, 100Al and 120 were austenitized at 1000°C for 300 s, and the steel 120Al was austenitized at 1000°C for 240 s. These time-temperature combinations were chosen to ensure complete austenitization with the least possible grain coarsening. 40MnB5 and Hadfield steel were used as reference steels. These steels were fully martensitic and fully austenitic, respectively. Their composition is also shown in Table 2. The Hadfield steel specimens had a lower carbon content than the usual 1.2 wt%. However, the steel was completely austenitic, and the lower carbon content led to an even greater wear resistance. [3] Two-Body Abrasion Tests: Unlubricated two-body abrasive wear tests were conducted using a Suga abrasion tester NUS-ISO3, Suga Test Instruments Co., Ltd. The tests were performed using standard SiC abrasive paper (46 μm average particle size or P320 grit) at wear loads of 8, 16, 24, and 32 N at a velocity of 26.7 m ms À1 . The wear track area was 12 mm in width and 30 mm in length. Figure 10 shows a schematic illustration of the experimental setup. The abrasive paper was attached to a wheel and pressed against the sample surface. After each reciprocating movement, which was performed by the sample, the wheel turned by 0.9°, thus the sample surface was in contact with unworn abrasive paper every stroke. Four hundred strokes (360°) completed one wear cycle (wear distance of 24 m) after which the weight loss of the sample was measured using an electronic balance (AE0.1 mg precision) and the abrasive paper is renewed. In total, four wear cycles (total wear distance of 96 m) had been completed for each sample and wear load.

Characterization:
The bulk microstructures of the as-quenched MAS were revealed by etching using Adler etchant (obtained from Crida Chemie) for 10-15 s. Before etching, the specimens were grounded and polished. A detailed preparation routine could be found in previous work. [20] Light microscopy images of the etched surfaces were taken using a Neophot 32 (Zeiss).
A confocal white light optical microscope (μSurf, NanoFocus AG) with a 20Â lens from Olympus was used to take a total of nine topography images that were stitched into one 2200 Â 2200 μm image. The presented roughness parameters are calculated utilizing the Gwyddion software. [30] To calculate the roughness parameters, three roughness profiles (each 2000 μm in length) perpendicular to the wear track direction were evaluated and averaged. According to ISO 21 920-3:2021, waviness was separated from the roughness by a cutoff value of 800 μm, which was appropriate for R a in the range of 0.1-2 μm. The surface parameter R a presented the average of the absolute asperity heights from the mean line, whereas R z accounted for the average absolute value of the five highest peaks and the five lowest valleys over the evaluation length. A background correction was performed by subtracting an area described by third-degree polynomials in the x and y directions. Displayed topographies were denoised by a mean value filter that was integrated over two neighboring pixels.
XRD measurements of as-polished and postwear surfaces were done using a D8 Discover diffractometer by Bruker. The measurements were conducted in Bragg-Brentano configuration utilizing Cu radiation. Moreover, primary and secondary 2.5°soller were mounted and a Ni-filter was used to reduce CuK β radiation. Scans were taken with 2Theta ranging from 40 to 98°, a step size of 0.03°per step, and a scan speed of 3 s per step. The fractions of austenite and martensite were determined by fullpattern Rietveld refinement using the software TOPAS V6 (Bruker). SEM (Ultra 55, Zeiss) was used to image the wear tracks.
Vickers hardness values were measured using a hardness tester KB30S (KB Prüftechnik GmbH). The mean value and standard deviation of five consecutive measurements were given.