Effect of carburizing and nitriding duplex treatment on the friction and wear properties of 20CrNi2Mo steel

In this study, the responses of 20CrNi2Mo steel to carburizing (C) and carburizing-nitriding (C + N) duplex treatment and the effects of these treatments on the friction and wear properties were systematically studied. The 20CrNi2Mo surface layers were characterized by optical microscopy (OM), laser scanning confocal microscopy (LSCM), x-ray diffraction (XRD), scanning electron microscopy (SEM), transmission electron microscopy (TEM), Vickers microhardness tests and high-speed reciprocating friction and wear testing. The results showed that after the carburizing and nitriding duplex treatment, an approximately 5 μm thick compound layer (CL) consisting of the γ′-Fe4N and ε-Fe2-3N phases had formed on the surface, and the diffusion layer depth was approximately 220 μm. The carbides (MC) stored during carburizing were easily converted to nitrides (MN) during the nitriding process. Furthermore, the surface microhardness (879 HV) was increased by a factor of 2 compared with the substrate (420 HV), the surface residual compressive stress (−652 MPa) was increased significantly, and the surface roughness also increased. Wear analyses under different loads showed that specimen C exhibited the worst wear resistance, and its wear mechanism was mainly abrasive wear. The C + N with CL samples showed excellent wear resistance under a 20-N load; after removing the CL samples (C + N with CL), they showed excellent wear resistance under a 40-N load, and the wear mechanism was mainly adhesive wear. This work showed that the wear resistance was improved significantly by the carburizing and nitriding duplex treatment, but the CL had different effects on the wear performance under different loads.


Introduction
The 20CrNi2Mo form of low-carbon steel has excellent properties, including high toughness, impact resistance and wear resistance. Due to its low cost and favorable mechanical properties, it is widely used in heavy-duty gears, bearings, excavator bucket teeth (pickaxe teeth), agricultural machinery wear parts and other components [1,2]. However, the surfaces of these parts can be damaged by both wear and fatigue, causing the performances of the moving contact machine parts to degrade [3]. In these instances, appropriate surface treatments are required to improve the tribological performance. Research on 20CrNi2Mo steel has been focused mainly on conventional thermochemical treatments, including gas carburizing [4], plasma nitriding [5,6], and carbonitriding [7,8]. However, these single methods are not sufficient to afford the surface performance required for industry applications. Research on composite processing techniques has therefore attracted widespread attention. For example, Tong [9] et al used SMAT as a pretreatment to surface-modify pure iron and 38CrMoAl steel before nitriding, and the results showed that the modified layer promoted the nitriding kinetics, greatly reduced the nitriding temperature, formed a compound layer composed of nanostructured nitrides, and thus improved the performance of the material. H Kovac [10] et al found that the diffusion layer thickness was double that observed with plasma nitriding alone, and the surface residual compressive stress and surface hardness both increased and led to better wear resistance. Much research [11,12] has been focused on composite treatment techniques combining mechanical and thermochemical surface modifications, but for many workpieces experiencing wear, these approaches cannot achieve the desired mechanical improvements. Therefore, composite thermochemical surface modification [13][14][15][16][17][18] is a more promising approach. Vacuum carburizing is superior to conventional carburizing in terms of throughput, control lability, surface property enhancement, etc. In addition, compared to gas nitriding and salt bath nitriding, the plasma nitriding treatment time is shorter, the microstructures are more uniform, the deformation magnitudes are smaller, the energy cost is lower, and there is less pollution; thus, this method has received much attention for thermochemical surface modifications of steel. Most studies [19][20][21][22][23][24] on the wear behaviors of nitrided steels have shown a significant improvement in wear resistance; however, there are conflicting results suggesting that the presence of a CL forms hard particles that reduce the resistance to frictional wear.
In the previous paper [25][26][27][28], we studied the duplex treatment technology of M50NiL high-alloy steel by combining carburizing and nitriding and studied the formation mechanism, morphology, composition and structure, as well as the wear and tension fatigue properties of the steel in detail. To the best knowledge of the authors, there are few reports on the biphasic process involving carburizing followed by plasma nitriding of 20CrNi2Mo low-alloy steel, and how these processes improve wear performance remains to be explored. Therefore, in this paper, carburizing and carburizing-nitriding duplex treatments were carried out on 20CrNi2Mo steel, and friction and wear tests were performed under 20-N and 40-N loads to study the effects of the different treatment processes on the friction and wear properties of the steel. In this respect, the following questions were investigated: acetylene and 50% nitrogen gases. During the pressurization phase, the gases were fed into the hot chamber with a total gas flow rate of 8 l min −1 . When the pressure reached 1500 Pa, the gas feed was stopped, and the pressure was maintained for 20 s. Then, the hot chamber pressure was purged to apply a vacuum. The above steps were repeated for 1 h, and then the diffusion stage was carried out for another 1 h by maintaining the hot chamber under vacuum at a temperature of 930°C. At the end of the diffusion stage, the temperature of the hot chamber was reduced to 820°C, each sample was removed, and the oil was quenched in a cold chamber for 8 min. Then, the samples were removed and divided into three groups. The first group was tempered at 200°C for 2 h, and these samples were denoted as 'carburized samples' (C). The second group was then nitrided in a plasma glow discharge pulse nitriding furnace (LDMC-30AZ, Wuhan Ande heat treatment technology Corporation, China) with pure ammonia gas, a 430°C temperature, 300 Pa pressure, and 10 h nitriding treatment; these samples were denoted as 'carburized-nitrided samples with CL' (C + N with CL). The last group of samples was then ground on a CNC to completely remove the CL, and these samples were denoted as 'carburized-nitrided sample without CL' (C + N without CL). Finally, friction and wear experiments were carried out on the three groups of samples.

Characterization
The cross-sections of the samples were etched with a 4% nitrate alcohol solution and then inspected by optical microscopy (OM, SDP TOP ICX4IM, Leica, Germany), laser scanning confocal microscopy (LSCM, OLS5000, Nishi-Shinjuku, Japan) and field emission scanning electron microscopy (SEM, SUPRA 40-41-90, Zeiss, Germany). The phase structure of a sample was characterized by x-ray diffraction (XRD, Rigaku Ultima IV, Japan) with Cu Kα radiation and scanning angles (2θ) ranging from 5°to 90°. An HVS-1000 hardness tester (Lunjie Motor Instrument Company, Shanghai, China) was used to measure the microhardness of a cross section, with an external load of 100 g, a load retention time of 10 s, and an average value reported across 5 measurement points. Laser scanning confocal microscopy (LSCM) was used to measure the surface roughness of random regions. The residual stress was determined with an x-ray residual stress meter (G.N.R. S.r.l. -Analytical Instruments Group, Milan, Italy). The precipitates in the diffusion layers were characterized by transmission electron microscopy (TEM, FEI Talos F200X, American FEI Company, USA). An HSR-2 M high-speed reciprocating friction and wear testing machine (HSR-2 M, Zhongke Kaihua Corporation, China) was used for friction and wear experiments. A tungsten carbide (WC) ball with a diameter of 6 mm was used as the counterpart. The test was carried out at room temperature and atmospheric pressure with a sliding speed of 0.08 m s −1 , a time of 3000 s and wear loads of 20 N and 40 N. The wear rate η was calculated as follows: is the wear amount and L (m) is the total sliding distance. The worn surfaces were subjected to laser confocal scanning microscopy (LCSM), SEM and energy-dispersive spectrometer (EDS) analyses.

Analysis of the carburizing layer and nitriding layer
The cross-sectional microstructures of the three groups of treated samples (C, C + N with CL, and C + N without CL) are shown in figure 1. Figure 1(a) shows the microstructure of the carburizing layer, which was composed mainly of tempered martensite, carbides and a small amount of residual austenite. The depth of the carburizing layer was approximately 1200 μm. The modified surface layer of the carburizing and nitriding duplex treatment sample was divided into three parts: a CL, a diffusion layer and a carburizing layer. The compound layer can also be called the white bright layer and was composed mainly of the γ'-Fe 4 N phase, the ε-Fe 2-3 N phase, and a mixture of the two phases. Its thickness was considered to be ∼5 μm. After polishing with sandpaper and a polishing cloth, it was uneven and easy to peel from the surface due to its brittleness, as shown in figure 1(b 1 ). Below the CL was the diffusion layer, in which nitrogen was mainly incorporated into the existing iron lattice as interstitial atoms or as finely dispersed nitrite alloy precipitates [29]. The diffusion layer was approximately 200 μm thick. As shown by the red arrow in figure 1(b 1 ), some vein-like structures were seen as having precipitated along the grain boundaries parallel to the surface. In many studies [30], the veined structures are regarded as nitride networks, whereas some other researchers [31,32] have used TEM analyses and found that veined structures are mainly carbon nitrides. By comparing figures 1(b 1 ) and (c 1 ), it is clear that the CL was removed by surface grinding. Thus, the line scan curve of N in figure 2(a) indicated a distinct N peak at the beginning, corresponding to a CL on the outer surface of the nitrided sample with a peak width of approximately 5 μm, consistent with the thickness of the CL observed in figure 1(b 1 ). However, the nitrided sample without CL in figure 2(b) had no peak value, proving that the CL was completely removed in this case.

Phase analysis
The XRD patterns measured after the three different treatments are shown in figure 3. The results show that there only α-Fe was found in the carburized sample (C), and the C + N duplex treatment samples (C + N with CL) were composed mainly of γ'-Fe 4 N, ε-Fe 2-3 N and small amounts of a nitrogen-iron solid mixture. Generally, XRD analyses provides information on the composition of the surface phase with a thickness of approximately 5 μm; since the CL thickness of the C + N with the CL sample was approximately 5 μm, XRD showed there was a small amount of a nitrogen-iron solids mixture phase in the diffusion layer. When the CL was removed from the C + N without CL samples, the surface phase structures were composed mainly of nitrogen-iron solids mixtures. Due to the presence of the nitrogen atoms, the crystal plane spacing of α-Fe decreased, which moved the peak for α-Fe to the right and broadened it relative to that of the carburized sample. The CL had a great influence on the friction and wear properties [33], so the presence or absence of the CL was crucial for optimizing these properties.   Figure 4 shows TEM observations of the precipitates in the diffusion layers of the C and C + N samples, as well as the corresponding selected area diffraction (SAED) and EDS mapping results of selected electron diffraction spectra. The results showed that M 23 C 6 and M 2 C carbides rich in Cr and Mo were precipitated mainly in granular forms in the carburized samples, and the edge lengths of the carbides were approximately 30-100 nm, as shown in figures 4(a)-(f). These results indicated that some carbides were not dissolved during the carburizing and quenching processes, and it was these second-phase carbide particles that had a strong pinning effect on the grain boundaries, making it difficult for grains to grow. According to previous studies [34], the preexisting large carbide precipitates are not easily converted to nitrides during nitriding, while preexisting small (nanoscale) carbides are easily converted to nitrides. However, at low nitriding temperatures (450°C), an increased nitrogen content made it easier for the preexisting carbides to absorb nitrogen atoms. Therefore, the MoN in the 20CrNi2Mo steel was generated from M 2 C carbides when the carburizing heat treatment was followed by nitriding.

Hardness analyses
The hardness curves measured after the three different treatment processes are shown in figure 5. The highest surface hardness after carburizing was 751.78 HV, and the threshold matrix hardness was 430 HV. The depth of the carburizing layer was approximately 1200 μm, which was largely consistent with the microstructure shown in figure 1(a). After the carburizing and nitriding duplex treatment, the surface hardness was further improved to between 850 and 900 HV. After the C + N treatment, because the carburizing layer was equivalent to the transition layer between the nitriding layer and the matrix, the depth of the nitriding layer (as determined by taking 751.78 HV as the threshold) was approximately 220 μm, and the hardness decreased gradually with increasing depth. The nitriding layer depth decreased until the hardnesses of all the matrix samples were approximately 430 HV, which was considered to be related to the changes in the carbon and nitrogen contents. Additionally, the nanoscale (M 23 C 6 and M 2 C) and microscale precipitates (MN) in the diffusion layer also showed a strengthening effect on the hardness. It is worth noting that the surface hardness of the C + N without CL sample was slightly less than that of the C + N with CL sample; this was mainly attributed to the fact that the CL was a hard and brittle phase. It is well known that higher hardness is beneficial to the wear properties of materials [35]. It has been reported [36] that the wear resistance of a nitrided sample increased with increasing hardness under a 20-100-N load.

Surface residual stress and surface roughness
The surface residual stress and surface roughness values are shown in table 1. The residual stress on the surface was compressive stress. The surface residual stress of sample C was −112 MPa. Qin [37] et al reported similar results after carburizing 18CrNiMo7-6 steel. This is because the carbon content in the surface layer of the carburized sample was higher than that in the core. During the quenching process, the martensite successively changed from the core to the surface, and the high carbon content on the surface converted some of the surface austenite into martensite, which eventually led to the surface compressive residual stress layer. However, lowtemperature tempering (at 200°C) promoted the release of the compressive residual stress and showed a low compressive residual stress. After the C + N treatment, the surface residual stress was −652 MPa. The main source of this residual stress was the volume expansion caused by the solid mixture of nitrogen in the iron lattice during nitriding and the formation of nitride precipitates in the steel matrix [38]. In contrast, carburizing had no effect on the surface roughness, which was 0.289 μm, while the surface roughness increased substantially to 1.422 μm after nitriding, mainly due to the surface relief caused by volumetric expansion associated with the formation of nitrides [39]. In addition, the sputtering and redeposition of specimens on the surface may have also affected the surface roughness, which is consistent with other reports [40,41].

. Friction coefficients and wear rates
To study the influence of the C, C + N with CL and C + N without CL treatment on the wear properties of the specimens, dry sliding wear tests were conducted under different loads, and the friction coefficient and wear time curves are shown in figure 6. All coefficients of friction first increased to high values during the running-in phase and then gradually stabilized. During dry sliding wear, the debris easily led to the formation of an oxide film that was brittle and reduced the friction coefficient. When the oxide film accumulated to a certain thickness and then was spalled off, the friction coefficient gradually increased and eventually remained stable [42]. In addition, stick-slip behaviors characterized by fluctuations or sudden drops in the coefficients of friction were observed during the stabilization phases, which was attributed to the 'oxidation-scrape-reoxidation' theory. In the wear process, the wear debris underwent alternating processes of formation, oxidation, compaction and delamination on the rubbed surfaces [43], which led to oscillation of the friction coefficients. The friction coefficient curve determined at 20 N showed that the carburized C sample was the first to reach a stable friction coefficient. Due to its relatively low hardness and low residual stress, it easily spalled by shear stress in the friction process, showing the highest friction coefficient. Both the C + N with CL and C + N without CL treatment provided similar hardness and compressive residual stresses, but the roughness of the C + N with CL treatment was significantly higher than that of the C + N without CL treatment. Additionally, a higher surface roughness led to a lower friction coefficient. The main reason is that the contact area between the wear surface and the WC ball surface decreased during the wear process. In addition, the presence of the CL significantly increased ductility and lubricity at the smaller 20 N load and reduced the coefficient of friction. Therefore, the C + N with CL sample had a lower friction coefficient. Furthermore, it was observed that at approximately 25 min during the friction process, the friction coefficient dropped sharply from the stable stage and then rose again; this may have been because wear occurred at the interface between the CL and the diffusion layer. However, under a load of 40 N, the oxide film produced by the grinding debris was easier to flake off, so the friction coefficient increased faster and remained stable. However, the friction coefficient of the C + N with CL sample also decreased sharply at 25 min, which was consistent with the observation for the 20-N load.
The wear rates of all samples are shown in figure 7. Increasing the load from 20 N to 40 N greatly increased the wear rate. Under a load of 20 N, the wear rate of the C + N with CL sample was the lowest at 0.575253 × 10 −5 mm 3 m −1 . When the load was increased to 40 N, the wear rate of the C + N without CL sample was the lowest, at 2.111443 × 10 −5 mm 3 m −1 , indicating that the C + N with CL sample had better wear resistance under the small 20 N load, while under a heavy load of 40 N, the C + N without CL sample had better wear resistance. Some studies [44] have indicated a direct relationship between the hardness and the wear rate. In other words, the greater the hardness was, the lower the wear rate. However, the C + N with CL and C + N without CL samples with the highest surface hardnesses showed different wear rates under different loads, indicating that other factors besides hardness affected the wear rate. It has been reported that the surface roughness affects the wear of the sample only in the initial stage of wear and has little effect on the final stage of wear. Therefore, considering the above results, the wear rate may be directly related to the CL, surface residual stress, etc. Cross-sectional graphs of these scratches showed accumulations of typical plastic deformations on both sides of the wear tracks, indicating that abrasive wear had occurred on all specimens. At the same time, the widths and depths of the wear marks that arose when applying the same loads to the different treatment processes were very different. The wear area of the C + N with CL specimen under a 20-N load was 690.304 μm 2 , and the wear depth was 6.958 μm. The wear area of the C specimen under a 40-N load was 5700.900 μm 2 , and the wear depth was 18.570 μm. It can be seen from the wear marks that when the load was increased from 20 N to 40 N, the abrasive wear of the sample became more serious, resulting in deep grooves in the wear marks. Additionally, accumulation and shedding on both sides increased during the wear process.

Wear surface and wear mechanism analyses
To analyze the wear mechanisms and wear processes of the three different samples under 20-N and 40-N loads in detail, SEM was used to observe the surface wear conditions, and the results are shown in figure 9. The wear morphology of the carburized (C) sample under a 20-N load is shown in figure 9(a) and mainly consisted of parallel furrows, spalling and an oxide film. The worn surface was filled with grooves and fish-scale wear marks caused by sliding. The main wear mechanisms were abrasive wear, oxidative wear and slight adhesive wear. Abrasive wear was mainly characterized by parallel grooves, soft phase adhesion and flaking caused by the oxidation of the original film. For materials with relatively low hardness, fish-scale wear has been reported to result from bulky abrasives falling off and coming together after sliding [45,46]. During dry slip wear, the surface roughness caused by fish-scale wear led to greater friction and an increased COF [47]. However, after the duplex treatment, the surface had high hardness, and the worn surface contained many nanoscale oxidation particles. The smooth glaze layer formed on the worn surface prevented formation of the rough fish scales and led to selflubrication, thus reducing the friction coefficient [42]. According to the EDS results shown in figure 10, O and W were detected in zone 'a' of sample C, which further confirmed the occurrence of oxidative wear and adhesive wear. The wear morphology of the C + N with CL sample is shown in figure 9(b). Slight parallel furrows can be clearly observed, and the worn surface in the high-resolution image was covered with adhesive and wear debris. As with the C + N without CL sample, in addition to parallel furrows, there was debris accumulated by wear. This was consistent with the light loading observed on other wear surfaces of nitrided alloys, in which adhesive wear was the dominant wear mechanism [48]. This was because CLs and diffusion layers with relatively high surface hardnesses are particularly effective in resisting wear over small loading ranges that do not damage the surface.
SEM images of the worn surface of the sample subjected to a 40-N load are shown in figure 11. As shown in figure 10, oxidative wear and adhesive wear still occurred under 40-N loading, and serious abrasive wear occurred. The higher the load was, the smaller the influence of the oxide film, and the more likely the corresponding WC ball is to destroy the oxide film and then directly contact the surface of the sample [34]. Therefore, under a load of 40 N, the main factor affecting the wear rate was the inherent wear resistance of the carburizing layer and the nitriding layer. The carburized C sample had a low hardness and a low residual compressive stress. Under a large load, it easily flaked off and formed a large amount of debris, which not only caused serious abrasive wear but also accelerated oxidative wear. Therefore, the carburized C specimens showed the highest wear rates and the most severe adhesive and oxidizing wear. Although the CL on the surface of the C + N with CL sample exhibited the highest hardness and the highest residual compressive stress, the hard and brittle CL began to fracture under the larger load of 40 N. Clear surface spalling traces (CL removal) were also observed on the worn surface. In addition, hard abrasive particles were generated in the tracks, which changed from the initial adhesive wear to abrasive wear, significantly increasing the wear between the WC ball and the sample [49]. Thus, the hard abrasive particles that formed due to the peeling CL remained in the contact zone, resulting in abrasive wear and accelerating the weight loss. This phenomenon of CL spalling has been found in previous studies and during the use of nitrided steel [33,50,51].
In addition, during application of the 40-N load wear, vibrations of the contact interface, the uneven distribution of the debris, and oxidation may have produced a large number of microcontacts on the contact surface [49]. This resulted in locally high contact pressures, which can lead to large subsurface shear stresses. The higher residual compressive stress on the surface of the C + N sample reduced the total stress. However, in addition to residual compressive stress, the C + N with CL sample exhibited deterioration of the surface roughness, which reduced the scratch resistance. After removal of the CL from the C + N without CL sample, the surface layer of the sample was the diffusion layer, and the most important characteristics of this layer were its high hardness, high toughness, significantly enhanced residual compressive stress and low roughness. The wear morphology mainly showed parallel trenches and a few scrapes, and the wear mechanism was mainly adhesive wear, as shown in figure 11(c). Therefore, the C + N without CL sample showed ideal wear resistance, especially under large loads. This explains why the wear performance of the C + N with CL sample was worse than that of the C + N without CL sample under 40-N loads.

Conclusion
In this study, the effects of the compound layer (CL), microhardness, surface residual stress and surface roughness on the wear properties and wear mechanism of 20CrNi2Mo steel were studied with a carburizing and nitriding duplex treatment. The carburizing and nitriding duplex treatment significantly improved the wear performance of the workpiece, and the presence of the compound layer (CL) had a great impact on the wear resistance. For a worn workpiece used under various working conditions, the wear resistance of the workpiece can be improved by adjusting the processing parameters. The corresponding conclusions are summarized as follows: (1)The carbonization layer of the carburized sample was mainly composed of tempered martensite, carbide and a small amount of residual austenite, and the depth of the carburizing layer was approximately 1200 μm. The duplex-treated sample had a CL of ∼5 μm on the surface and a diffusion layer of ∼220 μm. The CL comprised the γ'-Fe 4 N phase, ε-Fe 2-3 N phase and a mixture of the two phases. The diffusion layer was composed mainly of a nitrogen-iron solid mixture.
(2)The granular precipitates in the carburizing samples were mainly M 23 C 6 and M 2 C carbides rich in Cr and Mo. After the duplex treatment, the precipitates in the diffusion layer of the samples were rich in N and Mo, indicating a Mo nitride (MN). The MN was produced from the M 2 C carbides when the carburizing heat treatment was followed by nitriding. After carburizing, the highest surface hardness was 751.78 HV, and after C + N treatment, the surface hardness was further increased to between 850-900 HV. (3)Compared to the C sample, the C + N sample showed the higher hardness and compressive residual stress and an increased surface roughness. Under a load of 20 N, the wear rate of the C + N with CL sample was the lowest, at 0.5753 × 10 −5 mm 3 m −1 . When the load was 40 N, the wear rate of the C + N without CL sample was the lowest, at 2.1114 × 10 −5 mm 3 m −1 . At a low load of 20 N, CL significantly improved the surface hardness and wear resistance of material, but at a high load of 40 N, the brittle CL was prone to local cracking and spalling, while the surface of the C + N without CL sample was a diffusion layer with high hardness, high toughness and significantly enhanced compressive residual stress that also showed excellent wear resistance.
(4)When the load was increased from 20 N to 40 N, the wear mechanism controlling the wear rate changed from oxidative wear to abrasive wear.

Data availability statement
All data that support the findings of this study are included within the article (and any supplementary files).