Influence of heat treatment conditions of Hardox 500 steel on its resistance to abrasive wear

Among steels intended for mass application, metallic materials with a homogeneous martensitic microstructure or those designed for quenching in the stamping process – often referred to as boron steels for heat treatment – exhibit the highest possible tensile strength (R _m ) values. Despite the lack of available data in manufacturers’ catalog specifications, sources [1,2] indicate that their tensile strength (R _m ) can significantly exceed 2,000 Mpa. These mechanical properties, while maintaining satisfactory ductility parameters, contribute to high resistance to abrasive wear as well as the ability to withstand both static loads and impact energy absorption. The group of martensitic steels represents the most extensive category of materials designed for abrasive wear conditions, encompassing types such as Hardox and Raex (SSAB-Oxelösund), Perdur and TBL (ThyssenKrupp Steel Europe AG), Brinar and Durostat (Grobblech GmbH), Miilux (Miilukangas Group), Relia and Creusabro (Industeel), Abrazo (TATA Steel Group), and Dillidur (Dillinger Hütte GTS) [3].

A common characteristic of low-alloy martensitic steels is the presence of a microalloying addition of boron, which significantly increases the hardenability. Comparatively, its content in steel at a level of 0.001–0.003% by weight provides a hardening intensity equivalent to 0.6% Mn, 0.7% Cr, 0.5% Mo, or 1.5% Ni by weight [4]. Therefore, it can be concluded that an increase in hardenability occurs at boron concentrations that are considered trace levels for other alloying elements [5,6,7,8]. The production of a family of steels characterized by the aforementioned mechanical properties is justified by their high hardenability, which promotes deformation. Moreover, Hardox 500 steel, which exhibits a tensile strength (R _m ) of 1,653 Mpa [9], is also classified by the manufacturer as a steel with limited structural applications [10]. The authors’ experience indicates that quenching higher grades of Hardox steel in water increases the likelihood of cold cracking. It should be noted that data regarding the quenching methods of PHS steels are not commercially available and must be verified by the component manufacturer. Therefore, further research is necessary to examine the retention of specific functional properties in this material group while reducing the cooling rate after austenitization.

However, it should be noted that an increase in steel hardenability entails the risk of technological challenges. The amount of quenching-induced stresses is determined by the type of cooling medium used during the quenching process. The difference in cooling rates between the core and the surface of the object affects the extent of deformation, particularly in large-scale components with complex shapes. For example, the heat treatment of boron-alloyed martensitic steels is performed on profiled structural components such as plowshares or cultivator coulters. The critical cooling rate depends on the chemical composition of the material and increases as hardenability (most commonly defined by the carbon content) decreases. The optimal cooling rate minimizes the occurrence of potential deformations and cracks in the treated material, which is crucial for maintaining the required dimensions and shapes of finished products.

Water remains the most commonly used quenching medium. However, as the hardenability of the material increases, it is recommended to use synthetic or mineral oil for cooling. According to Rudnik [11], the cooling rate in mineral oil is four times lower than that in water, averaging 150°C/s within the temperature range of 650–550°C. A milder cooling medium is also provided by compressed air blast cooling.

In the study by Luo and Bai [12], it was demonstrated that air-cooled, medium-carbon, low-alloy MnCrB cast steel with a chromium content of 0.6% by weight exhibits a tensile strength (R _m ) exceeding 2,000 Mpa, ductility (A) of 4–6%, and impact toughness in the range of 111–164 J/cm². Its wear resistance is 30% higher than that of Hadfield steel in a three-body contact under a high load. In terms of scratch resistance and achieved hardness levels, cryogenic cooling of Hardox 500 steel appears to be a beneficial process, as this heat treatment condition resulted in a 27% increase in wear resistance compared to the as-delivered material. However, the negative aspects of cryogenic cooling must be considered, including the presence of residual stresses, which necessitate slower cooling to mitigate the risk of cracking [13]. A relevant example is Creusabro steels, characterized by a complex bainitic-martensitic microstructure with retained austenite and finely dispersed carbides. This microstructure is achieved through an appropriately tailored chemical composition and a controlled, slow quenching process [14]. Moreover, Creusabro 8000 steel exhibits a similar relative abrasion resistance coefficient (k _b ) compared to TBL PLUS and XAR 600 steels, despite having a hardness lower by 57 and 92 Brinell units, respectively [15]. According to George et al. [16], Usibor 1500P steel (C = 0.22% by weight), used for B-pillars in automotive applications where zones of lower hardness are required, develops a bainitic microstructure when cooled in a die at a rate of 30°C/s after preheating to 400°C. Based on continuous cooling transformation (CCT) diagrams, the same steel exhibits a fully martensitic microstructure at cooling rates of 45–2,200°C/s, whereas at a cooling rate of 25°C/s, 95% martensite and 5% bainite are observed [17]. For 22MnB5 steel, the resulting microstructure varies depending on the cooling medium: martensite forms during water quenching, tempered martensite and bainite develop during oil quenching, and a multiphase microstructure is obtained with air-blast cooling at a rate of 10°C/s [18,19]. According to equilibrium phase diagrams, Hardox 450 steel cooled at a rate of 10°C/s from 900°C does not exhibit ferrite precipitation; instead, the resulting structure consists of bainite and martensite [20].

For the study on the influence of cooling methods after austenitization on abrasive wear resistance, Hardox 500 steel was selected. The choice was motivated not only by its widespread industrial application but also by findings from previous field investigations [21,22], which indicated a tendency for accelerated abrasive wear in the vicinity of its welded joints [21]. Namely, field data have repeatedly shown that the heat-affected zones suffer from accelerated material loss under abrasive conditions, compromising component durability. For this reason, we aimed to reproduce microstructural variations – similar to those occurring within the heat-affected zone of welded joints – by applying different cooling rates after austenitization. By doing so, we can systematically induce and characterize the same microstructural heterogeneities. Unlike higher-grade steels such as Hardox 600 and Extreme [23,24], Hardox 500 is classified as bendable and weldable, which enhances its potential for industrial applications [25]. Hardox 500 is used in products such as liners, wear-resistant bars, cutting edges, and excavator buckets [26,27,28,29]. Its resistance to abrasive wear in soil environments may be comparable to that of Hardox Extreme steel [30]. According to Szala et al. [31], Hardox 500 also exhibits the lowest mass loss compared to S355JR, S355J2, and AISI304 steels in tribological tests involving garnet and silicon carbide (carborundum). Its properties may also be more favorable than those of 38GSA steel (hardness of 548 HBW), which is commonly used in Poland for plowshares [9,32]. Furthermore, the wear intensity of Hardox 500 is 12% lower than that of carburized 20MnCr5 steel [33].

The study utilized 10 mm thick sheets of Hardox 500 steel, supplied by the authorized distributor, STAL-HURT. The chemical composition analysis was performed using a spectral method with a Leco GDS500A glow discharge emission spectrometer. The following parameters were applied to ionize the inert gas: U = 1,250 V; I = 45 mA; 99.999% argon. The obtained results represent the arithmetic mean of at least five measurements.

The CCT diagram was obtained through computer simulations carried out using JmatPro software. All heat treatment operations were performed in gas-tight chamber furnaces (FCF 12SHM/R) manufactured by Czylok, using a protective atmosphere of 99.95% argon. The parameters of the heat treatment procedures are presented in detail in Table 1.

For microscopic examination, light microscopy (LM) was performed using a Nikon Eclipse MA200 light microscope. The samples were etched with a 5% HNO₃ solution, in accordance with ASTM E407. A Nikon DS-Fi2 digital camera, coupled with the microscope, and NIS Elements software from Nikon were utilized for recording and analyzing the captured images. Additional microstructure and worn surface images of the samples were conducted using a Phenom XL scanning electron microscope, employing secondary electron imaging or backscattered electron imaging at an accelerating voltage of 15 keV.

Hardness measurements of the base material were performed using a Zwick/Roell ZHU 187.5 universal hardness tester, applying the Brinell method, in accordance with ISO 6506-1:2014. A 2.5 mm diameter carbide ball was used under a load of 187.5 kgf (1838.7469 N), applied for 15 s.

Laboratory tests of abrasive wear resistance were carried out using a T-07 abrasive wear resistance tester in the presence of loose abrasive media, in accordance with GOST 23.208-79, a standard analogous to ASTM G65. The primary difference between the T-07 tester and the device described in ASTM standard G65 is that in the T-07 device, the tested material is positioned horizontally rather than vertically. The T-07 tribotester consists of a steel roller with a rubber ring of 50 mm (+0.2 mm) diameter and 15 mm (−0.1 mm) width, an abrasive feeder that allows for regulating the abrasive flow, and a lever with weights, which generates a vertical force pressing the sample against the roller. The hardness of the rubber coating on the roller is within the range of 78–85 ShA. The tests were conducted under a constant load of F = 44 N (∆F = 0.25 N). The abrasive medium used was electrocorundum #90, in accordance with PN-M-59115:1976. The test duration depended on the hardness of the tested material and was set to 30 min (1,800 rotational cycles). The selection of this testing methodology was dictated by its practical relevance – the analyzed group of low-alloy martensitic steels is commonly used in excavator bucket components and agricultural machinery parts, which operate under intense abrasive conditions during ore extraction and soil mass processing. The applied method has been confirmed to correlate with field tests [22,34]. Each test was performed in six repetitions, with sample dimensions of 30 mm × 30 mm × 3 mm. The relative abrasive wear resistance coefficient (k _b ) was used as a measure of abrasion resistance, calculated according to equation (1). (1) k b = Z w w ρ b N b Z w b ρ w N w , {k}_{b}=\frac{{Z}_{ww}{\rho }_{b}{N}_{b}}{{Z}_{wb}{\rho }_{w}{N}_{w}}, where k _b is the relative abrasive wear resistance coefficient [–], Z _ww is the mass loss of the reference material [g], Z _wb is the mass loss of the tested material [g], N _w is the number of counter-sample rotations during the reference material test, N _b is the number of counter-sample rotations during the tested material test, and ρ _w and ρ _b are densities of the reference and tested materials [g/cm³], respectively.

The reference sample was Hardox 500 steel, used in the as-delivered condition. Figure 1 presents the schematic diagram of the T-07 device, along with the positioning of the tested material sample.

Figure 1

Schematic diagram of the T-07 tribotester. 1 – sample, 2 – rubber-rimmed steel wheel, 3 – abrasive, 4 – load, and P1, P2, and P3 – regions of samples subjected to surface topography evaluation.

For the quantitative evaluation of sample surfaces, a HITACHI TM-3000, the second scanning electron microscope was used along with specific graphic-analytical software. Before roughness evaluation, the sample surfaces were cleaned with dry compressed air, inspected for mechanical damage, and marked at three locations: P1, P2, and P3 (Figure 1), where surface evaluation procedures were to be performed. All surface evaluation procedures were conducted at the same magnification, maintaining ten times the elementary segment length lr = 198.03 µm, i.e., 1980.26 µm, and with identical electron beam settings of 15 kV, ensuring precise observation and recording of surface details of steel samples. Before testing, a working distance (WD) calibration procedure was carried out using a roughness standard with a predefined profile.

The following roughness parameters were analyzed:

R _a – arithmetic mean deviation of the profile, R _p – height of the highest peak in the profile, and R _v – depth of the deepest valley in the profile.

Statistical analyses were performed using Statistica version 13. The key assumptions required for ANOVA are normality of distribution and homogeneity of variance (Table 2). In cases where these assumptions were not met, this did not necessarily prevent the use of parametric tests, as Lindman [35,36] demonstrated that the F statistic is robust to violations of variance homogeneity. Levene’s test was used to assess the homogeneity of variance, while Shapiro–Wilk’s test was applied to evaluate the normality of distributions. All datasets satisfied the normality assumption except those obtained after 5 bar air cooling. To determine which means differed significantly, Duncan’s post hoc test was used. For statistical analysis, mass wear was normalized per 1 m of friction distance. For all roughness parameters, normal distributions with homogeneous variance were obtained.

The article presents a comprehensive approach to the evaluation of heat-treated steels, with a particular focus on microstructural development aimed at achieving varied responses in terms of tribological resistance. In the course of this study, a broad spectrum of surface roughness parameters was selected for analysis. This decision was motivated, in part, by the absence of clear literature guidelines that would support the selection of a narrower set of parameters. The authors were concerned that limiting the scope might reduce the likelihood of detecting potential microstructural differences arising from distinct heat treatment processes, which could influence wear resistance. Therefore, an extended set of roughness parameters – R _a, R _q, R _z, R _p, and R _v – was employed to ensure more comprehensive evaluation. These parameters served not only as the basis for data interpretation but also as a reference for previous studies [3,22,28,50] and as a foundation for future work aimed at identifying a tailored set of parameters capable of distinguishing between different steel grades and heat treatment strategies. The experimental plan, which included a detailed assessment of the obtained roughness parameters – each reflecting specific surface features formed during abrasive wear – was intended to enable the identification of even subtle differences in the resulting microstructures. Previous studies had demonstrated varied outcomes regarding the correlation between the roughness parameters and the hardness of tested materials, reinforcing the need for the broad approach adopted in this work. Studies on the effect of grain size on the abrasive wear resistance of Hardox 500 steel have shown that lower austenitizing temperatures – particularly 850 and 900°C – produce smoother surfaces with lower roughness, whereas austenitizing at 1,000 and 1,200°C increases roughness parameters, especially R _z, R _p, and R _v, compared to the as-delivered condition [28]. An analysis of roughness on worn plowshares made of pearlitic versus martensitic steel likewise revealed that the pearlitic steel exhibited higher roughness values – most notably in the maximum profile peak height (R _p) – than its martensitic counterpart [50]. Roughness measurements on abraded Hardox 450 surfaces also demonstrated that microstructures with higher hardness correspond to the lower roughness parameter value [3]. In the as-delivered state, R _a was 0.393 µm, increasing after austenitization at 900 and 950°C; with a further increase in the austenitizing temperature, R _a decreased, a trend correlated with austenite grain growth and carbide-dissolution-induced microstructural changes. Similar trends were observed for R _p. It was found that as the austenite grain size increased, the wear weight-loss rate increased linearly – accompanied by a decline in abrasive wear resistance – yet no significant correlation between hardness and the wear-resistance index was detected. Finally, relationships between selected roughness parameters (R _p, R _v, and R _z) and surface hardness could be well approximated by a quadratic function.

In this study, despite conducting a comprehensive analysis of roughness parameters after various heat treatment conditions of Hardox 500 steel, it was observed that traditional roughness parameters do not fully reflect the surface states, which exhibit a high degree of variation. Therefore, the authors decided to address this issue in the discussion. Standard roughness parameters, such as R _a and R _q, provide only an average description of surface topography. However, for surfaces that are locally heterogeneous – containing deep pits, areas of intense wear, as well as relatively untouched regions or those formed by the wear of asperities – such averaging leads to a loss of critical information about localized defects. Statistical analysis of the obtained values did not show significant differences between the various heat treatment conditions, which contradicts microscopic observations, where clear local differences in wear mechanisms were noticeable. This discrepancy arises because roughness parameters characterize only the global features of the surface, without considering the spatial distribution of anomalies such as deep pits, material spallation, or localized plastic deformations. A critical aspect is that averaging roughness values masks these local variations. This approach can lead to an inadequate assessment of the actual surface condition, particularly in applications where local defects play a key role in the durability of working components. The literature frequently emphasizes that classical roughness parameters are insensitive to localized topographical variations, which may result in misinterpretation of surface conditions when analyzing materials with complex structures. Studies by other research groups suggest that conventional roughness parameters are also insufficient for predicting the structure of worn surfaces under sliding friction conditions [51]. Moreover, Kovalev et al. demonstrated the lack of correlation between initial roughness and the parameters of the worn surface.

Explanations for the above observations, as well as valuable supplementation to the experimental studies, are provided by numerical analyses of surface topography evolution, as proposed by Garcia-Suarez et al. [52]. The simulation results indicate that, regardless of the initial surface condition, wear processes gradually “erase” the original topographic features. As a result, the surface tends toward a steady-state roughness, reached after a transitional period during which surface parameters oscillate around an average value. Comparing these results with the Hardox 500 study allows for several conclusions. First, both experimental studies and numerical analyses suggest that global roughness parameters are insufficient for a comprehensive description of surface conditions, especially when local defects play a crucial role. Second, the transitional period observed in simulations, during which the surface “forgets” its initial state, reflects dynamic wear processes that may also occur in steel subjected to different heat treatment variants. Numerical simulations have shown that initially smooth surfaces reach a steady-state roughness more quickly, whereas initially rough surfaces require a longer transitional period. This observation aligns with the findings for Hardox 500, where differences in wear mechanisms visualized through SEM were not fully captured by averaged roughness parameters. This indicates that, under real operating conditions, the influence of roughness on wear may be more complex and dependent on local micromechanical wear mechanisms than suggested by standard parameters. Therefore, integrating experimental studies on Hardox 500 with numerical analyses in future research would enable a more comprehensive understanding of wear mechanisms. Advanced techniques, such as three-dimensional topography analysis, allow for a more precise identification of local defect distribution and their impact on overall surface conditions. Such a holistic methodology could contribute to the optimization of heat treatment processes, leading to increased wear resistance and improved durability of the material under operational conditions.

Additionally, analyses conducted by Bigerelle et al. [53] demonstrated that three primary wear mechanisms can be distinguished depending on the size of abrasive grains. For particles larger than 125 µm, wear occurs through microcutting, whereas for grains smaller than 10 µm, adhesion dominates. In the intermediate range, when the abrasive particle size is between 10 and 125 µm, a phenomenon known as the “grain size effect” is observed, where the extreme amplitude of peaks to valleys decreases with increasing abrasive particle size. This indicates the crucial role of grain size distribution in the surface degradation process. Bigerelle et al. also demonstrated that beyond a critical autocorrelation length (approximately 160 µm), the surface loses memory of its initial structure, meaning that its further evolution can be modeled using extreme value theory. Autocorrelation measures how roughness values at different points on the surface are related to one another. If there is a distinct correlation at a certain length, it indicates that the surface maintains an ordered structure on that scale. However, once the critical autocorrelation length (in this case, 160 µm) is exceeded, the surface loses this ordered relationship, and its further evolution becomes more random and independent of its previous state.

An extension of the above studies is provided by the results presented by Szala et al. [31], who investigated the wear mechanisms of S235JR, S355J2, C45, AISI 304, and Hardox 500 steels using three types of abrasives: garnet, corundum, and silicon carbide (carborundum). Their experiments demonstrated that the type of abrasive used has a decisive impact on wear mechanisms and local surface topography. For example, when silicon carbide – which is characterized by very high hardness and fine grain size – was used, the dominant wear mechanism for Hardox 500 steel was microcutting, resulting in lower values of R _k, R _pk, and R _vk parameters. In contrast, when using coarser abrasives such as garnet or corundum, microplowing and microfatigue mechanisms dominated, leading to higher averaged R _a and R _q values, which, however, did not fully capture the localized complexity of wear. The experimental results of Szala et al. [31] emphasize that both the steel microstructure and the properties of the abrasives used (e.g., hardness, particle size, and morphology) play a crucial role in wear processes and the development of surface topography. Analogous to the results of numerical analyses, where the surface evolves toward a steady-state roughness, regardless of its initial condition, the experiments of Szala et al. show that the dynamics of wear result from the complex interaction between the material’s microstructure and the abrasive properties. This effect highlights that traditional roughness measurement methods, which rely solely on averaged values, are insufficient for a comprehensive assessment of the surface condition, particularly in cases where local defects have a critical impact on the functionality of the worn component. In the context of Hardox 500 steel, this means that traditional roughness parameters do not allow for an accurate prediction of surface degradation mechanisms during friction, and their application in wear resistance evaluation may lead to incorrect conclusions. Therefore, it is necessary to employ more advanced surface topography analysis methods, which take into account not only global roughness characteristics but also the actual contact and microstructural transformations occurring during operation.

In conclusion, although traditional roughness parameters provide valuable information about the average surface condition, their limited sensitivity to local topographical variations makes them insufficient for analyzing materials with complex wear structures. Thus, it is essential to expand the research methodology to include techniques that enable a more precise description of local surface features, leading to a better understanding of wear mechanisms and the more effective adjustment of heat treatment parameters and material selection for specific operating conditions.

The obtained results emphasize the importance of selecting appropriate heat treatment parameters for components made from the analyzed material. The statistical analysis of the results confirmed significant differences between the examined heat treatment variants, indicating the material’s sensitivity to changes in cooling intensity, which directly influences microstructure formation. The research, conducted using the described methodology, allowed for the formulation of the following conclusions:

The use of high-intensity cooling media, such as water and oils, resulted in the formation of a martensitic microstructure, characterized by high hardness, which directly translated into significantly increased tribological resistance (k _b = 0.96–1.05). In contrast, cooling in lower-intensity media led to the formation of other hardened structures such as very fine pearlite, bainitic ferrite, and ferrite. The ferrite initially appeared acicular and later transformed into a polygonal form. These microstructures exhibited lower hardness, leading to a decline in tribological resistance (k _b = 0.68–0.79).

Authors state no funding involved.

Martyna Zemlik: Conceptualization, methodology, validation, formal analysis, investigation, resources, data curation, visualization, writing – original draft preparation, supervision. Beata Bialobrzeska: Conceptualization, methodology, validation, formal analysis, investigation, resources, data curation, visualization, writing – original draft preparation, writing – review and editing, supervision. Mateusz Stachowicz: methodology, validation, formal analysis, investigation, data curation. Lukasz Konat: formal analysis, resources, writing – review and editing.

Authors state no conflict of interest.

Data sharing is not applicable to this article as no datasets were generated or analysed during the current study.