2022 Volume 8 Article ID: 2022-0009
In this work, we performed reactive molecular dynamics-based sliding simulations of diamond-like carbon/Fe in the presence of H2O and O2 molecules to analyze tribochemical reaction processes and the atomic-scale wear mechanism. The atomic-scale wear amount in the O2 environment model is smaller than that in the H2O environment model. H2O molecules adsorbed on the surface prevent adhesion between surfaces. However, they are ejected from the contact surface when the high contact pressure was applied, allowing the direct convex-to-convex contact. On the other hand, O2 molecules reacted with the Fe surface, forming a chemically inert oxide layer, thereby leading to preventing atomic-scale adhesive wear.
Wear of sliding materials causes the destruction of machine systems and often leads to fatal accidents. Moreover, it was reported that about 3% of the world’s total energy consumption is used to remanufacture spare parts due to wear-related failures [1]. Therefore, an understanding of the wear mechanism of sliding parts is strongly demanded. In particular, the wear in automobiles engine is of particular interest not only for ensuring the safety but also for improving the durability of automobiles. Especially, the wear at the sliding interface between steel pistons and diamond-like carbon (DLC)-coated cylinders is an important issue because the frictional property of steel pistons and DLC-coated cylinders strongly affects the fuel efficiency of automobiles, thus sliding interface between steel and DLC has been widely studied [2].
The wear amount is often governed by mechanical factors such as contact pressure, sliding speed, surface topography, and lubricant [3, 4]. However, in addition to the mechanical factors, chemical reactions also affect the wear [5]. Especially, the chemical reactions of oxygen and water at the sliding interface are quite important because they always exist in the atmosphere. It was experimentally reported that the wear amount of steel increases as the relative humidity increases, while DLC hardly wears. It was also reported that DLC wears as well as steel in an oxygen environment [6]. These experimental results indicate that oxygen and water react with the sliding parts at the sliding interface and affect the wear. Therefore, it is important to elucidate the chemical reactions of H2O and O2 molecules at the sliding interface. However, it is difficult to reveal the atomic-scale phenomena at the sliding interface from experiments.
Computational simulation is very effective to clarify the atomic-scale phenomena [7,8,9,10,11,12,13,14,15]. Previous molecular dynamics (MD)-based sliding simulations of diamond/iron elucidated the dependency of the wear amount and scratch hardness on the grain size [7]. Density functional theory studies elucidated the initial oxidation process of Fe surface with O2 and H2O molecules [8, 9]. However, there has been no study on the chemical reactions of H2O and O2 at the sliding interface of DLC/Fe, in which chemical reactions and wear are involved.
Recently, reactive MD methods are widely used to analyze tribochemical reactions. J. Yeon et al. investigated the tribochemical reaction at the amorphous SiO2/Si sliding interface and found that the formation of covalent bonds varied depending on the amount of interfacial water [10]. Our group also investigated the wear mechanism of DLC in a hydrogen gas environment and found that the dissociative adsorption of hydrogen molecules on the DLC surface reduced DLC wear [11]. Thus, reactive MD is suitable for investigating the atomic-scale wear processes accompanying the chemical reactions.
In this study, we performed reactive MD-based sliding simulations to analyze the tribochemical reactions at the sliding interface of DLC/Fe in the presence of H2O and O2 molecules and their effect on atomic-scale wear.
Figure 1 shows the sliding simulation model of the hydrogen-free DLC/Fe sliding interface. Here, we used a nascent Fe substrate rather than iron oxide, because, usually, the iron oxide layer is removed due to the sliding by hard DLC. The DLC surface was terminated with H atoms. To take surface-to-surface contact into account, we used the substrates with a sinusoidal shape. For investigating the effects of water and oxygen environments on the atomic-scale wear of DLC and Fe, we prepared the following two models. One was the H2O environment model in which 2100 H2O molecules were placed at the sliding interface, and the other was the O2 environment model in which 2100 O2 molecules were placed at the sliding interface. In this paper, we refer to the former and the latter as the H2O model and O2 model, respectively. The system was equilibrated with a normal load of 1 GPa in the −z-direction up to 12.5 ps, then the sliding simulation was performed. Here we used rather higher contact pressure compared to typical Hertzian contact pressure in experiments to investigate the tribochemical reaction induced by the collision of the asperities [6]. The contact pressure is not uniform due to the microscale surface roughness, and higher where the asperities collide. In the sliding simulation, the lowermost part of the Fe substrate was fixed and the topmost part of the DLC substrate was slid along the x-direction at a speed of 100 m/s with the normal load of 1 GPa. The simulations were performed under the NVT ensemble (363 K). The MD program “Laich,” developed in our laboratory, was used for the simulations. To simulate the chemical reactions involving the chemical bond formation and dissociation, we employed the reactive force field, ReaxFF, developed by van Duin et al. [16].
Sliding simulation model of DLC/Fe.
Figure 2 shows the typical snapshots of the sliding simulations. In the H2O model, H2O molecules adsorbed on the Fe and DLC surfaces, preventing direct surface-to-surface contact during the equilibration process before the sliding simulation. The snapshot at 0 ps of Figure 2 (a) shows the final structure after the equilibration. However, direct contact of the surfaces occurred when the convex parts collided as shown in the snapshot at 75 ps of Figure 2 (a), and then the DLC substrate scraped off the Fe surface. Many Fe–O–H and Fe–H groups were formed on the Fe surface at 300 ps (Figure 2 (a) and Figure 2 (c)). On the other hand, in the O2 model, many O2 molecules reacted with the Fe surface during the equilibration process before the sliding simulation, forming a thick Fe-oxide layer on the Fe surface at 0 ps in Figure 2 (b). The oxide layer consists of Fe–O–Fe groups (Figure 2 (d)). In the sliding simulation, the Fe surface was deformed and flattened. Meanwhile, the O atoms diffused into the bulk region of the Fe substrate at 300 ps of Figure 2 (b).
Snapshots of the sliding simulation for (a) the H2O and (b) O2 models. (c) Fe–H and Fe–O–H on Fe surface of the H2O model at 300 ps (Enlargement of the orange frame in (a)). (d) Fe–O–Fe on Fe surface of the O2 model at 0 ps (Enlargement of the orange frame in (b)).
To investigate chemical reactions at the sliding interface, the number of H2O and O2 molecules, Fe–H, Fe–O–H, and Fe–O–Fe groups were analyzed (Figure 3). The number of C–H, C–O–H, and C–O–C is shown in Supplementary Materials Figure S1 because they are minor. In the H2O model, H2O molecules decreased, whereas Fe–O–H and Fe–O–Fe groups were formed. These changes indicate that the friction induces chemical reactions between H2O and Fe substrate. Besides, the rapid increase of Fe–H groups occurred due to the chemical reactions between the Fe surface and the H atom terminating the DLC surface. In the O2 model, the number of O2 molecules was about 700 at the beginning of the sliding simulation (0 ps), though 2100 O2 molecules were placed when the simulation model was constructed. The reduction of the O2 molecules is due to the chemical reactions with the Fe surface during the equilibration process before the sliding simulation. In consequence, about 2000 Fe–O–Fe groups were formed at the beginning of the sliding simulation (0 ps). The number of O2 molecules further decreased with the sliding and then, at 300ps almost all O2 molecules reacted. Finally, 2100 O2 molecules were transformed into about 4000 Fe–O–Fe groups. A small amount of Fe–H and Fe–O–H groups were also formed by reacting with the H atoms terminating DLC.
Time evolution of the number of H2O molecules (only for the H2O model), O2 molecules (only for the O2 model), Fe–H groups, Fe–O–H groups, and Fe–O–Fe groups in (a) the H2O and (b) O2 models.
In order to analyze the structure of the sliding interface in detail, we classified the oxygen atom belonging to the Fe–O–H group, Fe–O–Fe group, H2O molecule, and O2 molecule based on the chemical bonds of the oxygen atom (Figure 4) by color. In the H2O model, H2O molecules adsorbed on the surfaces and prevented the direct contact of the surfaces. On the other hand, in the O2 model, the oxide layer formed on the Fe surface which directly contacted with the DLC surface.
Enlarged snapshots of the contact area (the red box area in the small upper right Figure) of the sliding simulations for (a) the H2O and (b) O2 models at 50 ps. The O atoms are shown in different colors. Magenta, orange, blue, and purple balls indicate O atoms in the Fe–O–Fe group, Fe–O–H group, H2O molecule, and O2 molecule, respectively.
To evaluate the amount of atomic-scale adhesive wear, pull-up simulations were performed after the sliding simulations (Figure 5). In the pull-up simulation, the lowermost part of the Fe substrate was fixed and the topmost part of the DLC substrate was pulled up in the z-direction with a speed of 100 m/s. Adhesive wear is one of the most important wear processes at the sliding interface [17, 18]. When the DLC substrate was pulled up, Fe and C atoms transferred to the DLC surface and Fe surface respectively, which corresponds to the atomic-scale adhesive wear [9]. Table 1 shows the number of transferred atoms after the pull-up simulations. In both models, the transfer of Fe atoms to the DLC surface was dominant because the DLC is harder than Fe. The number of transferred Fe atoms in the O2 model is smaller than that in the H2O model. This result is consistent with the experimental fact that the wear amount of Fe increases in high humidity environments [6]. Meanwhile, the number of transferred C atoms in the O2 model is larger than that in the H2O model.
Snapshots of the pull-up simulation for (a) the H2O and (b) O2 models.
| model | Number of Fe atoms transferred to DLC substrate | Number of C atoms transferred to Fe substrate | Total |
| H2O | 916 | 119 | 1035 |
| O2 | 540 | 311 | 851 |
Finally, we discuss the reason for the different amounts of transferred atoms in these models. In the H2O model, H2O molecules adsorbed on the Fe and DLC surfaces and prevented surface-to-surface contact (Figure 4). However, when high contact pressure occurred due to the convex-to-convex contact, H2O molecules were ejected from the contact surface and the Fe and DLC surfaces came into contact. Consequently, the soft Fe surface was scraped away with strong adhesion to DLC. In the O2 model, while O2 molecules did not prevent surface contact (Figure 4), O2 molecules oxidized the Fe surface, forming the oxide layer. The oxide layer is chemically inert than the Fe surface, thus direct contact with the DLC surface is unlikely to result in adhesion. Therefore, the inert oxide layer formed by O2 molecules reduces the atomic-scale adhesive wear of the Fe surface. On the other hand, because the oxide layer is rather harder than the nascent Fe surface, the atomic-scale adhesive wear of the DLC surface is prompted in the O2 model.
In this work, we performed reactive MD-based sliding simulations of DLC/Fe in the presence of H2O and O2 molecules to analyze the tribochemical reaction processes and atomic-scale wear mechanisms. The atomic-scale wear amount in the O2 model is smaller than that in the H2O model. H2O molecules adsorbed on the surface prevent adhesion between surfaces. However, they are ejected from the contact surface when the high contact pressure was applied, allowing the direct convex-to-convex contact. On the other hand, O2 molecules reacted with the Fe surface, forming a chemically inert oxide layer, thereby leading to preventing atomic-scale adhesive wear.
This research was supported by the Research Association of Automotive Internal Combustion Engines (AICE), the Japan Science and Technology Agency CREST (Grant No. JPMJCR2191), and Japan Society for the Promotion of Science (JSPS) Grant-in-Aid for Scientific Research (C) (Grant No. 19K05380), Scientific Research (B) (Grant No. 21H01235), and Scientific Research (A) (Grant No. 18H03751). The simulation was performed with the MAterial science Supercomputing system for Advanced MUlti-scale simulations toward NExt-generation–Institute for Materials Research (MASAMUNE-IMR) of the Center for Computational Materials Science, Institute for Materials Research, Tohoku University (Proposal No. 202012-SCKXX-0502).
