MD simulation on the interactions between CH2 groups and the (001) surface of tungsten

This work studies the angle dependence of the interactions between impinging CH2 particles of 150 eV with the tungsten surface. The simulations show that the carbon atoms are much more easily bonded to the tungsten atoms than hydrogen atoms, though a few of the latter can also penetrate into the tungsten material. When the incidence angle is greater than 75°, the incident CH2 particles are reflected without break-ups. Below this angle, a W–C layer of about 0.5 nm is formed with another C, H-rich layer depositing on top of it. The molecular dynamics (MD) approach has proved to be a powerful tool to solve the structural problems at atomic length scale of various materials. Some of its possible applications to the railway track materials have also been discussed.


Introduction
The plasma-facing components (PFCs) are designed to work in an extreme environment exposing to high energy particles and heat flux from the plasma [1]. However, as the designed extreme conditions in the reaction chamber of ITER (International Thermonuclear Experimental Reactor) are still difficult to realize, theoretical simulations become crucial at this stage [2]. Tungsten steel, a composite material containing C and other transition metals, is now widely accepted as one candidate for the first-wall material because of its high melting point, low sputtering rate, good mechanical properties, resistance to radiation, etc. [2,3]. The performance of plasma-facing surface under intense transient thermal loads is another critical issue, since the formation of a melt layer may favor the generation of highly activated dust particles [1,4]. Graphite has been selected as one material in the middle and the lower part of the vertical target of the divertor and as the collection panel in the thermonuclear fusion reactor, for their good outgassing property, low Z character and capability of handling high heat fluxes [2,5]. However, graphite is prone to physical and chemical sputter due to the low sputtering energy threshold [6], which results in many dust particulates. The formation mechanisms, existing forms and their consequences of various kinds of dust particulates resulting from the plasma first-wall interactions have been systematically studied by many authors [7][8][9][10][11]. Among these dust particulates, hydrocarbon radicals have also attracted intensive studies [7][8][9][10][11]. These works shows that carbon atoms can easily combine with hydrogen or its isotope atoms in the plasma chamber leading to the formation of hydrocarbon groups H x C y , which reduces dramatically the performance of the first-wall materials [7-9, 11, 12]. Ohya et al. [13,14] have investigated the interactions of hydrocarbons with the surfaces of the possible first-wall materials (W, C, W-C), as well as their reflections, dissociations and the redepositions by using Monte Carlo (MC) method, and molecular dynamics (MD) method. Because of the torus-like shape of the plasma chamber, the confined highenergy particles are expected to fly in directions shear to the surface of the first wall. This work, thus focuses on the incident angle dependence of the interactions between CH 2 X. Feng Á X. He Á S. Deng (&) Á Y. Zhao Superconductivity and New Energy R & D Center, Southwest Jiaotong University, Chengdu 610031, China e-mail: s.deng@swjtu.edu.cn groups and tungsten surface using MD simulation method. In addition, we have used higher energy flux than the earlier work [13] to investigate the effects of incident energy of such radicals.

Computational details
In our MD calculations, all particles in the system are assumed to abide by Newtonian mechanics. The velocity-Verlet method [15] was used for the integration of Newton's equations of motion.
The force on each atom was calculated from the analytical derivative of the modified Tersoff-Brenner potential [16]. This potential consists of the attractive and repulsive terms as shown in Eqs. (1) and (2), respectively.
where D 0 , r 0 , S indicates the dimer bond energy, the dimer bond distance and the adjustable parameter, respectively. The parameter b in Eqs. (1), (2) can be derived from the ground-state oscillation frequency of the dimer. This potential function has been designed specifically for treating the W-C-H system in non-equilibrium state and has been tested in many cases [4,13,14,17]. An 8 9 8 9 10 supercell of the a -W (body-centered cubic) has been chosen as the initial configuration. This model contains 1,280 W atoms, which has a depth of about 3.165 nm and a top surface area of about 6.412 nm 2 . The (001) plane of a -W is the top surface, i.e., the plasmafacing surface. Periodic boundary conditions are assumed in the a, b directions of the supercell in solving the equations of motion. The bottom layers with a thickness of twounit cell length are fixed in the simulations. In the simulation, the coupling time between the incident particles and the sample is set to be 0.5 ps with a time step of 0.0005 ps. The initial sample temperature is set to be 300 K. In order to maintain a thermal equilibrium of the system, the BERENDSEN's heat bath technique was used with a BenerdsenTau of 0.01 ps [18]. In this work, the linear H-C-H configuration of CH 2 groups is assumed with the bond axes randomly distributed with respect to the incident direction. In all the cases, the incident energy of CH 2 particles is set to be 150 eV. The incident angle as defined in Fig. 1 is as 0°, 5°, 15°, 30°, 45°, 60°, 75°, and 85°, respectively. The incident fluence is chosen to be 4.09 9 10 16 cm -2 , a value comparable to those used in other similar simulation works [13,14].

Results and discussion
In Fig. 2, we show snapshots of the bombarded surface of the sample at the chosen incident fluence with three different incident angles. As shown in Fig. 2, both C atoms and H atoms can penetrate into the tungsten surface resulting in a new layer composed of W, C, and H atoms.
Our simulations indicate that when the incident angle is less than 45°, a thin ''hydrocarbon'' layer consisting of mainly C and H atoms is formed on top of the W-C-H layer. When the incident angle is greater than 45°, only W-C-H layer is formed without the upper covering ''hydrocarbon'' layer. Figure 3 shows the distributions of the various densities of bonds with respect to the depth in reference to the initial surface. As shown in Fig. 3, the original W surface is smeared with the occurrence of W-W bond distribution above the initial surface. The density of the W-C bonds shows peaks in all of the three cases above the initial surface, which drops rapidly below the initial surface. When the incident angle is less than 30°, the formation of C-C and C-H bonds becomes significant, which indicates the breaking of the incident CH 2 groups (see also Fig. 5). In all the cases, the H-H and W-H bonds are negligible in comparison with the other bonds. It is obvious that the bombardment process results in the displacement of W atoms (see Figs. 2, 3) and the formation of W-C bonding (see Fig. 3) etc. Figure 4 shows the relations between the numbers of deposited H (a), C (b) atoms and the number of respective incident particulates at different incident angles. As shown in Fig. 4, when the incident angle is smaller than or equal to 30°for H and 45°for C, respectively, the deposition is more effective. For both C and H atoms, when the incident angle is larger than 45°for H, 75°for C the deposition rate becomes negligibly small regardless of the density of the incident particles. These results indicate that when the incident direction leans toward the tangent of the surface of the W matrix, the kinetic energy of the incident particles exceeds their bound energy, which invalidates any trapping of the incident particles, and thus the reflection dominates. Figures 3 and 4 clearly show that C atoms are much more easily trapped by the W matrix indicating a stronger bonding between W and C atoms than that between W and H atoms. Figure 5 shows the incident angle dependence of the relative scattering ratio (RSR) of some important particles produced in the interaction process and the remained CH 2 groups. For clearance, the logarithmic ordinate is used for the RSRs of the indicated particles. As can be seen from Fig. 5, the RSR of C atoms is the lowest in a wide range of incident angles among the scattered particles. Furthermore it has very weak incident angle dependence. In contrast, the RSR of H atoms is the largest among the investigated particles, except for CH 2 radicals whose RSR surpasses all those of the other particles when the incident angle exceeds *35°, which indicates the dominance of reflection in this range. Considering the following relevant process in the Fig. 3 The bond density distribution along the vertical direction with respect to the initial W (001) surface interactions between the incident CH 2 groups and the W surface, One can expect that the RSR of C atoms should be approximately two times smaller than that of H atoms, if the above three processes have the similar probability and the relevant particulates are scattered out in the same efficiency. As can be seen from Fig. 5, the RSR of H atoms far exceeds that of C atoms by approximately one order. This may also be possible if the first process dominates, however, considering the results as shown in Figs. 3 and 4, it can be neglected. Taking into account the results shown in Fig. 3, we attribute the origin to be the stronger W-C bonding than the W-H boning. It should be pointed out that the dramatic drop of the scattering ratios for H, C, and CH particles at high incident angles ([75°) results from the reflectance of CH 2 particles, which reduces the source of all C, H, and CH particles.

Conclusions
Our simulations show that the deposition ratio of C and H atoms has a strong dependence on the incident angle. When the incident angle exceeds a threshold value the composite particle CH 2 is mainly reflected. The carbon atoms are much more easily bonded to the tungsten atoms than hydrogen atoms, though a few of the latter can also penetrate into the tungsten sample. When the incidence angle is greater than 75°, the kinetic energy of the particles exceeds significantly the bound energy. Below this angle, an effective W-C bonding occurs, which results in a W-C layer of about 0.5 nm with another C, H-rich layer depositing on top of it. It is clear that our studies on the interactions between the plasma and the first-wall materials are still not complete. Some interesting issues such as the consequences of choosing different hydrocarbon groups, the different crystal plane of a -W, e.g., (011), (111), etc., the different incident energy of the particulates, etc., are also very important in understanding the interactions between the plasma and the first-wall materials. It is thus worthy of further studies to clarify these issues.
It should be pointed that our preliminary study has mainly involved the interactions between the possible particles produced in the fusion chamber and the first-wall materials, i.e., tungsten steel. The role of carbon and the other elements in enhancing and improving the properties of a specific material has not been studied, though it is very important, otherwise we even do not have the concept of steel. The point raised here is that chemical composition does not uniquely decide the properties of a steel, in almost  The scattering ratio of the number of scattered particles to the total number of scattered particles all cases, the treatment processes such as heat treatment processes are also very crucial. This implies that the microstructure of a material is the dominating factor.
For example, the microstructure of rail steel is composed of pearlite, while that of low carbon steel used for railway sleepers consists of ferrite and cementite [19]. The properties, such as the hardness, wear resistance, tensile strength, etc., depend actually on the multi-scale structures, say, the composition, the proportion of different structural domains and the details of each domain. The MD simulation, a powerful tool to study the structure at the atomic length scale, has proved to be an efficient approach, as we have shown in this work, to reveal the relations between the microstructures of various complicated steels and their properties. Our future work will be done along this line for both fusion reactor materials and railway track materials.