- Email: [email protected]
Atomistic simulation of Al-graphene thin film growth on polycrystalline Al substrate
Accepted Manuscript Title: Atomistic simulation of Al-graphene thin film growth on polycrystalline Al substrate Authors: Lan Zhang, Yongchao Zhu, Na Li, Yan Rong, Huimin Xia, Huizhong Ma PII: DOI: Reference:
S0169-4332(17)32900-8 https://doi.org/10.1016/j.apsusc.2017.09.241 APSUSC 37320
To appear in:
APSUSC
Received date: Revised date: Accepted date:
14-7-2017 18-9-2017 28-9-2017
Please cite this article as: Lan Zhang, Yongchao Zhu, Na Li, Yan Rong, Huimin Xia, Huizhong Ma, Atomistic simulation of Al-graphene thin film growth on polycrystalline Al substrate, Applied Surface Science https://doi.org/10.1016/j.apsusc.2017.09.241 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Atomistic simulation of Al-Graphene thin film growth on polycrystalline Al substrate
Lan Zhang a,*, Yongchao Zhu a, Na Li a, Yan Rong a, Huimin Xia a, Huizhong Ma a,* a
School of Mechanics and Engineering Science, Zhengzhou University, Zhengzhou 450001, China
* Corresponding author. E-mail address: [email protected]
Highlights
We performed a molecular dynamics simulation of the growth of Al-Graphene composite coatings on polycrystalline Al substrate for the first time. The diffusion behaviors and growth process of Al adatoms on polycrystalline surface are investigated in detail, which give rise to the volmer-weber growth mode. The effects of the reinforcing phase in the deposition process are firstly studied in atomic scale by MD simulations. The changes in the morphology of composite coatings and grain sizes, which are consistent with some experimental reports, are made clear by the atomic details.
Abstract The growth of Al-Graphene composite coatings on polycrystalline Al substrate was investigated by using classical molecular dynamics (MD) simulations. Unlike the diffusion behaviors on single crystal surface, most of adatoms were easily bound by the steps on polycrystalline Al surface, owing to the local accelerated energy. Both Ehrlich-Schwoebel (ES) barriers and the steering effect backed up the volmer-weber growth mode, which was consistent with the dynamic growth process observed in the deposit. The morphology of composite coatings was significantly affected by graphene flakes. Enrichment of graphene flakes gave rise to an increase of the local thickness, and graphene flakes only existed in Al grain boundaries. The size of Al grains in the composite coating visibly decreased when compared with that in the pure Al coating. This grain refinement and the mechanical property can be reinforced by the increase of graphene flakes. Keywords: MD; graphene; deposition; steering effect; grain refinement 1. Introduction Metal matrix composite coatings containing nanosized particles used as reinforcement phase have attracted considerable interest in the field of material science and manufacturing, due to the improved mechanical properties. As known to all, the mechanical performances are directly related to the film structure, which is heavily influenced by the evolution of surface morphology
and interfacial properties in the deposit. Therefore, the understanding in the growth mechanism is of critical importance to the preparation of high-performance films. Experimentally, the characterization results of morphology and microstructure of deposited films can be demonstrated by scanning electron microscopy, energy diffraction spectrum, X-ray diffraction (XRD), etc. By use of these analytical techniques, Li et al. found that the insertion of graphene flakes in Al coatings changed the surface morphology observably with a number of asperities [1], and Zhang et al. revealed that graphene nano-platelets wrapped by Ni–Fe crystalline grains were embedded into the matrix metal acting as an effective reinforcement [2]. XRD analysis conducted by Chen et al. indicated that the average crystallite size in the Ni-graphene composite coating decreased with the increase of addition amount of graphene [3]. This grain refinement made a significant contribution to the increase in hardness and the improvement in elastic modulus [1-3]. Through X-ray photoelectron spectroscopy (XPS) analysis on Fe-Ni-Graphene composite, Subramanya et al. concluded that the addition of graphene increased the metallic nickel content in the deposit [4]. Nevertheless, these experimental investigations generally lay emphasis on specific results or phenomena. The dynamic growth processes can’t be detailed by experimentation, especially in atomic scale. By contrast, MD simulations are more suitable for exploring such processes. Kim et al. have studied the diffusion behaviors of Al on a Cu surface and found that these behaviors strongly depended on the substrate surface orientation [5]. Lee et al. have investigated Al and Ni thin film growth on Ni(111) surface and observed that sufficient kinetic energy can result in the smoother surface [6]. Cao et al. have demonstrated the deposition process of Al thin film on Cu substrate and argued that the morphology of deposited thin film was affected significantly by defects formed during deposition process [7]. Based on TB-SMA potential, Hong et al. revealed that the effect of the incident angle on the surface roughness was very small, but the surface roughness can be improved by increasing the incident energy and substrate temperature [8]. By combining MD and kinetic Monte Carlo (KMC) simulations, Seo et al. have indentified the steering effect during homoepitaxial growth of Cu on Cu(001), which led to preferential arrival of atoms on top of islands [9]. In general, these simulations concentrated upon the deposition process on single crystal substrate and effects of incident energy on surface morphology. Moreover, effects of the reinforcement phase in the deposition process have not been studied by MD simulations, as well as the complex structure of composite coatings. In this study, a large scale classical MD simulation is performed to study Al deposition on polycrystalline Al substrate, using the LAMMPS code. The effects of single layer graphene (SLG) flakes on metal coatings are also researched through static calculations of designed structures and measurements of the local crystal structure. The atomistic details of graphene in metal coatings can account for the changes of local structures and grain sizes, which are crucial to the mechanical property of metal-graphene composite coatings and are impossible to be observed visually by experimental works. So, this study is expected to avail the preparation of such high-performance films. 2. Molecular dynamics method 2.1 Simulation methodology for deposition process Fig. 1 shows atomistic model of deposition, which consists of an Al substrate, inserted Al atoms and SLG flakes. Here, a polycrystalline bulk created by Voronoi geometrical construction was used as the uniform substrate, which is comprised of 10 grains taking into account the periodic
boundary conditions (PBC). The dimension of this substrate in each direction is about 80 Å. In MD simulations, PBC is applied only to the lateral directions of X and Y to mimic the semi-infinite surface. The bottom atoms whose Z-coordinate value are less than 10 Å were fixed to prevent the substrate from moving, while all other layers are unconstrained and fully relaxed at 333 K in NVT ensemble. The deposition processes are simplified as the successive additions of a single Al atom and a SLG flake comprised of 22 carbon atoms (green network in Fig. 1). Every added atom or flake is randomly placed on a plane at a distance sufficiently far from the upper surface of the substrate. In every deposition process, 6000 Al atoms are deposited onto the same substrate in 12 ns with the different amount of SLG flakes (0, 20, 40, 60, 80, and 100), and the time intervals between two consecutive additions of Al atoms or SLG flakes are identical. It is important to note that the random number seeds of Al deposition in all deposition processes are the same. That is, the initial positions of added Al atoms in every deposition process are unchanged. The NVT ensemble is also adopted for the system to maintain 333 K, which is in the range of temperatures selected by electrodeposited experiments [1-4, 10]. The incident energy of each atom or flake is consistent with this temperature, and the incident direction is vertically downward. During all MD simulations, Newton’s equation of motion is integrated by using the velocity Verlet algorithm, and the time step is set to 1.0 femtosecond. The eam potential is applied to the interaction between Al atoms [11], the morse potential is used to simulate the deposition behavior in the Al-C system [12], and the airebo pair style is used to describe the interatomic interaction of carbon atoms [13]. 2.2 Static calculations The static calculations based on MD simulations are employed to calculate diffusion energy barriers for Al atoms. Single crystal Al bulks with orientations of (001), (011) and (111) are respectively used as the substrates for the present calculations. Take for example the (001) surface that is schematically shown in Fig. 2a. A half of atoms on the upper surface are removed to form an atomic terrace. To investigate the effect of SLG on diffusion behaviors of adatoms, a SLG flake is placed on the flat Al surfaces to gain new models after energy minimization (Fig. 2b). In any cases, the stable positions on these designed surfaces can be easily found, and 10 points along the kinetic path between two neighboring stable positions are selected. An added atom is placed at the points in turn, and the lowest potential energy of the entire structure at each point can be figured out by varying the height of the added atom, meanwhile the local accelerated energy for Al atoms at these points can be obtained. Additional points are checked to ensure the accuracy of the kinetic path. Thus, the diffusion energy barriers can be calculated from the lowest energy-path curve. 2.3 Nanoindentation process To investigate the mechanical property of achieved coatings, atomistic simulations of nanoindentation process were conducted. A diamond sphere of 50Å in radius was used as the indenter, which was located at the right above of the coatings. The interactions between the diamond indenter and Al-graphene coatings take the Lennard-Jones style. The LJ values for Al-Cdiamond is defined as σ = 3.01 Å, ε = 0.03438 eV [14]; and the LJ values for Cgraphene-Cdiamond is defined as σ = 3.4 Å, ε = 0.00284 eV [14]. During the indentation process, the loading force (Ⅰ) gradually increased from 0 to 50nN in 200ps, (Ⅱ) remained unchanged in another 200ps, and (Ⅲ) was completely unloaded in the third 200ps. On basis of the force-displacement curves of the indenter, the hardness (H) of coatings can be calculated as below [15]:
𝐻=
𝑃𝑚𝑎𝑥 𝐴𝑐
(1)
where Pmax represents the maximum indentation force and Ac is the contact area, as expressed by: 𝐴𝑐 = 𝜋𝑅ℎ𝑐 ℎ𝑐 = ℎ𝑚𝑎𝑥 − 0.72
(2) 𝑃𝑚𝑎𝑥 𝑆𝑚𝑎𝑥
(3)
where hmax is the maximum penetration depth and Smax is the maximum slope of the unloading curve. 3. Results and discussion 3.1 Deposition process of the pure Al film Fig. 3 shows the upper surface of polycrystalline Al substrate before and after relaxation. Because of inner stress in the initial configuration, the surface morphology of the substrate has changed obviously, giving rise to many surface defects. In the first deposition process, only Al atoms are deposited onto this substrate. Because the incident energy is very low, there is no insert atom penetrated into substrate. When Al atoms become attached to the substrate surface, the vacancies are apt to be filled with the adatoms (fig. 4a), and the steps are able to promote the agglomeration of adjacent adatoms (fig. 4b). Only a few single Al atoms deposited on some plat surfaces such as Al (001) seem not to diffuse actively (fig. 4c). The similar behaviors are also observed in Al deposition on single crystal Cu surface, which formed an atomic layer not an island in the end [5]. However, the deposition process proves that the growth of Al film on polycrystalline Al surface followed the volmer-weber mode, as shown in Fig. 5. In order to investigate this growth mechanism, we calculate the diffusion energy barriers of Al atoms near the step through molecular static calculations (Fig. 2a), and the energy barriers are summarized in table 1. According to the diffusion behaviors observed on Al (001) surface, the energy barrier of 0.2707 eV is high enough to constrain adatoms at 333 K. Thus, the surface morphology should reflect the randomness of the position of Al deposition. However, adatoms could move a certain distance before their kinetic energy depleted. Fortunately, the local accelerated energy at any positions, which is expected to be more than 2 eV for Al atoms on account of the static calculations, can supply considerable energy. So, most adatoms are captured by these steps, since there were a mass of steps on the polycrystalline surface and the site along steps are very stable by reason of the high diffusion barriers away from steps. However, it is more difficult for adatoms to diffuse over the steps, because all ES barriers are much higher than surface diffusion energy barriers, and this implies a significant uphill current. Furthermore, the adatom can be attracted to the top terrace and diffuses away from the step, as shown in Fig. 6. This steering effect of steps also plays an important role in epitaxial growth at normal incidence [9,16]. Thus, some Al islands are able to be formed on polycrystalline surface. 3.2 Deposition process of the Al-graphene film During the second deposition process, 40 graphene flakes are deposited with Al atoms. Fig. 7a suggests that many Al atoms can be absorbed by graphene flakes before they arrived on the substrate surface. Thus, the area where more flakes deposited should be thicker, since the random
number seeds of Al deposition in all deposition processes are the same. Moreover, Al atoms deposited on graphene flakes are inclined to form the close-packed plane (Fig. 7b), due to the honeycomb lattice of graphene. It means the arrangement of Al atoms near graphene flakes could be disorganized, leading to the formation of more defects such as grain boundary, which occupied more spaces than close packing. The effect of defects on thickness is also observed by changing the incident energy [7]. Fig. 8 presents the surface morphology of composite coating and the distribution of graphene flakes. The thickness in the place where more graphene flakes are deposited increases clearly, compared with the pure Al coating surface. Therefore, we expect that the enrichment of graphene flakes can result in an increase of the local thickness. A series of static simulations are employed to support the conclusion above. The diffusion energy barriers of Al atoms on graphene flakes which lie on Al substrate (Fig. 2b) are summarized in table. 2. Both the surface diffusion barriers and ES barriers in any case are very high. So, Al atoms on graphene flakes are stable enough to form new crystallographic orientation, which can hinder the growth of adjacent grains. Consequently, the arrangement of Al atoms surrounding the graphene is heavily disturbed, and the increase of disorganized atoms will form new grain boundaries in the end. Thus, the size of Al grains is limited, and graphene flakes just exist in the grain boundaries. The fact graphene nano-platelets can enhance nucleation sites by creating a disorder in the matrix and retard the crystal growth is also seen in the electro-deposition experiment [3]. Herein, the Centro-Symmetry Parameter (CSP) analysis is conducted to indentify the local atomic configuration which can indicate this grain refinement. As shown in Fig. 9, the substrate Al atoms are removed, the inner structures of coatings in the bottom view are colored by CSP values. It can be found that graphene flakes only appear in grain boundaries and the sizes of Al grains in the composite coating visibly decrease. 3.3 Atomic structures and mechanical property of Al-graphene films Al thin films formed with different numbers of SLG flakes are studied as well. It should be noted that these simulations are carried out regardless of the dispersibility of graphene. The columns in Fig. 10 present compositions of atomic structures in deposited coatings at different numbers of graphene flakes, based on the Common Neighbor Analysis (CNA). Atoms in local f.c.c. (face-centred cubic) order are considered as perfect grains, atoms in local h.c.p. (hexagonal close-packed) order are indentified as stacking faults, and all other atoms are classified as grain boundaries [17]. The increase of graphene flakes leads to an obvious decrease of Al (f.c.c) atoms in the thin films. In light of the inner structures of coatings (Fig. 9), the numbers of Al grains can be considered the same. So, the change in the quantity of Al f.c.c. atoms can represent that in the average volume of Al grains. As more graphene flakes existed in the coatings, the number of f.c.c. atoms will reduce from N0 to Ni. Thus, the relative crystalline size (lr) can be simply evaluated as follow: 𝑙𝑟 = 3√𝑁𝑖 / 3√𝑁0, and the changing curve of lr is also shown in Fig. 10. This trend could further illustrate the effect of the grain refinement, which is well consistent with the experiment studies on electrodeposited Al-graphene composite coatings [1-2]. In experimental studies, the Sherrer equation is generally used to evaluate the average crystalline size. Li etc. have prepared an electrodeposited Al-graphene composite coating with the crystalline size of 17.36 nm, which is about 20% smaller than that of pure Al coating [1]. In accordance of the curve of relative crystalline size, about 30 graphene flakes can achieve the same effect in MD simulations. The mechanical property of those composite coatings is measured by nanoindentation simulations. On account of the complex structure, three hardness data points are averaged for each
coating. The plane coordinates of the indenter centre are (0 Å, 0 Å), (20 Å, 20 Å), and (40 Å, 40 Å) respectively. The force-displacement curve at (40 Å, 40 Å) and the results of hardness are shown in Fig. 11. A small amount (such as 40) of graphene flakes brings about very small changes, the hardness of composite (2.08 GPa) is a little higher than that of pure aluminum coating (2.03 GPa). When the number of graphene flakes reaches 100, there is an obvious increase (about 20%) in hardness. It’s worth noting that this increase is accompanied by the inverse Hall–Petch effect [17-18], since the Al grains have reached a small enough size (much less than 10 nm) according to the dimensions and inner structures of such coatings (Fig. 9). Hence, synthesized Al-graphene composite coating with the crystalline size of 17.36 nm even can exhibit 3.8 folds increase in the hardness (Fig. 12) [1]. However, too many graphene flakes will lead to a decline in mechanical property because of the graphene agglomeration (Fig. 12) [2]. In these simulations, all graphene flakes are added at the same time interval, so there is no drop in hardness. Therefore, it’s safe to say that the growing of graphene numbers will make a remarkable advance in the mechanical property of composite coatings. 4 Conclusions Using MD method, Al deposition on polycrystalline Al substrate is investigated. The vacancies on the substrate surface could be filled with the adatoms. Most deposited atoms agglomerate along the steps, due to the local accelerated energy which can improve the mobility of deposited atoms at early stage. Because of ES barriers and the steering effect, Al atoms seem to follow the volmer-weber growth mode. The morphology of composite coatings is significantly affected by graphene flakes, which are able to absorb Al atoms during the deposit and gave rise to the formation of more grain boundaries. Static calculations and the atomic configuration based on CSP analysis can account for the grain refinement of graphene on Al thin film, and the CNA analysis indicates that the growing of graphene numbers can strengthen this grain refinement. Nanoindention results show that the increase of well dispersed graphene flakes will also make a great improvement to the mechanical property of Al-graphene composite films. Acknowledgements This work was supported by Program for New Century Excellent Talents in University (NCET-11-0951), National Natural Science Foundation of China(61350006), Zhengzhou Leading Talent Project (131PLJRC655), Basic and frontier technology research program of Henan Province (162300410026)
Reference [1] N Li, L Zhang, MT Xu, T Xia, XW Ruan, S Song, HZ Ma, Preparation and mechanical property of electrodeposited Al-graphene composite coating, Materials & Design 111(2016):522-527 [2] L Zhang, N Li, HM Xia, JY Zhang, PP Zhang, MT Xu, HZ Ma, Preparation and Mechanical Properties of (Ni–Fe)–Graphene Composite Coating, Advanced Engineering Materials 18 (2016):1716-1719 [3] J Chen, J Li, D Xiong, Y He, Y Ji, Y Qin, Preparation and tribological behavior of Ni-graphene compositecoating under room temperature, Applied Surface Science 361 (2016):49-56 [4] S Badrayyana, DK Bhat, S Shenoy, Y Ullal, AC Hegde, Novel Fe-Ni-Graphene composite electrode for hydrogen production, International Journal of Hydrogen Energy 40 (2015):10453-10462 [5] SP Kim, KR Lee, YC Chung, YK Kim, M Doi, Sahashi M, A Molecular Dynamics Study of the Deposition and the Diffusion Behaviors of Al on a Cu Surface, Journal- Korean Physical Society 52 (2008):68-35 [6] SG Lee, YC Chung, Atomic-level investigation of Al and Ni thin film growth on Ni(111) surface: Molecular dynamics simulation, Applied Surface Science 253 (2007):8896-8900 [7] Y Cao, J Zhang, T Sun, Y Yan, F Yu, Atomistic study of deposition process of Al thin film on Cu substrate, Applied Surface Science 256 (2010):5993-5997 [8] ZH Hong, SF Hwang, TH Fang, Atomic-level stress calculation and surface roughness of film deposition process using molecular dynamics simulation, Computational Materials Science 48 (2010):520-528 [9] J Seo, SM Kwon, HY Kim, JS Kim, Steering effect on the shape of islands for homoepitaxial growth of Cu on Cu(001), Physical Review B 67 (2002):181-183 [10] Kan HM, Zhu SS, Zhang N, Wang XY, Electrodeposition of aluminum and aluminum magnesium alloys at room temperature, Journal of Central South University 22(2015): 3689-3697 [11]KW Jacobsen, JK Norskov, MJ Puska, Interatomic interactions in the effective-medium theory, Phys Rev B Condens Matter 35 (1987):7423-7442 [12] P Peng, G Liao, T Shi, Z Tang, Y Gao, Molecular dynamic simulations of nanoindentation in aluminum thin film on silicon substrate, Applied Surface Science 256 (2010):6284-6290 [13] D.W. Brenner, O.A. Shenderova, J.A. Harrison, S.J. Stuart, B. Ni, S.B. Sinnott, A second-generation reactive empirical bond order (REBO) potential energy expression for hydrocarbons, J. Phys.-Condens. Matter 14 (2002):783–802 [14] Choi BK, Yoon GH, Lee S, Molecular dynamics studies of CNT-reinforced aluminum composites under uniaxial tensile loading, Composites Part B 91(2016):119-125 [15] Zhao YB, Peng XH, Fu T, Sun R, Feng C, Wang ZC, MD simulation of nanoindentation on (001) and (111) surfaces of Ag–Ni multilayers, Physica E 74(2015):481-488 [16] S. Van Dijken, L.C. Jorritsma, B. Poelsema, Steering-Enhanced Roughening during Metal Deposition at Grazing Incidence, Physical Review Letters 82(1999):4038-4041 [17] J Schiøtz, FDD Tolla, KW Jacobsen, Softening of nanocrystalline metals at very small grain sizes, Nature, 391 (1998):561-563 [18] H Conrad, J Narayan, On the grain size softening in nanocrystalline materials, Scripta Materialia, 42 (2000):1025-1030
Fig. 1. Atomistic model of the deposition process.
Fig. 2. The possible diffusion behaviors of Al adatoms on designed surfaces. Silver balls represent Al atoms in the bottom, yellow balls represent an Al terrace, and the blue ball represents an Al adatom; Red arrows stand for the kinetic path, and the green network stands for a SLG flake. (a) diffusion behaviors near a step (b) diffusion behaviors on SLG
Fig.3. The upper faces of the substrate before (a) and after (b) relaxation
(a)
(b)
(c)
Fig. 4. The diffusion behaviors observed on the polycrystalline Al substrate. Blue balls represent Al atoms in the substrate, silver balls represent deposited Al atoms, and other colorful balls stand for adatoms which have just been adsorbed. (a) a vacancy is filled; (b) agglomeration along a step that is marked by a blue line; (c) a stable atoms on Al(001) surface
Fig. 5. Atomic configurations in the deposition process. Blue balls represent Al atoms in the substrate which are not fully displayed here, silver balls represent deposited Al atoms, (a)-(c) the side views; (d) the surface morphology colored by height (in Å).
Fig. 6. The steering effect observed in the deposition process. The yellow ball represents an Al atom depositing onto the surface, and the blue line stands for a step. (a-c) snapshot and details in 3.32ns, (d-e) details in 3.33ns
(a)
(b)
Fig.7. The absorption of Al atoms on SLG flakes. (a)the adsorption before arrivals on surface, (b)the arrangement of Al atoms on a SLG flake
Fig.8. The effect of SLG flakes on surface morphology. (a)the distribution of SLG in composite coating. Points represent all Al atoms, and green networks marked by yellow circles represent SLG; (b) the surface morphology of composite coating; (c) the surface morphology of pure Al coating. (b) and (c) are colored by height (in Å).
Fig.9. Distribution of CSPs in the bottom view of Al coatings. (a) the inner structure of composite coating; (b) the inner structure of pure Al coating. Blue balls stand for Al atoms with bulk lattice, other colorful balls are classified as grain boundaries, and red networks in (a) represent SLG.
Fig.10. Compositions of atomic structure and the relative crystalline sizes with respect to the number of SLG in composite coatings
(a)
(b)
Fig.11. The nanoindention results with respect to the number of SLG. (a) force-displacement curve at (40 Å, 40 Å), (b) the value of hardness
Fig.12. The referenced microhardness of metal-graphene composite coatings synthetized in the bath solution at a variety of graphene concentration [1-2].
Table.1 the diffusion energy barriers of Al atoms near the step (in eV) Surface
Al (001)
Al (011)
Al (111)
Surface diffusion
0.2707
0.2996
0.0446
Away from the step
0.6330
0.4973
0.4815
ES barrier
0.4516
0.4531
0.2932
Table.2 The diffusion energy barriers of Al atoms on graphene flakes which lie on Al substrate (in eV) Surface Terrace edge
Surface diffusion
ES barrier
SLG on Al(001)
SLG on Al (011)
SLG on Al (111)
Armc
Armc
Armc
hair
Zigza g
0.742 3
0.677 9
0.995 7
hair
g 0.653
1 0.978
1
Zigza
0.492 3
0.909 7
hair
g 0.634
3 0.929
6
Zigza
0.765 0
0.663 7
1.040 8
S0169-4332(17)32900-8 https://doi.org/10.1016/j.apsusc.2017.09.241 APSUSC 37320
To appear in:
APSUSC
Received date: Revised date: Accepted date:
14-7-2017 18-9-2017 28-9-2017
Please cite this article as: Lan Zhang, Yongchao Zhu, Na Li, Yan Rong, Huimin Xia, Huizhong Ma, Atomistic simulation of Al-graphene thin film growth on polycrystalline Al substrate, Applied Surface Science https://doi.org/10.1016/j.apsusc.2017.09.241 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Atomistic simulation of Al-Graphene thin film growth on polycrystalline Al substrate
Lan Zhang a,*, Yongchao Zhu a, Na Li a, Yan Rong a, Huimin Xia a, Huizhong Ma a,* a
School of Mechanics and Engineering Science, Zhengzhou University, Zhengzhou 450001, China
* Corresponding author. E-mail address: [email protected]
Highlights
We performed a molecular dynamics simulation of the growth of Al-Graphene composite coatings on polycrystalline Al substrate for the first time. The diffusion behaviors and growth process of Al adatoms on polycrystalline surface are investigated in detail, which give rise to the volmer-weber growth mode. The effects of the reinforcing phase in the deposition process are firstly studied in atomic scale by MD simulations. The changes in the morphology of composite coatings and grain sizes, which are consistent with some experimental reports, are made clear by the atomic details.
Abstract The growth of Al-Graphene composite coatings on polycrystalline Al substrate was investigated by using classical molecular dynamics (MD) simulations. Unlike the diffusion behaviors on single crystal surface, most of adatoms were easily bound by the steps on polycrystalline Al surface, owing to the local accelerated energy. Both Ehrlich-Schwoebel (ES) barriers and the steering effect backed up the volmer-weber growth mode, which was consistent with the dynamic growth process observed in the deposit. The morphology of composite coatings was significantly affected by graphene flakes. Enrichment of graphene flakes gave rise to an increase of the local thickness, and graphene flakes only existed in Al grain boundaries. The size of Al grains in the composite coating visibly decreased when compared with that in the pure Al coating. This grain refinement and the mechanical property can be reinforced by the increase of graphene flakes. Keywords: MD; graphene; deposition; steering effect; grain refinement 1. Introduction Metal matrix composite coatings containing nanosized particles used as reinforcement phase have attracted considerable interest in the field of material science and manufacturing, due to the improved mechanical properties. As known to all, the mechanical performances are directly related to the film structure, which is heavily influenced by the evolution of surface morphology
and interfacial properties in the deposit. Therefore, the understanding in the growth mechanism is of critical importance to the preparation of high-performance films. Experimentally, the characterization results of morphology and microstructure of deposited films can be demonstrated by scanning electron microscopy, energy diffraction spectrum, X-ray diffraction (XRD), etc. By use of these analytical techniques, Li et al. found that the insertion of graphene flakes in Al coatings changed the surface morphology observably with a number of asperities [1], and Zhang et al. revealed that graphene nano-platelets wrapped by Ni–Fe crystalline grains were embedded into the matrix metal acting as an effective reinforcement [2]. XRD analysis conducted by Chen et al. indicated that the average crystallite size in the Ni-graphene composite coating decreased with the increase of addition amount of graphene [3]. This grain refinement made a significant contribution to the increase in hardness and the improvement in elastic modulus [1-3]. Through X-ray photoelectron spectroscopy (XPS) analysis on Fe-Ni-Graphene composite, Subramanya et al. concluded that the addition of graphene increased the metallic nickel content in the deposit [4]. Nevertheless, these experimental investigations generally lay emphasis on specific results or phenomena. The dynamic growth processes can’t be detailed by experimentation, especially in atomic scale. By contrast, MD simulations are more suitable for exploring such processes. Kim et al. have studied the diffusion behaviors of Al on a Cu surface and found that these behaviors strongly depended on the substrate surface orientation [5]. Lee et al. have investigated Al and Ni thin film growth on Ni(111) surface and observed that sufficient kinetic energy can result in the smoother surface [6]. Cao et al. have demonstrated the deposition process of Al thin film on Cu substrate and argued that the morphology of deposited thin film was affected significantly by defects formed during deposition process [7]. Based on TB-SMA potential, Hong et al. revealed that the effect of the incident angle on the surface roughness was very small, but the surface roughness can be improved by increasing the incident energy and substrate temperature [8]. By combining MD and kinetic Monte Carlo (KMC) simulations, Seo et al. have indentified the steering effect during homoepitaxial growth of Cu on Cu(001), which led to preferential arrival of atoms on top of islands [9]. In general, these simulations concentrated upon the deposition process on single crystal substrate and effects of incident energy on surface morphology. Moreover, effects of the reinforcement phase in the deposition process have not been studied by MD simulations, as well as the complex structure of composite coatings. In this study, a large scale classical MD simulation is performed to study Al deposition on polycrystalline Al substrate, using the LAMMPS code. The effects of single layer graphene (SLG) flakes on metal coatings are also researched through static calculations of designed structures and measurements of the local crystal structure. The atomistic details of graphene in metal coatings can account for the changes of local structures and grain sizes, which are crucial to the mechanical property of metal-graphene composite coatings and are impossible to be observed visually by experimental works. So, this study is expected to avail the preparation of such high-performance films. 2. Molecular dynamics method 2.1 Simulation methodology for deposition process Fig. 1 shows atomistic model of deposition, which consists of an Al substrate, inserted Al atoms and SLG flakes. Here, a polycrystalline bulk created by Voronoi geometrical construction was used as the uniform substrate, which is comprised of 10 grains taking into account the periodic
boundary conditions (PBC). The dimension of this substrate in each direction is about 80 Å. In MD simulations, PBC is applied only to the lateral directions of X and Y to mimic the semi-infinite surface. The bottom atoms whose Z-coordinate value are less than 10 Å were fixed to prevent the substrate from moving, while all other layers are unconstrained and fully relaxed at 333 K in NVT ensemble. The deposition processes are simplified as the successive additions of a single Al atom and a SLG flake comprised of 22 carbon atoms (green network in Fig. 1). Every added atom or flake is randomly placed on a plane at a distance sufficiently far from the upper surface of the substrate. In every deposition process, 6000 Al atoms are deposited onto the same substrate in 12 ns with the different amount of SLG flakes (0, 20, 40, 60, 80, and 100), and the time intervals between two consecutive additions of Al atoms or SLG flakes are identical. It is important to note that the random number seeds of Al deposition in all deposition processes are the same. That is, the initial positions of added Al atoms in every deposition process are unchanged. The NVT ensemble is also adopted for the system to maintain 333 K, which is in the range of temperatures selected by electrodeposited experiments [1-4, 10]. The incident energy of each atom or flake is consistent with this temperature, and the incident direction is vertically downward. During all MD simulations, Newton’s equation of motion is integrated by using the velocity Verlet algorithm, and the time step is set to 1.0 femtosecond. The eam potential is applied to the interaction between Al atoms [11], the morse potential is used to simulate the deposition behavior in the Al-C system [12], and the airebo pair style is used to describe the interatomic interaction of carbon atoms [13]. 2.2 Static calculations The static calculations based on MD simulations are employed to calculate diffusion energy barriers for Al atoms. Single crystal Al bulks with orientations of (001), (011) and (111) are respectively used as the substrates for the present calculations. Take for example the (001) surface that is schematically shown in Fig. 2a. A half of atoms on the upper surface are removed to form an atomic terrace. To investigate the effect of SLG on diffusion behaviors of adatoms, a SLG flake is placed on the flat Al surfaces to gain new models after energy minimization (Fig. 2b). In any cases, the stable positions on these designed surfaces can be easily found, and 10 points along the kinetic path between two neighboring stable positions are selected. An added atom is placed at the points in turn, and the lowest potential energy of the entire structure at each point can be figured out by varying the height of the added atom, meanwhile the local accelerated energy for Al atoms at these points can be obtained. Additional points are checked to ensure the accuracy of the kinetic path. Thus, the diffusion energy barriers can be calculated from the lowest energy-path curve. 2.3 Nanoindentation process To investigate the mechanical property of achieved coatings, atomistic simulations of nanoindentation process were conducted. A diamond sphere of 50Å in radius was used as the indenter, which was located at the right above of the coatings. The interactions between the diamond indenter and Al-graphene coatings take the Lennard-Jones style. The LJ values for Al-Cdiamond is defined as σ = 3.01 Å, ε = 0.03438 eV [14]; and the LJ values for Cgraphene-Cdiamond is defined as σ = 3.4 Å, ε = 0.00284 eV [14]. During the indentation process, the loading force (Ⅰ) gradually increased from 0 to 50nN in 200ps, (Ⅱ) remained unchanged in another 200ps, and (Ⅲ) was completely unloaded in the third 200ps. On basis of the force-displacement curves of the indenter, the hardness (H) of coatings can be calculated as below [15]:
𝐻=
𝑃𝑚𝑎𝑥 𝐴𝑐
(1)
where Pmax represents the maximum indentation force and Ac is the contact area, as expressed by: 𝐴𝑐 = 𝜋𝑅ℎ𝑐 ℎ𝑐 = ℎ𝑚𝑎𝑥 − 0.72
(2) 𝑃𝑚𝑎𝑥 𝑆𝑚𝑎𝑥
(3)
where hmax is the maximum penetration depth and Smax is the maximum slope of the unloading curve. 3. Results and discussion 3.1 Deposition process of the pure Al film Fig. 3 shows the upper surface of polycrystalline Al substrate before and after relaxation. Because of inner stress in the initial configuration, the surface morphology of the substrate has changed obviously, giving rise to many surface defects. In the first deposition process, only Al atoms are deposited onto this substrate. Because the incident energy is very low, there is no insert atom penetrated into substrate. When Al atoms become attached to the substrate surface, the vacancies are apt to be filled with the adatoms (fig. 4a), and the steps are able to promote the agglomeration of adjacent adatoms (fig. 4b). Only a few single Al atoms deposited on some plat surfaces such as Al (001) seem not to diffuse actively (fig. 4c). The similar behaviors are also observed in Al deposition on single crystal Cu surface, which formed an atomic layer not an island in the end [5]. However, the deposition process proves that the growth of Al film on polycrystalline Al surface followed the volmer-weber mode, as shown in Fig. 5. In order to investigate this growth mechanism, we calculate the diffusion energy barriers of Al atoms near the step through molecular static calculations (Fig. 2a), and the energy barriers are summarized in table 1. According to the diffusion behaviors observed on Al (001) surface, the energy barrier of 0.2707 eV is high enough to constrain adatoms at 333 K. Thus, the surface morphology should reflect the randomness of the position of Al deposition. However, adatoms could move a certain distance before their kinetic energy depleted. Fortunately, the local accelerated energy at any positions, which is expected to be more than 2 eV for Al atoms on account of the static calculations, can supply considerable energy. So, most adatoms are captured by these steps, since there were a mass of steps on the polycrystalline surface and the site along steps are very stable by reason of the high diffusion barriers away from steps. However, it is more difficult for adatoms to diffuse over the steps, because all ES barriers are much higher than surface diffusion energy barriers, and this implies a significant uphill current. Furthermore, the adatom can be attracted to the top terrace and diffuses away from the step, as shown in Fig. 6. This steering effect of steps also plays an important role in epitaxial growth at normal incidence [9,16]. Thus, some Al islands are able to be formed on polycrystalline surface. 3.2 Deposition process of the Al-graphene film During the second deposition process, 40 graphene flakes are deposited with Al atoms. Fig. 7a suggests that many Al atoms can be absorbed by graphene flakes before they arrived on the substrate surface. Thus, the area where more flakes deposited should be thicker, since the random
number seeds of Al deposition in all deposition processes are the same. Moreover, Al atoms deposited on graphene flakes are inclined to form the close-packed plane (Fig. 7b), due to the honeycomb lattice of graphene. It means the arrangement of Al atoms near graphene flakes could be disorganized, leading to the formation of more defects such as grain boundary, which occupied more spaces than close packing. The effect of defects on thickness is also observed by changing the incident energy [7]. Fig. 8 presents the surface morphology of composite coating and the distribution of graphene flakes. The thickness in the place where more graphene flakes are deposited increases clearly, compared with the pure Al coating surface. Therefore, we expect that the enrichment of graphene flakes can result in an increase of the local thickness. A series of static simulations are employed to support the conclusion above. The diffusion energy barriers of Al atoms on graphene flakes which lie on Al substrate (Fig. 2b) are summarized in table. 2. Both the surface diffusion barriers and ES barriers in any case are very high. So, Al atoms on graphene flakes are stable enough to form new crystallographic orientation, which can hinder the growth of adjacent grains. Consequently, the arrangement of Al atoms surrounding the graphene is heavily disturbed, and the increase of disorganized atoms will form new grain boundaries in the end. Thus, the size of Al grains is limited, and graphene flakes just exist in the grain boundaries. The fact graphene nano-platelets can enhance nucleation sites by creating a disorder in the matrix and retard the crystal growth is also seen in the electro-deposition experiment [3]. Herein, the Centro-Symmetry Parameter (CSP) analysis is conducted to indentify the local atomic configuration which can indicate this grain refinement. As shown in Fig. 9, the substrate Al atoms are removed, the inner structures of coatings in the bottom view are colored by CSP values. It can be found that graphene flakes only appear in grain boundaries and the sizes of Al grains in the composite coating visibly decrease. 3.3 Atomic structures and mechanical property of Al-graphene films Al thin films formed with different numbers of SLG flakes are studied as well. It should be noted that these simulations are carried out regardless of the dispersibility of graphene. The columns in Fig. 10 present compositions of atomic structures in deposited coatings at different numbers of graphene flakes, based on the Common Neighbor Analysis (CNA). Atoms in local f.c.c. (face-centred cubic) order are considered as perfect grains, atoms in local h.c.p. (hexagonal close-packed) order are indentified as stacking faults, and all other atoms are classified as grain boundaries [17]. The increase of graphene flakes leads to an obvious decrease of Al (f.c.c) atoms in the thin films. In light of the inner structures of coatings (Fig. 9), the numbers of Al grains can be considered the same. So, the change in the quantity of Al f.c.c. atoms can represent that in the average volume of Al grains. As more graphene flakes existed in the coatings, the number of f.c.c. atoms will reduce from N0 to Ni. Thus, the relative crystalline size (lr) can be simply evaluated as follow: 𝑙𝑟 = 3√𝑁𝑖 / 3√𝑁0, and the changing curve of lr is also shown in Fig. 10. This trend could further illustrate the effect of the grain refinement, which is well consistent with the experiment studies on electrodeposited Al-graphene composite coatings [1-2]. In experimental studies, the Sherrer equation is generally used to evaluate the average crystalline size. Li etc. have prepared an electrodeposited Al-graphene composite coating with the crystalline size of 17.36 nm, which is about 20% smaller than that of pure Al coating [1]. In accordance of the curve of relative crystalline size, about 30 graphene flakes can achieve the same effect in MD simulations. The mechanical property of those composite coatings is measured by nanoindentation simulations. On account of the complex structure, three hardness data points are averaged for each
coating. The plane coordinates of the indenter centre are (0 Å, 0 Å), (20 Å, 20 Å), and (40 Å, 40 Å) respectively. The force-displacement curve at (40 Å, 40 Å) and the results of hardness are shown in Fig. 11. A small amount (such as 40) of graphene flakes brings about very small changes, the hardness of composite (2.08 GPa) is a little higher than that of pure aluminum coating (2.03 GPa). When the number of graphene flakes reaches 100, there is an obvious increase (about 20%) in hardness. It’s worth noting that this increase is accompanied by the inverse Hall–Petch effect [17-18], since the Al grains have reached a small enough size (much less than 10 nm) according to the dimensions and inner structures of such coatings (Fig. 9). Hence, synthesized Al-graphene composite coating with the crystalline size of 17.36 nm even can exhibit 3.8 folds increase in the hardness (Fig. 12) [1]. However, too many graphene flakes will lead to a decline in mechanical property because of the graphene agglomeration (Fig. 12) [2]. In these simulations, all graphene flakes are added at the same time interval, so there is no drop in hardness. Therefore, it’s safe to say that the growing of graphene numbers will make a remarkable advance in the mechanical property of composite coatings. 4 Conclusions Using MD method, Al deposition on polycrystalline Al substrate is investigated. The vacancies on the substrate surface could be filled with the adatoms. Most deposited atoms agglomerate along the steps, due to the local accelerated energy which can improve the mobility of deposited atoms at early stage. Because of ES barriers and the steering effect, Al atoms seem to follow the volmer-weber growth mode. The morphology of composite coatings is significantly affected by graphene flakes, which are able to absorb Al atoms during the deposit and gave rise to the formation of more grain boundaries. Static calculations and the atomic configuration based on CSP analysis can account for the grain refinement of graphene on Al thin film, and the CNA analysis indicates that the growing of graphene numbers can strengthen this grain refinement. Nanoindention results show that the increase of well dispersed graphene flakes will also make a great improvement to the mechanical property of Al-graphene composite films. Acknowledgements This work was supported by Program for New Century Excellent Talents in University (NCET-11-0951), National Natural Science Foundation of China(61350006), Zhengzhou Leading Talent Project (131PLJRC655), Basic and frontier technology research program of Henan Province (162300410026)
Reference [1] N Li, L Zhang, MT Xu, T Xia, XW Ruan, S Song, HZ Ma, Preparation and mechanical property of electrodeposited Al-graphene composite coating, Materials & Design 111(2016):522-527 [2] L Zhang, N Li, HM Xia, JY Zhang, PP Zhang, MT Xu, HZ Ma, Preparation and Mechanical Properties of (Ni–Fe)–Graphene Composite Coating, Advanced Engineering Materials 18 (2016):1716-1719 [3] J Chen, J Li, D Xiong, Y He, Y Ji, Y Qin, Preparation and tribological behavior of Ni-graphene compositecoating under room temperature, Applied Surface Science 361 (2016):49-56 [4] S Badrayyana, DK Bhat, S Shenoy, Y Ullal, AC Hegde, Novel Fe-Ni-Graphene composite electrode for hydrogen production, International Journal of Hydrogen Energy 40 (2015):10453-10462 [5] SP Kim, KR Lee, YC Chung, YK Kim, M Doi, Sahashi M, A Molecular Dynamics Study of the Deposition and the Diffusion Behaviors of Al on a Cu Surface, Journal- Korean Physical Society 52 (2008):68-35 [6] SG Lee, YC Chung, Atomic-level investigation of Al and Ni thin film growth on Ni(111) surface: Molecular dynamics simulation, Applied Surface Science 253 (2007):8896-8900 [7] Y Cao, J Zhang, T Sun, Y Yan, F Yu, Atomistic study of deposition process of Al thin film on Cu substrate, Applied Surface Science 256 (2010):5993-5997 [8] ZH Hong, SF Hwang, TH Fang, Atomic-level stress calculation and surface roughness of film deposition process using molecular dynamics simulation, Computational Materials Science 48 (2010):520-528 [9] J Seo, SM Kwon, HY Kim, JS Kim, Steering effect on the shape of islands for homoepitaxial growth of Cu on Cu(001), Physical Review B 67 (2002):181-183 [10] Kan HM, Zhu SS, Zhang N, Wang XY, Electrodeposition of aluminum and aluminum magnesium alloys at room temperature, Journal of Central South University 22(2015): 3689-3697 [11]KW Jacobsen, JK Norskov, MJ Puska, Interatomic interactions in the effective-medium theory, Phys Rev B Condens Matter 35 (1987):7423-7442 [12] P Peng, G Liao, T Shi, Z Tang, Y Gao, Molecular dynamic simulations of nanoindentation in aluminum thin film on silicon substrate, Applied Surface Science 256 (2010):6284-6290 [13] D.W. Brenner, O.A. Shenderova, J.A. Harrison, S.J. Stuart, B. Ni, S.B. Sinnott, A second-generation reactive empirical bond order (REBO) potential energy expression for hydrocarbons, J. Phys.-Condens. Matter 14 (2002):783–802 [14] Choi BK, Yoon GH, Lee S, Molecular dynamics studies of CNT-reinforced aluminum composites under uniaxial tensile loading, Composites Part B 91(2016):119-125 [15] Zhao YB, Peng XH, Fu T, Sun R, Feng C, Wang ZC, MD simulation of nanoindentation on (001) and (111) surfaces of Ag–Ni multilayers, Physica E 74(2015):481-488 [16] S. Van Dijken, L.C. Jorritsma, B. Poelsema, Steering-Enhanced Roughening during Metal Deposition at Grazing Incidence, Physical Review Letters 82(1999):4038-4041 [17] J Schiøtz, FDD Tolla, KW Jacobsen, Softening of nanocrystalline metals at very small grain sizes, Nature, 391 (1998):561-563 [18] H Conrad, J Narayan, On the grain size softening in nanocrystalline materials, Scripta Materialia, 42 (2000):1025-1030
Fig. 1. Atomistic model of the deposition process.
Fig. 2. The possible diffusion behaviors of Al adatoms on designed surfaces. Silver balls represent Al atoms in the bottom, yellow balls represent an Al terrace, and the blue ball represents an Al adatom; Red arrows stand for the kinetic path, and the green network stands for a SLG flake. (a) diffusion behaviors near a step (b) diffusion behaviors on SLG
Fig.3. The upper faces of the substrate before (a) and after (b) relaxation
(a)
(b)
(c)
Fig. 4. The diffusion behaviors observed on the polycrystalline Al substrate. Blue balls represent Al atoms in the substrate, silver balls represent deposited Al atoms, and other colorful balls stand for adatoms which have just been adsorbed. (a) a vacancy is filled; (b) agglomeration along a step that is marked by a blue line; (c) a stable atoms on Al(001) surface
Fig. 5. Atomic configurations in the deposition process. Blue balls represent Al atoms in the substrate which are not fully displayed here, silver balls represent deposited Al atoms, (a)-(c) the side views; (d) the surface morphology colored by height (in Å).
Fig. 6. The steering effect observed in the deposition process. The yellow ball represents an Al atom depositing onto the surface, and the blue line stands for a step. (a-c) snapshot and details in 3.32ns, (d-e) details in 3.33ns
(a)
(b)
Fig.7. The absorption of Al atoms on SLG flakes. (a)the adsorption before arrivals on surface, (b)the arrangement of Al atoms on a SLG flake
Fig.8. The effect of SLG flakes on surface morphology. (a)the distribution of SLG in composite coating. Points represent all Al atoms, and green networks marked by yellow circles represent SLG; (b) the surface morphology of composite coating; (c) the surface morphology of pure Al coating. (b) and (c) are colored by height (in Å).
Fig.9. Distribution of CSPs in the bottom view of Al coatings. (a) the inner structure of composite coating; (b) the inner structure of pure Al coating. Blue balls stand for Al atoms with bulk lattice, other colorful balls are classified as grain boundaries, and red networks in (a) represent SLG.
Fig.10. Compositions of atomic structure and the relative crystalline sizes with respect to the number of SLG in composite coatings
(a)
(b)
Fig.11. The nanoindention results with respect to the number of SLG. (a) force-displacement curve at (40 Å, 40 Å), (b) the value of hardness
Fig.12. The referenced microhardness of metal-graphene composite coatings synthetized in the bath solution at a variety of graphene concentration [1-2].
Table.1 the diffusion energy barriers of Al atoms near the step (in eV) Surface
Al (001)
Al (011)
Al (111)
Surface diffusion
0.2707
0.2996
0.0446
Away from the step
0.6330
0.4973
0.4815
ES barrier
0.4516
0.4531
0.2932
Table.2 The diffusion energy barriers of Al atoms on graphene flakes which lie on Al substrate (in eV) Surface Terrace edge
Surface diffusion
ES barrier
SLG on Al(001)
SLG on Al (011)
SLG on Al (111)
Armc
Armc
Armc
hair
Zigza g
0.742 3
0.677 9
0.995 7
hair
g 0.653
1 0.978
1
Zigza
0.492 3
0.909 7
hair
g 0.634
3 0.929
6
Zigza
0.765 0
0.663 7
1.040 8