- Email: [email protected]
Numerical-experimental study on polishing of silicon wafer using magnetic abrasive finishing process
Wear 424–425 (2019) 143–150
Contents lists available at ScienceDirect
Wear journal homepage: www.elsevier.com/locate/wear
Numerical-experimental study on polishing of silicon wafer using magnetic abrasive finishing process Mohammad Mosavat, Abdolreza Rahimi
T
⁎
Department of Mechanical Engineering, Amirkabir University of Technology, 424 Hafez Ave., Tehran 15875-4413, Iran
A R T I C LE I N FO
A B S T R A C T
Keywords: Magnetic abrasive finishing Nano-finishing Silicon wafer Material removal Surface roughness Al2O3 FEM/SPH
Silicon wafer as a brittle material is extensively used in semiconductors. The surface quality of this material significantly affects the quality and efficiency of related components. In this study, the coupled algorithm of SPH/FEM is used to simulate the surface polishing of silicon wafers with Magnetic Abrasive Finishing process. The effects of rotational speed and machining gap on percent change in surface roughness (%ΔRa) and material removal (MR) are comprehensively analyzed with simulations and experiments. Furthermore, the material removal mechanism in wafers was investigated by using AFM. Our observations showed that both micro-fracture and micro-cutting mechanisms might happen and it highly depends on polishing parameters. Results of the simulations and experimental data showed that MR and %ΔRa value increase with increasing rotational speed and decreasing machining gap. According to our experimental findings, maximum %ΔRa and MR are 65% and 39.09 mg, respectively.
1. Introduction Advanced brittle materials such as glasses, silicon wafers, and so on are widely used in many high-tech industrious such as optoelectronics, microelectronics, infrared optics device, etc., and the importance of both surface finish and surface integrity of them has become extremely important [1–3]. This progress causes higher demand for excellent surface finishing capabilities for various industrial and engineering applications since the finishing process is an essential and expensive process of the manufacturer [4,5]. Because of the fast development in material science, industries require materials with specific properties such as high toughness, hardness, and fragility [6]. The primary challenge for the manufacturing industry is polishing of these materials by conventional finishing processes such as grinding, honing and lapping which generate micro/nano burrs, residual stresses and subsurface damage [7,8]. These techniques are not efficient and also incapable of finishing of brittle materials like silicon wafers. In recent decades, significant advances are obtained in the fine polishing of these materials. Magnetic Abrasive Finishing (MAF) process is one of these processes used for polishing of both ferrous and non-ferrous materials. In this technique, a magnetic force is applied in the finishing area to allow the abrasive particles to shear-off the material from the workpiece surface in microchips form.
⁎
Various scientists have studied the surface roughness based on MAF. Jain et al. investigated the effect of circumferential speed and working gap on the performance of MAF. They concluded that material removal decreases by decreasing the circumferential speed and increasing the machining gap [9]. Jain et al. in other work, studied the surface texture generated by flexible magnetic abrasive brush (FMAB) and they observed that the process creates micro scratches having the width less than 0.5 µm on the finished surface [10]. Pashmforoush et al. studied the finishing of BK7 glass with MAF. They optimized the parameters and achieved 23 nm surface roughness [11]. Saraeian et al. studied the polishing of AISI321 stainless steel with MAF. Their results showed 50% improvement compared with initial workpiece surface roughness [12]. Vahdati et al. studied polishing of freeform surface with MAF and they reported their optimum process parameters levels [13]. Pandey et al. studied the normal force and finishing torque during UAMAF. They concluded that either low machining gap or high supply current could strengthen FMAB [14]. Compared with the presented technologies, MAF is a low-speed finishing procedure that has various unclear effect parameters. Although several studies were done on this subject, there are limited studies on MR and surface roughness of the MAF process in brittle material like silicon wafer. Dominant factors must be investigated separately with high precision in this method, and optimal conditions should be obtained to achieve high efficiency in the finishing process.
Corresponding author. E-mail address: [email protected] (A. Rahimi).
https://doi.org/10.1016/j.wear.2019.02.007 Received 1 November 2018; Received in revised form 24 January 2019; Accepted 4 February 2019 Available online 05 February 2019 0043-1648/ © 2019 Published by Elsevier B.V.
Wear 424–425 (2019) 143–150
M. Mosavat and A. Rahimi
2.2. Basic principles of SPH
In this study, the primary focus is to investigate the machining gap and abrasive rotational speed through the MAF process. According to the finishing procedure and the Finite Element Method (FEM) and Smooth Particle Hydrodynamic (SPH) methods, numerous simulations and experimental investigations are done to reveal the primary relationship between mentioned parameters with the %ΔRa and MR. The present study is limited to a silicon wafer and Al2O3 polishing agent. Finally, specific modification for achieving high MR and the surface with high quality by using Al2O3 abrasive particles is presented. In this study, in addition to simulating MAF and validating the relevant results, mechanisms of the material removal was also studied during finishing of silicon wafers. Traditionally, mechanisms of material removal in the polishing process of brittle materials can be divided into two categories: ductile and brittle mode. In a ductile mode, finishing process contains the micro-cutting mechanism that results in the surface with high quality and less surface and subsurface damages [15,16]. On the contrary, in brittle mode, finishing associated with the growth of lateral cracks which are created under the abrasive particles and propagated to the workpiece surface, and this yields a damaged surface [17]. Given the two mentioned mechanisms, the finishing of brittle materials such as silicon wafer in ductile mode could be very successful. According to energy balance conception, there is a critical cutting depth, so that, if the cutting is performed at a lower critical depth, the micro-cutting will occur [18]. In this manner, the required energy for a plastic deformation is less than the required energy for the crack propagation. In addition to the mentioned concept, another concept describes the brittle material removal mode from the stress field point [19]. According to this hypothesis, when the thickness of uncut chip is significantly low, the critical stress is less than stress value required to suppress fracture [20]. In MAF technique, the force of process is so low that the process is expected to occur without the appearance of cracks. Furthermore, abrasive rake angle that negative in MAF results in sufficient hydrostatic pressure that facilitates plastic deformation. From another view, conventional finishing processes like lapping and grinding for finishing of brittle materials such as silicon wafers results in surface damages. So, one of the aims of this study is related to inspect the workpiece surface after the MAF process to investigate the mechanism of material removal that occurs in silicon wafers. For this goal, Atomic Force Microscope (AFM) was used to inspect the surface of the specimen.
In 1977, for the first time, Lucy [25], Gingold and Monaghan [26] introduced the SPH as a non-grid algorithm method. This method is used for computational fluid dynamic problems, especially continuum mechanics problems. This advantage of the SPH approach is highly significant compared to the FEM. In order to model surface polishing using the FEM method, it is essential to define parameters permitting the material failure and separation caused by the abrasives. Hence, adaptive remeshing and element deletion are applied. It should be mentioned that adaptive remeshing is time-consuming process while the element deletion leads to eliminate of the mass from the calculation. So, for minimizing the element deletion effects, the element size should be reasonably small. In the SPH approach, none of these techniques is required, and material deletion in the SPH is done directly through a loss of cohesion among particles. Thus, loss of the workpiece mass calculation is not affected by the calculation [27]. 2.3. Calculation of area roughness The workpiece in the SPH method is modeled with smooth particles. As shown in Fig. 1, the polishing surface modeled as a rectangular array contains grids in which every grid has a coordinate (xi, yj). After the finishing process is done, the smooth particles are removed, and a new surface with regenerated space position is achieved. According to the measuring mechanism of the AFM, roughness value of the area is calculated by using the formula as follows:
Sq =
1 PQ
p−1 Q−1
∑∑
(z(xi,yj))2
i=0 j=0
(1)
where Sq is roughness value, P and Q are the points number in x and ydirection respectively, and z (xi, yj) is the position change of grid (xi, yj). 3. Magnetic flux density analysis The normal force that applies to MAP during the MAF process is calculated from the following equation.
π 2 ⎞ F = P ⎛ DMAP ⎝4 ⎠ β2 ω ⎞ ⎛ P = 0.75 µ0 ⎝ ω + α ⎠ 3 α=1+ µs − 1
2. Theoretical background 2.1. Mechanism of MAF
(2)
where P is the machining pressure, DMAP the mean diameter of MAP, β the magnetic flux density, µ0 the permeability of vacuum, µs the relative permeability of iron, and ω is the volume fraction of iron powders [28]. Magnetic flux density itself depends on the machining gap, specimen magnetic properties, and the dimensions of permanent magnets.
MAF process was initiated by applying the magnetic field through the gap between magnet and workpiece which is called "machining gap". In the MAF process, to generate a magnetic field in the machining gap, the magnetic source which is usually an electromagnet or permanent magnet is used [21]. Inside the gap, a mixture of abrasive powders (SiC, Al2O3, CBN, etc.) and iron powders which called abrasive magnetic particle(s) (MAP(s)) is located. Previous studies showed that the mentioned mixture forms a chain-like structure in the working gap in which each chain contains a large number of MAPs [22]. By joining these chain-like structures, Flexible Magnetic Abrasive Brush (FMAB) as a multi-edge finishing tool is achieved [23]. In this method, the normal and tangential forces are mainly active in the finishing zone. The normal force effecting through FMAB chain induces micro-nano indentations on the surface of workpiece [24]. On the other side, the relative rotation between the magnet and workpiece produces the machining force. The mentioned forces can be managed by various parameters such as abrasive size, rotational speed, machining gap, magnetic flux density, magnet force, etc.
Fig. 1. Surface model of the workpiece being polished with SPH. 144
Wear 424–425 (2019) 143–150
M. Mosavat and A. Rahimi 310 1 mm machining gap 1.5 mm machining gap 2 mm machining gap
Magnetic flux density (mT)
300
290
280
270
260
250
240 -15
-10
-5
0
5
10
15
Distance from center (mm)
Fig. 4. Single Al2O3 particle polishing model.
Fig. 2. Distribution of magnetic flux density in the center line of the specimen surface.
Table 1 ΔSimulation results of variations in the number of SPH elements on the responses.
In this paper, the magnetic flux density in different distances of the magnet center and in various machining gaps was calculated by using Ansoft Maxwell. The Ansoft model consisted of three Nd–Fe–B permanent magnets with relative permeability of 1.0997 and the silicon wafer specimen with relative permeability of one. The boundary of the model was considered as vacuum with relative permeability of one. Fig. 2 illustrates the distribution of magnetic flux density in the center line for the machining gaps of 1 mm, 1.5 mm, and 2 mm. As shown in Fig. 2, by increasing the machining gap, magnetic flux density decreases. The maximum flux density obtained was 301 mT for the machining gap of 1 mm. Then, the magnetic normal force was estimated using Eq. (2), and the resulted force was applied to MAP during the MAF process simulation.
Experiment results
Percentage change in surface roughness (%ΔRa) Material removal (mg)
Simulation results with various number of SPH elements 43,851
51,623
59,132
65%
49%
58%
74%
39.09
31.12
36.05
48.68
In the simulation of this model, it was assumed that the silicon workpiece was fixed with a relative motion in some directions. Abrasive particles could move on a workpiece surface in different rotational speeds while all constraints are exerted from the right and bottom side of the workpiece. In our model, the left, upper surfaces and the interior nodes are allowed to move in all directions. The contact algorithm of CONTACT-AUTOMATIC-NODES-TO-SURFACE in LS-DYNA was applied due to the application of the SPH particles. This contact is used between the Al2O3 FEM elements as well as the SPH particles of the workpiece. In this study, the workpiece material is silicon and it is considered as a piecewise-linear elastic material in LS-DYNA, and full details are shown in Table 2. Al2O3 considered being rigid. Accordingly, by using the SPH/FEM, the machining gap parameters are set up separately to reveal the significance of machining gap on % ΔRa and MR. The influence of machining gap on %ΔRa and MR is studied by simulations and experimental examinations of the MAF polishing method with four various machining gaps under the same abrasive rotational speed of 1500 rpm. Also, to reveal the influence of abrasive rotational speed on mentioned outputs, simulations and experimental examinations are performed with four different rotational speeds under the same machining gap of 2 mm. Table 3 presents the groups of simulation schemes.
4. Polishing model It should be noted that the polishing process is a complicated task, and the MAP(s) have no systematic shape, and limited deformation occurs by the finishing process. Fig. 3 shows the MAP Scanning Electron Microscope (SEM) image, and we simplified the modeling of the polishing process for reducing the time of consumption. Fig. 4 illustrates the single abrasive 3D model of the MAF. It is clear that Al2O3 is much harder than silicon workpiece. Hence, Al2O3 particles are supposed as a rigid spherical body with Lagrange meshing of 10500 hexahedron elements. To simplify the model, a part of the circular workpiece is considered in the FEM/SPH model. Thus, the size of the block is 16 µm × 32 µm × 10 µm, and there are 51,623 smooth particles inside the domain. To investigate the adequacy of the number of SPH elements on the simulation results, three different grid sizes are simulated according to Table 1. Finally, the model with 51,623 grids resulted in the closest response.
Table 2 Properties of the silicon wafer.
Fig. 3. Magnetic abrasive particle SEM image. 145
Young's modulus (GPa)
160
yield stress (GPA) Poisson's ratio density (gr/cm3) The coefficient of thermal expansion (10–6/°C) Thermal conductivity (W/m K) Specific heat capacity (J/g K) melting point (°C)
7 0.22 2.4 2.6 157 0.7 1415
Wear 424–425 (2019) 143–150
M. Mosavat and A. Rahimi
5. Experimental verification
Table 3 Simulation scheme and results. Group
Machining gap (mm)
Rotational speed (rpm)
Percentage change in surface roughness %ΔRa
Material removal (mg)
1
1 1.5 2 2.5 2 2 2 2
1500 1500 1500 1500 800 1250 1500 2000
58 43 21 14 19 34 38 47
36.05 23.97 13.21 4.06 2.96 12.11 15.65 21.45
2
Fig. 5 shows the MAF experimental setup. Experiments were done on a 3-in.-diameter circular flat silicon wafer. An aluminum fixture was used to hold the wafer. The fixture connected to the vacuum pump for fixing the wafer on the fixture. The other fixture which connects to the motor attaches the three Nd-Fe-B permanent magnets with the thickness of 5 mm and the inner and outer diameter of 20 and 30 mm which located on the above of the wafer. The MAPs used in this study contain Al2O3 abrasives, iron powders, and glass powder as a binder. The particles were initially mixed to obtain a homogeneous mixture and compressed into a cylindrical manner and sintered in a vacuum furnace at 900 °C for 2 h. After the sintering process, the MAPs were crushed and sieved to obtain abrasive particles. Fig. 6 shows sintered MAPs scanned image with SEM. Al2O3 abrasives particles used in experiments was 1.0 µm. The weight ratio
Fig. 5. Experimental setup of MAF: (a) scheme figure (b) actual set up. 146
Wear 424–425 (2019) 143–150
Percentage change in surface roghness
M. Mosavat and A. Rahimi
70
Experiment Simulation
60
50
40
30
20
10 1
Fig. 6. Magnetic Abrasive Particle SEM images (Al2O3 + Fe).
1
2
2
2.5
Fig. 8. Simulation and experimental test results of percentage change in surface roughness (%ΔRa) at different machining gap, polishing time:10 min, rotational speed:1500 rpm.
Table 4 Experimental scheme and results. Group
1.5
Machining gap (mm)
Machining gap (mm)
Rotational speed (rpm)
Average percentage change in surface roughness (%ΔRa)
Average material removal (mg)
1 1.5 2 2.5 2 2 2 2
1500 1500 1500 1500 800 1250 1500 2000
65 41 21 14 16 31 36 42
39.09 26.97 14.21 6.98 3.11 7.61 9.95 11.23
between the Al2O3 and iron is 40:60, and the experimental environment temperature is about 25 °C. Each experiment was carried out four times, and after each test, the surface roughness and material removal were measured. The initial surface roughness of wafers was not equal, but all roughness values are within 0.11–0.39 µm. So, percentage changes in surface roughness which is given in Eq. (3). considered as a response of experiments.
Fig. 7. Simulation results of machining gap at: (a) 1 mm, (b) 1.5 mm, (c) 2 mm, (d) 2.5 mm. 147
Wear 424–425 (2019) 143–150
M. Mosavat and A. Rahimi
Percentage change in surface roughness
40
Material reomival (mg)
Experiment Simulation
30
20
10
1
1.5
2
2.5
45
Experiment Simulation
40
35
30
25
20
15 800
Machining gap (mm)
1000
1200
1400
1600
1800
2000
Rotational speed (rpm)
Fig. 9. Simulation and experimental test results of material removal at different machining gap, polishing time:10 min, rotational speed: 1500 rpm.
Fig. 11. Simulation and experimental test results of percentage change in surface roughness (%ΔRa) at different rotational speed, polishing time:10 min, machining gap:2 mm.
Initial Surface Roughness − Final Surface Roughness ⎞ %ΔR a = ⎜⎛ ⎟ × 100 Initial Surface Roughness ⎝ ⎠ (3)
Fig. 8 and Fig. 9, respectively. As shown in Figures, in both simulation and experimental curves, %ΔRa and MR decrease as the machining gap is increased. The obtained results indicate that the effect of magnetic flux density on abrasives reduced when the machining gap was increased. In this condition, the indentation force of cut reduces. In larger machining gaps, due to low magnetic flux density, the exerted force is not sufficient for effectively indentation and cutting to yield effective cuts on the workpiece surface. This also causes magnetic abrasive inefficiency and thus, changes in surface roughness and material removal value decline.
AFM is also used for measuring of wafers surface roughness. Finishing experiments are done similar to the simulation layout. Table 4 presents details of polishing experiments schemes and results of our tests. 6. Results and discussion 6.1. Effect of machining gap
6.2. Effect of abrasive rotational speed
The effects of the machining gap on %ΔRa and MR are studied through simulating and experimenting with the MAF. Fig. 7 shows the results of machining gap simulation at 1, 1.5, 2, and 2.5 mm. The effect of the machining gap on %ΔRa and MR is demonstrated in
The effect of abrasive rotational speed on %ΔRa and MR is also studied through simulating and experimenting during MAF. To do this,
Fig. 10. Simulation results of rotational speed at: (a) 800 rpm, (b) 1250 rpm, (c) 1500 rpm, (d) 2000 rpm. 148
Wear 424–425 (2019) 143–150
M. Mosavat and A. Rahimi
Experiment Simulation
20
Material removal (mg)
effect of work material may also be contributed to an increase in torque. As shown in Fig. 12, with increasing the abrasive rotational speeds, the difference between simulation and experiment results significantly increases and MR changes in the simulation are much more than experimental results. This is because of that in MAF technique, the normal force applied on MAPs consists of centrifugal and magnetic forces which are perpendicular to each other. When the abrasive rotational speed increases, the centrifugal force also rises. Increasing the centrifugal force applied on MAPs with growing the abrasive rotational speeds causes the MAP to leave the machining gap and run out of the circle. Thus, the total number of MAPs reduces. It should be noted that MAPs close to weak magnetic field first left the gap and this behavior intensifies within the zones with the stronger magnetic field. Finally, decreasing the MAPs that presents in the machining gap augments differences between simulation and experimental results as shown in Fig. 12. Finally, both simulation and experimental results confirm that the MR and %ΔRa increase by enlarging of abrasive particle size and our results are in good agreement with the results of Jain et al. [9].
15
10
5
800
1000
1200
1400
1600
1800
2000
Rotational speed (rpm) Fig. 12. Simulation and experimental test results of material removal at different rotational speed, polishing time:10 min, machining gap:2 mm.
6.3. Material removal mechanism To investigate the material removal mechanism during MAF, AFM was used for inspection of the surfaces of the specimen. Fig. 13a presents the surface roughness measurement of an experiment with 2.0 mm machining gap and 800 rpm rotational speed after 10 min of finishing. The results of 1.0 mm machining gap and 1500 rpm abrasive rotational speed with the same polishing times are depicted in Fig., 13b. As shown in Fig. 12a, polished surface with cut marks indicates that the microcutting material removal mechanism in ductile mode has occurred during the polishing process, and the resulted surface is free from surface damages and fractures. On the contrary, the topography of surface (as shown in Fig. 13b) indicates that side cracks achieved on the surface and micro-fracture mechanism occurs during the MAF. Results of surface topography clearly show that material removal has taken part in the brittle mode accompanied by lateral cracks, which initiated under the abrasive and propagate to the surface. Therefore, the quality of the surface of this case is lower than the previous one. According to our findings, polishing condition considerably influences on micro-fracture and micro-cutting mechanisms. Comparison of the surface roughness of our results confirms that surface with high roughness is achieved as the micro-fracture mechanism occurs. Meanwhile, micro-cutting mechanism resulted in a nano-level surface with no surface and subsurface damages. 7. Conclusions In this study, the algorithms of FEM and SPH are coupled to simulate the finishing process of monocrystalline silicon wafers with the MAF method. This work presents a comprehensive analysis of the effect of rotational speed and machining gap on %ΔRa and MR. Several experiments also performed on a 3-in.-diameter circular silicon wafer with various machining gaps and rotational speeds. Also, the material removal mechanisms were investigated in wafers by using AFM. Following conclusions could be derived from this study.
Fig. 13. Measurement results of surface roughness.
2.0 mm machining gap and 1 µm abrasive size with four different abrasive rotational speeds are considered. The simulation results of the abrasive rotational speed at 800, 1250, 1500, and 2000 rpm are shown in Fig. 10. The effects of abrasive rotational speed on %ΔRa and MR are shown in Fig. 11 and Fig. 12. As observed in figures, in both simulation and experiment curves, improvement of MR and %ΔRa increases with the rise of the rotational speed. The finishing torque increases when the abrasive rotational speed rises. Indeed, high abrasive rotational speeds produce more amounts of MAPs shear on the peaks of the workpiece in unit time, So, the change in momentum per unit time is high and this leads to high torque. On the other side, due to the high shear rate of peaks, the workpiece hardening
● Machining gap and rotational speed are dominant parameters during MAF process. ● Both simulation and experiments results show that MR value and % ΔRa increases with decreasing the machining gap. ● Both Results of simulation and experiments indicate that MR value and %ΔRa increases when the abrasive rotational speed increases. ● According to the results of experiments in the optimum condition, maximum MR and %ΔRa are 39.09 mg and 65%, respectively. ● The results of our experimental examinations indicate that the SPH method is reliable and robust in SPH/FEM coupled method for 149
Wear 424–425 (2019) 143–150
M. Mosavat and A. Rahimi
simulating the MAF process. Moreover, the simulation agrees well with experimental results and this confirms that the applied numerical technique is efficient and reliable. ● AFM observations indicate that polishing condition considerably influences on micro-fracture and micro-cutting mechanisms, and rougher surfaces achieve as the micro-fracture mechanism occurs. Meanwhile, micro-cutting mechanism resulted in a nano-level surface without surface and subsurface damages.
[12] P. Saraeian, H. Soleimani Mehr, B. Moradi, H. Tavakoli, O. Khalil Alrahmani, Study of magnetic abrasive finishing for AISI321 stainless steel, Mater. Manuf. Process. 31 (2016) 2023–2029. [13] M. Vahdati, S. Rasouli, Study of magnetic abrasive finishing on freeform surface, Trans. IMF 94 (2016) 294–302. [14] V.C. Shukla, P.M. Pandey, U.S. Dixit, A. Roy, V. Silberschmidt, Modeling of normal force and finishing torque considering shearing and ploughing effects in ultrasonic assisted magnetic abrasive finishing process with sintered magnetic abrasive powder, Wear 390 (2017) 11–22. [15] S. Malkin, T. Hwang, Grinding mechanisms for ceramics, CIRP Ann. 45 (1996) 569–580. [16] J. Feng, P. Chen, J. Ni, Prediction of surface generation in microgrinding of ceramic materials by coupled trajectory and finite element analysis, Finite Elem. Anal. Des. 57 (2012) 67–80. [17] D. Marshall, A.G. Evans, B.K. Yakub, J. Tien, G. Kino, The nature of machining damage in brittle materials, Proc. R. Soc. Lond. A 385 (1983) 461–475. [18] T.G. Bifano, T.A. Dow, R.O. Scattergood, Ductile-regime grinding: a new technology for machining brittle materials, J. Eng. Ind. 113 (1991) 184–189. [19] J. Yan, M. Yoshino, T. Kuriagawa, T. Shirakashi, K. Syoji, R. Komanduri, On the ductile machining of silicon for micro electro-mechanical systems (MEMS), optoelectronic and optical applications, Mater. Sci. Eng. A 297 (2001) 230–234. [20] T. Nakasuji, S. Kodera, S. Hara, H. Matsunaga, N. Ikawa, S. Shimada, Diamond turning of brittle materials for optical components, CIRP Ann.-Manuf. Technol. 39 (1990) 89–92. [21] H. Suzuki, S. Kodera, S. Hara, H. Matsunaga, T. Kurobe, Magnetic field-assisted polishing—application to a curved surface, Precis. Eng. 11 (1989) 197–202. [22] T. Mori, K. Hirota, Y. Kawashima, Clarification of magnetic abrasive finishing mechanism, J. Mater. Process. Technol. 143 (2003) 682–686. [23] V. Jain, Abrasive-based nano-finishing techniques: an overview, Mach. Sci. Technol. 12 (2008) 257–294. [24] V. Jain, Magnetic field assisted abrasive based micro-/nano-finishing, J. Mater. Process. Technol. 209 (2009) 6022–6038. [25] L.B. Lucy, A numerical approach to the testing of the fission hypothesis, Astron. J. 82 (1977) 1013–1024. [26] R.A. Gingold, J.J. Monaghan, Smoothed particle hydrodynamics: theory and application to non-spherical stars, Mon. Not. R. Astron. Soc. 181 (1977) 375–389. [27] M. Heinstein, D. Segalman, Simulation of orthogonal cutting with smooth particle hydrodynamics, Sandia Report, Livermore, California, USA, 1997. [28] A.B. Khairy, Aspects of surface and edge finish by magnetoabrasive particles, J. Mater. Process. Technol. 116 (2001) 77–83.
References [1] C. Chen, P. Zhang, C. Xiao, L. Chen, L. Qian, Effect of mechanical interaction on the tribochemical wear of bare silicon in water, Wear 376 (2017) 1307–1313. [2] B. Yu, H. He, L. Chen, L. Qian, Tribological behavior of monocrystalline silicon from single-to multiple-asperity scratch, Wear 374 (2017) 29–35. [3] C. Xiao, C. Chen, J. Guo, P. Zhang, L. Chen, L. Qian, Threshold contact pressure for the material removal on monocrystalline silicon by SiO2 microsphere, Wear 376 (2017) 188–193. [4] P. Zhang, C. Chen, C. Xiao, L. Chen, L. Qian, Comparison of wear methods at nanoscale: line scanning and area scanning, Wear 400 (2018) 137–143. [5] T. Shao, F. Ge, Y. Dong, K. Li, P. Li, D. Sun, F. Huang, Microstructural effect on the tribo-corrosion behaviors of magnetron sputtered CrSiN coatings, Wear 416 (2018) 44–53. [6] R. Knoblauch, D. Boing, W.L. Weingaertner, K. Wegener, F. Kuster, F.A. Xavier, Investigation of the progressive wear of individual diamond grains in wire used to cut monocrystalline silicon, Wear 414 (2018) 50–58. [7] A. Kumar, S. Kaminski, S.N. Melkote, C. Arcona, Effect of wear of diamond wire on surface morphology, roughness and subsurface damage of silicon wafers, Wear 364 (2016) 163–168. [8] Q. Zhang, Y. Fu, H. Su, Q. Zhao, S. To, Surface damage mechanism of monocrystalline silicon during single point diamond grinding, Wear 396 (2018) 48–55. [9] V. Jain, P. Kumar, P. Behera, S. Jayswal, Effect of working gap and circumferential speed on the performance of magnetic abrasive finishing process, Wear 250 (2001) 384–390. [10] D.K. Singh, V. Jain, V. Raghuram, R. Komanduri, Analysis of surface texture generated by a flexible magnetic abrasive brush, Wear 259 (2005) 1254–1261. [11] F. Pashmforoush, A. Rahimi, Nano-finishing of BK7 optical glass using magnetic abrasive finishing process, Appl. Opt. 54 (2015) 2199–2207.
150
Contents lists available at ScienceDirect
Wear journal homepage: www.elsevier.com/locate/wear
Numerical-experimental study on polishing of silicon wafer using magnetic abrasive finishing process Mohammad Mosavat, Abdolreza Rahimi
T
⁎
Department of Mechanical Engineering, Amirkabir University of Technology, 424 Hafez Ave., Tehran 15875-4413, Iran
A R T I C LE I N FO
A B S T R A C T
Keywords: Magnetic abrasive finishing Nano-finishing Silicon wafer Material removal Surface roughness Al2O3 FEM/SPH
Silicon wafer as a brittle material is extensively used in semiconductors. The surface quality of this material significantly affects the quality and efficiency of related components. In this study, the coupled algorithm of SPH/FEM is used to simulate the surface polishing of silicon wafers with Magnetic Abrasive Finishing process. The effects of rotational speed and machining gap on percent change in surface roughness (%ΔRa) and material removal (MR) are comprehensively analyzed with simulations and experiments. Furthermore, the material removal mechanism in wafers was investigated by using AFM. Our observations showed that both micro-fracture and micro-cutting mechanisms might happen and it highly depends on polishing parameters. Results of the simulations and experimental data showed that MR and %ΔRa value increase with increasing rotational speed and decreasing machining gap. According to our experimental findings, maximum %ΔRa and MR are 65% and 39.09 mg, respectively.
1. Introduction Advanced brittle materials such as glasses, silicon wafers, and so on are widely used in many high-tech industrious such as optoelectronics, microelectronics, infrared optics device, etc., and the importance of both surface finish and surface integrity of them has become extremely important [1–3]. This progress causes higher demand for excellent surface finishing capabilities for various industrial and engineering applications since the finishing process is an essential and expensive process of the manufacturer [4,5]. Because of the fast development in material science, industries require materials with specific properties such as high toughness, hardness, and fragility [6]. The primary challenge for the manufacturing industry is polishing of these materials by conventional finishing processes such as grinding, honing and lapping which generate micro/nano burrs, residual stresses and subsurface damage [7,8]. These techniques are not efficient and also incapable of finishing of brittle materials like silicon wafers. In recent decades, significant advances are obtained in the fine polishing of these materials. Magnetic Abrasive Finishing (MAF) process is one of these processes used for polishing of both ferrous and non-ferrous materials. In this technique, a magnetic force is applied in the finishing area to allow the abrasive particles to shear-off the material from the workpiece surface in microchips form.
⁎
Various scientists have studied the surface roughness based on MAF. Jain et al. investigated the effect of circumferential speed and working gap on the performance of MAF. They concluded that material removal decreases by decreasing the circumferential speed and increasing the machining gap [9]. Jain et al. in other work, studied the surface texture generated by flexible magnetic abrasive brush (FMAB) and they observed that the process creates micro scratches having the width less than 0.5 µm on the finished surface [10]. Pashmforoush et al. studied the finishing of BK7 glass with MAF. They optimized the parameters and achieved 23 nm surface roughness [11]. Saraeian et al. studied the polishing of AISI321 stainless steel with MAF. Their results showed 50% improvement compared with initial workpiece surface roughness [12]. Vahdati et al. studied polishing of freeform surface with MAF and they reported their optimum process parameters levels [13]. Pandey et al. studied the normal force and finishing torque during UAMAF. They concluded that either low machining gap or high supply current could strengthen FMAB [14]. Compared with the presented technologies, MAF is a low-speed finishing procedure that has various unclear effect parameters. Although several studies were done on this subject, there are limited studies on MR and surface roughness of the MAF process in brittle material like silicon wafer. Dominant factors must be investigated separately with high precision in this method, and optimal conditions should be obtained to achieve high efficiency in the finishing process.
Corresponding author. E-mail address: [email protected] (A. Rahimi).
https://doi.org/10.1016/j.wear.2019.02.007 Received 1 November 2018; Received in revised form 24 January 2019; Accepted 4 February 2019 Available online 05 February 2019 0043-1648/ © 2019 Published by Elsevier B.V.
Wear 424–425 (2019) 143–150
M. Mosavat and A. Rahimi
2.2. Basic principles of SPH
In this study, the primary focus is to investigate the machining gap and abrasive rotational speed through the MAF process. According to the finishing procedure and the Finite Element Method (FEM) and Smooth Particle Hydrodynamic (SPH) methods, numerous simulations and experimental investigations are done to reveal the primary relationship between mentioned parameters with the %ΔRa and MR. The present study is limited to a silicon wafer and Al2O3 polishing agent. Finally, specific modification for achieving high MR and the surface with high quality by using Al2O3 abrasive particles is presented. In this study, in addition to simulating MAF and validating the relevant results, mechanisms of the material removal was also studied during finishing of silicon wafers. Traditionally, mechanisms of material removal in the polishing process of brittle materials can be divided into two categories: ductile and brittle mode. In a ductile mode, finishing process contains the micro-cutting mechanism that results in the surface with high quality and less surface and subsurface damages [15,16]. On the contrary, in brittle mode, finishing associated with the growth of lateral cracks which are created under the abrasive particles and propagated to the workpiece surface, and this yields a damaged surface [17]. Given the two mentioned mechanisms, the finishing of brittle materials such as silicon wafer in ductile mode could be very successful. According to energy balance conception, there is a critical cutting depth, so that, if the cutting is performed at a lower critical depth, the micro-cutting will occur [18]. In this manner, the required energy for a plastic deformation is less than the required energy for the crack propagation. In addition to the mentioned concept, another concept describes the brittle material removal mode from the stress field point [19]. According to this hypothesis, when the thickness of uncut chip is significantly low, the critical stress is less than stress value required to suppress fracture [20]. In MAF technique, the force of process is so low that the process is expected to occur without the appearance of cracks. Furthermore, abrasive rake angle that negative in MAF results in sufficient hydrostatic pressure that facilitates plastic deformation. From another view, conventional finishing processes like lapping and grinding for finishing of brittle materials such as silicon wafers results in surface damages. So, one of the aims of this study is related to inspect the workpiece surface after the MAF process to investigate the mechanism of material removal that occurs in silicon wafers. For this goal, Atomic Force Microscope (AFM) was used to inspect the surface of the specimen.
In 1977, for the first time, Lucy [25], Gingold and Monaghan [26] introduced the SPH as a non-grid algorithm method. This method is used for computational fluid dynamic problems, especially continuum mechanics problems. This advantage of the SPH approach is highly significant compared to the FEM. In order to model surface polishing using the FEM method, it is essential to define parameters permitting the material failure and separation caused by the abrasives. Hence, adaptive remeshing and element deletion are applied. It should be mentioned that adaptive remeshing is time-consuming process while the element deletion leads to eliminate of the mass from the calculation. So, for minimizing the element deletion effects, the element size should be reasonably small. In the SPH approach, none of these techniques is required, and material deletion in the SPH is done directly through a loss of cohesion among particles. Thus, loss of the workpiece mass calculation is not affected by the calculation [27]. 2.3. Calculation of area roughness The workpiece in the SPH method is modeled with smooth particles. As shown in Fig. 1, the polishing surface modeled as a rectangular array contains grids in which every grid has a coordinate (xi, yj). After the finishing process is done, the smooth particles are removed, and a new surface with regenerated space position is achieved. According to the measuring mechanism of the AFM, roughness value of the area is calculated by using the formula as follows:
Sq =
1 PQ
p−1 Q−1
∑∑
(z(xi,yj))2
i=0 j=0
(1)
where Sq is roughness value, P and Q are the points number in x and ydirection respectively, and z (xi, yj) is the position change of grid (xi, yj). 3. Magnetic flux density analysis The normal force that applies to MAP during the MAF process is calculated from the following equation.
π 2 ⎞ F = P ⎛ DMAP ⎝4 ⎠ β2 ω ⎞ ⎛ P = 0.75 µ0 ⎝ ω + α ⎠ 3 α=1+ µs − 1
2. Theoretical background 2.1. Mechanism of MAF
(2)
where P is the machining pressure, DMAP the mean diameter of MAP, β the magnetic flux density, µ0 the permeability of vacuum, µs the relative permeability of iron, and ω is the volume fraction of iron powders [28]. Magnetic flux density itself depends on the machining gap, specimen magnetic properties, and the dimensions of permanent magnets.
MAF process was initiated by applying the magnetic field through the gap between magnet and workpiece which is called "machining gap". In the MAF process, to generate a magnetic field in the machining gap, the magnetic source which is usually an electromagnet or permanent magnet is used [21]. Inside the gap, a mixture of abrasive powders (SiC, Al2O3, CBN, etc.) and iron powders which called abrasive magnetic particle(s) (MAP(s)) is located. Previous studies showed that the mentioned mixture forms a chain-like structure in the working gap in which each chain contains a large number of MAPs [22]. By joining these chain-like structures, Flexible Magnetic Abrasive Brush (FMAB) as a multi-edge finishing tool is achieved [23]. In this method, the normal and tangential forces are mainly active in the finishing zone. The normal force effecting through FMAB chain induces micro-nano indentations on the surface of workpiece [24]. On the other side, the relative rotation between the magnet and workpiece produces the machining force. The mentioned forces can be managed by various parameters such as abrasive size, rotational speed, machining gap, magnetic flux density, magnet force, etc.
Fig. 1. Surface model of the workpiece being polished with SPH. 144
Wear 424–425 (2019) 143–150
M. Mosavat and A. Rahimi 310 1 mm machining gap 1.5 mm machining gap 2 mm machining gap
Magnetic flux density (mT)
300
290
280
270
260
250
240 -15
-10
-5
0
5
10
15
Distance from center (mm)
Fig. 4. Single Al2O3 particle polishing model.
Fig. 2. Distribution of magnetic flux density in the center line of the specimen surface.
Table 1 ΔSimulation results of variations in the number of SPH elements on the responses.
In this paper, the magnetic flux density in different distances of the magnet center and in various machining gaps was calculated by using Ansoft Maxwell. The Ansoft model consisted of three Nd–Fe–B permanent magnets with relative permeability of 1.0997 and the silicon wafer specimen with relative permeability of one. The boundary of the model was considered as vacuum with relative permeability of one. Fig. 2 illustrates the distribution of magnetic flux density in the center line for the machining gaps of 1 mm, 1.5 mm, and 2 mm. As shown in Fig. 2, by increasing the machining gap, magnetic flux density decreases. The maximum flux density obtained was 301 mT for the machining gap of 1 mm. Then, the magnetic normal force was estimated using Eq. (2), and the resulted force was applied to MAP during the MAF process simulation.
Experiment results
Percentage change in surface roughness (%ΔRa) Material removal (mg)
Simulation results with various number of SPH elements 43,851
51,623
59,132
65%
49%
58%
74%
39.09
31.12
36.05
48.68
In the simulation of this model, it was assumed that the silicon workpiece was fixed with a relative motion in some directions. Abrasive particles could move on a workpiece surface in different rotational speeds while all constraints are exerted from the right and bottom side of the workpiece. In our model, the left, upper surfaces and the interior nodes are allowed to move in all directions. The contact algorithm of CONTACT-AUTOMATIC-NODES-TO-SURFACE in LS-DYNA was applied due to the application of the SPH particles. This contact is used between the Al2O3 FEM elements as well as the SPH particles of the workpiece. In this study, the workpiece material is silicon and it is considered as a piecewise-linear elastic material in LS-DYNA, and full details are shown in Table 2. Al2O3 considered being rigid. Accordingly, by using the SPH/FEM, the machining gap parameters are set up separately to reveal the significance of machining gap on % ΔRa and MR. The influence of machining gap on %ΔRa and MR is studied by simulations and experimental examinations of the MAF polishing method with four various machining gaps under the same abrasive rotational speed of 1500 rpm. Also, to reveal the influence of abrasive rotational speed on mentioned outputs, simulations and experimental examinations are performed with four different rotational speeds under the same machining gap of 2 mm. Table 3 presents the groups of simulation schemes.
4. Polishing model It should be noted that the polishing process is a complicated task, and the MAP(s) have no systematic shape, and limited deformation occurs by the finishing process. Fig. 3 shows the MAP Scanning Electron Microscope (SEM) image, and we simplified the modeling of the polishing process for reducing the time of consumption. Fig. 4 illustrates the single abrasive 3D model of the MAF. It is clear that Al2O3 is much harder than silicon workpiece. Hence, Al2O3 particles are supposed as a rigid spherical body with Lagrange meshing of 10500 hexahedron elements. To simplify the model, a part of the circular workpiece is considered in the FEM/SPH model. Thus, the size of the block is 16 µm × 32 µm × 10 µm, and there are 51,623 smooth particles inside the domain. To investigate the adequacy of the number of SPH elements on the simulation results, three different grid sizes are simulated according to Table 1. Finally, the model with 51,623 grids resulted in the closest response.
Table 2 Properties of the silicon wafer.
Fig. 3. Magnetic abrasive particle SEM image. 145
Young's modulus (GPa)
160
yield stress (GPA) Poisson's ratio density (gr/cm3) The coefficient of thermal expansion (10–6/°C) Thermal conductivity (W/m K) Specific heat capacity (J/g K) melting point (°C)
7 0.22 2.4 2.6 157 0.7 1415
Wear 424–425 (2019) 143–150
M. Mosavat and A. Rahimi
5. Experimental verification
Table 3 Simulation scheme and results. Group
Machining gap (mm)
Rotational speed (rpm)
Percentage change in surface roughness %ΔRa
Material removal (mg)
1
1 1.5 2 2.5 2 2 2 2
1500 1500 1500 1500 800 1250 1500 2000
58 43 21 14 19 34 38 47
36.05 23.97 13.21 4.06 2.96 12.11 15.65 21.45
2
Fig. 5 shows the MAF experimental setup. Experiments were done on a 3-in.-diameter circular flat silicon wafer. An aluminum fixture was used to hold the wafer. The fixture connected to the vacuum pump for fixing the wafer on the fixture. The other fixture which connects to the motor attaches the three Nd-Fe-B permanent magnets with the thickness of 5 mm and the inner and outer diameter of 20 and 30 mm which located on the above of the wafer. The MAPs used in this study contain Al2O3 abrasives, iron powders, and glass powder as a binder. The particles were initially mixed to obtain a homogeneous mixture and compressed into a cylindrical manner and sintered in a vacuum furnace at 900 °C for 2 h. After the sintering process, the MAPs were crushed and sieved to obtain abrasive particles. Fig. 6 shows sintered MAPs scanned image with SEM. Al2O3 abrasives particles used in experiments was 1.0 µm. The weight ratio
Fig. 5. Experimental setup of MAF: (a) scheme figure (b) actual set up. 146
Wear 424–425 (2019) 143–150
Percentage change in surface roghness
M. Mosavat and A. Rahimi
70
Experiment Simulation
60
50
40
30
20
10 1
Fig. 6. Magnetic Abrasive Particle SEM images (Al2O3 + Fe).
1
2
2
2.5
Fig. 8. Simulation and experimental test results of percentage change in surface roughness (%ΔRa) at different machining gap, polishing time:10 min, rotational speed:1500 rpm.
Table 4 Experimental scheme and results. Group
1.5
Machining gap (mm)
Machining gap (mm)
Rotational speed (rpm)
Average percentage change in surface roughness (%ΔRa)
Average material removal (mg)
1 1.5 2 2.5 2 2 2 2
1500 1500 1500 1500 800 1250 1500 2000
65 41 21 14 16 31 36 42
39.09 26.97 14.21 6.98 3.11 7.61 9.95 11.23
between the Al2O3 and iron is 40:60, and the experimental environment temperature is about 25 °C. Each experiment was carried out four times, and after each test, the surface roughness and material removal were measured. The initial surface roughness of wafers was not equal, but all roughness values are within 0.11–0.39 µm. So, percentage changes in surface roughness which is given in Eq. (3). considered as a response of experiments.
Fig. 7. Simulation results of machining gap at: (a) 1 mm, (b) 1.5 mm, (c) 2 mm, (d) 2.5 mm. 147
Wear 424–425 (2019) 143–150
M. Mosavat and A. Rahimi
Percentage change in surface roughness
40
Material reomival (mg)
Experiment Simulation
30
20
10
1
1.5
2
2.5
45
Experiment Simulation
40
35
30
25
20
15 800
Machining gap (mm)
1000
1200
1400
1600
1800
2000
Rotational speed (rpm)
Fig. 9. Simulation and experimental test results of material removal at different machining gap, polishing time:10 min, rotational speed: 1500 rpm.
Fig. 11. Simulation and experimental test results of percentage change in surface roughness (%ΔRa) at different rotational speed, polishing time:10 min, machining gap:2 mm.
Initial Surface Roughness − Final Surface Roughness ⎞ %ΔR a = ⎜⎛ ⎟ × 100 Initial Surface Roughness ⎝ ⎠ (3)
Fig. 8 and Fig. 9, respectively. As shown in Figures, in both simulation and experimental curves, %ΔRa and MR decrease as the machining gap is increased. The obtained results indicate that the effect of magnetic flux density on abrasives reduced when the machining gap was increased. In this condition, the indentation force of cut reduces. In larger machining gaps, due to low magnetic flux density, the exerted force is not sufficient for effectively indentation and cutting to yield effective cuts on the workpiece surface. This also causes magnetic abrasive inefficiency and thus, changes in surface roughness and material removal value decline.
AFM is also used for measuring of wafers surface roughness. Finishing experiments are done similar to the simulation layout. Table 4 presents details of polishing experiments schemes and results of our tests. 6. Results and discussion 6.1. Effect of machining gap
6.2. Effect of abrasive rotational speed
The effects of the machining gap on %ΔRa and MR are studied through simulating and experimenting with the MAF. Fig. 7 shows the results of machining gap simulation at 1, 1.5, 2, and 2.5 mm. The effect of the machining gap on %ΔRa and MR is demonstrated in
The effect of abrasive rotational speed on %ΔRa and MR is also studied through simulating and experimenting during MAF. To do this,
Fig. 10. Simulation results of rotational speed at: (a) 800 rpm, (b) 1250 rpm, (c) 1500 rpm, (d) 2000 rpm. 148
Wear 424–425 (2019) 143–150
M. Mosavat and A. Rahimi
Experiment Simulation
20
Material removal (mg)
effect of work material may also be contributed to an increase in torque. As shown in Fig. 12, with increasing the abrasive rotational speeds, the difference between simulation and experiment results significantly increases and MR changes in the simulation are much more than experimental results. This is because of that in MAF technique, the normal force applied on MAPs consists of centrifugal and magnetic forces which are perpendicular to each other. When the abrasive rotational speed increases, the centrifugal force also rises. Increasing the centrifugal force applied on MAPs with growing the abrasive rotational speeds causes the MAP to leave the machining gap and run out of the circle. Thus, the total number of MAPs reduces. It should be noted that MAPs close to weak magnetic field first left the gap and this behavior intensifies within the zones with the stronger magnetic field. Finally, decreasing the MAPs that presents in the machining gap augments differences between simulation and experimental results as shown in Fig. 12. Finally, both simulation and experimental results confirm that the MR and %ΔRa increase by enlarging of abrasive particle size and our results are in good agreement with the results of Jain et al. [9].
15
10
5
800
1000
1200
1400
1600
1800
2000
Rotational speed (rpm) Fig. 12. Simulation and experimental test results of material removal at different rotational speed, polishing time:10 min, machining gap:2 mm.
6.3. Material removal mechanism To investigate the material removal mechanism during MAF, AFM was used for inspection of the surfaces of the specimen. Fig. 13a presents the surface roughness measurement of an experiment with 2.0 mm machining gap and 800 rpm rotational speed after 10 min of finishing. The results of 1.0 mm machining gap and 1500 rpm abrasive rotational speed with the same polishing times are depicted in Fig., 13b. As shown in Fig. 12a, polished surface with cut marks indicates that the microcutting material removal mechanism in ductile mode has occurred during the polishing process, and the resulted surface is free from surface damages and fractures. On the contrary, the topography of surface (as shown in Fig. 13b) indicates that side cracks achieved on the surface and micro-fracture mechanism occurs during the MAF. Results of surface topography clearly show that material removal has taken part in the brittle mode accompanied by lateral cracks, which initiated under the abrasive and propagate to the surface. Therefore, the quality of the surface of this case is lower than the previous one. According to our findings, polishing condition considerably influences on micro-fracture and micro-cutting mechanisms. Comparison of the surface roughness of our results confirms that surface with high roughness is achieved as the micro-fracture mechanism occurs. Meanwhile, micro-cutting mechanism resulted in a nano-level surface with no surface and subsurface damages. 7. Conclusions In this study, the algorithms of FEM and SPH are coupled to simulate the finishing process of monocrystalline silicon wafers with the MAF method. This work presents a comprehensive analysis of the effect of rotational speed and machining gap on %ΔRa and MR. Several experiments also performed on a 3-in.-diameter circular silicon wafer with various machining gaps and rotational speeds. Also, the material removal mechanisms were investigated in wafers by using AFM. Following conclusions could be derived from this study.
Fig. 13. Measurement results of surface roughness.
2.0 mm machining gap and 1 µm abrasive size with four different abrasive rotational speeds are considered. The simulation results of the abrasive rotational speed at 800, 1250, 1500, and 2000 rpm are shown in Fig. 10. The effects of abrasive rotational speed on %ΔRa and MR are shown in Fig. 11 and Fig. 12. As observed in figures, in both simulation and experiment curves, improvement of MR and %ΔRa increases with the rise of the rotational speed. The finishing torque increases when the abrasive rotational speed rises. Indeed, high abrasive rotational speeds produce more amounts of MAPs shear on the peaks of the workpiece in unit time, So, the change in momentum per unit time is high and this leads to high torque. On the other side, due to the high shear rate of peaks, the workpiece hardening
● Machining gap and rotational speed are dominant parameters during MAF process. ● Both simulation and experiments results show that MR value and % ΔRa increases with decreasing the machining gap. ● Both Results of simulation and experiments indicate that MR value and %ΔRa increases when the abrasive rotational speed increases. ● According to the results of experiments in the optimum condition, maximum MR and %ΔRa are 39.09 mg and 65%, respectively. ● The results of our experimental examinations indicate that the SPH method is reliable and robust in SPH/FEM coupled method for 149
Wear 424–425 (2019) 143–150
M. Mosavat and A. Rahimi
simulating the MAF process. Moreover, the simulation agrees well with experimental results and this confirms that the applied numerical technique is efficient and reliable. ● AFM observations indicate that polishing condition considerably influences on micro-fracture and micro-cutting mechanisms, and rougher surfaces achieve as the micro-fracture mechanism occurs. Meanwhile, micro-cutting mechanism resulted in a nano-level surface without surface and subsurface damages.
[12] P. Saraeian, H. Soleimani Mehr, B. Moradi, H. Tavakoli, O. Khalil Alrahmani, Study of magnetic abrasive finishing for AISI321 stainless steel, Mater. Manuf. Process. 31 (2016) 2023–2029. [13] M. Vahdati, S. Rasouli, Study of magnetic abrasive finishing on freeform surface, Trans. IMF 94 (2016) 294–302. [14] V.C. Shukla, P.M. Pandey, U.S. Dixit, A. Roy, V. Silberschmidt, Modeling of normal force and finishing torque considering shearing and ploughing effects in ultrasonic assisted magnetic abrasive finishing process with sintered magnetic abrasive powder, Wear 390 (2017) 11–22. [15] S. Malkin, T. Hwang, Grinding mechanisms for ceramics, CIRP Ann. 45 (1996) 569–580. [16] J. Feng, P. Chen, J. Ni, Prediction of surface generation in microgrinding of ceramic materials by coupled trajectory and finite element analysis, Finite Elem. Anal. Des. 57 (2012) 67–80. [17] D. Marshall, A.G. Evans, B.K. Yakub, J. Tien, G. Kino, The nature of machining damage in brittle materials, Proc. R. Soc. Lond. A 385 (1983) 461–475. [18] T.G. Bifano, T.A. Dow, R.O. Scattergood, Ductile-regime grinding: a new technology for machining brittle materials, J. Eng. Ind. 113 (1991) 184–189. [19] J. Yan, M. Yoshino, T. Kuriagawa, T. Shirakashi, K. Syoji, R. Komanduri, On the ductile machining of silicon for micro electro-mechanical systems (MEMS), optoelectronic and optical applications, Mater. Sci. Eng. A 297 (2001) 230–234. [20] T. Nakasuji, S. Kodera, S. Hara, H. Matsunaga, N. Ikawa, S. Shimada, Diamond turning of brittle materials for optical components, CIRP Ann.-Manuf. Technol. 39 (1990) 89–92. [21] H. Suzuki, S. Kodera, S. Hara, H. Matsunaga, T. Kurobe, Magnetic field-assisted polishing—application to a curved surface, Precis. Eng. 11 (1989) 197–202. [22] T. Mori, K. Hirota, Y. Kawashima, Clarification of magnetic abrasive finishing mechanism, J. Mater. Process. Technol. 143 (2003) 682–686. [23] V. Jain, Abrasive-based nano-finishing techniques: an overview, Mach. Sci. Technol. 12 (2008) 257–294. [24] V. Jain, Magnetic field assisted abrasive based micro-/nano-finishing, J. Mater. Process. Technol. 209 (2009) 6022–6038. [25] L.B. Lucy, A numerical approach to the testing of the fission hypothesis, Astron. J. 82 (1977) 1013–1024. [26] R.A. Gingold, J.J. Monaghan, Smoothed particle hydrodynamics: theory and application to non-spherical stars, Mon. Not. R. Astron. Soc. 181 (1977) 375–389. [27] M. Heinstein, D. Segalman, Simulation of orthogonal cutting with smooth particle hydrodynamics, Sandia Report, Livermore, California, USA, 1997. [28] A.B. Khairy, Aspects of surface and edge finish by magnetoabrasive particles, J. Mater. Process. Technol. 116 (2001) 77–83.
References [1] C. Chen, P. Zhang, C. Xiao, L. Chen, L. Qian, Effect of mechanical interaction on the tribochemical wear of bare silicon in water, Wear 376 (2017) 1307–1313. [2] B. Yu, H. He, L. Chen, L. Qian, Tribological behavior of monocrystalline silicon from single-to multiple-asperity scratch, Wear 374 (2017) 29–35. [3] C. Xiao, C. Chen, J. Guo, P. Zhang, L. Chen, L. Qian, Threshold contact pressure for the material removal on monocrystalline silicon by SiO2 microsphere, Wear 376 (2017) 188–193. [4] P. Zhang, C. Chen, C. Xiao, L. Chen, L. Qian, Comparison of wear methods at nanoscale: line scanning and area scanning, Wear 400 (2018) 137–143. [5] T. Shao, F. Ge, Y. Dong, K. Li, P. Li, D. Sun, F. Huang, Microstructural effect on the tribo-corrosion behaviors of magnetron sputtered CrSiN coatings, Wear 416 (2018) 44–53. [6] R. Knoblauch, D. Boing, W.L. Weingaertner, K. Wegener, F. Kuster, F.A. Xavier, Investigation of the progressive wear of individual diamond grains in wire used to cut monocrystalline silicon, Wear 414 (2018) 50–58. [7] A. Kumar, S. Kaminski, S.N. Melkote, C. Arcona, Effect of wear of diamond wire on surface morphology, roughness and subsurface damage of silicon wafers, Wear 364 (2016) 163–168. [8] Q. Zhang, Y. Fu, H. Su, Q. Zhao, S. To, Surface damage mechanism of monocrystalline silicon during single point diamond grinding, Wear 396 (2018) 48–55. [9] V. Jain, P. Kumar, P. Behera, S. Jayswal, Effect of working gap and circumferential speed on the performance of magnetic abrasive finishing process, Wear 250 (2001) 384–390. [10] D.K. Singh, V. Jain, V. Raghuram, R. Komanduri, Analysis of surface texture generated by a flexible magnetic abrasive brush, Wear 259 (2005) 1254–1261. [11] F. Pashmforoush, A. Rahimi, Nano-finishing of BK7 optical glass using magnetic abrasive finishing process, Appl. Opt. 54 (2015) 2199–2207.
150