- Email: [email protected]
On form accuracy and surface roughness in micro-ultrasonic machining of silicon microchannels
Precision Engineering xxx (xxxx) xxx–xxx
Contents lists available at ScienceDirect
Precision Engineering journal homepage: www.elsevier.com/locate/precision
On form accuracy and surface roughness in micro-ultrasonic machining of silicon microchannels Dungali Sreeharia,b, Apurbba Kumar Sharmab, a b
⁎
Department of Mechanical Engineering, National Institute of Technology, Uttarakhand, Srinagar, Pauri (Garhwal), Uttarakhand, 246174, India Department of Mechanical and Industrial Engineering, Indian Institute of Technology Roorkee, Roorkee, Haridwar, Uttarakhand, 247667, India
A R T I C LE I N FO
A B S T R A C T
Keywords: Surface roughness Overcut Stray cut Medium viscosity Feature plot
Accuracy in manufacturing microchannels is important in order to achieve their intended function. Smooth and high aspect ratio microchannels on silicon wafer substrate are needed in the heat removal application in various microelectronic components. Generally, etching techniques are used to fabricate silicon microchannels; however, the maximum achievable limit for the channel depth is a major concern. Micro-ultrasonic machining (micro-USM) is capable of machining high aspect ratio microchannels on hard and brittle material such as silicon, glass, ceramics, etc. However, achieving reasonable form accuracy and surface roughness of the microchannels is challenging. Overcut and edge damage (stray cut) are undesirable for precision machining while surface roughness of the microchannels can be set at an optimized value to attain maximum heat transfer. In the present study, silicon microchannels were fabricated using the micro-USM technique. In order to improve the precision and quality of the fabricated silicon microchannels in terms of surface roughness, overcut and stray cut; viscous fluids with different viscosities were considered for investigation in combination with other machining conditions. The experimental investigation revealed that using low viscous fluids yields better surface roughness compared to high viscous fluid; however, overcut and stray cut were minimized while using high viscous fluids. Machining at higher feed rates could minimize the surface roughness, over cut and stray cut irrespective of the abrasive concentration percentage. Possible interactions between the tool, abrasive and workpiece in the machining zone were analyzed vis-à-vis the experimental results.
1. Introduction Focus in improving the capabilities and accuracies in micromachining has been on the rise with an increase in the demand for miniaturized components in various applications such as lab-on-chips, micro heat exchangers, electronic systems, micro-reactors, micro-electromechanical systems (MEMS), micro-fluidic systems, etc. Fabrication of precise microchannels for the intended components is one of the most primary requirements in all such applications. Microchannels have attained prominence in miniaturization, especially in the electronic industry, due to their capability of removing high heat fluxes. Aluminum, copper, stainless steel and silicon are commonly used substrate materials to fabricate microchannels due to their good mechanical and thermal properties. However, silicon microchannels are popular in heat transfer applications in microelectronic devices. The first silicon microchannel was developed by Tuckerman and Pease in 1980’s by an orientation dependent etching technique [1]. These microchannels were used to remove a high heat flux (∼790 W/ cm2) from a small area with deionized water as a working fluid. Later, ⁎
research was mostly focused on fabrication of precise microchannels with different cross-sections, aspect ratios and surface conditions using different micromachining techniques such as micro-EDM, micro-ECM, micro-LBM, micro-USM, LIGA, photolithography, micro-cutting, microcasting, etc. [2,3]. Heat transfer characteristics were investigated experimentally by Qu et al. and compared the results with numerical data on trapezoidal silicon microchannels fabricated by anisotropic etching technique [4]. They reported significant difference between these results which was attributed to the surface roughness of the microchannel walls. Wu and Cheng had fabricated differently sized silicon microchannels of trapezoidal cross-section by varying surface conditions by the wet etching technique and studied the heat transfer characteristics [5]. The studies revealed that both the surface roughness and geometric parameters of the microchannels had significant effect on their heat transfer characteristics. Attempts were made to improve the surface finish of the walls and bottom of the silicon microchannels by (i) varying the temperature of silicon etchant, (ii) controlling the composition of the etchant and (iii) orientation of the masking pattern [6]. The wet and dry etching techniques are commonly used to fabricate
Corresponding author. E-mail address: [email protected] (A.K. Sharma).
https://doi.org/10.1016/j.precisioneng.2018.04.014 Received 14 January 2018; Received in revised form 10 March 2018; Accepted 16 April 2018 0141-6359/ © 2018 Elsevier Inc. All rights reserved.
Please cite this article as: Sreehari, D., Precision Engineering (2018), https://doi.org/10.1016/j.precisioneng.2018.04.014
Precision Engineering xxx (xxxx) xxx–xxx
D. Sreehari, A.K. Sharma
Fig. 1. (a) Schematic of the micro-USM setup used for the trials (inset: enlarged view of the machining zone), (b) Image of the experimental setup.
neglected or not given priority in the previous studies as in most of the investigations water was used in the slurry medium [16]. However, it was also reported that there was an improvement in the surface finish while using oil-based abrasive slurry than the water-based abrasive slurry [17]. It was further shown that the workpiece feed rate also had a considerable influence on surface roughness, overcut and stray cut of the glass microchannel fabricated by the micro-USM [18]. In the present study, the effect of slurry viscosity, abrasive slurry concentration and workpiece feed rate on surface quality, overcut and stray cut of the silicon microchannel were investigated. The experimental methodology adopted and the results obtained have been presented with relevant analyses.
silicon microchannels; however, these techniques are dependent on orientation and etch temperature. Further, etching techniques require clean room environment; the chemicals which are used for etching could create a hazardous environment. In photolithography technique, surface roughness may no longer be an issue, but the major limitation is the low aspect ratio. Practically, 3D microchannels cannot be machined using such techniques and it is difficult to control the desired depth of the channels. The 3D microchannels could be effectively machined by micro-USM technique as demonstrated by some authors [7–10]. The 3D microchannels fabricated using micro-USM on glass and silicon were investigated in terms of surface roughness, material removal rate and tool wear rate by varying different process parameters [10]. Surface roughness characteristics on (100) silicon wafers machined using micro-ultrasonic technique were studied and showed that the surface roughness depends mainly on the depth of cut and the size of the abrasive particles in the slurry mixture [11]. Several attempts were made to improve the surface finish of the profile generated by micro-USM. Abrasive Flow Machining (AFM) technique was used to improve the surface quality of the SiC microchannels fabricated by micro-USM and the authors had reported a significant improvement (approximately 59.3%) [12]. Few researchers had applied a hard wax coating on the substrate material to control the crack initiation and hence to control the surface quality while fabricating microholes on the glass [13]. The authors claimed that the wax coating on the substrate helps in protecting the surface by minimizing the crack initiation. Chemical assisted machining was explored by Choi et al. in order to improve the surface integrity of ultrasonic machined profile [14]. Form accuracy of the microchannels in terms of overcut and stray cut, along with surface quality, are important issues that need attention. In micro-USM, the surface quality, overcut and stray cut are significantly affected by the abrasive particle movement [15]. Further, abrasive movement depends on the viscosity of the slurry medium and on the slurry concentration. The viscosity effect of the slurry was either
2. Methodology 2.1. Selection of process parameters Abrasive slurry in USM acts as a coolant for the horn, tool and workpiece; it also supplies fresh abrasives to the cutting zone and removes debris from the cutting area [19]. It should also provide good acoustic bond allowing efficient energy transfer between the tool, abrasive and the workpiece. In general, water is used as a slurry medium. In the present study, three different slurry media were used to investigate the effect of viscosity of the slurry on the improvement of surface quality and form accuracy of the machined microchannel. Three different viscous fluids namely – palm oil, transformer oil and water, having relatively high, moderate and low viscosity, were taken as slurry medium for the trials, these media are low cost and easily available. The dynamic viscosities of these fluids at 30 °C are 57.85 mPa.s, 13.44 mPa.s and 0.7972 mPa.s, respectively [20]. It was reported that slurry concentration and workpiece feed rate are the other parameters that significantly affect the machining process [18]. Therefore, these two parameters were considered while investigating the performance of the three slurry types. 2
Precision Engineering xxx (xxxx) xxx–xxx
D. Sreehari, A.K. Sharma
Table 1 Input process parameters and their levels.
Overcut = (Actual width − nominal width), μm
S.No.
Input parameter
Level 1
Level 2
Level 3
1 2 3
Slurry medium Slurry concentration (%) Feed rate (mm/min)
Palm oil 20 10
Transformer oil 15 15
Water 10 20
(1)
The stray cut was defined as the damage at the edges of the microchannel due to lateral fracture. In an earlier work, an image processing technique was successfully used to quantify the stray cut; the percentage of stray cut was calculated as follows [18]:
Percentage of stary cut = 2.2. Experimentation
number of surface damage pixels × 100% number of actual channel pixels
In the present study, image processing technique was used to quantify the stray cut and was measured in terms of percentage of pixels; codes were developed using widely used tool (MATLAB, R2014b). The RGB image (Fig. 2(a)) taken under the microscope, was converted into black and white on which the edge damage could be easily visible. The damage was measured by counting the number of pixels on the damaged area and the total number of pixels present on the channel width (Fig. 2(b)) using the codes developed. The white area in Fig. 2(b) shows the stray cut and the black part between the two parallel lines is the microchannel width.
Machining trials were conducted using a micro-ultrasonic machine setup at 20 kHz and 800 W. The experimental setup was developed in the laboratory and is shown in Fig. 1. A solid cylindrical tungsten carbide rod of 600 μm diameter was taken as the tool. P – type single crystal silicon wafer having the crystal structure of (100) with 4–inch diameter and 525 μm thickness was cleaved into small workpieces of size 10 × 10 mm2. The workpiece was attached to the work holding device on the X-Y table having a linear resolution of 0.1 μm. The movement of the X-Y table was controlled using a customized CNC controller. Linear microchannels of rectangular cross-section (W × H = 600 μm × 300 μm) and length 10 mm were fabricated on the silicon wafers. A layer-by-layer machining approach was used to achieve a depth of 300 μm, using a constant step feed of 10 μm in Z – direction. Silicon carbide (SiC) particles of grit size #1800 were used as the abrasive. Full factorial methodology was followed to design and perform the experiments by varying three selected process parameters at three different levels each. The input parameters and their levels are shown in Table 1. The response parameters were considered as surface roughness, overcut and stray cut. The samples were cleaned in the ultrasonic bath after the trials. The average surface roughness (SR, Ra) was measured using a stylus-type surface roughness tester (Mitutoyo, Surftest SJ-410) with cut-off length 0.8 mm. The images of the machined microchannels taken under the microscope were analyzed to quantify overcut and stray cut using image processing technique. The overcut was defined as the difference between actual width and the nominal width of the microchannel to be fabricated (Fig. 2).
3. Interactions in the machining zone In abrasive based machining, the material removal and the dimensional accuracy of the machined part are related to the interactions of the tool, abrasives and the work surface in the machining zone. The stray cut, overcut and surface finish are significantly influenced by the interactions in the zone. In USM, the interactions are basically governed by (i) flow characteristics of the slurry, (ii) abrasive characteristics, (iii) tool characteristics and (iv) characteristics of the input energy. Considering that the input ultrasonic vibration at constant frequency and constant amplitude is applied in an USM setup where the abrasive particles are spherical and are being flown through the interaction zone in laminar motion, the possible interactions maybe analysed as follows. If relatively small particles are made to flow in a high viscous medium, then the viscosity is the dominating factor that determines the amount of resistance the particles encounter. If the fluid is flowing with a velocity ‘u’ and the abrasive particles are moving with a velocity ‘ua’ in the fluid then an opposite resisting force acts on the abrasive
Fig. 2. (a) Top view of a typical microchannel, (b) A black and white image of microchannel for straycut analysis. 3
Precision Engineering xxx (xxxx) xxx–xxx
D. Sreehari, A.K. Sharma
Fig. 3. Various components of forces induced on the abrasive particle in the slurry medium.
Now, the total momentum force (F) of a particle can be divided into two components i.e., horizontal (Fh) and vertical (Fv). In addition, two other forces −FB (=buoyancy force) and Fg (=gravity force) also act on the particles. Therefore, at an instant of time (t = t1), an abrasive particle experiences the forces Fh, Fv, Fd, FB & Fg (Fig. 3(a)). Due to these forces, the resultant force FR1 acts on the abrasive particle.
Table 2 Experimental values of the response parameters. S.No.
Slurry medium
Abrasive Conc. (%)
Feed rate (mm/ min)
SR, Ra (μm)
Overcut (μm)
Stray cut (%)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27
Palm Oil Palm Oil Palm Oil Palm Oil Palm Oil Palm Oil Palm Oil Palm Oil Palm Oil Transformer Transformer Transformer Transformer Transformer Transformer Transformer Transformer Transformer Water Water Water Water Water Water Water Water Water
20 20 20 15 15 15 10 10 10 20 20 20 15 15 15 10 10 10 20 20 20 15 15 15 10 10 10
10 15 20 10 15 20 10 15 20 10 15 20 10 15 20 10 15 20 10 15 20 10 15 20 10 15 20
0.91 0.87 0.76 0.93 0.89 0.64 0.91 0.65 0.63 0.65 0.52 0.40 0.44 0.39 0.35 0.59 0.51 0.38 0.76 0.59 0.56 0.64 0.56 0.52 0.64 0.52 0.43
16.09 9.22 10.83 28.52 22.98 17.21 22.98 16.27 13.51 24.80 11.44 8.22 13.33 13.33 5.56 17.07 14.93 10.79 28.80 17.80 9.07 46.40 42.50 37.15 29.33 21.36 10.29
7.98 7.49 6.87 6.24 7.53 5.01 14.97 10.66 5.33 11.06 10.78 9.64 11.38 12.04 10.92 10.16 9.89 8.18 14.22 13.29 8.64 25.05 22.77 21.89 19.87 11.05 8.02
Oil Oil Oil Oil Oil Oil Oil Oil Oil
FR1 =
Y = a/2 sin (2πft)
(6)
Therefore, velocity Y˙ = πaf cos (2πft )
(7)
Maximum velocity at an instant of time ‘t’ occurs at cos(2πft) = 1
i.e., Y˙max = πaf
(8)
At t = t3 (Fig. 3(b)), the tool hits the abrasive particle with the maximum velocity (Y˙max ) and the abrasive particle is acted upon by an impact force (Fi) in the vertical direction. Further, a resultant force ‘FR2’ acts on the abrasive particle due to the forces ‘Fh − Fd’ and ‘Fi’.
FR2 =
(Fh − Fd )2 + Fi 2
(9)
The particle then hits the workpiece with FR2 which is basically responsible for the median and lateral cracks (Fig. 3(c)) and indentation. At t = t4 (Fig. 3(c)), when the particle hits the workpiece, an equal and opposite reaction force (Fn) acts on the abrasive particle in the direction normal to the workpiece. As the Fn helps in rebounding of the abrasive which is eventually carried away by the flow velocity. If the resultant force FR2 is large, the abrasives hit the surface, rebounds and leave the machining zone creating a crack on it. Otherwise, if FR2 is too small to cause an impact, abrasives may get accumulated in the machining zone due to lack of rebounding, presence of high drag and its inertia. Therefore, based on the magnitude of FR2 two possibilities may arise – (i) the abrasives may get accumulated in the machining zone, or (ii) move freely after hitting the machining surface. However, the damage may occur on the bottom surface of the channel
(3)
where, μ – dynamic viscosity of the fluid, R – average radius of the particle and v – relative velocity of the abrasive particle. According to Stokes’s law the drag force is given by [21], Fd = − 6πμRv
(5)
If the abrasive particle is assumed to be moving in the horizontal direction (i.e., direction of flow) and was not settling down due to its gravitaional force, then the horizontal component of the forces was dominating and hence the vertical component of forces could be neglected (Fv, Fg & FB → 0). At an instant of time t = t2, where t2 > t1, it comes into the working gap i.e., in between the tool and the workpiece (Fig. 3(a)) with a force of (Fh − Fd). Let us consider that the tool vibrates with an amplitude ‘a/2′ microns, frequency ‘f’ kHz, for time ‘t’ seconds; then the displacement (Y) of the tool in the vertical direction is given by
particles with the drag force (Fd), which depends on the viscosity of fluid and size, shape and velocity (v = u − ua) of the particles relative to the fluid. Fd = f (μ, R, v)
(Fh − Fd )2 + (Fv + Fg − FB )2
(4)
Here, negative sign indicates that the drag force acts in the opposite direction to flow velocity. Fig. 3 explains different components of forces acting on the abrasive particles in the slurry medium at different instants of time (i.e., t0, t1, t2, t3, t4, t5). Let us consider the average mass of an abrasive particle is ‘m’. 4
Precision Engineering xxx (xxxx) xxx–xxx
D. Sreehari, A.K. Sharma
Fig. 4. Effect of feed rate on surface roughness of the microchannels.
illustrates the effect of the three slurry media on surface roughness at 20% abrasive concentration. Here, machining in high viscous slurry medium exhibited higher surface roughness while compared to other fluids at each feed rate. The free movement of abrasive particles decreases as the resisting force (Fd) on the abrasive particles is more in high viscous fluids as per the Eq. (4). As a result, the abrasive particles may get accumulated between the tool and workpiece. The tool hits on the accumulated abrasives repeatedly due to hammering action, and creates several grooves on the work surface leading to higher surface roughness (Fig. 5(a)). Similar result of surface damage was observed by Yu et al. due to the accumulation of debris in the machining zone [22]. As the viscosity of slurry medium decreases the resisting force on the abrasive particles decreases (Eq. (4)) which makes the abrasives to move more freely in the machining zone. Due to low drag, the abrasives hit the surface with a higher resultant force (Eq. (9)) which could also lead to more surface damage. In water based abrasive machining, the material removal takes place by the impact of free moving abrasive particles on the workpiece because of low drag on abrasives. From Fig. 4(a–c) it was observed that the surface roughness in the water based abrasive medium is next to the high viscous fluid (palm oil based abrasive slurry). Therefore, in high viscous fluids as well as low viscous fluids significant surface damage was observed. Accumulation of abrasives/debris and the relatively free movement of abrasives in the machining zone owing to less drag were the reasons for the observed surface damage in high and low viscous fluids, respectively. However, a
and on the channel wall edges (stray cut). As the viscosity of fluid changes (increases/decreases), the resisting force on the abrasives also gets affected (increases/decreases) which in turn causes the horizontal force (Fh) to change (decrease/increase) and hence a corresponding variation in FR2 occurs. Thus, it is clear that the velocity of the abrasive particles and the magnitude of the force which they interact with the workpiece surface influences the quality of the machined part in terms of crack formation, stray cut and overcut to a large extent. The motion of the particles inside a fluid medium can be significantly affected/controlled by using suitable viscosity of the medium. The experimental investigations to this effect were carried out in the context of the present machining conditions and are presented in the following section. 4. Results and discussion Trials were conducted with the selected process parameters using full factorial experimentation approach. Each trial was conducted thrice and the average responses in terms of surface roughness, overcut and stray cut were presented in Table 2. 4.1. Effect of process parameters on surface roughness Fig. 4(a–c) show the variations in surface roughness while varying feed rate with three different levels of abrasive concentrations. Fig. 4(a) 5
Precision Engineering xxx (xxxx) xxx–xxx
D. Sreehari, A.K. Sharma
Fig. 5. (a) Schematic diagram of the location for the SEM imaging; SEM images of machined surface using (b) palm oil (c) transformer oil (d) water slurry medium.
better surface finish was observed while machining with transformer oil based abrasive slurry (Fig. 4(a–c)) having a moderate dynamic viscosity with respect to palm oil and water. The reason for this observation is that the abrasive accumulation in the machining zone was reduced, also free movement of abrasives was restricted due to the moderate viscosity of the transformer oil. Fig. 4(c) represents the plot of surface roughness values at 10% abrasive concentration at three levels of feed rates. Here, it was observed that the surface roughness values of the microchannels machined using transformer oil and water are nearly similar. This was, therefore, evident that the abrasive concentration also plays a role in it though there was viscosity effect on the surface roughness. The result indicates that the accumulation of particles in the working gap was also decreasing as the abrasive concentration was decreased. A similar decreasing trend was observed in all three conditions (Fig. 4(a–c)) with increasing feed rate. It was also observed that with the moderate viscous fluid – transformer oil, the surface roughness values were minimum irrespective of the percentage of abrasive concentration and feed rate. While the high viscous medium leads to apparent clogging and increased effect of hammering, the low viscosity fluids facilitate more intensive impacts that cause more cracks resulting in inferior surface quality.
bottom surface of the machined microchannel under palm oil slurry with 20% abrasive concentration. Several grooves were observed on the surface due to the repeated indentation of abrasives on the workpiece. Though the samples were cleaned in an ultrasonic bath after machining, minute abrasive particles were observed embedded on the surface. Machining in high viscous medium got the abrasives pushed hard inside the work material when they got clogged in the machining gap. Hammering action on the trapped abrasives made the material to deform plastically rather than flaking under impact. Plastically deformed grooves are clearly seen in Fig. 5(b). On the other hand, Fig. 5(c and d) show typical SEM images of microchannel surface machined in transformer oil and water slurry media, respectively. Less embedded particles were seen on the surface and the grooves formed due to repeated indentations were also observed less; the surface presents a more uniformly brittle fractured appearance (Fig. 5(c)). However, in Fig. 5(d), too many dents were observed on the surface. Use of low viscous medium would have resulted in less drag force on the abrasive particles and as a result they could move freely in the slurry medium. Therefore, in low viscous medium the material removal by micro chipping dominates than the hammering action of abrasive particles which creates more number of shallow dents on the bottom surface of the machined microchannel which could have contributed to observed reduction in monitored surface roughness (most of the Ra values are ≤0.65 μm, Fig. 5(c–d)).
4.2. SEM analysis of surface roughness Morphology of the machined surface was also studied using scanning electron microscope (SEM). Fig. 5(b) shows SEM image of the 6
Precision Engineering xxx (xxxx) xxx–xxx
D. Sreehari, A.K. Sharma
Fig. 6. Variation of overcut with feed rate.
4.3. Effect of process parameters on overcut
4.4. Effect of process parameters on stray cut
As defined earlier, overcut was measured as the difference between the actual width and the nominal width of the microchannels. The overcut was observed more while machining with water based slurry (low viscous medium). As viscous drag force was less in low viscous fluids, the abrasives enter and leave the working gap with less resistance enlarging the width of the channel. This is evident from the plot between overcut and feed rate at different abrasive concentrations (Fig. 6(a–c)). Machining with high viscous slurry showed minimum overcut in the given conditions (Fig. 6(a)). As explained earlier, movement of the abrasive particles was constrained in the high viscous slurry resulting in the reduced cut. The monitored data confirms that the moderate viscosity transformer oil results in reduced overcut as it does not allow the abrasives to cause cutting while rebounding due to less energy available with them. Overcut was minimal at higher feed rates compared to lower workpiece feed rate irrespective of the abrasive concentration. The reason is that at a lower feed rate, the interaction time between three bodies – the workpiece, abrasive and the tool is more and hence three body abrasion wear took place in the machining zone [23]. Due to the 3–body abrasive wear, there is a possibility of an increase in the width of microchannel at the lower feed rate which is evident from the plots (Fig. 6(a–c)).
The stray cut data of the machined microchannels are presented in Fig. 7. It was evident from Fig. 7(a–c) that the machining with low viscous fluid (water) slurry results in more stray cut than the high viscous fluids (oils). As viscous drag forces are less in low viscous fluids, the resultant force FR2 will be more and will tend to move away from the normal which would result in the abrasives to move freely hitting the edges of microchannel while entering and leaving the machining zone. The stray cut is insignificant with palm oil based slurry compared to water based slurry machining. This is obvious that in high viscous fluid, abrasives tend to get accumulated in the machining zone; enters and leaves the machining zone with less FR2 due to which edge damage (stray cut) may reduce. Though the abrasive concentration of the slurry media was varied, the viscosity effect was observed significant on the stray cut in Fig. 7. Decreasing trend in the stray cut was evidenced at different abrasive concentrations with changing workpiece feed rate from level 1 (10 mm/min) to level 3 (20 mm/min). The reason could be that at lower feed rates the interaction time between tool, abrasive and work piece was more and hence, a higher number of abrasives hit the edges of the microchannel while entering and leaving the machining zone leading to observed higher stray cut. The above discussion, thus clearly indicates that the experimental observations follow the theoretical analyses provided in Section 3. The 7
Precision Engineering xxx (xxxx) xxx–xxx
D. Sreehari, A.K. Sharma
Fig. 7. Variation of stray cut with feed rate. Table 3 Theoretical model verses experimental observations. Theoretical model The resultant force FR2 (Eq. (9)) is responsible for the effect on the response parameters. FR2 = (Fh − Fd )2 + Fi 2 As μ increases, Fd will increase (Eq. (4)); thus FR2 tends to be less. It results in Higher surface roughness Lower overcut Lower stray cut
• • •
Experimental observations The experimental results showed similar variations in the response parameters with respect to μ as seen in the theoretical model. 1. Highest surface roughness obtained – palm oil (μpalm oil = 57.85 mPa.s > μtransformer oil = 13.44 mPa.s > μwater = 0.7972 mPa.s) 2. Highest overcut obtained – water (μwater = 0.7972 mPa.s < μtransformer oil = 13.44 mPa.s < μpalm oil = 57.85 mPa.s) 3. Highest stray cut obtained – water (μwater = 0.7972 mPa.s < μtransformer oil = 13.44 mPa.s < μpalm oil = 57.85 mPa.s)
parameters i.e., overcut and stray cut. Fig. 6(a–c), Fig. 7(a–c) show that at 15% abrasive concentration (Figs. 6(b) and 7(b)), there were large variations of the monitored stray cut and overcut in different slurry media at different feed rates. At 20% abrasive concentration, the movement of abrasives may get restricted due to high concentration, the variation observed in the stray cut and overcut was less and at 10% abrasive concentration, the movement of abrasives may not be restricted due to less concentration, but the variation observed was less due to presence of low number of abrasives in the machining zone. However, at 15% abrasive concentration, the restriction of movement
effects of the resultant force FR2 as derived in Eq. (9) can be explained in terms of the surface roughness, overcut and stray cut during machining. A summary of the discussion is presented in Table 3. 4.5. Feature plots In order to fabricate a precise microchannel, it is necessary to minimize both the imprecision measures – overcut and stray cut. In order to find the best possible input process parameters ‘Feature plots’ were drawn (Figs. 8 and 9) correlating the two main output process
8
Precision Engineering xxx (xxxx) xxx–xxx
D. Sreehari, A.K. Sharma
Fig. 8. Feature plot between stray cut and overcut at 15% abrasive concentration (images of stray cut at specific conditions are shown in the insets).
Fig. 9. Feature plot between stray cut and overcut at 20 mm/min feed rate (inset: images of stray cut at specific machining conditions).
relationship between overcut and stray cut. The plots shown in the clusters can be divided into four regions – R1, R2, R3 and R4 as illustrated in Figs. 8 and 9. The region R1 represents higher values of stray cut and overcut, which has to be avoided for precision machining of microchannels. The regions R2 & R3 represent lower values of stray cut at high and low values of overcut, respectively. The cluster 2 fall in R2 was considered if the stray cut is to be significant than overcut and cluster 3 which is in R3 was selected if the overcut and stray cut both be significantly small. This was evident from the microscopic images (Fig. 8 Inset). The data point in the region R1 was having an overcut 46.40 μm and stray cut 25.05% and the region R3 was having overcut 5.56 μm and stray cut 10.92%. The data points in the region R1 were corresponding to 15% abrasive concentration and water as a slurry medium at different feed rates. Therefore, machining
of abrasives was less compared to 20% and also the number of abrasives present were more compared to 10%. Hence, the variation in the stray cut and overcut was observed high at 15% abrasive concentration. Further, the monitored overcut and stray cut values were within the similar range in cases of 20% and 10% concentrations for machining with other oils barring water. Water being the least viscous among the three media used it allows relatively higher movements to the rebounding abrasives which were reflected in the higher variations in overcut and stray cut as observed in Figs. 6(b) and 7(b). Therefore, a feature plot (Fig. 8) was drawn to correlate the stray cut and overcut values at an abrasive concentration of 15% considering all three viscous fluids and feed rates. The feature plots indicate clear clusters based on the response parameter values. Data points in the cluster represent strong
9
Precision Engineering xxx (xxxx) xxx–xxx
D. Sreehari, A.K. Sharma
References
using water based slurry results in more edge damage compared to high viscous fluids. In the plots between the output process parameters with respect to feed rate, it was found that better results were obtained at higher feed rate (20 mm/min). Therefore, another feature plot (Fig. 9) was generated correlating the stray cut and the overcut corresponding to 20 mm/ min feed rate. Here, it was observed that a more concentrated cluster of data points was formed in the region R3. It explains that combination of input process parameters (slurry medium and abrasive concentration) at 20 mm/min feed rate gives minimum stray cut and minimum overcut. The findings can be evidenced by the microscopic images; the image shown in region R1 was having overcut 37.15 μm and stray cut 21.89% while the image in region R3 was showing an overcut of 13.51 μm while the corresponding stray cut was 5.33%. In practice, thus, it is implied that for precision applications where surface roughness may not be critical with respect to overcut and stray cut (where leakage is more critical than smooth flow), higher viscosity fluid be preferred to water. Further, the study also confirms that high viscosity fluids also result is better machining with higher feed rate which is indicative of higher productivity.
[1] Tuckerman DB, Pease RFW. High-performance heat sinking for VLSI. IEEE Electron Device Lett 1981;2:126–9. [2] Masuzawa T. State of the art of micromachining. CIRP Ann – Manuf Technol 2000;49:473–88. [3] Gentili E, Tabaglio L, Aggogeri F. Review on micromachining techniques. AMST’05 Adv Manuf Syst Technol 2005:387–96. [4] Qu W, Mala GM, Li D. Heat transfer for water flow in trapezoidal silicon microchannels. Int J Heat Mass Transf 2000;43:3925–36. [5] Wu HY, Cheng P. An experimental study of convective heat transfer in silicon microchannels with different surface conditions. Int J Heat Mass Transf 2003;46:2547–56. [6] Dwivedi VK, Gopal R, Ahmad S. Fabrication of very smooth walls and bottoms of silicon microchannels for heat dissipation of semiconductor devices. Microelectron J 2000;31:405–10. [7] Sun X-Q, Masuzawa T, Fujino M. Micro ultrasonic machining and self-aligned multilayer machining/assembly technologies for 3D micromachines. Proceedings of Ninth International Workshop on Micro Electromechanical Systems 1996. [8] Yu ZY, Rajurkar KP, Tandon A. Study of 3D micro-Ultrasonic machining. J Manuf Sci Eng 2004;126:727–32. [9] Jain V, Sharma AK, Kumar P. Developments and research issues in microultrasonic machining. ISRN Mech Eng 2011:1–15. 2011. [10] Jain V. Investigations on development of microchannels using micro ultrasonic machining Indian Institute of Technology Roorkee; 2012. Doctoral thesis. [11] Klopfstein MJ, Ghisleni R, Lucca DA, Brinksmeier E. Surface characteristics of micro-ultrasonically machined (1 0) silicon. Int J Mach Tools Manuf 2008;48:473–6. [12] Venkatesh G, Singh T, Sharma AK, Dvivedi A. Finishing of micro-channels using abrasive flow machining. Proceedings of the International Conference on Research and Innovations in Mechanical Engineering 2014. [13] Baek DK, Ko TJ, Yang SH. Enhancement of surface quality in ultrasonic machining of glass using a sacrificing coating. J Mater Process Tech 2013;213:553–9. [14] Choi JP, Jeon BH, Kim BH. Chemical-assisted ultrasonic machining of glass. J Mater Process Technol 2007;191:153–6. [15] Pei W, Yu Z, Li J, Ma C, Xu W, Wang X, et al. Influence of abrasive particle movement in micro USM. Procedia CIRP 2013;6:551–5. [16] Miller GE. Special theory of ultrasonic machining. J Appl Phys 1957;28:149–56. [17] Sundaram MM, Cherku S, Rajurkar KP. Micro ultrasonic machining using oil based abrasive slurry. Proceedings of International Manufacturing Science and Engineering Conference 2008. [18] Cheema MS. An ultrasonic micromachining approach to fabricate microchannels on glass Indian Institute of Technology Roorkee; 2015. Doctoral thesis. [19] Thoe TB, Aspinwall DK, Wise ML. Review on ultrasonic machining. Int J Mach Tools Manuf 1998;38:239–55. [20] Kothandaraman CP. Heat and mass transfer data book. New Age International; 2004. [21] White FM. Fluid mechanics. McGraw-Hill Book Company; 2011. [22] Yu Z, Hu X, Rajurkar KP. Influence of debris accumulation on material removal and surface roughness in micro ultrasonic machining of silicon. CIRP Ann – Manuf Technol 2006;55:201–4. [23] Cheema MS, Singh PK, Tyagi O, Dvivedi A, Sharma AK. Tool wear and form accuracy in ultrasonically machined microchannels. Measurement 2016;81:85–94.
5. Conclusion The study was aimed to fabricate precise and quality silicon microchannel by minimizing the undesirable response parameters such as surface roughness, overcut and stray cut. The following conclusions were drawn from the present investigation: a. Ultrasonic micromachining outputs are significantly influenced by the slurry related factors like medium viscosity. Higher viscosity offers constrained movement to the abrasives which tends to deteriorate surface finish while low viscosity allows the abrasives to stray around more freely causing an increase in stray cut and overcut. b. Higher workpiece feed yields better machined parts owing to shorter abrasive-tool-workpiece interaction. Surface finish and form accuracy improves at higher feed rate. c. Use of moderate viscosity medium for abrasive slurry offers effective machining ranges to choose from to achieve better machining control as evidenced by the feature plots. d. Feature plot representation is an effective technique to identify the nature of the interacting parameters and in adopting suitable machining strategy to reduce inaccuracies.
10
Contents lists available at ScienceDirect
Precision Engineering journal homepage: www.elsevier.com/locate/precision
On form accuracy and surface roughness in micro-ultrasonic machining of silicon microchannels Dungali Sreeharia,b, Apurbba Kumar Sharmab, a b
⁎
Department of Mechanical Engineering, National Institute of Technology, Uttarakhand, Srinagar, Pauri (Garhwal), Uttarakhand, 246174, India Department of Mechanical and Industrial Engineering, Indian Institute of Technology Roorkee, Roorkee, Haridwar, Uttarakhand, 247667, India
A R T I C LE I N FO
A B S T R A C T
Keywords: Surface roughness Overcut Stray cut Medium viscosity Feature plot
Accuracy in manufacturing microchannels is important in order to achieve their intended function. Smooth and high aspect ratio microchannels on silicon wafer substrate are needed in the heat removal application in various microelectronic components. Generally, etching techniques are used to fabricate silicon microchannels; however, the maximum achievable limit for the channel depth is a major concern. Micro-ultrasonic machining (micro-USM) is capable of machining high aspect ratio microchannels on hard and brittle material such as silicon, glass, ceramics, etc. However, achieving reasonable form accuracy and surface roughness of the microchannels is challenging. Overcut and edge damage (stray cut) are undesirable for precision machining while surface roughness of the microchannels can be set at an optimized value to attain maximum heat transfer. In the present study, silicon microchannels were fabricated using the micro-USM technique. In order to improve the precision and quality of the fabricated silicon microchannels in terms of surface roughness, overcut and stray cut; viscous fluids with different viscosities were considered for investigation in combination with other machining conditions. The experimental investigation revealed that using low viscous fluids yields better surface roughness compared to high viscous fluid; however, overcut and stray cut were minimized while using high viscous fluids. Machining at higher feed rates could minimize the surface roughness, over cut and stray cut irrespective of the abrasive concentration percentage. Possible interactions between the tool, abrasive and workpiece in the machining zone were analyzed vis-à-vis the experimental results.
1. Introduction Focus in improving the capabilities and accuracies in micromachining has been on the rise with an increase in the demand for miniaturized components in various applications such as lab-on-chips, micro heat exchangers, electronic systems, micro-reactors, micro-electromechanical systems (MEMS), micro-fluidic systems, etc. Fabrication of precise microchannels for the intended components is one of the most primary requirements in all such applications. Microchannels have attained prominence in miniaturization, especially in the electronic industry, due to their capability of removing high heat fluxes. Aluminum, copper, stainless steel and silicon are commonly used substrate materials to fabricate microchannels due to their good mechanical and thermal properties. However, silicon microchannels are popular in heat transfer applications in microelectronic devices. The first silicon microchannel was developed by Tuckerman and Pease in 1980’s by an orientation dependent etching technique [1]. These microchannels were used to remove a high heat flux (∼790 W/ cm2) from a small area with deionized water as a working fluid. Later, ⁎
research was mostly focused on fabrication of precise microchannels with different cross-sections, aspect ratios and surface conditions using different micromachining techniques such as micro-EDM, micro-ECM, micro-LBM, micro-USM, LIGA, photolithography, micro-cutting, microcasting, etc. [2,3]. Heat transfer characteristics were investigated experimentally by Qu et al. and compared the results with numerical data on trapezoidal silicon microchannels fabricated by anisotropic etching technique [4]. They reported significant difference between these results which was attributed to the surface roughness of the microchannel walls. Wu and Cheng had fabricated differently sized silicon microchannels of trapezoidal cross-section by varying surface conditions by the wet etching technique and studied the heat transfer characteristics [5]. The studies revealed that both the surface roughness and geometric parameters of the microchannels had significant effect on their heat transfer characteristics. Attempts were made to improve the surface finish of the walls and bottom of the silicon microchannels by (i) varying the temperature of silicon etchant, (ii) controlling the composition of the etchant and (iii) orientation of the masking pattern [6]. The wet and dry etching techniques are commonly used to fabricate
Corresponding author. E-mail address: [email protected] (A.K. Sharma).
https://doi.org/10.1016/j.precisioneng.2018.04.014 Received 14 January 2018; Received in revised form 10 March 2018; Accepted 16 April 2018 0141-6359/ © 2018 Elsevier Inc. All rights reserved.
Please cite this article as: Sreehari, D., Precision Engineering (2018), https://doi.org/10.1016/j.precisioneng.2018.04.014
Precision Engineering xxx (xxxx) xxx–xxx
D. Sreehari, A.K. Sharma
Fig. 1. (a) Schematic of the micro-USM setup used for the trials (inset: enlarged view of the machining zone), (b) Image of the experimental setup.
neglected or not given priority in the previous studies as in most of the investigations water was used in the slurry medium [16]. However, it was also reported that there was an improvement in the surface finish while using oil-based abrasive slurry than the water-based abrasive slurry [17]. It was further shown that the workpiece feed rate also had a considerable influence on surface roughness, overcut and stray cut of the glass microchannel fabricated by the micro-USM [18]. In the present study, the effect of slurry viscosity, abrasive slurry concentration and workpiece feed rate on surface quality, overcut and stray cut of the silicon microchannel were investigated. The experimental methodology adopted and the results obtained have been presented with relevant analyses.
silicon microchannels; however, these techniques are dependent on orientation and etch temperature. Further, etching techniques require clean room environment; the chemicals which are used for etching could create a hazardous environment. In photolithography technique, surface roughness may no longer be an issue, but the major limitation is the low aspect ratio. Practically, 3D microchannels cannot be machined using such techniques and it is difficult to control the desired depth of the channels. The 3D microchannels could be effectively machined by micro-USM technique as demonstrated by some authors [7–10]. The 3D microchannels fabricated using micro-USM on glass and silicon were investigated in terms of surface roughness, material removal rate and tool wear rate by varying different process parameters [10]. Surface roughness characteristics on (100) silicon wafers machined using micro-ultrasonic technique were studied and showed that the surface roughness depends mainly on the depth of cut and the size of the abrasive particles in the slurry mixture [11]. Several attempts were made to improve the surface finish of the profile generated by micro-USM. Abrasive Flow Machining (AFM) technique was used to improve the surface quality of the SiC microchannels fabricated by micro-USM and the authors had reported a significant improvement (approximately 59.3%) [12]. Few researchers had applied a hard wax coating on the substrate material to control the crack initiation and hence to control the surface quality while fabricating microholes on the glass [13]. The authors claimed that the wax coating on the substrate helps in protecting the surface by minimizing the crack initiation. Chemical assisted machining was explored by Choi et al. in order to improve the surface integrity of ultrasonic machined profile [14]. Form accuracy of the microchannels in terms of overcut and stray cut, along with surface quality, are important issues that need attention. In micro-USM, the surface quality, overcut and stray cut are significantly affected by the abrasive particle movement [15]. Further, abrasive movement depends on the viscosity of the slurry medium and on the slurry concentration. The viscosity effect of the slurry was either
2. Methodology 2.1. Selection of process parameters Abrasive slurry in USM acts as a coolant for the horn, tool and workpiece; it also supplies fresh abrasives to the cutting zone and removes debris from the cutting area [19]. It should also provide good acoustic bond allowing efficient energy transfer between the tool, abrasive and the workpiece. In general, water is used as a slurry medium. In the present study, three different slurry media were used to investigate the effect of viscosity of the slurry on the improvement of surface quality and form accuracy of the machined microchannel. Three different viscous fluids namely – palm oil, transformer oil and water, having relatively high, moderate and low viscosity, were taken as slurry medium for the trials, these media are low cost and easily available. The dynamic viscosities of these fluids at 30 °C are 57.85 mPa.s, 13.44 mPa.s and 0.7972 mPa.s, respectively [20]. It was reported that slurry concentration and workpiece feed rate are the other parameters that significantly affect the machining process [18]. Therefore, these two parameters were considered while investigating the performance of the three slurry types. 2
Precision Engineering xxx (xxxx) xxx–xxx
D. Sreehari, A.K. Sharma
Table 1 Input process parameters and their levels.
Overcut = (Actual width − nominal width), μm
S.No.
Input parameter
Level 1
Level 2
Level 3
1 2 3
Slurry medium Slurry concentration (%) Feed rate (mm/min)
Palm oil 20 10
Transformer oil 15 15
Water 10 20
(1)
The stray cut was defined as the damage at the edges of the microchannel due to lateral fracture. In an earlier work, an image processing technique was successfully used to quantify the stray cut; the percentage of stray cut was calculated as follows [18]:
Percentage of stary cut = 2.2. Experimentation
number of surface damage pixels × 100% number of actual channel pixels
In the present study, image processing technique was used to quantify the stray cut and was measured in terms of percentage of pixels; codes were developed using widely used tool (MATLAB, R2014b). The RGB image (Fig. 2(a)) taken under the microscope, was converted into black and white on which the edge damage could be easily visible. The damage was measured by counting the number of pixels on the damaged area and the total number of pixels present on the channel width (Fig. 2(b)) using the codes developed. The white area in Fig. 2(b) shows the stray cut and the black part between the two parallel lines is the microchannel width.
Machining trials were conducted using a micro-ultrasonic machine setup at 20 kHz and 800 W. The experimental setup was developed in the laboratory and is shown in Fig. 1. A solid cylindrical tungsten carbide rod of 600 μm diameter was taken as the tool. P – type single crystal silicon wafer having the crystal structure of (100) with 4–inch diameter and 525 μm thickness was cleaved into small workpieces of size 10 × 10 mm2. The workpiece was attached to the work holding device on the X-Y table having a linear resolution of 0.1 μm. The movement of the X-Y table was controlled using a customized CNC controller. Linear microchannels of rectangular cross-section (W × H = 600 μm × 300 μm) and length 10 mm were fabricated on the silicon wafers. A layer-by-layer machining approach was used to achieve a depth of 300 μm, using a constant step feed of 10 μm in Z – direction. Silicon carbide (SiC) particles of grit size #1800 were used as the abrasive. Full factorial methodology was followed to design and perform the experiments by varying three selected process parameters at three different levels each. The input parameters and their levels are shown in Table 1. The response parameters were considered as surface roughness, overcut and stray cut. The samples were cleaned in the ultrasonic bath after the trials. The average surface roughness (SR, Ra) was measured using a stylus-type surface roughness tester (Mitutoyo, Surftest SJ-410) with cut-off length 0.8 mm. The images of the machined microchannels taken under the microscope were analyzed to quantify overcut and stray cut using image processing technique. The overcut was defined as the difference between actual width and the nominal width of the microchannel to be fabricated (Fig. 2).
3. Interactions in the machining zone In abrasive based machining, the material removal and the dimensional accuracy of the machined part are related to the interactions of the tool, abrasives and the work surface in the machining zone. The stray cut, overcut and surface finish are significantly influenced by the interactions in the zone. In USM, the interactions are basically governed by (i) flow characteristics of the slurry, (ii) abrasive characteristics, (iii) tool characteristics and (iv) characteristics of the input energy. Considering that the input ultrasonic vibration at constant frequency and constant amplitude is applied in an USM setup where the abrasive particles are spherical and are being flown through the interaction zone in laminar motion, the possible interactions maybe analysed as follows. If relatively small particles are made to flow in a high viscous medium, then the viscosity is the dominating factor that determines the amount of resistance the particles encounter. If the fluid is flowing with a velocity ‘u’ and the abrasive particles are moving with a velocity ‘ua’ in the fluid then an opposite resisting force acts on the abrasive
Fig. 2. (a) Top view of a typical microchannel, (b) A black and white image of microchannel for straycut analysis. 3
Precision Engineering xxx (xxxx) xxx–xxx
D. Sreehari, A.K. Sharma
Fig. 3. Various components of forces induced on the abrasive particle in the slurry medium.
Now, the total momentum force (F) of a particle can be divided into two components i.e., horizontal (Fh) and vertical (Fv). In addition, two other forces −FB (=buoyancy force) and Fg (=gravity force) also act on the particles. Therefore, at an instant of time (t = t1), an abrasive particle experiences the forces Fh, Fv, Fd, FB & Fg (Fig. 3(a)). Due to these forces, the resultant force FR1 acts on the abrasive particle.
Table 2 Experimental values of the response parameters. S.No.
Slurry medium
Abrasive Conc. (%)
Feed rate (mm/ min)
SR, Ra (μm)
Overcut (μm)
Stray cut (%)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27
Palm Oil Palm Oil Palm Oil Palm Oil Palm Oil Palm Oil Palm Oil Palm Oil Palm Oil Transformer Transformer Transformer Transformer Transformer Transformer Transformer Transformer Transformer Water Water Water Water Water Water Water Water Water
20 20 20 15 15 15 10 10 10 20 20 20 15 15 15 10 10 10 20 20 20 15 15 15 10 10 10
10 15 20 10 15 20 10 15 20 10 15 20 10 15 20 10 15 20 10 15 20 10 15 20 10 15 20
0.91 0.87 0.76 0.93 0.89 0.64 0.91 0.65 0.63 0.65 0.52 0.40 0.44 0.39 0.35 0.59 0.51 0.38 0.76 0.59 0.56 0.64 0.56 0.52 0.64 0.52 0.43
16.09 9.22 10.83 28.52 22.98 17.21 22.98 16.27 13.51 24.80 11.44 8.22 13.33 13.33 5.56 17.07 14.93 10.79 28.80 17.80 9.07 46.40 42.50 37.15 29.33 21.36 10.29
7.98 7.49 6.87 6.24 7.53 5.01 14.97 10.66 5.33 11.06 10.78 9.64 11.38 12.04 10.92 10.16 9.89 8.18 14.22 13.29 8.64 25.05 22.77 21.89 19.87 11.05 8.02
Oil Oil Oil Oil Oil Oil Oil Oil Oil
FR1 =
Y = a/2 sin (2πft)
(6)
Therefore, velocity Y˙ = πaf cos (2πft )
(7)
Maximum velocity at an instant of time ‘t’ occurs at cos(2πft) = 1
i.e., Y˙max = πaf
(8)
At t = t3 (Fig. 3(b)), the tool hits the abrasive particle with the maximum velocity (Y˙max ) and the abrasive particle is acted upon by an impact force (Fi) in the vertical direction. Further, a resultant force ‘FR2’ acts on the abrasive particle due to the forces ‘Fh − Fd’ and ‘Fi’.
FR2 =
(Fh − Fd )2 + Fi 2
(9)
The particle then hits the workpiece with FR2 which is basically responsible for the median and lateral cracks (Fig. 3(c)) and indentation. At t = t4 (Fig. 3(c)), when the particle hits the workpiece, an equal and opposite reaction force (Fn) acts on the abrasive particle in the direction normal to the workpiece. As the Fn helps in rebounding of the abrasive which is eventually carried away by the flow velocity. If the resultant force FR2 is large, the abrasives hit the surface, rebounds and leave the machining zone creating a crack on it. Otherwise, if FR2 is too small to cause an impact, abrasives may get accumulated in the machining zone due to lack of rebounding, presence of high drag and its inertia. Therefore, based on the magnitude of FR2 two possibilities may arise – (i) the abrasives may get accumulated in the machining zone, or (ii) move freely after hitting the machining surface. However, the damage may occur on the bottom surface of the channel
(3)
where, μ – dynamic viscosity of the fluid, R – average radius of the particle and v – relative velocity of the abrasive particle. According to Stokes’s law the drag force is given by [21], Fd = − 6πμRv
(5)
If the abrasive particle is assumed to be moving in the horizontal direction (i.e., direction of flow) and was not settling down due to its gravitaional force, then the horizontal component of the forces was dominating and hence the vertical component of forces could be neglected (Fv, Fg & FB → 0). At an instant of time t = t2, where t2 > t1, it comes into the working gap i.e., in between the tool and the workpiece (Fig. 3(a)) with a force of (Fh − Fd). Let us consider that the tool vibrates with an amplitude ‘a/2′ microns, frequency ‘f’ kHz, for time ‘t’ seconds; then the displacement (Y) of the tool in the vertical direction is given by
particles with the drag force (Fd), which depends on the viscosity of fluid and size, shape and velocity (v = u − ua) of the particles relative to the fluid. Fd = f (μ, R, v)
(Fh − Fd )2 + (Fv + Fg − FB )2
(4)
Here, negative sign indicates that the drag force acts in the opposite direction to flow velocity. Fig. 3 explains different components of forces acting on the abrasive particles in the slurry medium at different instants of time (i.e., t0, t1, t2, t3, t4, t5). Let us consider the average mass of an abrasive particle is ‘m’. 4
Precision Engineering xxx (xxxx) xxx–xxx
D. Sreehari, A.K. Sharma
Fig. 4. Effect of feed rate on surface roughness of the microchannels.
illustrates the effect of the three slurry media on surface roughness at 20% abrasive concentration. Here, machining in high viscous slurry medium exhibited higher surface roughness while compared to other fluids at each feed rate. The free movement of abrasive particles decreases as the resisting force (Fd) on the abrasive particles is more in high viscous fluids as per the Eq. (4). As a result, the abrasive particles may get accumulated between the tool and workpiece. The tool hits on the accumulated abrasives repeatedly due to hammering action, and creates several grooves on the work surface leading to higher surface roughness (Fig. 5(a)). Similar result of surface damage was observed by Yu et al. due to the accumulation of debris in the machining zone [22]. As the viscosity of slurry medium decreases the resisting force on the abrasive particles decreases (Eq. (4)) which makes the abrasives to move more freely in the machining zone. Due to low drag, the abrasives hit the surface with a higher resultant force (Eq. (9)) which could also lead to more surface damage. In water based abrasive machining, the material removal takes place by the impact of free moving abrasive particles on the workpiece because of low drag on abrasives. From Fig. 4(a–c) it was observed that the surface roughness in the water based abrasive medium is next to the high viscous fluid (palm oil based abrasive slurry). Therefore, in high viscous fluids as well as low viscous fluids significant surface damage was observed. Accumulation of abrasives/debris and the relatively free movement of abrasives in the machining zone owing to less drag were the reasons for the observed surface damage in high and low viscous fluids, respectively. However, a
and on the channel wall edges (stray cut). As the viscosity of fluid changes (increases/decreases), the resisting force on the abrasives also gets affected (increases/decreases) which in turn causes the horizontal force (Fh) to change (decrease/increase) and hence a corresponding variation in FR2 occurs. Thus, it is clear that the velocity of the abrasive particles and the magnitude of the force which they interact with the workpiece surface influences the quality of the machined part in terms of crack formation, stray cut and overcut to a large extent. The motion of the particles inside a fluid medium can be significantly affected/controlled by using suitable viscosity of the medium. The experimental investigations to this effect were carried out in the context of the present machining conditions and are presented in the following section. 4. Results and discussion Trials were conducted with the selected process parameters using full factorial experimentation approach. Each trial was conducted thrice and the average responses in terms of surface roughness, overcut and stray cut were presented in Table 2. 4.1. Effect of process parameters on surface roughness Fig. 4(a–c) show the variations in surface roughness while varying feed rate with three different levels of abrasive concentrations. Fig. 4(a) 5
Precision Engineering xxx (xxxx) xxx–xxx
D. Sreehari, A.K. Sharma
Fig. 5. (a) Schematic diagram of the location for the SEM imaging; SEM images of machined surface using (b) palm oil (c) transformer oil (d) water slurry medium.
better surface finish was observed while machining with transformer oil based abrasive slurry (Fig. 4(a–c)) having a moderate dynamic viscosity with respect to palm oil and water. The reason for this observation is that the abrasive accumulation in the machining zone was reduced, also free movement of abrasives was restricted due to the moderate viscosity of the transformer oil. Fig. 4(c) represents the plot of surface roughness values at 10% abrasive concentration at three levels of feed rates. Here, it was observed that the surface roughness values of the microchannels machined using transformer oil and water are nearly similar. This was, therefore, evident that the abrasive concentration also plays a role in it though there was viscosity effect on the surface roughness. The result indicates that the accumulation of particles in the working gap was also decreasing as the abrasive concentration was decreased. A similar decreasing trend was observed in all three conditions (Fig. 4(a–c)) with increasing feed rate. It was also observed that with the moderate viscous fluid – transformer oil, the surface roughness values were minimum irrespective of the percentage of abrasive concentration and feed rate. While the high viscous medium leads to apparent clogging and increased effect of hammering, the low viscosity fluids facilitate more intensive impacts that cause more cracks resulting in inferior surface quality.
bottom surface of the machined microchannel under palm oil slurry with 20% abrasive concentration. Several grooves were observed on the surface due to the repeated indentation of abrasives on the workpiece. Though the samples were cleaned in an ultrasonic bath after machining, minute abrasive particles were observed embedded on the surface. Machining in high viscous medium got the abrasives pushed hard inside the work material when they got clogged in the machining gap. Hammering action on the trapped abrasives made the material to deform plastically rather than flaking under impact. Plastically deformed grooves are clearly seen in Fig. 5(b). On the other hand, Fig. 5(c and d) show typical SEM images of microchannel surface machined in transformer oil and water slurry media, respectively. Less embedded particles were seen on the surface and the grooves formed due to repeated indentations were also observed less; the surface presents a more uniformly brittle fractured appearance (Fig. 5(c)). However, in Fig. 5(d), too many dents were observed on the surface. Use of low viscous medium would have resulted in less drag force on the abrasive particles and as a result they could move freely in the slurry medium. Therefore, in low viscous medium the material removal by micro chipping dominates than the hammering action of abrasive particles which creates more number of shallow dents on the bottom surface of the machined microchannel which could have contributed to observed reduction in monitored surface roughness (most of the Ra values are ≤0.65 μm, Fig. 5(c–d)).
4.2. SEM analysis of surface roughness Morphology of the machined surface was also studied using scanning electron microscope (SEM). Fig. 5(b) shows SEM image of the 6
Precision Engineering xxx (xxxx) xxx–xxx
D. Sreehari, A.K. Sharma
Fig. 6. Variation of overcut with feed rate.
4.3. Effect of process parameters on overcut
4.4. Effect of process parameters on stray cut
As defined earlier, overcut was measured as the difference between the actual width and the nominal width of the microchannels. The overcut was observed more while machining with water based slurry (low viscous medium). As viscous drag force was less in low viscous fluids, the abrasives enter and leave the working gap with less resistance enlarging the width of the channel. This is evident from the plot between overcut and feed rate at different abrasive concentrations (Fig. 6(a–c)). Machining with high viscous slurry showed minimum overcut in the given conditions (Fig. 6(a)). As explained earlier, movement of the abrasive particles was constrained in the high viscous slurry resulting in the reduced cut. The monitored data confirms that the moderate viscosity transformer oil results in reduced overcut as it does not allow the abrasives to cause cutting while rebounding due to less energy available with them. Overcut was minimal at higher feed rates compared to lower workpiece feed rate irrespective of the abrasive concentration. The reason is that at a lower feed rate, the interaction time between three bodies – the workpiece, abrasive and the tool is more and hence three body abrasion wear took place in the machining zone [23]. Due to the 3–body abrasive wear, there is a possibility of an increase in the width of microchannel at the lower feed rate which is evident from the plots (Fig. 6(a–c)).
The stray cut data of the machined microchannels are presented in Fig. 7. It was evident from Fig. 7(a–c) that the machining with low viscous fluid (water) slurry results in more stray cut than the high viscous fluids (oils). As viscous drag forces are less in low viscous fluids, the resultant force FR2 will be more and will tend to move away from the normal which would result in the abrasives to move freely hitting the edges of microchannel while entering and leaving the machining zone. The stray cut is insignificant with palm oil based slurry compared to water based slurry machining. This is obvious that in high viscous fluid, abrasives tend to get accumulated in the machining zone; enters and leaves the machining zone with less FR2 due to which edge damage (stray cut) may reduce. Though the abrasive concentration of the slurry media was varied, the viscosity effect was observed significant on the stray cut in Fig. 7. Decreasing trend in the stray cut was evidenced at different abrasive concentrations with changing workpiece feed rate from level 1 (10 mm/min) to level 3 (20 mm/min). The reason could be that at lower feed rates the interaction time between tool, abrasive and work piece was more and hence, a higher number of abrasives hit the edges of the microchannel while entering and leaving the machining zone leading to observed higher stray cut. The above discussion, thus clearly indicates that the experimental observations follow the theoretical analyses provided in Section 3. The 7
Precision Engineering xxx (xxxx) xxx–xxx
D. Sreehari, A.K. Sharma
Fig. 7. Variation of stray cut with feed rate. Table 3 Theoretical model verses experimental observations. Theoretical model The resultant force FR2 (Eq. (9)) is responsible for the effect on the response parameters. FR2 = (Fh − Fd )2 + Fi 2 As μ increases, Fd will increase (Eq. (4)); thus FR2 tends to be less. It results in Higher surface roughness Lower overcut Lower stray cut
• • •
Experimental observations The experimental results showed similar variations in the response parameters with respect to μ as seen in the theoretical model. 1. Highest surface roughness obtained – palm oil (μpalm oil = 57.85 mPa.s > μtransformer oil = 13.44 mPa.s > μwater = 0.7972 mPa.s) 2. Highest overcut obtained – water (μwater = 0.7972 mPa.s < μtransformer oil = 13.44 mPa.s < μpalm oil = 57.85 mPa.s) 3. Highest stray cut obtained – water (μwater = 0.7972 mPa.s < μtransformer oil = 13.44 mPa.s < μpalm oil = 57.85 mPa.s)
parameters i.e., overcut and stray cut. Fig. 6(a–c), Fig. 7(a–c) show that at 15% abrasive concentration (Figs. 6(b) and 7(b)), there were large variations of the monitored stray cut and overcut in different slurry media at different feed rates. At 20% abrasive concentration, the movement of abrasives may get restricted due to high concentration, the variation observed in the stray cut and overcut was less and at 10% abrasive concentration, the movement of abrasives may not be restricted due to less concentration, but the variation observed was less due to presence of low number of abrasives in the machining zone. However, at 15% abrasive concentration, the restriction of movement
effects of the resultant force FR2 as derived in Eq. (9) can be explained in terms of the surface roughness, overcut and stray cut during machining. A summary of the discussion is presented in Table 3. 4.5. Feature plots In order to fabricate a precise microchannel, it is necessary to minimize both the imprecision measures – overcut and stray cut. In order to find the best possible input process parameters ‘Feature plots’ were drawn (Figs. 8 and 9) correlating the two main output process
8
Precision Engineering xxx (xxxx) xxx–xxx
D. Sreehari, A.K. Sharma
Fig. 8. Feature plot between stray cut and overcut at 15% abrasive concentration (images of stray cut at specific conditions are shown in the insets).
Fig. 9. Feature plot between stray cut and overcut at 20 mm/min feed rate (inset: images of stray cut at specific machining conditions).
relationship between overcut and stray cut. The plots shown in the clusters can be divided into four regions – R1, R2, R3 and R4 as illustrated in Figs. 8 and 9. The region R1 represents higher values of stray cut and overcut, which has to be avoided for precision machining of microchannels. The regions R2 & R3 represent lower values of stray cut at high and low values of overcut, respectively. The cluster 2 fall in R2 was considered if the stray cut is to be significant than overcut and cluster 3 which is in R3 was selected if the overcut and stray cut both be significantly small. This was evident from the microscopic images (Fig. 8 Inset). The data point in the region R1 was having an overcut 46.40 μm and stray cut 25.05% and the region R3 was having overcut 5.56 μm and stray cut 10.92%. The data points in the region R1 were corresponding to 15% abrasive concentration and water as a slurry medium at different feed rates. Therefore, machining
of abrasives was less compared to 20% and also the number of abrasives present were more compared to 10%. Hence, the variation in the stray cut and overcut was observed high at 15% abrasive concentration. Further, the monitored overcut and stray cut values were within the similar range in cases of 20% and 10% concentrations for machining with other oils barring water. Water being the least viscous among the three media used it allows relatively higher movements to the rebounding abrasives which were reflected in the higher variations in overcut and stray cut as observed in Figs. 6(b) and 7(b). Therefore, a feature plot (Fig. 8) was drawn to correlate the stray cut and overcut values at an abrasive concentration of 15% considering all three viscous fluids and feed rates. The feature plots indicate clear clusters based on the response parameter values. Data points in the cluster represent strong
9
Precision Engineering xxx (xxxx) xxx–xxx
D. Sreehari, A.K. Sharma
References
using water based slurry results in more edge damage compared to high viscous fluids. In the plots between the output process parameters with respect to feed rate, it was found that better results were obtained at higher feed rate (20 mm/min). Therefore, another feature plot (Fig. 9) was generated correlating the stray cut and the overcut corresponding to 20 mm/ min feed rate. Here, it was observed that a more concentrated cluster of data points was formed in the region R3. It explains that combination of input process parameters (slurry medium and abrasive concentration) at 20 mm/min feed rate gives minimum stray cut and minimum overcut. The findings can be evidenced by the microscopic images; the image shown in region R1 was having overcut 37.15 μm and stray cut 21.89% while the image in region R3 was showing an overcut of 13.51 μm while the corresponding stray cut was 5.33%. In practice, thus, it is implied that for precision applications where surface roughness may not be critical with respect to overcut and stray cut (where leakage is more critical than smooth flow), higher viscosity fluid be preferred to water. Further, the study also confirms that high viscosity fluids also result is better machining with higher feed rate which is indicative of higher productivity.
[1] Tuckerman DB, Pease RFW. High-performance heat sinking for VLSI. IEEE Electron Device Lett 1981;2:126–9. [2] Masuzawa T. State of the art of micromachining. CIRP Ann – Manuf Technol 2000;49:473–88. [3] Gentili E, Tabaglio L, Aggogeri F. Review on micromachining techniques. AMST’05 Adv Manuf Syst Technol 2005:387–96. [4] Qu W, Mala GM, Li D. Heat transfer for water flow in trapezoidal silicon microchannels. Int J Heat Mass Transf 2000;43:3925–36. [5] Wu HY, Cheng P. An experimental study of convective heat transfer in silicon microchannels with different surface conditions. Int J Heat Mass Transf 2003;46:2547–56. [6] Dwivedi VK, Gopal R, Ahmad S. Fabrication of very smooth walls and bottoms of silicon microchannels for heat dissipation of semiconductor devices. Microelectron J 2000;31:405–10. [7] Sun X-Q, Masuzawa T, Fujino M. Micro ultrasonic machining and self-aligned multilayer machining/assembly technologies for 3D micromachines. Proceedings of Ninth International Workshop on Micro Electromechanical Systems 1996. [8] Yu ZY, Rajurkar KP, Tandon A. Study of 3D micro-Ultrasonic machining. J Manuf Sci Eng 2004;126:727–32. [9] Jain V, Sharma AK, Kumar P. Developments and research issues in microultrasonic machining. ISRN Mech Eng 2011:1–15. 2011. [10] Jain V. Investigations on development of microchannels using micro ultrasonic machining Indian Institute of Technology Roorkee; 2012. Doctoral thesis. [11] Klopfstein MJ, Ghisleni R, Lucca DA, Brinksmeier E. Surface characteristics of micro-ultrasonically machined (1 0) silicon. Int J Mach Tools Manuf 2008;48:473–6. [12] Venkatesh G, Singh T, Sharma AK, Dvivedi A. Finishing of micro-channels using abrasive flow machining. Proceedings of the International Conference on Research and Innovations in Mechanical Engineering 2014. [13] Baek DK, Ko TJ, Yang SH. Enhancement of surface quality in ultrasonic machining of glass using a sacrificing coating. J Mater Process Tech 2013;213:553–9. [14] Choi JP, Jeon BH, Kim BH. Chemical-assisted ultrasonic machining of glass. J Mater Process Technol 2007;191:153–6. [15] Pei W, Yu Z, Li J, Ma C, Xu W, Wang X, et al. Influence of abrasive particle movement in micro USM. Procedia CIRP 2013;6:551–5. [16] Miller GE. Special theory of ultrasonic machining. J Appl Phys 1957;28:149–56. [17] Sundaram MM, Cherku S, Rajurkar KP. Micro ultrasonic machining using oil based abrasive slurry. Proceedings of International Manufacturing Science and Engineering Conference 2008. [18] Cheema MS. An ultrasonic micromachining approach to fabricate microchannels on glass Indian Institute of Technology Roorkee; 2015. Doctoral thesis. [19] Thoe TB, Aspinwall DK, Wise ML. Review on ultrasonic machining. Int J Mach Tools Manuf 1998;38:239–55. [20] Kothandaraman CP. Heat and mass transfer data book. New Age International; 2004. [21] White FM. Fluid mechanics. McGraw-Hill Book Company; 2011. [22] Yu Z, Hu X, Rajurkar KP. Influence of debris accumulation on material removal and surface roughness in micro ultrasonic machining of silicon. CIRP Ann – Manuf Technol 2006;55:201–4. [23] Cheema MS, Singh PK, Tyagi O, Dvivedi A, Sharma AK. Tool wear and form accuracy in ultrasonically machined microchannels. Measurement 2016;81:85–94.
5. Conclusion The study was aimed to fabricate precise and quality silicon microchannel by minimizing the undesirable response parameters such as surface roughness, overcut and stray cut. The following conclusions were drawn from the present investigation: a. Ultrasonic micromachining outputs are significantly influenced by the slurry related factors like medium viscosity. Higher viscosity offers constrained movement to the abrasives which tends to deteriorate surface finish while low viscosity allows the abrasives to stray around more freely causing an increase in stray cut and overcut. b. Higher workpiece feed yields better machined parts owing to shorter abrasive-tool-workpiece interaction. Surface finish and form accuracy improves at higher feed rate. c. Use of moderate viscosity medium for abrasive slurry offers effective machining ranges to choose from to achieve better machining control as evidenced by the feature plots. d. Feature plot representation is an effective technique to identify the nature of the interacting parameters and in adopting suitable machining strategy to reduce inaccuracies.
10