- Email: [email protected]
A study on separating of a silicon wafer with moving laser beam by using thermal stress cleaving technique
Journal of Materials Processing Technology 223 (2015) 252–261
Contents lists available at ScienceDirect
Journal of Materials Processing Technology journal homepage: www.elsevier.com/locate/jmatprotec
A study on separating of a silicon wafer with moving laser beam by using thermal stress cleaving technique Alias Mohd Saman a,d,∗ , Tatsuaki Furumoto b , Takashi Ueda c , Akira Hosokawa b a Division of Innovative Technology and Science, Graduate School of Natural Science and Technology, Kanazawa University, Kakuma-machi, Kanazawa, Ishikawa 920-1192, Japan b Institute of Science and Engineering, Kanazawa University, Japan c Department of Mechanical Science and Engineering, Nagoya University, Japan d Faculty of Mechanical Engineering, Universiti Teknologi MARA (UiTM), Malaysia
a r t i c l e
i n f o
Article history: Received 3 July 2014 Received in revised form 15 February 2015 Accepted 1 April 2015 Available online 13 April 2015 Keywords: Thermal stress cleaving Two-color pyrometer Acoustic emission (AE) Silicon wafer Cleaving temperature Finite-element method (FEM)
a b s t r a c t This study describes the characteristics of separating a silicon wafer with a moving Nd:YAG laser beam by using a thermal stress cleaving technique. The applied laser energy produces a thermal stress that causes the wafer to split along the irradiation path. The wafer separation is similar to crack extension. In this study, the micro-groove was prepared at the leading edge of the silicon wafer to facilitate the fracture. In order to study the thermal effect in the separating process, the temperature at the laser spot was measured by using a two-color pyrometer with an optical fiber, and the mechanism of crack propagation was observed by using an acoustic emission (AE) sensor. The influence of the micro-groove length and depth was also examined. Thermal stress distribution was calculated using the finite-element method (FEM) by considering the temperature from the experimental result. The result indicates that the wafer separation occurred in two stages, fracture initiation and intermittent crack propagation. A higher temperature resulted in faster fracture initiation and higher repetition of the crack propagation signal. The wave mark on the cleaved surface was consistent with the AE signal. The influences of laser power, temperature and the groove parameters to the fracture initiation, crack propagation and cleaved surface features are explained based on the experimental results, while the thermal stress condition is clarified with FEM analysis. © 2015 Elsevier B.V. All rights reserved.
1. Introduction Laser cutting is today the most common industrial application of laser material processing. Besides the need for additional investment in a tough industry, laser cutting happens to be able to cut faster with higher quality than the other processes (Steen and Mazumder, 2010). Laser cutting is becoming an upcoming potential process substitution into an established market. One of the latest technologies that has been evolving rapidly is thermal stress cleaving using a laser beam, which is used in separating the substrates from brittle material such as silicon wafer, ceramic and glass (Wang and Lin, 2007). Thermal stress cleaving is a process used for separating a brittle material by irradiating a laser beam onto a small area of the
∗ Corresponding author at: Division of Innovative Technology and Science, Graduate School of Natural Science and Technology, Kanazawa University, Kakuma-machi, Kanazawa, Ishikawa 920-1192, Japan. Tel.: +81 9094482207. E-mail address: [email protected] (A.M. Saman). http://dx.doi.org/10.1016/j.jmatprotec.2015.04.001 0924-0136/© 2015 Elsevier B.V. All rights reserved.
substrate. The laser energy absorbed generates the thermal gradient that creates a compressive stress at the laser spot area and tensile stress around it. A fracture on the wafer that is caused by circumferential tensile stress will propagate toward the laser spot. By introducing a flaw, such as a groove on the material, the crack is started from the groove tip, and the fracture can be controlled. By moving the laser beam, the crack will propagate along the laser path, causing separation of wafer. The material separation process is similar to crack extension. This process is preferred over the conventional mechanical dicing process due to its non-contact cutting process. The process does not need a coolant and production of chips is eliminated. Furthermore, this method is capable of producing a good surface finish. Garibotti (1963) first proposed the application of a laser in separating brittle material. A laser was used to scribe the grooves along the desired line before broken by ultrasonic energy. Lumley (1969) proposed an important laser cutting process that has high potential for separating brittle materials such as alumina ceramic and glass by using a controlled fracture technique. This procedure uses less laser power and enables high cutting speeds compared to
A.M. Saman et al. / Journal of Materials Processing Technology 223 (2015) 252–261
Fig. 1. Schematic illustration of silicon wafer specimen with the micro-groove and references axis.
150
Groove depth, µm
conventional laser cutting methods. Lambert et al. (1976) developed the laser cutting process to separate glass by using two lasers. The first laser beam was used to create a shallow groove via heating and vaporizing procedures. The second laser beam generated thermal stress at the groove tip to separate the material. Tsai and Liou (2003) investigated the fracture mechanism of laser cutting of alumina by using a single laser. A shallow groove was created by using an evaporative procedure, and heat produced from the laser beam generated a time-dependent thermal stress along the moving path. Tsai and Huang (2008) reported on diamond scribing and laser breaking for LCD glass substrate. A groove-crack was created along the cutting path using a scoring tool, before separated by applying a defocused CO2 laser beam. Several studies have been performed on the thermal cleaving process of a silicon wafer. Ueda et al. (2002) found that the temperature at the area irradiated with a pulsed laser is an important factor in controlling the propagation of the crack and in achieving high cleaving accuracy with low thermal damage. Yamada et al. (2006) recommended a cleaving process with a pulsed laser and the use of refrigerating chuck for cleaving with a continuous wave (CW) laser to reduce thermal damage during the cleaving processes. Takeda et al. (2009) reported that separation in the same direction of the wafer cleavage plane can be achieved by lower laser energy, furthermore improving the cleaving surface. Both researchers conducted their experiments on thermal stress cleaving of a silicon wafer with the preparation of the initial crack by using the Vickers indenter impressions. However, the impressions created damage to the indentation area and a lateral crack in the direction perpendicular to the cleaving path. Ishikawa et al. (2012) introduced the use of a micro-groove to facilitate the fracture initiation during the thermal stress cleaving process. The unnecessary crack and damage were eliminated during the initial crack preparation, subsequently improving the quality of the material specimen. Previous studies on the thermal stress cleaving process of a silicon wafer were focused on the cleaving temperature, thermal damage and surface finish of the material’s separated surface. However, the mechanism of the thermal stress cleaving process of a silicon wafer had not been analyzed in detail, and the relation between laser condition and cleaving performance is not fully understood. In this paper, the separating characteristics of a silicon wafer with a moving laser beam by using the thermal stress cleaving process is investigated experimentally and computationally. The micro-groove was prepared at the leading edge of the silicon wafer to facilitate the fracture. The effect of laser energy on the cleaving mechanism was analyzed, and the influence of the groove parameters such as groove length and depth were also studied. The temperature of the moving laser spot was measured by using a two-color pyrometer with an optical fiber, and the AE signal was assessed to observe the mechanism of propagations. The cleaved surface was observed using scanning electron microscope (SEM). Further, the influences of laser power, temperature and the groove parameters to the fracture initiation, crack propagation and cleaved surface features are explained based on the experimental results. In order to explain the three-dimensional thermal stress distribution, the finite-element method (FEM) software, ANSYS, is applied by considering the temperature ascertained from the experimental result.
253
100 Laser : Nd:YAG (SHG) λ=532 nm P=280 mW f = 10 Hz
50 0 0
300
600
900
1200
Laser pulse , N Fig. 2. Relationship between laser pulse number and groove depth.
Fig. 3. Images of micro-groove with dimensions.
2. Experimental method
schematic illustration of the silicon wafer specimen with a size of 20 mm × 10 mm × 0.5 mm is shown in Fig. 1. The micro-groove was created initially by focusing the pulsed laser beam through the micro-lens at the edge of the silicon wafer (Ishikawa et al., 2012). A second harmonic generation (SHG) Nd:YAG laser with a wavelength of 532 nm, frequency of 10 Hz and pulse width of 5 ns was used in the specimen preparation. The laser energy was focused onto the specimen resulting in material heating and evaporating, thus forming the groove. By using this technique, the groove size can be controlled, and the area of damage can be minimized. In contrast with the initial crack created by the indenter, the unnecessary crack was eliminated; consequently improved the quality of the separated surface. The depth of the groove was adjusted by varying the laser pulse number while the groove length was controlled by focusing the laser beam only onto the required exposed area. Fig. 2 shows the relationship between the groove depth and laser pulse number. The images of the pre-prepared micro-groove are shown in Fig. 3.
2.1. Silicon wafer specimen
2.2. Temperature measurement
The specimen used in the experiments was a silicon wafer of 1 0 0 crystal orientation and with a thickness of 0.5 mm. The
The temperature gradient produced from the laser beam generates thermal stress in the material substrate. Therefore, it is
254
A.M. Saman et al. / Journal of Materials Processing Technology 223 (2015) 252–261
Fig. 4. Schematic illustration of experimental set-up.
important to know the temperature during the cleaving process. It is difficult to measure the temperature of the laser spot during the thermal cleaving process because the laser beam is moving, and the size of the laser beam used is very small. Ignatiev et al. (1996) reported on the measurement of pulsed laser action on metallic material using two silicon photo diodes. Ueda et al. (1997) monitored the surface temperature of ceramic materials in real time during laser beam irradiation using an infrared radiation pyrometer with a fused fiber coupler. Furumoto et al. (2009) used a pyrometer composed of an optical fiber and InAs and InSb infrared detectors to observe the temperature during laser consolidation of metal powder. In this paper, the temperature during laser irradiation was measured using a two-color pyrometer with different infrared detectors, namely InSb and MCT; and NSEG chalcogenide optical fiber. The fundamental structure of a two-color pyrometer is shown in Fig. 4. The infrared energy radiated from the object was captured by a chalcogenide optical fiber and transmitted to a two-color pyrometer consisting of InSb and MCT detectors. The InSb detector was mounted over an MCT detector in a sandwich configuration. The InSb detector responds to radiation from 3.1 to 5.5 m of wavelength and transmits waves larger than 5.5 m, and the MCT detector responds to radiation from 5.5 to 11 m. The infrared energy was converted to an amplified electric signal. By calculating the ratio of output voltages from these two detectors, the
Temperature, oC
600
temperature of an object can be obtained by using the calibration curve. Fig. 5 shows the calibration curve of the pyrometer, which was obtained experimentally by observing the incidence surface of the silicon wafer heated to the known constant temperature electrically. The frequency characteristic of the pyrometer has a flat response to about 100 kHz, which was sufficient for the experiment conducted in this paper (Ueda et al., 1998). An NSEG chalcogenide optical fiber with a core diameter of 380 m and acceptance angle of 60◦ was used to transmit the infrared rays of a wavelength from 3 to 11 m. Therefore, the fiber functions as an optical filter for filtering the laser beam of the YAG laser beam of 1.06 m reflected from the surface of the wafer. 2.3. Acoustic emission (AE) signal measurement The AE signal was measured to monitor the characteristics of fracture initiation and crack propagation during laser irradiation along the cleaving path. An AE sensor was fixed on the wafer’s surface at a distance of 5 mm from the micro-groove as shown in Fig. 4. Details of the AE signal measurement are summarized in Table 1. 2.4. Experimental procedure The experimental arrangement of the laser cleaving is illustrated in Fig. 4. The Nd:YAG laser with the wavelength of 1.06 m was used as a heat source, and the elliptical beam spot with a major axis length of 0.25 mm and minor axis length of 0.19 mm was defocused on the wafer’s surface. A chalcogenide optical fiber was set at a distance of 3 mm from the laser spot and at an angle of 45◦ from the optical axis. The center of the laser spot was always on the center
400 Table 1 Measuring system of AE signal.
Experiment
200
Units
Theoretical curve 0 0
1
2
3
4
5
Output ratio (InSb/MCT) Fig. 5. Calibration curve.
6
Acoustic emission (AE) sensor Size Frequency band (−10 dB)
Ø5 × 3.2 1–4
mm MHz
PreAmplifier Gain Frequency band (−10 dB)
40 0.1–20
dB MHz
A.M. Saman et al. / Journal of Materials Processing Technology 223 (2015) 252–261
255
Table 2 Experimental condition. Laser
Nd:YAG
Units
Wavelength, Irradiation mode Laser power, P Laser beam diameter, D Moving speed, v
1064 Continuous wave 50–70 0.25 × 0.19 2
nm
Work piece
Silicon wafer
Crystal orientation Size
1 0 0 10 × 20 × 0.5
mm3
Pyrometer Optical fiber Detector Fiber core diameter, Ø
Two-color NSEG InSb/MCT 380
m
W mm2 mm/s
axis of the fiber. In the experiments, the specimen was set on a moving table (X–Y stage). The work table moved at a constant speed of 2 mm/s, and the laser beam relatively scanned the work material at the same speed. The laser beam and optical fiber remained in a static position while the AE sensor was moving together with the work material. Before the X–Y stage was moving, the laser beam was set at a position of 2 mm from the silicon wafer’s leading edge. After the X–Y stage has been moving for one second, the center of the laser beam will reach the wafer’s leading edge. Therefore, the voltage output of the pyrometer detectors and the AE sensor in relation to the irradiation time can be determined. Consequently, the temperature distribution and crack occurrence during laser irradiation along the laser beam path on the work material can be verified. The output voltages from the pyrometer detectors and AE sensor were recorded in a digital oscilloscope. The laser power used for this experiment varied between 50 and 70 W, while the groove length and the depth were maintained at 500 and 80 m accordingly. A summary of the experimental conditions is shown in Table 2. After the specimen was separated into two pieces, the cleaving surface of the specimen was analyzed by using a scanning electron microscope (SEM).
3. Experimental result and discussion 3.1. Output signal from pyrometer and AE sensor Fig. 6 shows the typical output profiles of the pyrometer and AE sensor. Fig. 6(a) and (b) represents the output voltages of the two detectors, while Fig. 6(c) shows the temperature history of the laser spot along the irradiation path. Fig. 6(d) shows the signal from the AE sensor. The laser beam reached the wafer’s leading edge after moving for one second. Therefore, the distributions of InSb, MCT and AE signals per unit time along the cleaving path on the work piece can be recognized. The signal from the pyrometer detectors shown was a result of chopped signals that were produced by using a chopper in the pyrometer set-up. The InSb and MCT signals were increased rapidly after the laser beam reached the wafer’s leading edge, later declined proportionally until they attained a stable voltage output at the center area of the wafer. The laser beam was emitted on the wafer surface with Gaussian’s intensity distribution (Ueda et al., 2002). Hence, the pyrometer detectors’ signals increased earlier, before the laser beam center reached the wafer’s leading edge. The signal size increases smoothly when the laser beam was approaching the wafer’s trailing edge and declines quickly before the laser beam passes the wafer’s trailing edge. AE signals in Fig. 6(d) show the characteristics of the thermal cleaving process during laser irradiation. When the crack occurs, the AE sensor will trigger the signals. A large signal occurred a few
Fig. 6. Output wave of pyrometer and AE sensor.
milliseconds after the laser beam reached the wafer’s leading edge, continuing with the repeating signals. 3.2. Temperature along the irradiation path The variation of temperature was obtained by the ratio of output signals from the InSb and MCT signals. The temperature history during laser irradiation in Fig. 6(c) shows a higher reading at a position near to edge of the silicon wafer. The time it takes for the center of the laser beam to reach the wafer’s leading edge is considered as t = 0 s. A higher temperature of 385 ◦ C was recorded approximately at t = 0.27 s, later reduced to a constant temperature of approximately 205 ◦ C, then rising to 340 ◦ C approximately at t = 0.27 s before the laser beam reached the wafer’s trailing edge at t = 5 s. The results indicate that the laser spot was maintained at a constant lower temperature at the center of the laser path. Silicon wafer has a high thermal conductivity of 156 W/m K (Liu et al., 2011; Yamada et al., 2006). Hence, the energy produced by laser irradiation was circumferentially diffused when the laser beam was at the center area of the wafer compared to when the laser beam was closest to the edge of the wafer. The variation of maximum temperatures at the leading edge, trailing edge and average temperature at the center of the specimen with the laser power are shown in Fig. 7. The temperature at all three positions is increased with the increase of the laser power. From the results, it is clear that the temperature of the moving laser spot was dependent on the energy supplied by the laser beam. Higher laser power resulted in an increasing of temperature. 3.3. Crack occurrence Voltage output signals from the AE sensor (Fig. 6(d)) demonstrates that the crack occurrences appeared in two stages, namely fracture initiation and intermittent crack propagation. AE signals
256
A.M. Saman et al. / Journal of Materials Processing Technology 223 (2015) 252–261
Fig. 7. Influence of laser power on temperature.
Fig. 8. Influence of laser power on crack initiation time.
were produced when a rapid release of energy from localized sources within the material occurred. Energy release happened as a result of crack extension due to thermal stress generated by the laser beam. The first signal of the AE wave shows that fracture initiation occurred with large signal size. Subsequently, a series of smaller AE signals indicates that crack propagation occurred intermittently. A large signal indicates that a higher energy was released during fracture initiation. It was shown that the energy required to start the fracture is high, in contrast with the energy during the crack propagation. This was because the fracture initiation started at the groove tip with a blunt notch (Fig. 3(a)) and a sharp crack tip was produced after the first fracture occurred. Greater stress concentration developed at the sharper crack tip in contrast to the blunt groove. Therefore, larger energy was required to increase the local stress at the blunt groove to reach the fracture strength of the material.
Fig. 10. Relationship between laser power and fracture initiation distance.
3.3.1. Fracture initiation The laser beam energy generates a thermal gradient, which creates a compressive thermal stress at the laser beam area and tensile stress area around it. When the laser beam reaches the leading edge, compressive stress generated at this point due to the heat induced from the laser beam. When the laser beam moves forward, thermal stress at the leading edge of the silicon wafer is transformed into tensile stress and concentrated at the groove tip (x = 0, y = 0, z = −80 m). When the tensile stress reaches the fracture strength of the material, the fracture will be initiated. Fig. 8 shows the influence of laser power on the fracture initiation time. The specimen was prepared with a groove length and depth of approximately 500 and 80 m respectively; and the laser beam scan speed was set constantly at 2 mm/s. It was found that the fracture initiation time decreased with an increase in laser power. The average fracture initiation time for laser power of 50 W was attained approximately at 0.42 s compared to 0.27 s for laser power of 70 W. The stress at which a specimen fails via fracture was achieved faster with higher laser energy. As a result, earlier fracture initiation was achieved at higher laser power. Fig. 9 shows the SEM images of smooth surfaces on the separated plane as observed at the leading edge area of the silicon wafer. A smooth, flat surface was produced for a width of a few micrometers from the wafer’s edge followed by a periodic wave marking. The variation of the extent of the smooth surface in relation to laser power is shown in Fig. 10. It was found that the extent of the smooth surface increased with the decrease in laser power. The distances of the smooth surfaces were obtained approximately 185 m at a laser power of 70 W, in comparison with the 295 and 420 m for the laser power of 60 W and 50 W respectively. The result shows that fracture initiation occurs at longer travel distance when a lower laser power was used. As a result, a longer smooth cleaving surface was produced. However, if the laser power used is too low, the fracture might not occur due to insufficient
Fig. 9. SEM images at the starting section of cleaving surface.
A.M. Saman et al. / Journal of Materials Processing Technology 223 (2015) 252–261
257
Fig. 14. Relationship between cleaving mark interval and laser power. Fig. 11. AE signals.
with the relaxation of local stress. The crack will develop again when the sufficient stress was attained. The stress recovery rate was increased with the increased of laser power. Therefore, crack signals appeared rapidly. Fig. 13 shows the SEM images of the cleaving marks pattern on the center section of the separated surface. The wave marks were developed due to crack advancement occurrences during the intermittent crack propagation phenomena. The variation between the mark interval and the laser power is given in Fig. 14. The interval between the marks was found to be approximately 32 m for the laser power of 50 W in comparison with 27 and 22 m for 60 W and 70 W respectively. The increase of laser power caused higher repetition of the cleaving mark. The mark interval was comparable to the AE signal results in Fig. 12. 3.4. The influence of groove length and depth Fig. 12. The relationship between laser power and AE signal interval.
stress concentration at the groove tip. In contrast, if the laser power used is high, thermal damage might occur, as shown in Fig. 9(a). 3.3.2. Intermittent crack propagation Fig. 11 shows the typical output signals of an AE sensor measured at the middle of the laser path. The result shows that a repetitive AE signal appeared with a starting and stopping phenomenon at regular intervals. This occurrence revealed that the crack was extended and relaxed frequently at a perpetual distance, as the laser beam moved along the laser path. The variation of laser power with the AE signal interval is shown in Fig. 12. The increase of laser power produced shorter intervals between the crack propagation signals. The intervals were found to be constant and proportionate to the laser power. For instance, the interval in signals found for a laser power of P = 50 W was about 16.2 ms or 32.4 m in distance while the signal interval for P = 70 W was about 10.5 ms or 21 m in distance. The crack in the material grows
To understand the influence of groove parameters, such as groove length and groove depth on cleaving performance, the experiments were performed by irradiating a constant laser power of 70 W and scan speed of 2 mm/s. In the case of varied groove lengths, the depth was fixed at 80 m and in the case of varied groove depths, the length was fixed at 500 m. The AE signal was used to monitor the fracture initiation following the procedure stated in the previous section. Table 3 shows the summary of experimental conditions for varied groove length and depth accordingly. Fig. 15 shows the effects of groove length on fracture initiation time, whereas the variation of groove depth with fracture initiation time is given in Fig. 16. The result shows that the time for fracture initiation was decreased with the enlarging of groove length. For instance, the average fracture initiation time obtained for a groove length of 100 m was about 450 ms compared with 270 ms for a 500 m groove length. Furthermore, fracture initiation time was also found to be faster for deeper groove depth. The average fracture initiation time for a groove depth of 50 m was reached
Fig. 13. SEM images at the center section of cleaving surface.
258
A.M. Saman et al. / Journal of Materials Processing Technology 223 (2015) 252–261
Table 3 Experiment conditions of varied groove length and groove depth. Laser Wavelength, Irradiation mode Laser power, P Laser beam diameter, D Moving speed, v Experiment of varies groove length Groove length Groove depth Experiment of varies groove depth Groove length Groove depth
Nd:YAG
Units
1064 Continuous wave 70 0.25 × 0.19 2
nm
100–500 80
m m
W mm2 mm/s Fig. 17. The cross-section of the specimen with a coordinate system.
500 20–150
m m
Fig. 15. Effect of groove length on fracture initiation time.
analysis was performed to investigate the thermal stress during fracture initiation. Finite-element software, ANSYS, was used to analyze the behavior of thermal stress cleaving during the laser irradiation by considering the finite boundary effect. The temperature and thermal stress distribution around the laser spot diverged along the thickness direction of the substrate. Therefore, the finite-element model should be considered as a three-dimensional problem. The specimen size used in this study was 20 mm × 10 mm × 0.5 mm. The half-part of the specimen with a coordinate system is shown in Fig. 17. The following basic assumptions for the thermo-mechanical analysis were considered. The material properties such as the heattransfer properties and mechanical properties were assumed to be constant as the temperature range was not high. The stress–strain relationship of the silicon wafer was assumed to be perfectly elastic. Convection conditions were assumed to exist on all boundaries except at the surface of x–z plane (y = 0). This surface was considered to be in adiabatic condition. The energy absorbed by the silicon wafer during laser irradiation varies along the material thickness. The absorption characteristic is represented by Beer Lambert’s equation below: ˇ=
Fig. 16. Effect of groove depth on fracture initiation time.
approximately at 340 ms compared to 270 ms for a groove depth of 130 m. From the finite-element analysis results, it is demonstrated that higher tensile stress concentration was resulted at a longer groove length (Fig. 21) and at a deeper groove depth (Fig. 23). The results explained why fracture initiation time was reached earlier at a longer groove length and a deeper groove depth. It does become apparent that the groove has become the weakness point in the material. Local stress concentrated at the groove tip could rise to several times that of the applied stress, and the stress level is increased with the increase of groove size.
− ln T x
(1)
where T is spectral transmittance (T = 23% as measured experimentally), ˇ is the absorption coefficient and x is the thickness of the wafer. The boundary condition of the stress analysis is traction-free at all surfaces, and the x–z surface is set as a fixed plane which has zero displacement in the y-direction. During the laser irradiation, a moving laser beam with the power output of 70 W and laser beam spot size of 0.25 mm × 0.19 mm was set on the wafer surface above the focal plane. When the center of the laser beam reached the wafer leading edge (i.e. x = y = z = 0), the time was considered as t = 0 s, later moving toward the x-direction at a constant speed of v = 2 mm/s. The initial temperature of the silicon wafer substrate and the surrounding temperature were set to 20 ◦ C. The analysis was carried out into two phases. The transient temperature analysis was calculated when the moving laser beam was irradiated onto the silicon wafer. Then, the steady state stress analysis was performed using the temperature distribution results at a specific point in time. The temperature and stress distribution adjacent to the laser spot vary along the thickness direction during the cleaving process; therefore, the calculation was considered as a three-dimensional problem. The summary of analysis condition and properties of the silicon wafer are shown in Table 4. 4.1. Temperature and thermal stress distribution result
4. Finite-element analysis The computational analysis was performed to understand the actual phenomenon of fracture during the thermal cleaving process. The temperature and AE signal of crack behavior have been monitored during the experimental work. However, the mechanisms of fracture have not been clarified. In this section, the
The transient temperature distributions were calculated at points in time from 0 s to 0.495 s. The temperature distribution at various times is shown in Fig. 18. The temperature was measured on the surface of the material along the x-axis (y = 0, z = 0). The maximum temperature reached at the times of 0 s, 0.275 s and 0.495 s was 230 ◦ C, 380 ◦ C and 320 ◦ C accordingly. The calculated
A.M. Saman et al. / Journal of Materials Processing Technology 223 (2015) 252–261
259
Table 4 Analysis condition and properties of silicon wafer. Units Analysis conditions Workpiece size Laser beam diameter (oval) Heat transfer coefficient, h Initial temperature, To Laser beam scan speed, v Laser power, P
20 × 10 × 0.5 0.25 × 0.19 8 20 2 70
mm3 mm/s W m−2 K−1 C mm/s W
Properties Thermal conductivity, K Specific heat, C Thermal expansion coef., ˛ Young modulus, E Poisson ratio, Density,
156 761 2.62 × 10−6 169 0.262 2340
W/m K J/kg K K−1 GPa kg/m3
Fig. 20. Thermal stress yy distribution on x–z plane (y = 0).
Fig. 18. The temperature distribution of moving laser.
transformed from compression to tensile stress in a short time as the laser beam moved forward in the x-axis direction. As the laser beam moved further, large tensile stress was accumulated at the groove tip (x = 0). This large tensile stress yy may induce a fracture when the stress reaches the fracture strength of the material. The fracture could be induced along the thickness direction (z-axis) and extended through the material. The tensile stress occurred after the laser beam was passing through; therefore, the fracture will follow the laser spot with a short lagging distance. 4.2. The influence of groove length
Fig. 19. Thermal stress yy , distribution at various times (z = −80 m).
temperature results were consistent with the temperature from the experimental result (Fig. 7). The thermal stress yy distribution, along the cleaving plane, at the time of 0.01 s, 0.055 s, 0.165 s, 0.275 s, 0.385 s and 0.495 s is shown in Fig. 19. When the laser beam was at the position of x = 0, and the time of t = 0 s, the stress at the groove tip (z = −80 m) on the edge was a compressive stress. The thermal stress then
According to Fig. 15, the experimental result indicated that in the case of a wafer with a groove length of 500 m, the fracture initiation time was attained at approximately t = 0.275 s. At this time, the laser beam reached x = 0.55 mm. For the purpose of assessment, the analyses of stress distribution for various cases of groove length were performed at the same time of 0.275 s. Fig. 20 shows the thermal stress yy distribution on the x–z plane and y = 0 for the cases of a groove length of 100, 300 and 500 m at the time of t = 0.275 s. The thermal stress yy distribution in the x-axis direction for all cases was almost similar, except at the area under the groove. Compressive stress was centered at the laser spot area and a tensile stress resulted at the area near to the edge of the material, where x = 0 mm. However, the maximum tensile stress was located at the groove tip (z = −80 m) with more than 60 MPa value of tensile stress concentrated on the groove tip at the edge of the material (x = 0 mm). The figures on the left of Fig. 20(a)–(c) illustrate the thermal stress distribution, yy , along the z-axis (x = 0, y = 0). The results indicate that the thermal stress, yy , was at its maximum at the
260
A.M. Saman et al. / Journal of Materials Processing Technology 223 (2015) 252–261
Fig. 23. Relationship between thermal stress yy and groove depth. Fig. 21. Relationship between thermal stress yy and groove length.
that the thermal stress, yy , is distributed along the x-direction. Compressive stress is focused under the laser beam position (x = 0.55 mm) and tensile stress resulted at the edge of the material (x = 0 mm). Large tensile stress of more than 60 MPa was concentrated on the groove tip (z = 20 m, x = 0 mm). Moreover, this phenomenon was repeated in the case of a groove depth of 50 and 80 m as illustrated in Fig. 22(b) and (c) accordingly. The maximum tensile stress was concentrated on the groove tip for all cases. The figures on the left of Fig. 22(a)–(c) illustrate the thermal stress distribution, yy , along the z-axis (x = 0, y = 0). The results indicate that the thermal stress, yy , is highest at the groove tip and reduces rapidly to the constant tensile stress value of approximately 30 MPa along the −z direction. The relationship between the maximum thermal stress at the groove tip and groove depth is shown in Fig. 23. A deeper groove resulted in higher thermal stress concentration. 5. Separating mechanism of silicon wafer
Fig. 22. Thermal stress yy distribution on x–z plane (y = 0).
groove tip (z = −80 m), and reduces rapidly when it moves along the −z direction before reaching approximately 30 MPa of constant tensile stress value. The relationship between the maximum thermal stress and groove length is shown in Fig. 21. It was found that the proportion of the tensile stress concentrated on the groove tip was dependent on the groove length. The longer groove length resulted in a higher concentration of thermal stress, yy . 4.3. The influence of groove depth Fig. 22 shows a comparison of thermal stress, yy , distribution on the x–z plane (y = 0) of the wafer with a groove depth of 20, 50, and 80 m at the time of 0.275 s. In Fig. 22(a), it is illustrated
Based on the experimental and computational analysis results, the mechanism of separating a silicon wafer by applying a laser beam using the thermal stress cleaving process can be understood clearly. Fig. 24 illustrates the separating mechanism of the silicon wafer with a pre-prepared micro-groove, by utilizing the moving laser beam. The cleaving process can be divided into two stages, namely, fracture initiation and intermittent crack propagation. As shown in Fig. 24(a), when the laser reaches the edge of the wafer, the fracture does not occur yet. This is due to the fact that the laser beam induces the heat at the laser spot area; hence, generated a compressive stress at this point. After the laser beam moves further, the compressive stress transforms into tensile stress. The tensile stress concentrates at the groove tip, and it increases with the increase in temperature at the laser spot and the distance relative to the leading edge of the wafer. When the intense stress at the groove tip reached the material fracture strength, a fracture initiated through the material thickness and extended toward the laser spot in a very short time. As illustrated in Fig. 24(b), the fracture stopped before the laser beam spot and created a new crack tip. Continuous heat supplied by the laser beam caused the tensile stress to increase on the newly created crack tip, and the crack propagated and stopped again before the laser spot. As shown in Fig. 24(c), the cleaving mark was created as the crack stopped. This process was replicated again and again until the laser beam reached the end of the wafer. It is demonstrated that fracture initiation and intermittent crack propagation occurrence are influenced by the heat supplied from the laser beam during the irradiation process. An increase in laser power improved the fracture initiation time and increased crack propagation recurrence. However, the laser power should be controlled to avoid any thermal damage to the silicon wafer.
A.M. Saman et al. / Journal of Materials Processing Technology 223 (2015) 252–261
261
Fig. 24. Thermal stress cleaving mechanism of silicon wafer with microgroove.
6. Conclusion The separating characteristics of a silicon wafer with a moving Nd:YAG laser by using the thermal stress cleaving technique were successfully monitored and analyzed by the utilization of a twocolor pyrometer with an optical fiber and AE sensor. Computational analysis by the finite-element method described the phenomenon of stress distribution during the process. Based on the study, the mechanism of the thermal stress cleaving of a silicon wafer with a laser beam was explained. The separating process of the silicon wafer occurred in two stages. The first stage is the fracture initiation, where the fracture occurs due to tensile stress concentrated at the groove tip on the edge of specimen. The second is intermittent crack propagation which occurs at constant intervals in time. The temperature at the laser irradiation area was related to the separating characteristics of a silicon wafer by the cleaving technique. The temperature influenced the cleaving achievement and cleaving surface quality. Fracture initiation and crack propagation occurrences produced the cleaving mark waves on the separating surface. The cleaving mark wave was consistent with the AE signals. The finite-element analysis explained the stress distribution during laser irradiation. The results correspond well with the experimental results. References Furumoto, T., Ueda, T., Kobayashi, N., Yassin, A., Hosokawa, A., Abe, S., 2009. Study on laser consolidation of metal powder with Yb:fiber laser—evaluation of line consolidation structure. J. Mater. Process. Technol. 209 (18–19), 5973–5980.
Garibotti, D.J., 1963. Dicing of micro-semiconductors. U.S. Patent No. 3,112,850. Ignatiev, M.B., Smurov, Yu, I., Flamant, G., Senchenko, V.N., 1996. Surface temperature measurement during pulsed laser action on metallic and ceramic materials. Appl. Surf. Sci. 96–98, 505–512. Ishikawa, T., Ueda, T., Furumoto, T., Hosokawa, A., Tanaka, R., 2012. Pre-groove development for laser cleaving of brittle materials by using micro-lens. J. Adv. Mech. Des. Syst. Manuf. 6 (6), 841–848. Lambert, E., Lambert, J.L., Longueville, B.D., 1976. Severing of a glass or vitrocrystalline bodies. U.S. Patent No. 3,935,419. Liu, J., Lu, J., Ni Xi Dai, G., Zhang, L., Chen, Y., 2011. Numerical study on thermal stress cutting of silicon wafer using two-point pulsed laser. Opt. Appl. 41 (1), 247–255. Lumley, R.M., 1969. Controlled separation of brittle materials using a laser. Am. Ceram. Soc. Bull. 48 (1), 850–854. Steen, W.M., Mazumder, J., 2010. Laser Material Processing, fourth ed. SpringerVerlag, London, pp. 131–193. Takeda, R., Ueda, T., Furumoto, T., Hosokawa, A., Tanaka, R., 2009. Study on cleaving mechanism of silicon wafer by laser beam irradiation. In: Proc. Leading Edge Manufacturing in 21st Century (5), The Japan Society of Mechanical Engineers, pp. 589–592. Tsai, C.H., Huang, B.W., 2008. Diamond scribing and laser breaking for LCD glass substrates. J. Mater. Process. Technol. 198 (1–3), 350–358. Tsai, C.H., Liou, C.S., 2003. Fracture mechanism of laser cutting with controlled fracture. J. Manuf. Sci. Eng. ASME 125 (3), 519–528. Ueda, T., Yamada, K., Nakayama, K., 1997. Temperature of work materials irradiated with CO2 laser. CIRP Ann. Manuf. Technol. 46 (1), 117–122. Ueda, T., Sato, M., Nakayama, K., 1998. The temperature of a single crystal diamond tool in turning. CIRP Ann. Manuf. Technol. 47 (1), 2–5. Ueda, T., Yamada, K., Oiso, K., Hosokawa, A., 2002. Thermal stress cleaving of brittle materials by laser beam. CIRP Ann. Manuf. Technol. 51 (1), 149–152. Wang, Y.Z., Lin, J., 2007. Characterization of the laser cleaving on glass sheets with a line-shape laser beam. Opt. Laser Technol. 39 (5), 892–899. Yamada, K., Ueda, T., Hosokawa, A., Yamane, Y., Sekiya, K., 2006. Thermal damage of silicon wafer in thermal cleaving process with pulsed laser and CW laser. In: Proc. SPIE Laser-based Micro Packaging, vol. 6107, pp. 61070H–61070H–10.
Contents lists available at ScienceDirect
Journal of Materials Processing Technology journal homepage: www.elsevier.com/locate/jmatprotec
A study on separating of a silicon wafer with moving laser beam by using thermal stress cleaving technique Alias Mohd Saman a,d,∗ , Tatsuaki Furumoto b , Takashi Ueda c , Akira Hosokawa b a Division of Innovative Technology and Science, Graduate School of Natural Science and Technology, Kanazawa University, Kakuma-machi, Kanazawa, Ishikawa 920-1192, Japan b Institute of Science and Engineering, Kanazawa University, Japan c Department of Mechanical Science and Engineering, Nagoya University, Japan d Faculty of Mechanical Engineering, Universiti Teknologi MARA (UiTM), Malaysia
a r t i c l e
i n f o
Article history: Received 3 July 2014 Received in revised form 15 February 2015 Accepted 1 April 2015 Available online 13 April 2015 Keywords: Thermal stress cleaving Two-color pyrometer Acoustic emission (AE) Silicon wafer Cleaving temperature Finite-element method (FEM)
a b s t r a c t This study describes the characteristics of separating a silicon wafer with a moving Nd:YAG laser beam by using a thermal stress cleaving technique. The applied laser energy produces a thermal stress that causes the wafer to split along the irradiation path. The wafer separation is similar to crack extension. In this study, the micro-groove was prepared at the leading edge of the silicon wafer to facilitate the fracture. In order to study the thermal effect in the separating process, the temperature at the laser spot was measured by using a two-color pyrometer with an optical fiber, and the mechanism of crack propagation was observed by using an acoustic emission (AE) sensor. The influence of the micro-groove length and depth was also examined. Thermal stress distribution was calculated using the finite-element method (FEM) by considering the temperature from the experimental result. The result indicates that the wafer separation occurred in two stages, fracture initiation and intermittent crack propagation. A higher temperature resulted in faster fracture initiation and higher repetition of the crack propagation signal. The wave mark on the cleaved surface was consistent with the AE signal. The influences of laser power, temperature and the groove parameters to the fracture initiation, crack propagation and cleaved surface features are explained based on the experimental results, while the thermal stress condition is clarified with FEM analysis. © 2015 Elsevier B.V. All rights reserved.
1. Introduction Laser cutting is today the most common industrial application of laser material processing. Besides the need for additional investment in a tough industry, laser cutting happens to be able to cut faster with higher quality than the other processes (Steen and Mazumder, 2010). Laser cutting is becoming an upcoming potential process substitution into an established market. One of the latest technologies that has been evolving rapidly is thermal stress cleaving using a laser beam, which is used in separating the substrates from brittle material such as silicon wafer, ceramic and glass (Wang and Lin, 2007). Thermal stress cleaving is a process used for separating a brittle material by irradiating a laser beam onto a small area of the
∗ Corresponding author at: Division of Innovative Technology and Science, Graduate School of Natural Science and Technology, Kanazawa University, Kakuma-machi, Kanazawa, Ishikawa 920-1192, Japan. Tel.: +81 9094482207. E-mail address: [email protected] (A.M. Saman). http://dx.doi.org/10.1016/j.jmatprotec.2015.04.001 0924-0136/© 2015 Elsevier B.V. All rights reserved.
substrate. The laser energy absorbed generates the thermal gradient that creates a compressive stress at the laser spot area and tensile stress around it. A fracture on the wafer that is caused by circumferential tensile stress will propagate toward the laser spot. By introducing a flaw, such as a groove on the material, the crack is started from the groove tip, and the fracture can be controlled. By moving the laser beam, the crack will propagate along the laser path, causing separation of wafer. The material separation process is similar to crack extension. This process is preferred over the conventional mechanical dicing process due to its non-contact cutting process. The process does not need a coolant and production of chips is eliminated. Furthermore, this method is capable of producing a good surface finish. Garibotti (1963) first proposed the application of a laser in separating brittle material. A laser was used to scribe the grooves along the desired line before broken by ultrasonic energy. Lumley (1969) proposed an important laser cutting process that has high potential for separating brittle materials such as alumina ceramic and glass by using a controlled fracture technique. This procedure uses less laser power and enables high cutting speeds compared to
A.M. Saman et al. / Journal of Materials Processing Technology 223 (2015) 252–261
Fig. 1. Schematic illustration of silicon wafer specimen with the micro-groove and references axis.
150
Groove depth, µm
conventional laser cutting methods. Lambert et al. (1976) developed the laser cutting process to separate glass by using two lasers. The first laser beam was used to create a shallow groove via heating and vaporizing procedures. The second laser beam generated thermal stress at the groove tip to separate the material. Tsai and Liou (2003) investigated the fracture mechanism of laser cutting of alumina by using a single laser. A shallow groove was created by using an evaporative procedure, and heat produced from the laser beam generated a time-dependent thermal stress along the moving path. Tsai and Huang (2008) reported on diamond scribing and laser breaking for LCD glass substrate. A groove-crack was created along the cutting path using a scoring tool, before separated by applying a defocused CO2 laser beam. Several studies have been performed on the thermal cleaving process of a silicon wafer. Ueda et al. (2002) found that the temperature at the area irradiated with a pulsed laser is an important factor in controlling the propagation of the crack and in achieving high cleaving accuracy with low thermal damage. Yamada et al. (2006) recommended a cleaving process with a pulsed laser and the use of refrigerating chuck for cleaving with a continuous wave (CW) laser to reduce thermal damage during the cleaving processes. Takeda et al. (2009) reported that separation in the same direction of the wafer cleavage plane can be achieved by lower laser energy, furthermore improving the cleaving surface. Both researchers conducted their experiments on thermal stress cleaving of a silicon wafer with the preparation of the initial crack by using the Vickers indenter impressions. However, the impressions created damage to the indentation area and a lateral crack in the direction perpendicular to the cleaving path. Ishikawa et al. (2012) introduced the use of a micro-groove to facilitate the fracture initiation during the thermal stress cleaving process. The unnecessary crack and damage were eliminated during the initial crack preparation, subsequently improving the quality of the material specimen. Previous studies on the thermal stress cleaving process of a silicon wafer were focused on the cleaving temperature, thermal damage and surface finish of the material’s separated surface. However, the mechanism of the thermal stress cleaving process of a silicon wafer had not been analyzed in detail, and the relation between laser condition and cleaving performance is not fully understood. In this paper, the separating characteristics of a silicon wafer with a moving laser beam by using the thermal stress cleaving process is investigated experimentally and computationally. The micro-groove was prepared at the leading edge of the silicon wafer to facilitate the fracture. The effect of laser energy on the cleaving mechanism was analyzed, and the influence of the groove parameters such as groove length and depth were also studied. The temperature of the moving laser spot was measured by using a two-color pyrometer with an optical fiber, and the AE signal was assessed to observe the mechanism of propagations. The cleaved surface was observed using scanning electron microscope (SEM). Further, the influences of laser power, temperature and the groove parameters to the fracture initiation, crack propagation and cleaved surface features are explained based on the experimental results. In order to explain the three-dimensional thermal stress distribution, the finite-element method (FEM) software, ANSYS, is applied by considering the temperature ascertained from the experimental result.
253
100 Laser : Nd:YAG (SHG) λ=532 nm P=280 mW f = 10 Hz
50 0 0
300
600
900
1200
Laser pulse , N Fig. 2. Relationship between laser pulse number and groove depth.
Fig. 3. Images of micro-groove with dimensions.
2. Experimental method
schematic illustration of the silicon wafer specimen with a size of 20 mm × 10 mm × 0.5 mm is shown in Fig. 1. The micro-groove was created initially by focusing the pulsed laser beam through the micro-lens at the edge of the silicon wafer (Ishikawa et al., 2012). A second harmonic generation (SHG) Nd:YAG laser with a wavelength of 532 nm, frequency of 10 Hz and pulse width of 5 ns was used in the specimen preparation. The laser energy was focused onto the specimen resulting in material heating and evaporating, thus forming the groove. By using this technique, the groove size can be controlled, and the area of damage can be minimized. In contrast with the initial crack created by the indenter, the unnecessary crack was eliminated; consequently improved the quality of the separated surface. The depth of the groove was adjusted by varying the laser pulse number while the groove length was controlled by focusing the laser beam only onto the required exposed area. Fig. 2 shows the relationship between the groove depth and laser pulse number. The images of the pre-prepared micro-groove are shown in Fig. 3.
2.1. Silicon wafer specimen
2.2. Temperature measurement
The specimen used in the experiments was a silicon wafer of 1 0 0 crystal orientation and with a thickness of 0.5 mm. The
The temperature gradient produced from the laser beam generates thermal stress in the material substrate. Therefore, it is
254
A.M. Saman et al. / Journal of Materials Processing Technology 223 (2015) 252–261
Fig. 4. Schematic illustration of experimental set-up.
important to know the temperature during the cleaving process. It is difficult to measure the temperature of the laser spot during the thermal cleaving process because the laser beam is moving, and the size of the laser beam used is very small. Ignatiev et al. (1996) reported on the measurement of pulsed laser action on metallic material using two silicon photo diodes. Ueda et al. (1997) monitored the surface temperature of ceramic materials in real time during laser beam irradiation using an infrared radiation pyrometer with a fused fiber coupler. Furumoto et al. (2009) used a pyrometer composed of an optical fiber and InAs and InSb infrared detectors to observe the temperature during laser consolidation of metal powder. In this paper, the temperature during laser irradiation was measured using a two-color pyrometer with different infrared detectors, namely InSb and MCT; and NSEG chalcogenide optical fiber. The fundamental structure of a two-color pyrometer is shown in Fig. 4. The infrared energy radiated from the object was captured by a chalcogenide optical fiber and transmitted to a two-color pyrometer consisting of InSb and MCT detectors. The InSb detector was mounted over an MCT detector in a sandwich configuration. The InSb detector responds to radiation from 3.1 to 5.5 m of wavelength and transmits waves larger than 5.5 m, and the MCT detector responds to radiation from 5.5 to 11 m. The infrared energy was converted to an amplified electric signal. By calculating the ratio of output voltages from these two detectors, the
Temperature, oC
600
temperature of an object can be obtained by using the calibration curve. Fig. 5 shows the calibration curve of the pyrometer, which was obtained experimentally by observing the incidence surface of the silicon wafer heated to the known constant temperature electrically. The frequency characteristic of the pyrometer has a flat response to about 100 kHz, which was sufficient for the experiment conducted in this paper (Ueda et al., 1998). An NSEG chalcogenide optical fiber with a core diameter of 380 m and acceptance angle of 60◦ was used to transmit the infrared rays of a wavelength from 3 to 11 m. Therefore, the fiber functions as an optical filter for filtering the laser beam of the YAG laser beam of 1.06 m reflected from the surface of the wafer. 2.3. Acoustic emission (AE) signal measurement The AE signal was measured to monitor the characteristics of fracture initiation and crack propagation during laser irradiation along the cleaving path. An AE sensor was fixed on the wafer’s surface at a distance of 5 mm from the micro-groove as shown in Fig. 4. Details of the AE signal measurement are summarized in Table 1. 2.4. Experimental procedure The experimental arrangement of the laser cleaving is illustrated in Fig. 4. The Nd:YAG laser with the wavelength of 1.06 m was used as a heat source, and the elliptical beam spot with a major axis length of 0.25 mm and minor axis length of 0.19 mm was defocused on the wafer’s surface. A chalcogenide optical fiber was set at a distance of 3 mm from the laser spot and at an angle of 45◦ from the optical axis. The center of the laser spot was always on the center
400 Table 1 Measuring system of AE signal.
Experiment
200
Units
Theoretical curve 0 0
1
2
3
4
5
Output ratio (InSb/MCT) Fig. 5. Calibration curve.
6
Acoustic emission (AE) sensor Size Frequency band (−10 dB)
Ø5 × 3.2 1–4
mm MHz
PreAmplifier Gain Frequency band (−10 dB)
40 0.1–20
dB MHz
A.M. Saman et al. / Journal of Materials Processing Technology 223 (2015) 252–261
255
Table 2 Experimental condition. Laser
Nd:YAG
Units
Wavelength, Irradiation mode Laser power, P Laser beam diameter, D Moving speed, v
1064 Continuous wave 50–70 0.25 × 0.19 2
nm
Work piece
Silicon wafer
Crystal orientation Size
1 0 0 10 × 20 × 0.5
mm3
Pyrometer Optical fiber Detector Fiber core diameter, Ø
Two-color NSEG InSb/MCT 380
m
W mm2 mm/s
axis of the fiber. In the experiments, the specimen was set on a moving table (X–Y stage). The work table moved at a constant speed of 2 mm/s, and the laser beam relatively scanned the work material at the same speed. The laser beam and optical fiber remained in a static position while the AE sensor was moving together with the work material. Before the X–Y stage was moving, the laser beam was set at a position of 2 mm from the silicon wafer’s leading edge. After the X–Y stage has been moving for one second, the center of the laser beam will reach the wafer’s leading edge. Therefore, the voltage output of the pyrometer detectors and the AE sensor in relation to the irradiation time can be determined. Consequently, the temperature distribution and crack occurrence during laser irradiation along the laser beam path on the work material can be verified. The output voltages from the pyrometer detectors and AE sensor were recorded in a digital oscilloscope. The laser power used for this experiment varied between 50 and 70 W, while the groove length and the depth were maintained at 500 and 80 m accordingly. A summary of the experimental conditions is shown in Table 2. After the specimen was separated into two pieces, the cleaving surface of the specimen was analyzed by using a scanning electron microscope (SEM).
3. Experimental result and discussion 3.1. Output signal from pyrometer and AE sensor Fig. 6 shows the typical output profiles of the pyrometer and AE sensor. Fig. 6(a) and (b) represents the output voltages of the two detectors, while Fig. 6(c) shows the temperature history of the laser spot along the irradiation path. Fig. 6(d) shows the signal from the AE sensor. The laser beam reached the wafer’s leading edge after moving for one second. Therefore, the distributions of InSb, MCT and AE signals per unit time along the cleaving path on the work piece can be recognized. The signal from the pyrometer detectors shown was a result of chopped signals that were produced by using a chopper in the pyrometer set-up. The InSb and MCT signals were increased rapidly after the laser beam reached the wafer’s leading edge, later declined proportionally until they attained a stable voltage output at the center area of the wafer. The laser beam was emitted on the wafer surface with Gaussian’s intensity distribution (Ueda et al., 2002). Hence, the pyrometer detectors’ signals increased earlier, before the laser beam center reached the wafer’s leading edge. The signal size increases smoothly when the laser beam was approaching the wafer’s trailing edge and declines quickly before the laser beam passes the wafer’s trailing edge. AE signals in Fig. 6(d) show the characteristics of the thermal cleaving process during laser irradiation. When the crack occurs, the AE sensor will trigger the signals. A large signal occurred a few
Fig. 6. Output wave of pyrometer and AE sensor.
milliseconds after the laser beam reached the wafer’s leading edge, continuing with the repeating signals. 3.2. Temperature along the irradiation path The variation of temperature was obtained by the ratio of output signals from the InSb and MCT signals. The temperature history during laser irradiation in Fig. 6(c) shows a higher reading at a position near to edge of the silicon wafer. The time it takes for the center of the laser beam to reach the wafer’s leading edge is considered as t = 0 s. A higher temperature of 385 ◦ C was recorded approximately at t = 0.27 s, later reduced to a constant temperature of approximately 205 ◦ C, then rising to 340 ◦ C approximately at t = 0.27 s before the laser beam reached the wafer’s trailing edge at t = 5 s. The results indicate that the laser spot was maintained at a constant lower temperature at the center of the laser path. Silicon wafer has a high thermal conductivity of 156 W/m K (Liu et al., 2011; Yamada et al., 2006). Hence, the energy produced by laser irradiation was circumferentially diffused when the laser beam was at the center area of the wafer compared to when the laser beam was closest to the edge of the wafer. The variation of maximum temperatures at the leading edge, trailing edge and average temperature at the center of the specimen with the laser power are shown in Fig. 7. The temperature at all three positions is increased with the increase of the laser power. From the results, it is clear that the temperature of the moving laser spot was dependent on the energy supplied by the laser beam. Higher laser power resulted in an increasing of temperature. 3.3. Crack occurrence Voltage output signals from the AE sensor (Fig. 6(d)) demonstrates that the crack occurrences appeared in two stages, namely fracture initiation and intermittent crack propagation. AE signals
256
A.M. Saman et al. / Journal of Materials Processing Technology 223 (2015) 252–261
Fig. 7. Influence of laser power on temperature.
Fig. 8. Influence of laser power on crack initiation time.
were produced when a rapid release of energy from localized sources within the material occurred. Energy release happened as a result of crack extension due to thermal stress generated by the laser beam. The first signal of the AE wave shows that fracture initiation occurred with large signal size. Subsequently, a series of smaller AE signals indicates that crack propagation occurred intermittently. A large signal indicates that a higher energy was released during fracture initiation. It was shown that the energy required to start the fracture is high, in contrast with the energy during the crack propagation. This was because the fracture initiation started at the groove tip with a blunt notch (Fig. 3(a)) and a sharp crack tip was produced after the first fracture occurred. Greater stress concentration developed at the sharper crack tip in contrast to the blunt groove. Therefore, larger energy was required to increase the local stress at the blunt groove to reach the fracture strength of the material.
Fig. 10. Relationship between laser power and fracture initiation distance.
3.3.1. Fracture initiation The laser beam energy generates a thermal gradient, which creates a compressive thermal stress at the laser beam area and tensile stress area around it. When the laser beam reaches the leading edge, compressive stress generated at this point due to the heat induced from the laser beam. When the laser beam moves forward, thermal stress at the leading edge of the silicon wafer is transformed into tensile stress and concentrated at the groove tip (x = 0, y = 0, z = −80 m). When the tensile stress reaches the fracture strength of the material, the fracture will be initiated. Fig. 8 shows the influence of laser power on the fracture initiation time. The specimen was prepared with a groove length and depth of approximately 500 and 80 m respectively; and the laser beam scan speed was set constantly at 2 mm/s. It was found that the fracture initiation time decreased with an increase in laser power. The average fracture initiation time for laser power of 50 W was attained approximately at 0.42 s compared to 0.27 s for laser power of 70 W. The stress at which a specimen fails via fracture was achieved faster with higher laser energy. As a result, earlier fracture initiation was achieved at higher laser power. Fig. 9 shows the SEM images of smooth surfaces on the separated plane as observed at the leading edge area of the silicon wafer. A smooth, flat surface was produced for a width of a few micrometers from the wafer’s edge followed by a periodic wave marking. The variation of the extent of the smooth surface in relation to laser power is shown in Fig. 10. It was found that the extent of the smooth surface increased with the decrease in laser power. The distances of the smooth surfaces were obtained approximately 185 m at a laser power of 70 W, in comparison with the 295 and 420 m for the laser power of 60 W and 50 W respectively. The result shows that fracture initiation occurs at longer travel distance when a lower laser power was used. As a result, a longer smooth cleaving surface was produced. However, if the laser power used is too low, the fracture might not occur due to insufficient
Fig. 9. SEM images at the starting section of cleaving surface.
A.M. Saman et al. / Journal of Materials Processing Technology 223 (2015) 252–261
257
Fig. 14. Relationship between cleaving mark interval and laser power. Fig. 11. AE signals.
with the relaxation of local stress. The crack will develop again when the sufficient stress was attained. The stress recovery rate was increased with the increased of laser power. Therefore, crack signals appeared rapidly. Fig. 13 shows the SEM images of the cleaving marks pattern on the center section of the separated surface. The wave marks were developed due to crack advancement occurrences during the intermittent crack propagation phenomena. The variation between the mark interval and the laser power is given in Fig. 14. The interval between the marks was found to be approximately 32 m for the laser power of 50 W in comparison with 27 and 22 m for 60 W and 70 W respectively. The increase of laser power caused higher repetition of the cleaving mark. The mark interval was comparable to the AE signal results in Fig. 12. 3.4. The influence of groove length and depth Fig. 12. The relationship between laser power and AE signal interval.
stress concentration at the groove tip. In contrast, if the laser power used is high, thermal damage might occur, as shown in Fig. 9(a). 3.3.2. Intermittent crack propagation Fig. 11 shows the typical output signals of an AE sensor measured at the middle of the laser path. The result shows that a repetitive AE signal appeared with a starting and stopping phenomenon at regular intervals. This occurrence revealed that the crack was extended and relaxed frequently at a perpetual distance, as the laser beam moved along the laser path. The variation of laser power with the AE signal interval is shown in Fig. 12. The increase of laser power produced shorter intervals between the crack propagation signals. The intervals were found to be constant and proportionate to the laser power. For instance, the interval in signals found for a laser power of P = 50 W was about 16.2 ms or 32.4 m in distance while the signal interval for P = 70 W was about 10.5 ms or 21 m in distance. The crack in the material grows
To understand the influence of groove parameters, such as groove length and groove depth on cleaving performance, the experiments were performed by irradiating a constant laser power of 70 W and scan speed of 2 mm/s. In the case of varied groove lengths, the depth was fixed at 80 m and in the case of varied groove depths, the length was fixed at 500 m. The AE signal was used to monitor the fracture initiation following the procedure stated in the previous section. Table 3 shows the summary of experimental conditions for varied groove length and depth accordingly. Fig. 15 shows the effects of groove length on fracture initiation time, whereas the variation of groove depth with fracture initiation time is given in Fig. 16. The result shows that the time for fracture initiation was decreased with the enlarging of groove length. For instance, the average fracture initiation time obtained for a groove length of 100 m was about 450 ms compared with 270 ms for a 500 m groove length. Furthermore, fracture initiation time was also found to be faster for deeper groove depth. The average fracture initiation time for a groove depth of 50 m was reached
Fig. 13. SEM images at the center section of cleaving surface.
258
A.M. Saman et al. / Journal of Materials Processing Technology 223 (2015) 252–261
Table 3 Experiment conditions of varied groove length and groove depth. Laser Wavelength, Irradiation mode Laser power, P Laser beam diameter, D Moving speed, v Experiment of varies groove length Groove length Groove depth Experiment of varies groove depth Groove length Groove depth
Nd:YAG
Units
1064 Continuous wave 70 0.25 × 0.19 2
nm
100–500 80
m m
W mm2 mm/s Fig. 17. The cross-section of the specimen with a coordinate system.
500 20–150
m m
Fig. 15. Effect of groove length on fracture initiation time.
analysis was performed to investigate the thermal stress during fracture initiation. Finite-element software, ANSYS, was used to analyze the behavior of thermal stress cleaving during the laser irradiation by considering the finite boundary effect. The temperature and thermal stress distribution around the laser spot diverged along the thickness direction of the substrate. Therefore, the finite-element model should be considered as a three-dimensional problem. The specimen size used in this study was 20 mm × 10 mm × 0.5 mm. The half-part of the specimen with a coordinate system is shown in Fig. 17. The following basic assumptions for the thermo-mechanical analysis were considered. The material properties such as the heattransfer properties and mechanical properties were assumed to be constant as the temperature range was not high. The stress–strain relationship of the silicon wafer was assumed to be perfectly elastic. Convection conditions were assumed to exist on all boundaries except at the surface of x–z plane (y = 0). This surface was considered to be in adiabatic condition. The energy absorbed by the silicon wafer during laser irradiation varies along the material thickness. The absorption characteristic is represented by Beer Lambert’s equation below: ˇ=
Fig. 16. Effect of groove depth on fracture initiation time.
approximately at 340 ms compared to 270 ms for a groove depth of 130 m. From the finite-element analysis results, it is demonstrated that higher tensile stress concentration was resulted at a longer groove length (Fig. 21) and at a deeper groove depth (Fig. 23). The results explained why fracture initiation time was reached earlier at a longer groove length and a deeper groove depth. It does become apparent that the groove has become the weakness point in the material. Local stress concentrated at the groove tip could rise to several times that of the applied stress, and the stress level is increased with the increase of groove size.
− ln T x
(1)
where T is spectral transmittance (T = 23% as measured experimentally), ˇ is the absorption coefficient and x is the thickness of the wafer. The boundary condition of the stress analysis is traction-free at all surfaces, and the x–z surface is set as a fixed plane which has zero displacement in the y-direction. During the laser irradiation, a moving laser beam with the power output of 70 W and laser beam spot size of 0.25 mm × 0.19 mm was set on the wafer surface above the focal plane. When the center of the laser beam reached the wafer leading edge (i.e. x = y = z = 0), the time was considered as t = 0 s, later moving toward the x-direction at a constant speed of v = 2 mm/s. The initial temperature of the silicon wafer substrate and the surrounding temperature were set to 20 ◦ C. The analysis was carried out into two phases. The transient temperature analysis was calculated when the moving laser beam was irradiated onto the silicon wafer. Then, the steady state stress analysis was performed using the temperature distribution results at a specific point in time. The temperature and stress distribution adjacent to the laser spot vary along the thickness direction during the cleaving process; therefore, the calculation was considered as a three-dimensional problem. The summary of analysis condition and properties of the silicon wafer are shown in Table 4. 4.1. Temperature and thermal stress distribution result
4. Finite-element analysis The computational analysis was performed to understand the actual phenomenon of fracture during the thermal cleaving process. The temperature and AE signal of crack behavior have been monitored during the experimental work. However, the mechanisms of fracture have not been clarified. In this section, the
The transient temperature distributions were calculated at points in time from 0 s to 0.495 s. The temperature distribution at various times is shown in Fig. 18. The temperature was measured on the surface of the material along the x-axis (y = 0, z = 0). The maximum temperature reached at the times of 0 s, 0.275 s and 0.495 s was 230 ◦ C, 380 ◦ C and 320 ◦ C accordingly. The calculated
A.M. Saman et al. / Journal of Materials Processing Technology 223 (2015) 252–261
259
Table 4 Analysis condition and properties of silicon wafer. Units Analysis conditions Workpiece size Laser beam diameter (oval) Heat transfer coefficient, h Initial temperature, To Laser beam scan speed, v Laser power, P
20 × 10 × 0.5 0.25 × 0.19 8 20 2 70
mm3 mm/s W m−2 K−1 C mm/s W
Properties Thermal conductivity, K Specific heat, C Thermal expansion coef., ˛ Young modulus, E Poisson ratio, Density,
156 761 2.62 × 10−6 169 0.262 2340
W/m K J/kg K K−1 GPa kg/m3
Fig. 20. Thermal stress yy distribution on x–z plane (y = 0).
Fig. 18. The temperature distribution of moving laser.
transformed from compression to tensile stress in a short time as the laser beam moved forward in the x-axis direction. As the laser beam moved further, large tensile stress was accumulated at the groove tip (x = 0). This large tensile stress yy may induce a fracture when the stress reaches the fracture strength of the material. The fracture could be induced along the thickness direction (z-axis) and extended through the material. The tensile stress occurred after the laser beam was passing through; therefore, the fracture will follow the laser spot with a short lagging distance. 4.2. The influence of groove length
Fig. 19. Thermal stress yy , distribution at various times (z = −80 m).
temperature results were consistent with the temperature from the experimental result (Fig. 7). The thermal stress yy distribution, along the cleaving plane, at the time of 0.01 s, 0.055 s, 0.165 s, 0.275 s, 0.385 s and 0.495 s is shown in Fig. 19. When the laser beam was at the position of x = 0, and the time of t = 0 s, the stress at the groove tip (z = −80 m) on the edge was a compressive stress. The thermal stress then
According to Fig. 15, the experimental result indicated that in the case of a wafer with a groove length of 500 m, the fracture initiation time was attained at approximately t = 0.275 s. At this time, the laser beam reached x = 0.55 mm. For the purpose of assessment, the analyses of stress distribution for various cases of groove length were performed at the same time of 0.275 s. Fig. 20 shows the thermal stress yy distribution on the x–z plane and y = 0 for the cases of a groove length of 100, 300 and 500 m at the time of t = 0.275 s. The thermal stress yy distribution in the x-axis direction for all cases was almost similar, except at the area under the groove. Compressive stress was centered at the laser spot area and a tensile stress resulted at the area near to the edge of the material, where x = 0 mm. However, the maximum tensile stress was located at the groove tip (z = −80 m) with more than 60 MPa value of tensile stress concentrated on the groove tip at the edge of the material (x = 0 mm). The figures on the left of Fig. 20(a)–(c) illustrate the thermal stress distribution, yy , along the z-axis (x = 0, y = 0). The results indicate that the thermal stress, yy , was at its maximum at the
260
A.M. Saman et al. / Journal of Materials Processing Technology 223 (2015) 252–261
Fig. 23. Relationship between thermal stress yy and groove depth. Fig. 21. Relationship between thermal stress yy and groove length.
that the thermal stress, yy , is distributed along the x-direction. Compressive stress is focused under the laser beam position (x = 0.55 mm) and tensile stress resulted at the edge of the material (x = 0 mm). Large tensile stress of more than 60 MPa was concentrated on the groove tip (z = 20 m, x = 0 mm). Moreover, this phenomenon was repeated in the case of a groove depth of 50 and 80 m as illustrated in Fig. 22(b) and (c) accordingly. The maximum tensile stress was concentrated on the groove tip for all cases. The figures on the left of Fig. 22(a)–(c) illustrate the thermal stress distribution, yy , along the z-axis (x = 0, y = 0). The results indicate that the thermal stress, yy , is highest at the groove tip and reduces rapidly to the constant tensile stress value of approximately 30 MPa along the −z direction. The relationship between the maximum thermal stress at the groove tip and groove depth is shown in Fig. 23. A deeper groove resulted in higher thermal stress concentration. 5. Separating mechanism of silicon wafer
Fig. 22. Thermal stress yy distribution on x–z plane (y = 0).
groove tip (z = −80 m), and reduces rapidly when it moves along the −z direction before reaching approximately 30 MPa of constant tensile stress value. The relationship between the maximum thermal stress and groove length is shown in Fig. 21. It was found that the proportion of the tensile stress concentrated on the groove tip was dependent on the groove length. The longer groove length resulted in a higher concentration of thermal stress, yy . 4.3. The influence of groove depth Fig. 22 shows a comparison of thermal stress, yy , distribution on the x–z plane (y = 0) of the wafer with a groove depth of 20, 50, and 80 m at the time of 0.275 s. In Fig. 22(a), it is illustrated
Based on the experimental and computational analysis results, the mechanism of separating a silicon wafer by applying a laser beam using the thermal stress cleaving process can be understood clearly. Fig. 24 illustrates the separating mechanism of the silicon wafer with a pre-prepared micro-groove, by utilizing the moving laser beam. The cleaving process can be divided into two stages, namely, fracture initiation and intermittent crack propagation. As shown in Fig. 24(a), when the laser reaches the edge of the wafer, the fracture does not occur yet. This is due to the fact that the laser beam induces the heat at the laser spot area; hence, generated a compressive stress at this point. After the laser beam moves further, the compressive stress transforms into tensile stress. The tensile stress concentrates at the groove tip, and it increases with the increase in temperature at the laser spot and the distance relative to the leading edge of the wafer. When the intense stress at the groove tip reached the material fracture strength, a fracture initiated through the material thickness and extended toward the laser spot in a very short time. As illustrated in Fig. 24(b), the fracture stopped before the laser beam spot and created a new crack tip. Continuous heat supplied by the laser beam caused the tensile stress to increase on the newly created crack tip, and the crack propagated and stopped again before the laser spot. As shown in Fig. 24(c), the cleaving mark was created as the crack stopped. This process was replicated again and again until the laser beam reached the end of the wafer. It is demonstrated that fracture initiation and intermittent crack propagation occurrence are influenced by the heat supplied from the laser beam during the irradiation process. An increase in laser power improved the fracture initiation time and increased crack propagation recurrence. However, the laser power should be controlled to avoid any thermal damage to the silicon wafer.
A.M. Saman et al. / Journal of Materials Processing Technology 223 (2015) 252–261
261
Fig. 24. Thermal stress cleaving mechanism of silicon wafer with microgroove.
6. Conclusion The separating characteristics of a silicon wafer with a moving Nd:YAG laser by using the thermal stress cleaving technique were successfully monitored and analyzed by the utilization of a twocolor pyrometer with an optical fiber and AE sensor. Computational analysis by the finite-element method described the phenomenon of stress distribution during the process. Based on the study, the mechanism of the thermal stress cleaving of a silicon wafer with a laser beam was explained. The separating process of the silicon wafer occurred in two stages. The first stage is the fracture initiation, where the fracture occurs due to tensile stress concentrated at the groove tip on the edge of specimen. The second is intermittent crack propagation which occurs at constant intervals in time. The temperature at the laser irradiation area was related to the separating characteristics of a silicon wafer by the cleaving technique. The temperature influenced the cleaving achievement and cleaving surface quality. Fracture initiation and crack propagation occurrences produced the cleaving mark waves on the separating surface. The cleaving mark wave was consistent with the AE signals. The finite-element analysis explained the stress distribution during laser irradiation. The results correspond well with the experimental results. References Furumoto, T., Ueda, T., Kobayashi, N., Yassin, A., Hosokawa, A., Abe, S., 2009. Study on laser consolidation of metal powder with Yb:fiber laser—evaluation of line consolidation structure. J. Mater. Process. Technol. 209 (18–19), 5973–5980.
Garibotti, D.J., 1963. Dicing of micro-semiconductors. U.S. Patent No. 3,112,850. Ignatiev, M.B., Smurov, Yu, I., Flamant, G., Senchenko, V.N., 1996. Surface temperature measurement during pulsed laser action on metallic and ceramic materials. Appl. Surf. Sci. 96–98, 505–512. Ishikawa, T., Ueda, T., Furumoto, T., Hosokawa, A., Tanaka, R., 2012. Pre-groove development for laser cleaving of brittle materials by using micro-lens. J. Adv. Mech. Des. Syst. Manuf. 6 (6), 841–848. Lambert, E., Lambert, J.L., Longueville, B.D., 1976. Severing of a glass or vitrocrystalline bodies. U.S. Patent No. 3,935,419. Liu, J., Lu, J., Ni Xi Dai, G., Zhang, L., Chen, Y., 2011. Numerical study on thermal stress cutting of silicon wafer using two-point pulsed laser. Opt. Appl. 41 (1), 247–255. Lumley, R.M., 1969. Controlled separation of brittle materials using a laser. Am. Ceram. Soc. Bull. 48 (1), 850–854. Steen, W.M., Mazumder, J., 2010. Laser Material Processing, fourth ed. SpringerVerlag, London, pp. 131–193. Takeda, R., Ueda, T., Furumoto, T., Hosokawa, A., Tanaka, R., 2009. Study on cleaving mechanism of silicon wafer by laser beam irradiation. In: Proc. Leading Edge Manufacturing in 21st Century (5), The Japan Society of Mechanical Engineers, pp. 589–592. Tsai, C.H., Huang, B.W., 2008. Diamond scribing and laser breaking for LCD glass substrates. J. Mater. Process. Technol. 198 (1–3), 350–358. Tsai, C.H., Liou, C.S., 2003. Fracture mechanism of laser cutting with controlled fracture. J. Manuf. Sci. Eng. ASME 125 (3), 519–528. Ueda, T., Yamada, K., Nakayama, K., 1997. Temperature of work materials irradiated with CO2 laser. CIRP Ann. Manuf. Technol. 46 (1), 117–122. Ueda, T., Sato, M., Nakayama, K., 1998. The temperature of a single crystal diamond tool in turning. CIRP Ann. Manuf. Technol. 47 (1), 2–5. Ueda, T., Yamada, K., Oiso, K., Hosokawa, A., 2002. Thermal stress cleaving of brittle materials by laser beam. CIRP Ann. Manuf. Technol. 51 (1), 149–152. Wang, Y.Z., Lin, J., 2007. Characterization of the laser cleaving on glass sheets with a line-shape laser beam. Opt. Laser Technol. 39 (5), 892–899. Yamada, K., Ueda, T., Hosokawa, A., Yamane, Y., Sekiya, K., 2006. Thermal damage of silicon wafer in thermal cleaving process with pulsed laser and CW laser. In: Proc. SPIE Laser-based Micro Packaging, vol. 6107, pp. 61070H–61070H–10.