- Email: [email protected]
MOPA pulsed fiber laser for silicon scribing
Optics & Laser Technology 80 (2016) 67–71
Contents lists available at ScienceDirect
Optics & Laser Technology journal homepage: www.elsevier.com/locate/optlastec
Full length article
MOPA pulsed fiber laser for silicon scribing Limei Yang, Wei Huang n, Mengmeng Deng, Feng Li Key Laboratory of Nanodevices and Applications, Suzhou Institute of Nano-Tech and Nano-Bionics, Chinese Academy of Sciences (CAS), Suzhou, China
art ic l e i nf o
a b s t r a c t
Article history: Received 1 September 2015 Received in revised form 14 December 2015 Accepted 4 January 2016
A 1064 nm master oscillator power amplifier (MOPA) pulsed fiber laser is developed with flexible control over the pulse width, repetition frequency and peak power, and it is used to investigate the dependence of mono-crystalline silicon scribe depth on the laser pulse width, scanning speed and repeat times. Experimental results indicate that long pulses with low peak powers lead to deep ablation depths. We also demonstrate that the ablation depth grows fast with the scanning repeat times at first and progressively tends to be saturated when the repeat times reach a certain level. A thermal model considering the laser pulse overlapping effect that predicts the silicon temperature variation and scribe depth is employed to verify the experimental conclusions with reasonably close agreement. These conclusions are of great benefits to the optimization of the laser material processing with high efficiency. & 2016 Elsevier Ltd. All rights reserved.
Keywords: MOPA Fiber laser Laser ablation Thermal model Silicon Surface temperature Subject classification codes: 140.3538 140.3615 140.6810 140.3390
1. Introduction Pulsed lasers find more and more micromachining applications ranging from ablation, cutting, drilling and marking. A wide array of publications have discussed the laser materials processing using a variety of lasers [1–3]. It is noted that the laser machining process is defined by the specific lasers used, and the optimization of a machining process involves the adjustment of multiple laser parameters, e.g., pulse width, average power, pulse energy and repetition frequency [4–6]. However, a variety of pulsed lasers in use, such as diode-pumped Q-switched lasers comprising an intra-cavity Q-switch element, adjusting one of the laser's parameters would invariably lead to changes in other parameters [7,8]. As a result, the improvement of the performance of laser micromachining is limited by the operation capability of these pulsed lasers. High power pulsed fiber lasers offer a compact, electrically efficient and stable alternative to conventional bulky solid-state lasers, and they are being widely used in industrial manufacturing. Among various pulsed fiber laser solutions, the MOPA approach is a particularly attractive regime, and it offers unprecedented levels of performance and flexibility. That is, the laser pulse parameters n
Corresponding author. E-mail address: [email protected] (W. Huang).
http://dx.doi.org/10.1016/j.optlastec.2016.01.006 0030-3992/& 2016 Elsevier Ltd. All rights reserved.
can be adjusted independently, while maintaining a constant beam quality [9–11]. In this study, a MOPA fiber laser at the level of nanosecond is developed, and it is used to conduct experiments that evaluate the effects of pulse width and laser scanning parameters on the ablation process of mono-crystalline silicon, while maintaining a constant average power. A theoretical thermal model is utilized to analyze the scribe depths in terms of temperature variation of silicon material, and the theoretical predictions agree well with the experimental results. It indicates that laser pulse width and scanning parameters have great influences on the material removal rate and debris accumulation.
2. Laser configuration and marking system The schematic of MOPA nanosecond fiber laser system is shown in Fig. 1. The system consists of two amplifying stages, e.g., preamplifier and power amplifier. The modulated seed source is a pulsed laser diode at 1064 nm, with power, pulse width and repetition frequency that are variable and independently controlled. Both stages are based on Yb-doped fiber amplifier (YDFA). The preamplifier stage is forward pumped by a 915 nm, 10 W laser diode coupled with Yb-doped double-clad fiber (YDDCF) using a combiner, and it provides an amplification of the seed laser with gain
68
L. Yang et al. / Optics & Laser Technology 80 (2016) 67–71
Fig. 1. Schematic of nano-second MOPA fiber laser system.
of 24 dB. A 2 2 fiber coupler, which is applied before the preamplifier stage, is for monitoring signal and back-reflection light without dismantling. The power-amplifier stage is cladding pumped by two 915 nm, 24 W laser diodes. Seed source and preamplifier stage, pre-amplifier stage and power-amplifier stage are separated with isolators which prevent the backward reflections and amplified spontaneous emission (ASE). The unabsorbed pump light from the two stages is rejected by self-made cladding-mode strippers. The system is terminated with an optically isolated fiber collimator, ensuring the reflections due to mismatch do not cause oscillations. All the optical components are fixed on an aluminum plate for heat dissipation, and the whole system is enclosed in a box. As a result, a pulsed fiber laser is configured to emit pulses with an average power of 20 W, peak power up to 10 kW, pulse width ranging from 2 to 500 ns and repetition frequency of 20 to 500 kHz. The fiber laser is arranged in a marking station, consisting of a beam expander, scanner and projection f-theta lens, for delivery of laser beam on the target surface. The diameter of focused spot on the target surface is measured to be about 60 μm, which leads to peak irradiance level of about 0.35 GW/cm2 and fluence level of about 7.1 J/cm2 for pulse of 20 ns width and 100 kHz repetition frequency. It is worth noting that the MOPA pulsed fiber laser can be operated at a wide array of pulse configurations, and specific pulses with varying pulse widths are chosen in this work, while maintaining the same pulse energy (e.g., having the same average power 20 W and repetition frequency of 100 kHz). The scribing experiments are conducted on mono-crystalline silicon. The peak irradiance levels of pulses used in the experiment are well above the corresponding ablation threshold of silicon (the peak irradiance levels for pulse width of 20 ns, 40 ns, 100 ns, and 200 ns are 0.35 GW/cm2, 0.18 GW/cm2, 70 MW/cm2, and 35 MW/cm2, which are all above the corresponding ablation threshold of 27 MW/cm2, 19.41 MW/cm2, 10 MW/cm2, and 8.68 MW/cm2, respectively).
3. Theoretical model A detailed modeling of the laser ablation process is quite complex as complicated mechanisms are involved in the ablation process. When processing with nanosecond pulse laser, the laser can be described as a heat source, and the heat equation can be applied to calculate the temperature evolution in the substrate. It is assumed that the laser radiation losses during scribing can be neglected, the heat conduction equation is approximately described by [12,13]
ρC
∂T (r , z, t ) = K ·∇2T (r , z, t ) + α (T )(1 − R) I0 (t ) e−zα (T ) ∂t
(1)
Where Κ is the thermal conductivity (K ¼ κρC), κ is the thermal diffusivity, ρ is the mass density, C is the thermal heat capacity, Τ is the temperature as a function of position (r, z) and time t, R is the Fresnel reflectivity of the surface, I0 is the time-dependent incident laser intensity, and α is the absorption coefficient. The general formulation for the temperature cannot be evaluated explicitly in Eq. (1). It is noted that the surface temperature would be well above the room temperature as the spatial overlap of the consecutive optical pulses being considered for the irradiated surface during the scanning process. Since the absorption coefficient of the silicon increases rapidly as the temperature rises [12], such that it is reasonable to assume that the absorption coefficient of the silicon is high enough to ensure most of the laser energy absorbed at the silicon surface. As a result, the temperature rise of the silicon due to absorption of an incoming Gaussian laser pulse, is given by [14] 1
∆T (r , z, t ) =
Imax γw 2 ⎛⎜ κ ⎞⎟ 2 ⎝ π⎠ K
∫0
τ
p (τ − t )
(
)
t 4κt + w 2
⎛ z2 ⎞ 2 − r ⎜− ⎟ 4κt 4κt + w 2 ⎠ dt
e⎝
(2)
Where Imax is the peak power per unit area at the center of the beam spot, w is the beam radius, τ is the pulse width; γ being equal to 1 R, is the fraction of the pulse energy that is absorbed by the material, and p(t) is the laser pulse temporal profile. Eq. (2) is used to predict the temperature distribution at the cross-section of the silicon as a function of depth z and radial distance r with respect to the center of the laser beam, considering that the radiation of a single pulse with a certain width is absorbed. During the simulation, p(t) is assumed to be of a square shape for the MOPA fiber laser, which is quite reasonable especially for pulses with long widths. Other parameters used in the simulation are listed as follows, w ¼30 μm, κ ¼0.907 10 4 m2/s, K ¼145.3 W/m K, γ ¼0.69, the boiling temperature TB ¼ 2628 K [15]. Fig. 2(a)–(d) shows the calculated temperature distribution on the cross-section of silicon with pulses widths of 20 ns, 40 ns, 100 ns and 200 ns. The temperature contour follows the Gaussian distribution, and the temperature decreases as the depth and radial distance increases. The white line in Fig. 2 represents the boiling temperature of silicon at different pulse widths. Above the white line, the temperatures are high enough to sublimate of the material and well-defined ablation pits form. It also shows that a longer pulse with lower power can ablate a deeper pit than a shorter one with higher power, which is illustrated in Fig. 3. It is quite straightforward to calculate the ablation depth z(r) for a single pulse by setting the temperature rise ΔT being equal to the boiling temperature of the silicon minus the initial temperature in
L. Yang et al. / Optics & Laser Technology 80 (2016) 67–71
69
Fig. 2. Calculated temperature distribution on the cross section of silicon for pulse with width of 20 ns (a), 40 ns (b), 100 ns (c), and 200 ns (d).
next pulse arrives, the previous ablation region has cooled down to room temperature, the effective ablation depth can be expressed as,
d (r ) = z (r + v·f ) + ⋯ + z (r + n·v·f ) + z (r ) + z (v·f − r ) + ⋯ + z (n·v·f − r )
(3)
where –worow, v is the scanning speed, f is the pulse repetition frequency, and n is the number of sequential overlapping pulses.
4. Results and discussion
Fig. 3. The estimated ablation depth versus pulse width.
Eq. (2). For the most practical applications, multiple pulses overlap spatially as the pulsed laser beam scans across the material surface. In this case, one needs to take into account the depths caused by all the overlapping pulses. Based on the assumption that and when the
Fig. 4 shows SEM images of the silicon scribe results using pulse with widths of 20 ns, 40 ns, 100 ns, and 200 ns at scanning speeds of 1.2 m/s and 1.8 m/s, respectively. All the applied pulses have constant average power of 20 W and repetition frequency of 100 kHz. As seen in Fig. 4, the cross-sections of all the scribe grooves have similar Gaussian profiles as that of delivered laser beam. All the laser pulses employed have irradiance above the threshold one for producing the plasma [14], which drives the strong shock wave. For short pulse with high irradiance, the ablation is dominated by the latent heat of vaporization with less melt. Consequently, less molten silicon particles ejected by the shock wave resolidify at the edges of the groove comparing to that of long pulse (the zoomed SEM images in Fig. 5). We measured the depths of these scribed grooves at the center
70
L. Yang et al. / Optics & Laser Technology 80 (2016) 67–71
Fig. 4. SEM images of scribed grooves for pulse width ranging from 20 ns to 200 ns at scanning speed of 1.2 m/s and 1.8 m/s.
of the beam (e.g., r¼0) using the scanning electron microscope (SEM), and calculated the corresponding theoretical values according to (Eqs. (2) and 3). The results were plotted against pulse width with inset of zoomed SEM images of ablated grooves as shown in Fig. 5. The ablation depth increases as pulse width increases from 40 ns to 200 ns at both scanning speeds of 1.2 m/s and 1.8 m/s. Longer pulses with lower peak power lead to deeper scribing depths, while shorter pulses with higher peak power have shallower depths. Physically, the dominant factor is the latent heat of vaporization when laser pulse with short duration interacts with the silicon surface, and much of the pulse energy is devoted to heat the vaporized silicon to a higher temperature (i.e., producing plasma), and distribution of the heat to a large volume of silicon material is limited by the short pulse length. Whereas for pulse with longer duration and low peak power, the amount vaporized depends more on the thermal conductivity of silicon. The heat has a better chance to flow into the interior of the silicon, which lead to a deep ablation depth [14]. An overall good agreement is found between the theoretical predictions and experimental results, despite small discrepancies at pulse widths of 40 ns and 200 ns, which might be
attributed to lack of consideration of the dependence of some crucial parameters on the temperature during simulation, e.g., silicon absorption coefficient, conductivity and diffusivity [15], which are regarded as constants during the calculations. The error of the measurement data comes from the surface roughness at the bottom of the trench. Indeed, a single scan cannot meet the industrial demands, and a repeated scanning process is always necessary to achieve a deep penetration depth. Fig. 6 shows the silicon scribe depths as a function of repeat times at 100 ns. For both scanning speeds of 1.2 m/s and 1.8 m/s, the scribe depth increases with the repeat times, but it tends to be saturated when the repeat times reach 25. One of the possible reasons is the decrease of laser energy at the ablation region as the depth goes deeper. Physically, laser beam defocuses and laser power density decreases due to the divergence as it propagates into the deeper position of silicon, and this reduces the ablation rate. Moreover, EDX measurement results show that the laser-irradiated silicon oxidizes during the ablation process, which indicates that oxidation of silicon is probably another possible reason for the saturation effect as the repeat times increase.
L. Yang et al. / Optics & Laser Technology 80 (2016) 67–71
71
width increases from 20 ns to 200 ns at both scanning speeds of 1.2 m/s and 1.8 m/s. A theoretical ablation model is used to simulate the silicon temperature distribution and the ablation depth. Simulation results indicate pulse with shorter width leads to a higher temperature, shallower depth, and less melting and recasting of the material, which shows a good match to the experimental conclusions. Moreover, experimental results indicate it is impracticable to increase a deep ablation depth by increasing the repeat times all the time, as the ablation depth progressively grows slow with the repeat times as it reaches a certain level. The conclusions and analytical method demonstrated in the paper provide strategic guidance to optimize the laser parameters and scanning parameters toward a high throughput and efficient laser ablation process for specific materials.
Acknowledgments
Fig. 5. Scribe depths of silicon versus pulse widths at scanning speeds of 1.2 m/s and 1.8 m/s.
This work was supported by the National Natural Science Foundation of China (Grant no. 61505240), the Jiangsu Natural Science Foundation (Grant no. BK20140393), the Jiangsu IndustryAcademia-Research Joint Innovation Foundation (Grant no. BY2014065), the Applied Basic Research Programs of Suzhou City (Grant no. SYG201414), and the Youth Innovation Promotion Association CAS (Grant no. 2015258).
References
Fig. 6. Scribe depths as a function of the scanning repeat times at 100 ns.
5. Conclusions In conclusion, a MOPA pulsed fiber laser consisting of two amplifier stages is developed and configured to emit pulses with an average power of 20 W, pulses width of 2–500 ns, and repetition frequency of 20–500 kHz, and high flexibility in pulse controllability makes it an ideal tool for laser material processing. This MOPA fiber laser is used to evaluate the effect of pulse characteristic, scanning speed, scanning repeat times on silicon ablation. SEM pictures of the scribe lines are taken and the depths are measured. It shows that the ablation depth increases as the pulse
[1] K. Li, M. Sparkes, W. O’Neill, IEEE J. Sel. Top. Quantum Electron. 20 (2014) 0900807. [2] S.T. Hendow, J. Sousa, P.T. Guerreiro, N. Schilling, J. Rabe, In: Proceedings of the 28th International Congress on Applications of Lasers and Electro-Optics, Paper M209, Orlando, Florida, USA, November 2–5, 2009. [3] Y. Zhou, B. Wu, S. Tao, A. Forsman, Y. Gao, Appl. Surf. Sci. 257 (2011) 2886–2890. [4] K. Li, W. O’Neill, Int. J. Precis. Eng. Manuf. 13 (2012) 641–648. [5] V.V. Osipov, V.V. Lisenkov, V.V. Platonov, Appl. Phys. A Mater. Sci. Process. 118 (2015) 1133–1144. [6] S. McCulloch, A. Hassey, P. Harrison, Fiber Laser performance in industrial applications, In: Proceedings of SPIE, San Francisco, California, United States, 2013, pp. 86030C-1–86030C-14. [7] H. Aman, M. Rafiq, Opt. Spectrosc. 115 (2013) 457–461. [8] S.A. Solokhin, A.A. Sirotkin, S.V. Garnov, Laser Phys. 21 (2011) 1145–1149. [9] W. O’Neill, K. Li, IEEE J. Sel. Top. Quantum Electron. 15 (2009) 462–470. [10] P.T. Guerreiro, J.M. Sousa, R. Romero, M. Silva, A. Duarte, J.R. Salcedo, MOPA fiberlaser with controlled pulse width and peak power for optimizing micromachining applications, In: Proceedings of SPIE, San Francisco, California, United States, 2010, pp. 8785417-1–8785417-6. [11] E. Williams, E.B. Brousseau, A. Rees, Int. J. Adv. Manuf. Technol. 74 (2014) 769–780. [12] P.J. Chernek, J.A. Orson, A simple thermal response model for a p-doped silicon substrate irradiated by 1.06 and 1.32 μm lasers, In: Proceedings of SPIE, Boulder, Colorado, USA, 2002, pp. 186–197. [13] P.P. Pronko, S.K. Dutta, D. Du, R.K. Singh, J. Appl. Phys. 78 (1995) 6233–6240. [14] J.F. Ready, Effects of High-Power Laser Radiation, Academic Press, New York, 1971. [15] C.J. Glassbrenner, G.A. Slack, Phys. Rev. 134 (1964) 1058–1069.
Contents lists available at ScienceDirect
Optics & Laser Technology journal homepage: www.elsevier.com/locate/optlastec
Full length article
MOPA pulsed fiber laser for silicon scribing Limei Yang, Wei Huang n, Mengmeng Deng, Feng Li Key Laboratory of Nanodevices and Applications, Suzhou Institute of Nano-Tech and Nano-Bionics, Chinese Academy of Sciences (CAS), Suzhou, China
art ic l e i nf o
a b s t r a c t
Article history: Received 1 September 2015 Received in revised form 14 December 2015 Accepted 4 January 2016
A 1064 nm master oscillator power amplifier (MOPA) pulsed fiber laser is developed with flexible control over the pulse width, repetition frequency and peak power, and it is used to investigate the dependence of mono-crystalline silicon scribe depth on the laser pulse width, scanning speed and repeat times. Experimental results indicate that long pulses with low peak powers lead to deep ablation depths. We also demonstrate that the ablation depth grows fast with the scanning repeat times at first and progressively tends to be saturated when the repeat times reach a certain level. A thermal model considering the laser pulse overlapping effect that predicts the silicon temperature variation and scribe depth is employed to verify the experimental conclusions with reasonably close agreement. These conclusions are of great benefits to the optimization of the laser material processing with high efficiency. & 2016 Elsevier Ltd. All rights reserved.
Keywords: MOPA Fiber laser Laser ablation Thermal model Silicon Surface temperature Subject classification codes: 140.3538 140.3615 140.6810 140.3390
1. Introduction Pulsed lasers find more and more micromachining applications ranging from ablation, cutting, drilling and marking. A wide array of publications have discussed the laser materials processing using a variety of lasers [1–3]. It is noted that the laser machining process is defined by the specific lasers used, and the optimization of a machining process involves the adjustment of multiple laser parameters, e.g., pulse width, average power, pulse energy and repetition frequency [4–6]. However, a variety of pulsed lasers in use, such as diode-pumped Q-switched lasers comprising an intra-cavity Q-switch element, adjusting one of the laser's parameters would invariably lead to changes in other parameters [7,8]. As a result, the improvement of the performance of laser micromachining is limited by the operation capability of these pulsed lasers. High power pulsed fiber lasers offer a compact, electrically efficient and stable alternative to conventional bulky solid-state lasers, and they are being widely used in industrial manufacturing. Among various pulsed fiber laser solutions, the MOPA approach is a particularly attractive regime, and it offers unprecedented levels of performance and flexibility. That is, the laser pulse parameters n
Corresponding author. E-mail address: [email protected] (W. Huang).
http://dx.doi.org/10.1016/j.optlastec.2016.01.006 0030-3992/& 2016 Elsevier Ltd. All rights reserved.
can be adjusted independently, while maintaining a constant beam quality [9–11]. In this study, a MOPA fiber laser at the level of nanosecond is developed, and it is used to conduct experiments that evaluate the effects of pulse width and laser scanning parameters on the ablation process of mono-crystalline silicon, while maintaining a constant average power. A theoretical thermal model is utilized to analyze the scribe depths in terms of temperature variation of silicon material, and the theoretical predictions agree well with the experimental results. It indicates that laser pulse width and scanning parameters have great influences on the material removal rate and debris accumulation.
2. Laser configuration and marking system The schematic of MOPA nanosecond fiber laser system is shown in Fig. 1. The system consists of two amplifying stages, e.g., preamplifier and power amplifier. The modulated seed source is a pulsed laser diode at 1064 nm, with power, pulse width and repetition frequency that are variable and independently controlled. Both stages are based on Yb-doped fiber amplifier (YDFA). The preamplifier stage is forward pumped by a 915 nm, 10 W laser diode coupled with Yb-doped double-clad fiber (YDDCF) using a combiner, and it provides an amplification of the seed laser with gain
68
L. Yang et al. / Optics & Laser Technology 80 (2016) 67–71
Fig. 1. Schematic of nano-second MOPA fiber laser system.
of 24 dB. A 2 2 fiber coupler, which is applied before the preamplifier stage, is for monitoring signal and back-reflection light without dismantling. The power-amplifier stage is cladding pumped by two 915 nm, 24 W laser diodes. Seed source and preamplifier stage, pre-amplifier stage and power-amplifier stage are separated with isolators which prevent the backward reflections and amplified spontaneous emission (ASE). The unabsorbed pump light from the two stages is rejected by self-made cladding-mode strippers. The system is terminated with an optically isolated fiber collimator, ensuring the reflections due to mismatch do not cause oscillations. All the optical components are fixed on an aluminum plate for heat dissipation, and the whole system is enclosed in a box. As a result, a pulsed fiber laser is configured to emit pulses with an average power of 20 W, peak power up to 10 kW, pulse width ranging from 2 to 500 ns and repetition frequency of 20 to 500 kHz. The fiber laser is arranged in a marking station, consisting of a beam expander, scanner and projection f-theta lens, for delivery of laser beam on the target surface. The diameter of focused spot on the target surface is measured to be about 60 μm, which leads to peak irradiance level of about 0.35 GW/cm2 and fluence level of about 7.1 J/cm2 for pulse of 20 ns width and 100 kHz repetition frequency. It is worth noting that the MOPA pulsed fiber laser can be operated at a wide array of pulse configurations, and specific pulses with varying pulse widths are chosen in this work, while maintaining the same pulse energy (e.g., having the same average power 20 W and repetition frequency of 100 kHz). The scribing experiments are conducted on mono-crystalline silicon. The peak irradiance levels of pulses used in the experiment are well above the corresponding ablation threshold of silicon (the peak irradiance levels for pulse width of 20 ns, 40 ns, 100 ns, and 200 ns are 0.35 GW/cm2, 0.18 GW/cm2, 70 MW/cm2, and 35 MW/cm2, which are all above the corresponding ablation threshold of 27 MW/cm2, 19.41 MW/cm2, 10 MW/cm2, and 8.68 MW/cm2, respectively).
3. Theoretical model A detailed modeling of the laser ablation process is quite complex as complicated mechanisms are involved in the ablation process. When processing with nanosecond pulse laser, the laser can be described as a heat source, and the heat equation can be applied to calculate the temperature evolution in the substrate. It is assumed that the laser radiation losses during scribing can be neglected, the heat conduction equation is approximately described by [12,13]
ρC
∂T (r , z, t ) = K ·∇2T (r , z, t ) + α (T )(1 − R) I0 (t ) e−zα (T ) ∂t
(1)
Where Κ is the thermal conductivity (K ¼ κρC), κ is the thermal diffusivity, ρ is the mass density, C is the thermal heat capacity, Τ is the temperature as a function of position (r, z) and time t, R is the Fresnel reflectivity of the surface, I0 is the time-dependent incident laser intensity, and α is the absorption coefficient. The general formulation for the temperature cannot be evaluated explicitly in Eq. (1). It is noted that the surface temperature would be well above the room temperature as the spatial overlap of the consecutive optical pulses being considered for the irradiated surface during the scanning process. Since the absorption coefficient of the silicon increases rapidly as the temperature rises [12], such that it is reasonable to assume that the absorption coefficient of the silicon is high enough to ensure most of the laser energy absorbed at the silicon surface. As a result, the temperature rise of the silicon due to absorption of an incoming Gaussian laser pulse, is given by [14] 1
∆T (r , z, t ) =
Imax γw 2 ⎛⎜ κ ⎞⎟ 2 ⎝ π⎠ K
∫0
τ
p (τ − t )
(
)
t 4κt + w 2
⎛ z2 ⎞ 2 − r ⎜− ⎟ 4κt 4κt + w 2 ⎠ dt
e⎝
(2)
Where Imax is the peak power per unit area at the center of the beam spot, w is the beam radius, τ is the pulse width; γ being equal to 1 R, is the fraction of the pulse energy that is absorbed by the material, and p(t) is the laser pulse temporal profile. Eq. (2) is used to predict the temperature distribution at the cross-section of the silicon as a function of depth z and radial distance r with respect to the center of the laser beam, considering that the radiation of a single pulse with a certain width is absorbed. During the simulation, p(t) is assumed to be of a square shape for the MOPA fiber laser, which is quite reasonable especially for pulses with long widths. Other parameters used in the simulation are listed as follows, w ¼30 μm, κ ¼0.907 10 4 m2/s, K ¼145.3 W/m K, γ ¼0.69, the boiling temperature TB ¼ 2628 K [15]. Fig. 2(a)–(d) shows the calculated temperature distribution on the cross-section of silicon with pulses widths of 20 ns, 40 ns, 100 ns and 200 ns. The temperature contour follows the Gaussian distribution, and the temperature decreases as the depth and radial distance increases. The white line in Fig. 2 represents the boiling temperature of silicon at different pulse widths. Above the white line, the temperatures are high enough to sublimate of the material and well-defined ablation pits form. It also shows that a longer pulse with lower power can ablate a deeper pit than a shorter one with higher power, which is illustrated in Fig. 3. It is quite straightforward to calculate the ablation depth z(r) for a single pulse by setting the temperature rise ΔT being equal to the boiling temperature of the silicon minus the initial temperature in
L. Yang et al. / Optics & Laser Technology 80 (2016) 67–71
69
Fig. 2. Calculated temperature distribution on the cross section of silicon for pulse with width of 20 ns (a), 40 ns (b), 100 ns (c), and 200 ns (d).
next pulse arrives, the previous ablation region has cooled down to room temperature, the effective ablation depth can be expressed as,
d (r ) = z (r + v·f ) + ⋯ + z (r + n·v·f ) + z (r ) + z (v·f − r ) + ⋯ + z (n·v·f − r )
(3)
where –worow, v is the scanning speed, f is the pulse repetition frequency, and n is the number of sequential overlapping pulses.
4. Results and discussion
Fig. 3. The estimated ablation depth versus pulse width.
Eq. (2). For the most practical applications, multiple pulses overlap spatially as the pulsed laser beam scans across the material surface. In this case, one needs to take into account the depths caused by all the overlapping pulses. Based on the assumption that and when the
Fig. 4 shows SEM images of the silicon scribe results using pulse with widths of 20 ns, 40 ns, 100 ns, and 200 ns at scanning speeds of 1.2 m/s and 1.8 m/s, respectively. All the applied pulses have constant average power of 20 W and repetition frequency of 100 kHz. As seen in Fig. 4, the cross-sections of all the scribe grooves have similar Gaussian profiles as that of delivered laser beam. All the laser pulses employed have irradiance above the threshold one for producing the plasma [14], which drives the strong shock wave. For short pulse with high irradiance, the ablation is dominated by the latent heat of vaporization with less melt. Consequently, less molten silicon particles ejected by the shock wave resolidify at the edges of the groove comparing to that of long pulse (the zoomed SEM images in Fig. 5). We measured the depths of these scribed grooves at the center
70
L. Yang et al. / Optics & Laser Technology 80 (2016) 67–71
Fig. 4. SEM images of scribed grooves for pulse width ranging from 20 ns to 200 ns at scanning speed of 1.2 m/s and 1.8 m/s.
of the beam (e.g., r¼0) using the scanning electron microscope (SEM), and calculated the corresponding theoretical values according to (Eqs. (2) and 3). The results were plotted against pulse width with inset of zoomed SEM images of ablated grooves as shown in Fig. 5. The ablation depth increases as pulse width increases from 40 ns to 200 ns at both scanning speeds of 1.2 m/s and 1.8 m/s. Longer pulses with lower peak power lead to deeper scribing depths, while shorter pulses with higher peak power have shallower depths. Physically, the dominant factor is the latent heat of vaporization when laser pulse with short duration interacts with the silicon surface, and much of the pulse energy is devoted to heat the vaporized silicon to a higher temperature (i.e., producing plasma), and distribution of the heat to a large volume of silicon material is limited by the short pulse length. Whereas for pulse with longer duration and low peak power, the amount vaporized depends more on the thermal conductivity of silicon. The heat has a better chance to flow into the interior of the silicon, which lead to a deep ablation depth [14]. An overall good agreement is found between the theoretical predictions and experimental results, despite small discrepancies at pulse widths of 40 ns and 200 ns, which might be
attributed to lack of consideration of the dependence of some crucial parameters on the temperature during simulation, e.g., silicon absorption coefficient, conductivity and diffusivity [15], which are regarded as constants during the calculations. The error of the measurement data comes from the surface roughness at the bottom of the trench. Indeed, a single scan cannot meet the industrial demands, and a repeated scanning process is always necessary to achieve a deep penetration depth. Fig. 6 shows the silicon scribe depths as a function of repeat times at 100 ns. For both scanning speeds of 1.2 m/s and 1.8 m/s, the scribe depth increases with the repeat times, but it tends to be saturated when the repeat times reach 25. One of the possible reasons is the decrease of laser energy at the ablation region as the depth goes deeper. Physically, laser beam defocuses and laser power density decreases due to the divergence as it propagates into the deeper position of silicon, and this reduces the ablation rate. Moreover, EDX measurement results show that the laser-irradiated silicon oxidizes during the ablation process, which indicates that oxidation of silicon is probably another possible reason for the saturation effect as the repeat times increase.
L. Yang et al. / Optics & Laser Technology 80 (2016) 67–71
71
width increases from 20 ns to 200 ns at both scanning speeds of 1.2 m/s and 1.8 m/s. A theoretical ablation model is used to simulate the silicon temperature distribution and the ablation depth. Simulation results indicate pulse with shorter width leads to a higher temperature, shallower depth, and less melting and recasting of the material, which shows a good match to the experimental conclusions. Moreover, experimental results indicate it is impracticable to increase a deep ablation depth by increasing the repeat times all the time, as the ablation depth progressively grows slow with the repeat times as it reaches a certain level. The conclusions and analytical method demonstrated in the paper provide strategic guidance to optimize the laser parameters and scanning parameters toward a high throughput and efficient laser ablation process for specific materials.
Acknowledgments
Fig. 5. Scribe depths of silicon versus pulse widths at scanning speeds of 1.2 m/s and 1.8 m/s.
This work was supported by the National Natural Science Foundation of China (Grant no. 61505240), the Jiangsu Natural Science Foundation (Grant no. BK20140393), the Jiangsu IndustryAcademia-Research Joint Innovation Foundation (Grant no. BY2014065), the Applied Basic Research Programs of Suzhou City (Grant no. SYG201414), and the Youth Innovation Promotion Association CAS (Grant no. 2015258).
References
Fig. 6. Scribe depths as a function of the scanning repeat times at 100 ns.
5. Conclusions In conclusion, a MOPA pulsed fiber laser consisting of two amplifier stages is developed and configured to emit pulses with an average power of 20 W, pulses width of 2–500 ns, and repetition frequency of 20–500 kHz, and high flexibility in pulse controllability makes it an ideal tool for laser material processing. This MOPA fiber laser is used to evaluate the effect of pulse characteristic, scanning speed, scanning repeat times on silicon ablation. SEM pictures of the scribe lines are taken and the depths are measured. It shows that the ablation depth increases as the pulse
[1] K. Li, M. Sparkes, W. O’Neill, IEEE J. Sel. Top. Quantum Electron. 20 (2014) 0900807. [2] S.T. Hendow, J. Sousa, P.T. Guerreiro, N. Schilling, J. Rabe, In: Proceedings of the 28th International Congress on Applications of Lasers and Electro-Optics, Paper M209, Orlando, Florida, USA, November 2–5, 2009. [3] Y. Zhou, B. Wu, S. Tao, A. Forsman, Y. Gao, Appl. Surf. Sci. 257 (2011) 2886–2890. [4] K. Li, W. O’Neill, Int. J. Precis. Eng. Manuf. 13 (2012) 641–648. [5] V.V. Osipov, V.V. Lisenkov, V.V. Platonov, Appl. Phys. A Mater. Sci. Process. 118 (2015) 1133–1144. [6] S. McCulloch, A. Hassey, P. Harrison, Fiber Laser performance in industrial applications, In: Proceedings of SPIE, San Francisco, California, United States, 2013, pp. 86030C-1–86030C-14. [7] H. Aman, M. Rafiq, Opt. Spectrosc. 115 (2013) 457–461. [8] S.A. Solokhin, A.A. Sirotkin, S.V. Garnov, Laser Phys. 21 (2011) 1145–1149. [9] W. O’Neill, K. Li, IEEE J. Sel. Top. Quantum Electron. 15 (2009) 462–470. [10] P.T. Guerreiro, J.M. Sousa, R. Romero, M. Silva, A. Duarte, J.R. Salcedo, MOPA fiberlaser with controlled pulse width and peak power for optimizing micromachining applications, In: Proceedings of SPIE, San Francisco, California, United States, 2010, pp. 8785417-1–8785417-6. [11] E. Williams, E.B. Brousseau, A. Rees, Int. J. Adv. Manuf. Technol. 74 (2014) 769–780. [12] P.J. Chernek, J.A. Orson, A simple thermal response model for a p-doped silicon substrate irradiated by 1.06 and 1.32 μm lasers, In: Proceedings of SPIE, Boulder, Colorado, USA, 2002, pp. 186–197. [13] P.P. Pronko, S.K. Dutta, D. Du, R.K. Singh, J. Appl. Phys. 78 (1995) 6233–6240. [14] J.F. Ready, Effects of High-Power Laser Radiation, Academic Press, New York, 1971. [15] C.J. Glassbrenner, G.A. Slack, Phys. Rev. 134 (1964) 1058–1069.