Open Access
Add to BibSonomyAbstract
The effect of the spot diameter in nanosecond excimer laser percussion drilling of through via in silicon wafer is presented. Experimental results show a ten fold increase of the ablation efficiency when decreasing the spot diameter from 220 μm to 9 μm at constant fluence in the range 7.5 J/cm2 to 13.2 J/cm2. Such effect is absent when using 60 ps deep-UV laser pulses. A model is developed that explain the findings in terms of plume shielding effect on the laser pulse. The model is successfully applied also on previously published data on deep-UV laser drilling of Polyimide and Alumina.
© 2010 Optical Society of America
1. Introduction
Since the development of high power lasers, much research has been done in order to investigate the laser-solid interaction. Nowadays, the use of high power lasers in micromachining is well established due to both the availability of reliable industrial grade lasers and the good theoretical knowledge of the physical processes involved. Besides accurate micromachining of polymers with deep-UV excimer laser, already established since decades, the interest is growing in the use of lasers to drill, dice and in general process semiconductors, in particular silicon, in view the flexibility and accuracy of laser micromachining.
Laser percussion drilling of through micro-vias in silicon substrates is suitable for the production of 3D stacked electronic and micro-electro-mechanical chips [1,2], as well as for the development of metal-wrap-through and emitter-wrap-through technologies to increase the solar cell efficiency [3]. Through vias in silicon can be implemented also in the recently reported Silicon-on-Diamond chips [4] to produce the electrical interconnection between the diamond and the silicon substrates. Forseen applications for these devices are in particles detectors for hash environment, as well as in the development of novel bio-chip with improved performances [5].
A lot of research has been done in order to characterize the drilling of silicon with respect to efficiency and quality as function of the laser parameters such wavelength, pulse time duration, pulse repetition rate, and pulse fluence [6–9]. However, systematic study on the spot size effect in percussion laser drilling of silicon is missing. Such information is of great value in order to optimize the laser parameters with respect of the diameter of the drilled via.
In this article a systematic investigation of the laser percussion drilling efficency of silicon as function of the spot diameter is presented. The striking result is a ten fold increase in the drilling efficiency at costant fluence when decreasing the spot diameter from 220 μm to 9 μm. A model based on the shielding effect of the ablation plume on the laser pulse is developed resulting in an analytical formula that very well explain the experimental findings.
Similar investigations have been reported for other materials [10–12], and the developed model is validated also against published data for Polyimide (PI) and Alumina.
2. Experiments
Percussion drilling of a 50 μm thick Si wafer is performed in air with a 20 ns full-width at half-maximum (FWHM) KrF excimer laser (Coherent-CompexPro 110) at 248 nm, and a 60 ps FWHM frequency quadrupled Nd:YAG laser (Continuum-Leopard SS-20-SV) at 266 nm, both running at 20 Hz. The lasers are alternatively coupled with a micromachining workstation (Optec-MicroMaster). The sample is irradiated using the projection mask technique with a 0.1 numerical aperture projection lens set at a demagnification of 12 with optical resolution of 1.5 μm. A single hole is drilled at the time, and different spot diameters in the range from 9 μm to 220 μm are irradiated at the target using a stainless steel stencil mask with circular apertures of diameters from 108 μm to 2.64 mm. The correct positioning of the target in the image plane is achieved by properly set the lens position using a through-the-lens vision system based on the what-you-see-is-what-you-get technique. This is possible since the projection lens is a special multiplet corrected for chromatic aberration both in the working wavelength range and in the red part of the visible spectrum.
The Excimer laser pulse on the mask has energy up to 350 mJ and a rectangular beam profile with flat-top distribution on the long axis (FWHM of 30 mm) and a gaussian distribution on the short axis (FWHM of 12 mm). The ps laser is equipped with a serrated aperture system to produce flat top profile of the fundamental beam, and the fourth-harmonic beam on the mask has energy up to 15 mJ with a diameter of 7 mm. The unifrom irradiation of the spot on the silicon target is routinely checked using a beam profiling routine of the micromachining system. The fluence distribution on the target plane is also checked after each mask change by visual inspection of the ablation pattern created on a polymeric substarted irradiated with a fluence just above ablation thereshold. Only when using the ps laser with mask size bigger than 1.3 mm the illumination on the part is not uniform due to inhomogeneity of the laser beam profile on the mask. The fluence on the silicon target during drilling experiment is estimated to be uniform within 5%.
The shot dose (SD), i.e., the number of shots necessary to drill a through via, is measured for fluence on the silicon target in the range 7.5 J/cm2 to 13.2 J/cm2. The fluence on the silicon target is controlled by changing the laser pulse energy on the mask using a motorized and calibrated variable attenuator present within the micromaching system. The Si wafer is suspended few hundred μm on top a microscope glass slide and the SD is evaluated by monitoring, either by eye or using a camera, the fluorescence from the glass slide produced when the high energy laser pulse is transmitted through the hole. In Fig. 1 the experimental results are shown. The data points reported are the average over at least three drilling experiments, while the error bars are the standard deviation. It is evident a large dependence of the SD on the spot size for a fixed laser fluence. The SD decreases by one order of magnitude when the spot diameter is reduced from 200 μm to 9 μm, with a sharp decrease below 100 μm.
Fig. 1 Shot dose as function of spot diameter for different laser fluence of the 20 ns pulses. The dashed lines are the result of the fitting with the model function. The inset shows the results obtained with 60 ps pulses, and the dashed line represent the average value.
Typical SEM images of the produced through vias are shown in Fig. 2. Scattered debris and a recast layer are evident around the entrance hole, which are typical features of the thermal ablation process in Si, and reduce the entrance hole diameter by few microns compared to the irradiated spot diameter. In contrast, the output hole is clear from significant debris or recast.
Fig. 2 SEM images of drilled through holes: (a) top view, spot size 20 μm, fluence of 11 J/cm2; (b) bottom view, spot size 20 μm, fluence of 11 J/cm2; (c) top view, spot size 40 μm, fluence of 11 J/cm2; (d) top view, spot size 70 μm, fluence of 11 J/cm2.
The taper angle is found to be small, up to about 5 degrees for the smallest holes. Only when drilling a 9 μm via at the lowest fluence of 7.5 J/cm2, the exit hole diameter is below 1 μm and the SD increases above 80, because the ablated material is not efficiently removed from the bottom of the hole.
When using deep-UV ps pluses the spot size effect is not observed. In the inset of Fig. 1 the SD for a fluence of 5.4 J/cm2 is presented, and it is found to be constant over the spot size range from 110 μm to 20 μm.
3. Model
The experimental results obtained for nanosecond laser drilling can be qualitatively explained in terms of the shielding of the incoming laser pulse by the ablation plume. The ablation plume is expanding with a certain angle distribution over the surface of the target and only the central portion of the plume will interact with the nanosecond laser pulse. In general, the smaller the ablated spot size the smaller the relative part of the ablated material that falls within the laser beam path, thus the smaller the averall shielding effect. The plume expands with ns time scale [13]. With tens of picosecond long pulses this effect is minimized if not absent, in accordance with the findings for picosecond drilling shown in the inset of Fig. 1.
Starting from these general arguments a model is derived to quantitatively explain the experimental results obtained for nanosecond laser drilling. The model is based on few assumptions, that are elucidated in Fig. 3, and presented here after:
- The plume is expanding with a fixed aperture angle β which is only dependent upon the laser fluence and is independent on the spot diameter d; this assumption is based on experimental findings that show a cos4 distribution of the plume shortly after its creation [14].
- During the laser pulse the plume will expand up to a hight h, that is assumed to be independent from the spot size but dependent on the laser fluence.
- The beam is assumed to be collimated orthogonally to the target surface. This assumption is justified by the use of a low numerical apeture lens of 87 mm focal length, and by a few mm spot size on the projection lens, resulting in a few hundred micron depth of focus in the image plane.
The shot dose is given by SD = L/T, where L is the sample thickness, and combining Eqs. (1)–(4) it can be expressed as
where is the ablation efficiency without plume shielding effect, , and G = 2h tan(β) is a geometrical coefficient related to the plume expansion.4. Discussion
The curves resulting from the least-square fit on the experimental data using Eq. (5) are reported in Fig. 1, and very well follow the data points over the entire spot diameter range.
In general the thermal ablation efficiency is related to the thermal diffusion length , where D is the thermal diffusion coefficient, and τ is the pulse time duration. For silicon D = 0.85 cm2/s, resulting in a thermal diffusion length of 2.6 μm and 0.14 μm for 20ns and 60 ps pulses respectively. For 20 ns pulses the ablation efficiency is found to approach the thermal diffusion length only for the smaller spot diameter and at fluence above 10 J/cm2. For 60 ps pulses an ablation efficiency of about 0.6 μm is found, which is probably due to the onset of explosive boiling ablation regime at the very high peak intensity of 90 GW/cm2 [15].
In Fig. 4 the ablation efficiency without plume shielding, T0, is shown, along with experimental values for spot diameter of 9 μm and 70 μm reported for comparison. Fitting T0 with a logarithmic function leads to the values K =7.9 μm and Ft = 6 J/cm2. The value of K is larger than the thermal diffusion length, and the ablation threshold Ft is higher than the typical value for vaporization ablation threshold of silicon with deep-UV ns pulsed that is between 1–2 J/cm2 [14]. This indicates that without plume shielding a more efficient ablation mechanism may take place in the fluence range used in the experiments.
Using 3 ns pulse at 266 nm with a spot of 50 μm it has indeed been observed a sharp increase of the ablation rate of silicon at a fluence of 60 J/cm2 [15], that has been attributed to the onset of explosive boiling of the super-heated liquid. Calculation of threshold irradiance for explosive boiling in silicon results in a value of about 9 J/cm2, and the discrepancy with the experimental value is ascribed to absorption/scattering of the laser pulse by the ablation plume [16]. This calculated value is compatible with the fluence range used in the present experiments. Thus the large values of K = and Ft found from the present model can be ascribed to the onset of explosive boiling, which is hampered by plume shielding during actual drilling. This explanation is supported also by finding of large laser plume interaction during silicon ablation with deep-UV ns pulses starting at fluence of 5.8 J/cm2 as reported in [14].
The phenomenological coefficient α is in the range 8–10×103 cm2/g. Its interpretation is strongly dependent on the plume composition during its interaction with the laser pulse. In general a plasma is present in the plume, and electron-ion as well as electron-neutral inverse bremsstrahlung gives the major contribution to absorption. However, in thermal ablation there are also clusters and large debris that may absorb and scatter laser light.
The geometrical coefficient G resulting from the fit is in the range 35 μm to 129 μm increasing with decreasing fluence. Assuming an angular plume distribution of cos4 and taking as β the angle at half-maximum, i.e., tan(β) = 0.64, the resulting value of h is between 27 μm to 101 μm. Thus, for a laser pulse duration of 20 ns, the average plume expansion velocity is estimated between 103–104 m/s, well matching values reported in the literature [14].
The developed model is also applied to already published data on percussion drilling of through holes using excimer lasers with the mask projection technique, as shown in Fig. 5. The experiments with PI and Alumina have been conducted on 125 μm and 650 μm thick samples respectively, while in Fig. 5 the ablation efficiency data are plotted in SD to drill a 50 μm film for better comparison. Again the fitting of the data point with the model function given in Eq. (5) is very good. In the case of PI, see Fig. 5(a), the coefficients T0 obtained from the two curves results in KPI = 0.26 μm and Ft,PI = 50 mJ/cm2. The geometrical coefficient is about 65 μm, again corresponding to a velocity of about 103–104 m/s. All these values are in agreement with data on PI ablation reported in the literature [10].
Fig. 5 Analytical model applied to data on percussion drilling with Excimer laser reported in the literature: (a) Polyimide [10]; (b) Alumina [11]. The data are reported as shot dose to drill a 50 μm thick sample for better comparison.
5. Conclusion
In conclusions, it is shown experimentally that the spot size strongly influence the ablation efficiency in silicon when drilling holes with diameter smaller than 200 μm using ns laser pulses. These findings are explained in terms of plume shielding effect, and a model is developed that very well fits the presented experimental data as well as data reported in the literature for percussion drilling of other materials.
The presented results indicate that the spot size should be taken into great account when performing laser percussion drilling in silicon with ns pulses.
References and links
1. Y. H. Lee and K. J. Choi, “Analysis of silicon via hole drilling for wafer level chip stacking by UV laser,” Int. J. Precis. Eng. Manuf. 11501–507 (2010) [CrossRef]
2. C. W. Lin, H. A. Yang, W. C. Wang, and W. Fang, “Implementation of three-dimesional SOI-MEMS wafer-level packaging using through-wafer interconnections,” J. Micromech. Microeng. 171200–1205 (2007) [CrossRef]
3. K. P. Stolberg, B. Kremser, S. Friedel, and Y. Atsuta, “Systematic optimization of process parameter in laser drilling of 200 μm photovoltaic silicon wafer using new kind of nanosecond IR laser,” J. Laser Micro/Nanoeng. 4, 231–233 (2009). [CrossRef]
4. S. Lagomarsino, G. Parrini, S. Sciortino, M. Santoro, M. Citroni, M. Vannoni, A. Fossati, F. Gorelli, G. Molesini, and A. Scorzoni, “Silicon-on-diamond material by pulsed laser technique,” Appl. Phys. Lett. 96, 031901 (2010). [CrossRef]
5. S. Lagomarsino, G. Parrini, S. Sciortino, A. Fossati, M. Citroni, G. Ferrari, F. Gorelli, M. Santoro, G. Molesini, M. Vannoni, A. Marras, A. Scorzoni, A. Ranieri, L. Berdondini, F. Brandi, R. Carzino, A. Diaspro, M. Scotto, and B. Torre, “New prospectives for the silicon-on-diamond material,” Proceedings of Science RD09, 029 (2009).
6. B. Tan and K. Venkatakrishnan, “Nd-YAG laser microvia drilling for interconnection application,” J. Micromech. Microeng. 17, 1511–1517 (2007). [CrossRef]
7. B. Tan, S. Panchatsharam, and K. Venkatakrishnan, “High repetition rate femtosecond laser forming sub-10μm diameter interconnection vias,” J. Phys. D: Appl. Phys. 42, 065102 (2009). [CrossRef]
8. B. Tan, “Deep micro hole drilling in a silicon substrate using multi-bursts of nanosecond UV laser pulses,” J. Micromech. Microeng. 16, 109–112 (2006). [CrossRef]
9. S. T. Hendow and S. A. Shakir, “Structuring materials with nanosecond laser pulses,” Opt. Express 18, 10188–10199 (2010). [CrossRef] [PubMed]
10. H. Schmidt, J. Ihlemann, B. Wolff-Rottke, K. Luther, and J. Troe, “Ultraviolet laser ablation of polymers: spot size, pulse duration, and plume attenuation effects explained,” J. Appl. Phys. 83, 5458–5468 (1998). [CrossRef]
11. “High precision drilling of ceramics,” Highlights LambdaPhysik 34, 2–5 (1992).
12. M. Eyett and D. Bauerle, “Influence of beam spot size on ablation rates in pulsed-laser processing,” Appl. Phys. Lett. 51, 2054–2055 (1987). [CrossRef]
13. X. Zeng, X. L. Mao, R. Greif, and R. E. Russo, “Experimental investigation of ablation efficiency and plasma expansion during femtosecond and nanosecond laser ablation of silicon,” Appl. Phys. A 80, 237–241 (2005). [CrossRef]
14. Y. F. Lu, M. H. Hong, and T. S. Low, “Laser plasma interaction at an early stage of laser ablation,” J. Appl. Phys. 85, 2899–2903 (1999). [CrossRef]
15. Q. Lu, S. S. Mao, X. Mao, and E. Russo, “Delayed phase explosion during high-power nanosecond laser ablation of silicon,” Appl. Phys. Lett. 80, 3072–3074 (2002). [CrossRef]
16. J. H. Yoo, S. H. Jeong, R. Greif, and E. Russo, “Explosive change in crater properties during high power nanosecond laser ablation of silicon,” J. Appl. Phys. 88, 1638–1649 (2000). [CrossRef]
