- Email: [email protected]
Ultrafast-laser-assisted chemical restructuring of silicon and germanium surfaces
Applied Surface Science 253 (2007) 6580–6583 www.elsevier.com/locate/apsusc
Ultrafast-laser-assisted chemical restructuring of silicon and germanium surfaces Barada K. Nayak a, Mool C. Gupta a,*, Kurt W. Kolasinski b a
Department of Electrical & Computer Engineering, University of Virginia, Charlottesville, VA 22904, United States b Department of Chemistry, West Chester University, West Chester, PA 19383, United States Available online 30 January 2007
Abstract This article reports a comparative study on texturing in silicon and germanium surfaces after exposure to femtosecond laser irradiation in the gaseous environments of sulfur hexafluoride (SF6) and hydrogen chloride (HCl). The surface texturing results from the combined effect of laserassisted chemical etching and laser ablation. Optimized processing conditions have produced features on the order of nanometers in size. We demonstrate for the first time that regular conical pillars can be formed in Ge and that HCl can be used to form regular conical pillars in Si. # 2007 Elsevier B.V. All rights reserved. Keywords: Femtosecond laser; Surface texturing; Ultrafast lasers; Chemical restructuring; Silicon and germanium surfaces
1. Introduction Silicon and germanium are two of the most important electronic materials used in the semiconductor industry. Restructuring the surface of silicon has been an active area of research due to several reasons including: (1) it enhances the surface area and hence the effective active area of a device increases, (2) it can effectively trap more light for fabrication of high efficiency optoelectronic devices, (3) it increases catalytic actions due to large surface area, and (4) it can have possible biomedical applications. Surface texturing by porosification of silicon has been extensively studied over decades and recently ultrafast laserassisted surface texturing [1–8] has become progressively more popular due to several advantages: (1) unlike porous structures, regular sharp pillars can be formed in which the entire surface area is exposed as exterior surface, (2) the structural, optical and electronic properties can be well tailored by controlling the processing conditions, and (3) large area samples can be processed with automation. Nonetheless, the formation of regular conical pillars has only been demonstrated previously in silicon. Furthermore, while it is known unambiguously that the
* Corresponding author. Tel.: +1 434 924 6167; fax: +1 434 924 8818. E-mail address: [email protected] (M.C. Gupta). 0169-4332/$ – see front matter # 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.apsusc.2007.01.079
initial stages (i.e. during the first few laser shots) of surface texturing exhibit virtually no dependence on the presence and nature of the process gas, it is equally well known that the ultimate structures formed after hundreds of laser shots depend sensitively on the presence and nature (i.e. comparing vacuum to air to He to SF6) of the process gas. The specific role of the chemical interactions that lead to structure formation are shrouded in mystery. In this study we report the laser texturing of silicon and germanium surfaces and compare their surface morphologies. 2. Experimental procedure The Si(1 0 0) (B doped) and undoped Ge(1 0 0) wafers were cleaved into small chips and cleaned with acetone and methanol. Silicon and germanium chips were put successively on a stage inside a vacuum chamber (with base pressure 1 mbar) mounted on a high precision computer controlled X– Y stage. The chamber was backfilled with 400 mbar SF6 or HCl after rinsing several times with the process gas. The samples were then exposed to 1.4 mJ pulses of 800 nm/130 fs light at a repetition rate of 1 kHz from a regeneratively amplified Spectra Physics Ti-sapphire laser system. The laser beam was focused by a 1 m focal length coated lens and incident normal to the sample surface. The laser fluence was controlled by using a calcite Glan-laser polarizer.
B.K. Nayak et al. / Applied Surface Science 253 (2007) 6580–6583
6581
The spatial profile of the laser pulse was nearly Gaussian, though elongated in one axis compared to the other creating an elliptical profile, and the fluence was calculated using the spot size determined by exposing a small area on the sample surface to thousands of shots. In order to scan an area bigger than the laser spot size, the samples were translated using a motorized X–Y stage. Scanning also assists to make more uniform surface structures by smoothing out any shot-to-shot irregularities in the beam profile. By varying the scanning speed of the X–Y stage, the number of laser pulses impinging on the sample surface at a particular spot was controlled. The spot size was 0.3 mm along the minor axis and 0.6 mm along the major axis. Scanning was performed parallel to the minor axis. Samples were produced either with isolated single line scans or with large areas created by overlapping several line scans. The step size between scan lines was chosen to be sufficiently small (generally 0.38 mm) such that successive lines overlapped substantially, improving homogeneity. Homogeneity is further enhanced by performing two overlapping scans in orthogonal directions rather than one overlapping scan with an exposure of the same total number of shots. After laser processing, the samples were analyzed with a field emission scanning electron microscope (Zeiss SUPRA 40). 3. Results and discussion Fig. 1. shows the SEM images of sharp conical pillars formed on the silicon surface produced with 240 shots of 130 fs laser pulses at a fluence of 0.6 J cm 2 in 400 mbar of SF6 (Fig. 1a) and HCl (Fig. 1b), viewed 458 from the surface normal. The pillar size varies across the scanned laser line indicating the intensity variation of the laser fluence across the spatial profile of the laser pulse. We overlapped the scan lines to minimize height variation over a large area on the sample surface. The spikes that are obtained in SF6 are 15 mm tall and around 5 mm at the base. On the other hand, under similar conditions, the structures formed in the presence of HCl are composed of taller pillars ( 20 mm height) with a base dimension around 7 mm. It is evident from Fig. 1a and b that the pillars which are formed in SF6 are narrower towards the tips compared to pillars formed in HCl. Although the pillar density is comparable in both cases, the pillars formed in HCl look stronger and relatively blunt towards the tip indicating that differences in the etch chemistry in the F/Si system relative to that of Cl/Si system are responsible for significant structural differences. The role of chemistry in structure formation is not well understood and it is not a trivial result that regular conical pillars are formed during fs irradiation in SF6 as well as HCl. For example, regular conical pillars can be formed during ns irradiation of Si in the presence of SF6 but they are not formed for ns irradiation in the presence of HCl [9]. During ns pulsed irradiation of Si, HCl produces significantly blunter tips, the pillars are much taller (>50 mm) and they are porous rather than solid core [9]. This is a further indication that the mechanisms of pillar formation are not the same for fs and ns irradiation. This result is also significant because chemical impurities, namely sulfur,
Fig. 1. SEM images of pillars, viewed 458 from the surface normal, formed in silicon surface by 240 laser pulses of 130 fs duration, 0.6 J cm 2 fluence in the gaseous environment of 400 mbar of (a) SF6 and (b) HCl.
incorporated during laser processing have been implicated in changing the optical and electronic properties of the textured surfaces [10]. By expanding the range of gases in which pillar formation is possible, we should be able to disentangle the effects on optical and electronic properties of geometric parameters from chemical parameters. Fig. 2 displays SEM images taken for germanium samples treated under similar laser and gaseous conditions as in the case of silicon described above. A very different surface texturing was observed in germanium as is clear from Fig. 2. In the case of germanium processed in SF6, we noticed two distinct features not observed under conditions that are similar to those used for silicon: (a) conical pillar formation with higher cone angle up to a neck and (b) atop these pillars very sharp spikes up to 2 mm in length with 100 nm tip radius. Nanoclusters formed during laser ablation are more abundant in germanium compared to silicon under similar laser conditions, which might be due to the lower melting temperature of the former. On the other hand, the germanium structures formed in the presence of HCl (Fig. 1b) are less densely populated, shorter, high cone angle, wider base, and exhibit greater variations in pillar height compared to silicon (Fig. 1a). It is interesting to note that gas phase silicon etching is anisotropic in HCl and the etch rate is 20 mm min 1 for the Si(1 0 0) plane at temperatures in the
6582
B.K. Nayak et al. / Applied Surface Science 253 (2007) 6580–6583
and character of the pillars formed in silicon have been observed to depend critically on the temporal pulse width of the laser beam. Whereas pillar production with pulse durations in the range of 100 fs - 20 ns has been demonstrated with much smaller pillars being formed in the case of fs irradiation [10,13], we were unable to produce pillars with a 300 ns Nd:VO4 laser. It appears that pulses significantly shorter than the melt lifetime ( 250 ns) are a requisite characteristic of the laser for sharp pillar formation. 4. Conclusion
Fig. 2. SEM images of pillars, viewed 458 from the surface normal, formed in germanium surface by 240 laser pulses of 130 fs duration, 0.6 J cm 2 fluence in the gaseous environment of 400 mbar of (a) SF6 and (b) HCl.
range of 1050–1250 8C [11], whereas in germanium the etch rate is around 3 mm min 1, the etch rate is independent of temperature beyond 800 8C, and it leads to the development of square pits for Ge(1 0 0) [12]. The associated differences in etch chemistry are involved in determining the different structure formation dynamics that result in more regular spike formation in silicon compared to germanium in HCl. The details of how these differences affect structure formation will form the basis of future investigations. Laser induced surface texturing that produces regular structures on materials in the presence of reactive species is a complex process involving: (1) plume/process gas and plume/ surface interactions (decomposition of reactive gas, chemical etching enhanced by laser heating of the material and the formation of radicals, deposition out of the plume, etc.), (2) laser ablation of material, and (3) optical phenomena that initiate the formation of regular structures. The laser parameters, gaseous environment and the nature of the substrate, all play a role in the final outcome of the surface morphology. Structure formation is ultimately a convoluted dynamical process involving all three factors. The dimensions
We have for the first time demonstrated ultrafast laser induced pillar formation in Ge using SF6 and HCl gaseous environments. We have also demonstrated that ultrafast laser induced pillar formation is not limited to SF6 but other gases such as HCl can be used. In this article we have also reported a comparative study of ultrafast laser induced pillar formation between Ge and Si. Based on our observations, we make the following points regarding structure formation in silicon and germanium: (a) under similar laser and gas conditions the aspect ratios (height to base ratio) of silicon pillars is higher compared to germanium in both SF6 and HCl environments, (b) there is a spike formation atop pillars that occurs in germanium, whereas a bulb like structure forms atop silicon pillars when processed in SF6, (c) silicon pillars formed in HCl are taller and sturdier compared to SF6, (d) in germanium the structures formed in SF6 are more regular, sharp, well defined features compared to structures formed in HCl, (e) for the formation of regular conical pillars, the pulse length of the laser must be significantly shorter than the melt lifetime. Acknowledgements We gratefully acknowledge the financial support of the National Science Foundation under grant ECS-0100243, NSF I/ UCRC center grant, IGERT Program Grant 9972790 and CREST supplement Program Grant 0520208. This work was also supported by the University of Virginia and West Chester University. References [1] T.-H. Her, R.J. Finlay, C. Wu, S. Deliwala, E. Mazur, Appl. Phys. Lett. 73 (1998) 1673. [2] A.J. Pedraza, J.D. Fowlkes, D.H. Lowndes, Appl. Phys. Lett. 74 (1999) 2322. [3] T.-H. Her, R.J. Finlay, C. Wu, E. Mazur, Appl. Phys. A 70 (2000) 388. [4] J. Bonse, S. Baudach, J. Kruger, W. Kautek, M. Lenzner, Appl. Phys. A 74 (2002) 19. [5] M.Y. Shen, C.H. Crouch, J.E. Carey, R. Younkin, E. Mazur, M. Sheehy, C.M. Friend, Appl. Phys. Lett. 82 (2003) 1715. [6] D. Riedel, J.L. Hernandez-Pozos, R.E. Palmer, K.W. Kolasinski, Appl. Phys. A 78 (2004) 381. [7] B.K. Nayak, M.C. Gupta, Mater. Res. Soc. Symp. Proc. 850 (2005) (MM1.8.1). [8] D. Mills, K.W. Kolasinski, J. Phys. D: Appl. Phys. 38 (2005) 1.
B.K. Nayak et al. / Applied Surface Science 253 (2007) 6580–6583 [9] K.W. Kolasinski, D. Mills, M. Nahidi, J. Vac. Sci. Technol. A 24 (2006) 1474. [10] C.H. Crouch, J.E. Carey, J.M. Warrender, M.J. Aziz, E. Mazur, F.Y. Ge´nin, Appl. Phys. Lett. 2004 (1850) 84. [11] M. Druminski, R. Gessner, J. Cryst. Growth 31 (1975) 312.
6583
[12] J.A. Amick, E.A. Roth, H. Gossenberger, RCA Rev. 24 (1963) 473. [13] K.W. Kolasinski, in: S.G. Pandalai (Ed.), Recent Research Advances in Applied Physics, vol. 7, Transworld Research Network, Kerala, India, 2004, p. 267.
Ultrafast-laser-assisted chemical restructuring of silicon and germanium surfaces Barada K. Nayak a, Mool C. Gupta a,*, Kurt W. Kolasinski b a
Department of Electrical & Computer Engineering, University of Virginia, Charlottesville, VA 22904, United States b Department of Chemistry, West Chester University, West Chester, PA 19383, United States Available online 30 January 2007
Abstract This article reports a comparative study on texturing in silicon and germanium surfaces after exposure to femtosecond laser irradiation in the gaseous environments of sulfur hexafluoride (SF6) and hydrogen chloride (HCl). The surface texturing results from the combined effect of laserassisted chemical etching and laser ablation. Optimized processing conditions have produced features on the order of nanometers in size. We demonstrate for the first time that regular conical pillars can be formed in Ge and that HCl can be used to form regular conical pillars in Si. # 2007 Elsevier B.V. All rights reserved. Keywords: Femtosecond laser; Surface texturing; Ultrafast lasers; Chemical restructuring; Silicon and germanium surfaces
1. Introduction Silicon and germanium are two of the most important electronic materials used in the semiconductor industry. Restructuring the surface of silicon has been an active area of research due to several reasons including: (1) it enhances the surface area and hence the effective active area of a device increases, (2) it can effectively trap more light for fabrication of high efficiency optoelectronic devices, (3) it increases catalytic actions due to large surface area, and (4) it can have possible biomedical applications. Surface texturing by porosification of silicon has been extensively studied over decades and recently ultrafast laserassisted surface texturing [1–8] has become progressively more popular due to several advantages: (1) unlike porous structures, regular sharp pillars can be formed in which the entire surface area is exposed as exterior surface, (2) the structural, optical and electronic properties can be well tailored by controlling the processing conditions, and (3) large area samples can be processed with automation. Nonetheless, the formation of regular conical pillars has only been demonstrated previously in silicon. Furthermore, while it is known unambiguously that the
* Corresponding author. Tel.: +1 434 924 6167; fax: +1 434 924 8818. E-mail address: [email protected] (M.C. Gupta). 0169-4332/$ – see front matter # 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.apsusc.2007.01.079
initial stages (i.e. during the first few laser shots) of surface texturing exhibit virtually no dependence on the presence and nature of the process gas, it is equally well known that the ultimate structures formed after hundreds of laser shots depend sensitively on the presence and nature (i.e. comparing vacuum to air to He to SF6) of the process gas. The specific role of the chemical interactions that lead to structure formation are shrouded in mystery. In this study we report the laser texturing of silicon and germanium surfaces and compare their surface morphologies. 2. Experimental procedure The Si(1 0 0) (B doped) and undoped Ge(1 0 0) wafers were cleaved into small chips and cleaned with acetone and methanol. Silicon and germanium chips were put successively on a stage inside a vacuum chamber (with base pressure 1 mbar) mounted on a high precision computer controlled X– Y stage. The chamber was backfilled with 400 mbar SF6 or HCl after rinsing several times with the process gas. The samples were then exposed to 1.4 mJ pulses of 800 nm/130 fs light at a repetition rate of 1 kHz from a regeneratively amplified Spectra Physics Ti-sapphire laser system. The laser beam was focused by a 1 m focal length coated lens and incident normal to the sample surface. The laser fluence was controlled by using a calcite Glan-laser polarizer.
B.K. Nayak et al. / Applied Surface Science 253 (2007) 6580–6583
6581
The spatial profile of the laser pulse was nearly Gaussian, though elongated in one axis compared to the other creating an elliptical profile, and the fluence was calculated using the spot size determined by exposing a small area on the sample surface to thousands of shots. In order to scan an area bigger than the laser spot size, the samples were translated using a motorized X–Y stage. Scanning also assists to make more uniform surface structures by smoothing out any shot-to-shot irregularities in the beam profile. By varying the scanning speed of the X–Y stage, the number of laser pulses impinging on the sample surface at a particular spot was controlled. The spot size was 0.3 mm along the minor axis and 0.6 mm along the major axis. Scanning was performed parallel to the minor axis. Samples were produced either with isolated single line scans or with large areas created by overlapping several line scans. The step size between scan lines was chosen to be sufficiently small (generally 0.38 mm) such that successive lines overlapped substantially, improving homogeneity. Homogeneity is further enhanced by performing two overlapping scans in orthogonal directions rather than one overlapping scan with an exposure of the same total number of shots. After laser processing, the samples were analyzed with a field emission scanning electron microscope (Zeiss SUPRA 40). 3. Results and discussion Fig. 1. shows the SEM images of sharp conical pillars formed on the silicon surface produced with 240 shots of 130 fs laser pulses at a fluence of 0.6 J cm 2 in 400 mbar of SF6 (Fig. 1a) and HCl (Fig. 1b), viewed 458 from the surface normal. The pillar size varies across the scanned laser line indicating the intensity variation of the laser fluence across the spatial profile of the laser pulse. We overlapped the scan lines to minimize height variation over a large area on the sample surface. The spikes that are obtained in SF6 are 15 mm tall and around 5 mm at the base. On the other hand, under similar conditions, the structures formed in the presence of HCl are composed of taller pillars ( 20 mm height) with a base dimension around 7 mm. It is evident from Fig. 1a and b that the pillars which are formed in SF6 are narrower towards the tips compared to pillars formed in HCl. Although the pillar density is comparable in both cases, the pillars formed in HCl look stronger and relatively blunt towards the tip indicating that differences in the etch chemistry in the F/Si system relative to that of Cl/Si system are responsible for significant structural differences. The role of chemistry in structure formation is not well understood and it is not a trivial result that regular conical pillars are formed during fs irradiation in SF6 as well as HCl. For example, regular conical pillars can be formed during ns irradiation of Si in the presence of SF6 but they are not formed for ns irradiation in the presence of HCl [9]. During ns pulsed irradiation of Si, HCl produces significantly blunter tips, the pillars are much taller (>50 mm) and they are porous rather than solid core [9]. This is a further indication that the mechanisms of pillar formation are not the same for fs and ns irradiation. This result is also significant because chemical impurities, namely sulfur,
Fig. 1. SEM images of pillars, viewed 458 from the surface normal, formed in silicon surface by 240 laser pulses of 130 fs duration, 0.6 J cm 2 fluence in the gaseous environment of 400 mbar of (a) SF6 and (b) HCl.
incorporated during laser processing have been implicated in changing the optical and electronic properties of the textured surfaces [10]. By expanding the range of gases in which pillar formation is possible, we should be able to disentangle the effects on optical and electronic properties of geometric parameters from chemical parameters. Fig. 2 displays SEM images taken for germanium samples treated under similar laser and gaseous conditions as in the case of silicon described above. A very different surface texturing was observed in germanium as is clear from Fig. 2. In the case of germanium processed in SF6, we noticed two distinct features not observed under conditions that are similar to those used for silicon: (a) conical pillar formation with higher cone angle up to a neck and (b) atop these pillars very sharp spikes up to 2 mm in length with 100 nm tip radius. Nanoclusters formed during laser ablation are more abundant in germanium compared to silicon under similar laser conditions, which might be due to the lower melting temperature of the former. On the other hand, the germanium structures formed in the presence of HCl (Fig. 1b) are less densely populated, shorter, high cone angle, wider base, and exhibit greater variations in pillar height compared to silicon (Fig. 1a). It is interesting to note that gas phase silicon etching is anisotropic in HCl and the etch rate is 20 mm min 1 for the Si(1 0 0) plane at temperatures in the
6582
B.K. Nayak et al. / Applied Surface Science 253 (2007) 6580–6583
and character of the pillars formed in silicon have been observed to depend critically on the temporal pulse width of the laser beam. Whereas pillar production with pulse durations in the range of 100 fs - 20 ns has been demonstrated with much smaller pillars being formed in the case of fs irradiation [10,13], we were unable to produce pillars with a 300 ns Nd:VO4 laser. It appears that pulses significantly shorter than the melt lifetime ( 250 ns) are a requisite characteristic of the laser for sharp pillar formation. 4. Conclusion
Fig. 2. SEM images of pillars, viewed 458 from the surface normal, formed in germanium surface by 240 laser pulses of 130 fs duration, 0.6 J cm 2 fluence in the gaseous environment of 400 mbar of (a) SF6 and (b) HCl.
range of 1050–1250 8C [11], whereas in germanium the etch rate is around 3 mm min 1, the etch rate is independent of temperature beyond 800 8C, and it leads to the development of square pits for Ge(1 0 0) [12]. The associated differences in etch chemistry are involved in determining the different structure formation dynamics that result in more regular spike formation in silicon compared to germanium in HCl. The details of how these differences affect structure formation will form the basis of future investigations. Laser induced surface texturing that produces regular structures on materials in the presence of reactive species is a complex process involving: (1) plume/process gas and plume/ surface interactions (decomposition of reactive gas, chemical etching enhanced by laser heating of the material and the formation of radicals, deposition out of the plume, etc.), (2) laser ablation of material, and (3) optical phenomena that initiate the formation of regular structures. The laser parameters, gaseous environment and the nature of the substrate, all play a role in the final outcome of the surface morphology. Structure formation is ultimately a convoluted dynamical process involving all three factors. The dimensions
We have for the first time demonstrated ultrafast laser induced pillar formation in Ge using SF6 and HCl gaseous environments. We have also demonstrated that ultrafast laser induced pillar formation is not limited to SF6 but other gases such as HCl can be used. In this article we have also reported a comparative study of ultrafast laser induced pillar formation between Ge and Si. Based on our observations, we make the following points regarding structure formation in silicon and germanium: (a) under similar laser and gas conditions the aspect ratios (height to base ratio) of silicon pillars is higher compared to germanium in both SF6 and HCl environments, (b) there is a spike formation atop pillars that occurs in germanium, whereas a bulb like structure forms atop silicon pillars when processed in SF6, (c) silicon pillars formed in HCl are taller and sturdier compared to SF6, (d) in germanium the structures formed in SF6 are more regular, sharp, well defined features compared to structures formed in HCl, (e) for the formation of regular conical pillars, the pulse length of the laser must be significantly shorter than the melt lifetime. Acknowledgements We gratefully acknowledge the financial support of the National Science Foundation under grant ECS-0100243, NSF I/ UCRC center grant, IGERT Program Grant 9972790 and CREST supplement Program Grant 0520208. This work was also supported by the University of Virginia and West Chester University. References [1] T.-H. Her, R.J. Finlay, C. Wu, S. Deliwala, E. Mazur, Appl. Phys. Lett. 73 (1998) 1673. [2] A.J. Pedraza, J.D. Fowlkes, D.H. Lowndes, Appl. Phys. Lett. 74 (1999) 2322. [3] T.-H. Her, R.J. Finlay, C. Wu, E. Mazur, Appl. Phys. A 70 (2000) 388. [4] J. Bonse, S. Baudach, J. Kruger, W. Kautek, M. Lenzner, Appl. Phys. A 74 (2002) 19. [5] M.Y. Shen, C.H. Crouch, J.E. Carey, R. Younkin, E. Mazur, M. Sheehy, C.M. Friend, Appl. Phys. Lett. 82 (2003) 1715. [6] D. Riedel, J.L. Hernandez-Pozos, R.E. Palmer, K.W. Kolasinski, Appl. Phys. A 78 (2004) 381. [7] B.K. Nayak, M.C. Gupta, Mater. Res. Soc. Symp. Proc. 850 (2005) (MM1.8.1). [8] D. Mills, K.W. Kolasinski, J. Phys. D: Appl. Phys. 38 (2005) 1.
B.K. Nayak et al. / Applied Surface Science 253 (2007) 6580–6583 [9] K.W. Kolasinski, D. Mills, M. Nahidi, J. Vac. Sci. Technol. A 24 (2006) 1474. [10] C.H. Crouch, J.E. Carey, J.M. Warrender, M.J. Aziz, E. Mazur, F.Y. Ge´nin, Appl. Phys. Lett. 2004 (1850) 84. [11] M. Druminski, R. Gessner, J. Cryst. Growth 31 (1975) 312.
6583
[12] J.A. Amick, E.A. Roth, H. Gossenberger, RCA Rev. 24 (1963) 473. [13] K.W. Kolasinski, in: S.G. Pandalai (Ed.), Recent Research Advances in Applied Physics, vol. 7, Transworld Research Network, Kerala, India, 2004, p. 267.