Tailoring the crystal structure of individual silicon nanowires by ...
Home
Search
Collections
Journals
About
Contact us
My IOPscience
Tailoring the crystal structure of individual silicon nanowires by polarized laser annealing
This article has been downloaded from IOPscience. Please scroll down to see the full text article. 2011 Nanotechnology 22 305709 (http://iopscience.iop.org/0957-4484/22/30/305709) View the table of contents for this issue, or go to the journal homepage for more
Download details: IP Address: 128.125.121.187 The article was downloaded on 01/07/2011 at 21:00
Please note that terms and conditions apply.
IOP PUBLISHING
NANOTECHNOLOGY
Nanotechnology 22 (2011) 305709 (7pp)
doi:10.1088/0957-4484/22/30/305709
Tailoring the crystal structure of individual silicon nanowires by polarized laser annealing Chia-Chi Chang1 , Haitian Chen2 , Chun-Chung Chen2, Wei-Hsuan Hung3 , I-Kai Hsu3 , Jesse Theiss2 , Chongwu Zhou1,2 and Stephen B Cronin1,2,4 1
Department of Physics, University of Southern California, Los Angeles, CA 90089, USA Department of Electrical Engineering, University of Southern California, Los Angeles, CA 90089, USA 3 Department of Materials Science, University of Southern California, Los Angeles, CA 90089, USA 2
E-mail: [email protected]
Received 20 April 2011, in final form 6 June 2011 Published 1 July 2011 Online at stacks.iop.org/Nano/22/305709 Abstract We study the effect of polarized laser annealing on the crystalline structure of individual crystalline–amorphous core–shell silicon nanowires (NWs) using Raman spectroscopy. The crystalline fraction of the annealed spot increases dramatically from 0 to 0.93 with increasing incident laser power. We observe Raman lineshape narrowing and frequency hardening upon laser annealing due to the growth of the crystalline core, which is confirmed by high resolution transmission electron microscopy (HRTEM). The anti-Stokes:Stokes Raman intensity ratio is used to determine the local heating temperature caused by the intense focused laser, which exhibits a strong polarization dependence in Si NWs. The most efficient annealing occurs when the laser polarization is aligned along the axis of the NWs, which results in an amorphous–crystalline interface less than 0.5 μm in length. This paper demonstrates a new approach to control the crystal structure of NWs on the sub-micron length scale. (Some figures in this article are in colour only in the electronic version)
at 450 ◦ C [14]. Cui et al developed a one-step growth method of core–shell NWs at higher temperature (485 ◦ C) on stainless steel substrates in which transition metals, like Ni and Fe, act as catalysts [11]. In this method, the diameter of the c-Si core shows no dependence on the growth time, however, the thickness of the a-Si shell increases linearly with growth time. Post-growth treatments to modify silicon crystal structure have also been investigated. Colli et al demonstrated that the amorphization due to ion implantations can be fully recovered by thermal annealing at 800 ◦ C [7]. Liu et al have studied the amorphization of c-Si NWs via dry oxygen annealing at 800 ◦ C [15]. During the oxidation process, lattice distortioninduced stress disintegrates the c-Si into small crystal grains embedded in a-Si. By increasing the oxidation time the NWs would be completely amorphized with a thick layer of oxide. In this paper, we use Raman spectroscopy and high resolution transmission electron microscopy (HRTEM) to
The ability to control and quantify the crystal structure of silicon nanowires (Si NWs) is crucial for obtaining a fundamental understanding of their optical [1–4] and electrical [5–7] properties, and for evaluating their potential applications [8–11]. A variety of growth approaches have been studied, and the crystallinity of NWs can be varied significantly by adjusting the growth conditions [5, 8, 12]. Crystalline– amorphous core–shell Si NWs have been used in solar cells [8], transistors [13, 14], and lithium ion batteries [11] due to their special chemical and electrical properties. Dong et al have grown core–shell Si NWs using a two-step chemical vapor deposition (CVD) process with a precursor of SiH4 . In this processes, Au catalyzed crystalline Si (c-Si) NWs were first grown at 435 ◦ C and then an amorphous Si (a-Si) was deposited 4 Address for correspondence: Department of Electrical Engineering, University of Southern California, Powell Hall of Engineering PHE 624, Los Angeles, CA 90089-0271, USA.
0957-4484/11/305709+07$33.00
1
© 2011 IOP Publishing Ltd Printed in the UK & the USA
Nanotechnology 22 (2011) 305709
C-C Chang et al
Figure 1. (a) SEM image of as-grown Si NWs. (b) TEM images of individual NWs. (c) Optical image of a silicon nanowire with focused laser spot. (d) AFM image of the NW with a diameter of 78 nm. (e) Raman spectra of the NW before and after annealing at the same spot.
study the crystallization process of core–shell Si NWs under laser irradiation in an argon environment. A continuous (CW) 532 nm wavelength laser is focused through a 100× objective lens (0.9 NA) and used as a local heat source to modify the crystallinity of core–shell Si NWs. In our Raman spectrometer, a holographic notch filter is used, which enables us to take both Stokes and anti-Stokes Raman spectra. A half-wavelength polarizer is placed before the notch filter to rotate the laser polarization. In situ Raman spectra are taken during the annealing process with the same incident laser, which give information about the heating temperature based on the anti-Stokes:Stokes Raman intensity ratio [16] and the temperature-induced Raman lineshape broadening and frequency downshifting [17–19]. Low laser power (0.5 mW) Raman spectra with an integration time of 60 s are taken before and after annealing to reveal changes in the crystallinity due to the local annealing process. A strong polarization dependence of this local heating is observed in the core–shell NWs. By using focused laser annealing, we are able to control the crystallinity of core–shell NWs by varying the laser power and/or laser polarization, and, thus, create amorphous–single crystalline junctions in individual NWs. Core–shell Si NWs were synthesized by the thermal CVD method. A 2 nm film of gold was first evaporated on Si(100) substrates and annealed at 530 ◦ C for 30 min to produce isolated gold nanoparticles, which serve as catalysts for the NW growth. The growth was then carried out at 530 ◦ C for 30 min while flowing 20 SCCM hydrogen (H2 ) and 111 SCCM silane (SiH4 ). The resulting nanowires have a diameter distribution extending from 40 to 100 nm based on TEM and atomic force microscopy (AFM) characterizations, as shown in figure 1. As-grown nanowires were sonicated in isopropyl alcohol and then transferred onto substrates (0.25 mm thick glass or 100 nm thick silicon nitride membranes). The Raman
spectrometer was calibrated with a single crystal (100) silicon substrate routinely before each experiment to ensure that the transverse optical (TO) phonon mode of silicon exhibits a Raman peak with a frequency of 520.5 cm−1 and a FWHM of 4.5 cm−1 . Figures 1(a) and (b) show electron microscope images of Si NWs grown in this study. Metal strips with grid markers were deposited on top of one of the NWs, as shown in figure 1(c). AFM was used to determine the diameter of this NW to be 78 nm, as shown in figure 1(d). Figure 1(e) shows the Raman spectra of an individual Si NW before and after laser annealing at 50 mW, while the laser polarization is aligned along the NW axis. In the following, we align the laser polarization along the axis of the NWs, unless stated otherwise. The laser annealing process is carried out in an argon environment at 1 atmosphere to prevent oxidation during annealing. Figure 2(a) shows the evolution of the Raman spectra taken after 30 s irradiations with different laser powers at the same focused laser spot. Before annealing, a broad peak centered at 480 cm−1 , the Raman signature of amorphous silicon, is observed in this NW. As the incident laser power is increased, this peak can be seen to upshift and become narrower due to the reduction in phonon confinement effects [1–4, 20, 21]. The raw spectra in figure 2(a) were fitted with two Lorentzian peaks, as shown in the inset of figure 2(b). The broad peak is attributed to amorphous Si, and the sharp peak is attributed to crystalline Si. The crystalline fraction can be obtained from the relative peak intensities by following the equation: X c = Ic /(Ic + γ Iα ); where Ic and Iα are the integrated Raman intensities of the crystalline and amorphous components, respectively. γ is the scattering ratio of crystalline and amorphous silicon, which ranges from 0.8 to 1 [22–25]. We use γ = 1 here. In figure 2(b), the crystallinity increases with increasing annealing power during 2
Nanotechnology 22 (2011) 305709
C-C Chang et al
Figure 2. (a) Raman spectra before and after laser annealing at different powers at the same laser spot. These spectra have been artificially offset for clarity. (b) Crystalline fraction X c , (c) Raman linewidth, and (d) Raman shift relative to single crystal bulk and estimated mean crystalline size after laser annealing. The inset in (c) shows the crystalline fraction plotted as a function of laser irradiation time at two constant laser powers.
the crystallization process. Also, the lineshape and frequency of the crystalline component becomes narrower and upshifts as the annealing power is increased. These observations indicate that the crystalline core of the NW grows, while the amorphous shell becomes thinner due to the crystallization process. Note that the crystalline Raman peak becomes visible after laser annealing at 20 mW but is downshifted from the single crystal value to 513 cm−1 and significantly broadened to 20 cm−1 . We attribute this to the effect of phonon quantum confinement because of the small crystalline core. As the crystalline core increases due to successive annealing, the crystalline peak blue shifts to 520 cm−1 , while the linewidth narrows to 7 cm−1 , which is closer to the values of single crystal bulk Si. The linewidth and Raman shift ω of the crystalline peak are plotted as a function of annealing power in figures 2(c) and (d), where the peak downshift is from 520.5 cm−1 . The crystalline size D can be calculated from the bond polarization model: ω = B(a/D)γ , where B = 2.0 cm−1 , γ = 1.44, and a is the silicon lattice constant (a = 0.543 nm) [26]. Various models [1–4, 20, 21] have been developed to obtain the mean crystal grain size from these Raman features. However, they are all in agreement when the crystal size is above 3 nm [20]. The mechanism of CW laser-induced crystallization is different from that of a pulsed laser. The crystallization takes place in the solid phase under CW laser irradiation. The annealing temperature depends on the laser power and the heat loss to the substrate. The crystalline core acts as a seed for crystallization and the growth of the crystalline core is expected to depend on the annealing time at a given temperature. However, the crystalline fraction of the annealed
NWs does not show any time dependence, as shown in the inset of figure 2(c). This is because the irradiated nanowire reaches thermal equilibrium on a time scale of microseconds, and the resulting changes in crystallinity take place far below our minimum time resolution. Figure 3(a) shows the Raman spectra of a core–shell Si NW before and after annealing with 50 mW. A Raman spatial map scan is used to resolve the crystallinity changes along the NW after the single spot laser annealing. The integrated crystalline peak intensity Ic plotted as a function of position gives a full width at half maximum of 680 nm by fitting with a Lorentzian function. This is slightly larger than the nominal laser spot size (500 nm) obtained with the 100× objective lens. We believe that the slightly wider profile of the crystalline peak intensity is due to the finite thermal gradient produced while annealing, which results in different levels of crystallization that have occurred near the annealing spot. The integrated amorphous peak intensity Iα follows the opposite trend of the crystalline peak, as expected in the crystallization process. Further spatial mapping of the Raman intensity is plotted in figure 3(c), which shows the modulations of both a-Si and c-Si peaks along the length of the NW. Nanowires were also deposited on 100 nm thick silicon nitride (Si x N y ) membranes (SPI, Inc.), enabling HRTEM imaging to be taken after laser annealing. Figure 4(a) shows a TEM image of a locally annealed NW. Although the same laser conditions (50 mW with polarization along NW axis) were used, the annealed region is twice as large as that of the NW deposited on the glass substrate. This is because the heat conduction through the Six N y membrane is far less than that 3
Nanotechnology 22 (2011) 305709
C-C Chang et al
the NW. However, in the annealed region (figure 4(d)), the crystalline core fills nearly the entire NW cross section, with a thin silicon oxide layer. The same uniform crystalline structure as in figure 4(d) is observed within 1 μm of the annealed region, indicated by a gray box in figure 4(a). The axial modulation in crystal structure is also reflected in the spatial Raman mapping (intensity, Raman shift, and linewidth), as shown in figures 4(f) and (g). The uniformity of Raman shift and linewidth within the annealed region agrees with the TEM inspection and further corroborates the high crystal quality. The Raman Stokes (S) and anti-Stokes (AS) spectra shown in figure 4(e) were taken during the laser annealing process. These Raman modes downshift and broaden to 496.6 and 18 cm−1 due to the high temperature reached [17–19]. The annealing temperature is estimated to be 1103 K, based ¯ 0 /kB T , where on the AS/S intensity ratio, IAS /IS = C e−hω C is a calibrated coefficient from room temperature data, h¯ ω0 = 64.6 meV is the phonon energy, and kB is Boltzmann’s constant. We also observe a strong polarization dependence of the laser-induced heating. The Raman intensity follows the expression I (ϕ) = c cos2 (ϕ), where ϕ is the angle between laser polarization and NW axis as shown in figure 5(a). The temperature profile obtained from the Raman AS/S intensity ratio over a wide range of polarization angles, in figure 5(b), shows a strong preferential heating effect when the laser polarization is aligned at ϕ = 0◦ and 180◦ . There is a maximum temperature difference of 200 K under 40 mW irradiation. In figure 5(c), the Raman linewidth oscillates between 5 and 8 cm−1 and the Raman shift varies from 520 to 517 cm−1 , which can also be understood by the preferential heating. The polarization dependence of the Raman intensity has been reported in crystalline silicon films [27, 25]. The intensity of Raman scattering depends on the orientation of the incident laser polarization relative to the crystal axes followed by the simplified expression I (θ ) = c cos2 (2θ ), where θ is the angle between laser polarization and crystal axes. However, we do not observe other oscillation patterns of Raman intensity rather than the one in figure 5(c), which implies that the former effect dominates in the NW system. The polarization dependence of laser heating is attributed to the different degrees of absorption of the polarized laser. A recent finite-difference time-domain (FDTD) study on light absorption in a 50 nm Si NW shows that the absorption per unit volume of transverse magnetic (TM) light is almost 20-fold larger than that of transverse electric (TE) light at a wavelength of 532 nm [28]. In figure 6, we demonstrate the effect of polarizationdependent annealing on a 12 μm-long core–shell NW with a fixed laser power of 50 mW at various laser polarization angles specified on the top of figure 6(a). Each annealing spot is separated by 1 μm along the wire axis to avoid proximity heating from the Gaussian laser beam. Prior to annealing, spatial Raman mapping was performed and correlated with the polarization angles to reveal the uniformity of crystallinity over the length of the NW. After preferential annealing, the same spatial Raman mapping was performed again to determine the crystallinity modulation along the NW, as plotted in figure 6(c).
Figure 3. (a) Raman spectra before and after focused laser annealing. (b) Integrated intensity of the crystalline peak ( Ic ) as a function of position, plotted together with the intensity of the amorphous peak ( Iα ). (c) Raman intensity mapping along the NW.
of the glass substrate, which also increases the amorphous– crystalline interface to 0.5 μm in length. The drawing in figure 4(b) illustrates the cross section at the boundary between the unannealed and annealed regions. The unannealed region remains as a core–shell structure, as seen in figure 4(c). Here, the crystalline structure can only been seen near the core of 4
Nanotechnology 22 (2011) 305709
C-C Chang et al
Figure 4. (a) TEM image of a locally annealed Si NW in which the annealed area is 1 μm indicated by the box. (b) Schematic diagram of the NW boundary between annealed and unannealed regions. Corresponding HRTEM images in (c) and (d) are indicated by the dashed boxes. (c) HRTEM image of the unannealed area near the core of the NW. (d) HRTEM image of the annealed segment near the NW surface. (e) Stokes and anti-Stokes Raman spectra taken during the laser annealing process. (f) Crystalline peak intensity, (g) Raman shift and linewidth plotted as a function of position along the NW.
Figure 5. (a) Experimental setup of polarization-dependent laser heating on crystalline NWs. (b) AS/S intensity ratio and corresponding temperature as a function of polarization angle. (c) Raman intensity, linewidth, and Raman shift plotted as a function of polarization angle with respect to the NW axis.
5
Nanotechnology 22 (2011) 305709
C-C Chang et al
Figure 6. (a) Schematic diagram of polarization-dependent laser-induced preferential annealing. Blue areas represent crystalline silicon. (b) Spatial Raman intensity mapping of a laser annealed NW at various polarization angles. (c) Crystalline fraction X c before and after annealing and calibrated annealing temperature plotted as a function of polarization angle and position.
Center funded by the US Department of Energy, Office of Science, Office of Basic Energy Sciences under Award Number DE-SC0001013. This research was also supported in part by DOE Award No. DE-FG02-07ER46376 (IK-H) and NSF Award No. CBET-08467265.
In figure 6(b), visible a-Si peaks are observed at integer position numbers, which are unannealed spots. The intensity modulation of a-Si peaks is merely due to the polarization angle effect described in the previous paragraph. At the annealed spots (half-integer position numbers), the a-Si peaks either diminish or become very weak. For these c-Si peaks, there are two factors responsible for the intensity modulation. Besides the polarization angle effect, the crystalline fractions after annealing are different, as depicted in figure 6(a). The drawing illustrates the crystal structure modulations in both axial and radial directions. Figure 6(c) also shows the corresponding annealing temperatures as a function of polarization angle. The improvement of crystallinity shows strong polarization dependence due to preferential heating. In conclusion, Raman spectroscopy was used to study the effect of laser annealing on crystalline–amorphous core– shell Si NWs grown by the CVD method. The crystalline fraction was found to increase gradually with increasing laser power based on the local Raman spectra. The Raman shift and linewidth show upshifting and sharpening, respectively, due to the reduction of phonon confinement effects, as the crystalline core of the NW increases. Strong preferential heating due to the laser polarization was utilized to control the crystallinity of individual NWs. Using the focused laser annealing, we are able to create an amorphous–crystalline interface less than 0.5 μm in length.
References [1] Li B, Yu D and Zhang S-L 1999 Raman spectral study of silicon nanowires Phys. Rev. B 59 1645 [2] Wang R-P, Zhou G-W, Liu Y-L, Pan S-H, Zhang H-Z, Yu D-P and Zhang Z 2000 Raman spectral study of silicon nanowires: high-order scattering and phonon confinement effects Phys. Rev. B 61 16827 [3] Piscanec S, Cantoro M, Ferrari A C, Zapien J A, Lifshitz Y, Lee S T, Hofmann S and Robertson J 2003 Raman spectroscopy of silicon nanowires Phys. Rev. B 68 241312 [4] Gupta R, Xiong Q, Adu C K, Kim U J and Eklund P C 2003 Laser-induced Fano resonance scattering in silicon nanowires Nano Lett. 3 627–31 [5] Gudiksen M S, Lauhon L J, Wang J, Smith D C and Lieber C M 2002 Growth of nanowire superlattice structures for nanoscale photonics and electronics Nature 415 617–20 [6] Cui Y and Lieber C M 2001 Functional nanoscale electronic devices assembled using silicon nanowire building blocks Science 291 851–3 [7] Colli A, Fasoli A, Ronning C, Pisana S, Piscanec S and Ferrari A C 2008 Ion beam doping of silicon nanowires Nano Lett. 8 2188–93 [8] Tian B, Zheng X, Kempa T J, Fang Y, Yu N, Yu G, Huang J and Lieber C M 2007 Coaxial silicon nanowires as solar cells and nanoelectronic power sources Nature 449 885–9 [9] Chan C K, Peng H, Liu G, McIlwrath K, Zhang X F, Huggins R A and Cui Y 2008 High-performance lithium battery anodes using silicon nanowires Nat. Nano 3 31–5
Acknowledgments C-C Chang, C-C Chen, and JT were supported as part of the Center for Energy Nanoscience, an Energy Frontier Research 6
Nanotechnology 22 (2011) 305709
C-C Chang et al
[10] Ruffo R, Hong S S, Chan C K, Huggins R A and Cui Y 2009 Impedance analysis of silicon nanowire lithium ion battery anodes J. Phys. Chem. C 113 11390–8 [11] Cui L-F, Ruffo R, Chan C K, Peng H and Cui Y 2009 Crystalline–amorphous core–shell silicon nanowires for high capacity and high current battery electrodes Nano Lett. 9 491–5 [12] Wacaser B A, Reuter M C, Khayyat M M, Wen C-Y, Haight R, Guha S and Ross F M 2009 Growth system, structure, and doping of aluminum-seeded epitaxial silicon nanowires Nano Lett. 9 3296–301 [13] Goldberger J, Hochbaum A I, Fan R and Yang P 2006 Silicon vertically integrated nanowire field effect transistors Nano Lett. 6 973–7 [14] Dong Y, Yu G, McAlpine M C, Lu W and Lieber C M 2008 Si/a-Si core/shell nanowires as nonvolatile crossbar switches Nano Lett. 8 386–91 [15] Liu L Z, Wu X L, Zhang Z Y, Li T H and Chu P K 2009 Raman investigation of oxidation mechanism of silicon nanowires Appl. Phys. Lett. 95 093109–3 [16] Scheel H, Reich S, Ferrari A C, Cantoro M, Colli A and Thomsen C 2006 Raman scattering on silicon nanowires: the thermal conductivity of the environment determines the optical phonon frequency Appl. Phys. Lett. 88 233114 [17] Su Z, Sha J, Pan G, Liu J, Yang D, Dickinson C and Zhou W 2006 Temperature-dependent Raman scattering of silicon nanowires J. Phys. Chem. B 110 1229–34 [18] Chen Y, Peng B and Wang B 2007 Raman spectra and temperature-dependent Raman scattering of silicon nanowires J. Phys. Chem. C 111 5855–8 [19] Scheel H, Khachadorian S, Cantoro M, Colli A, Ferrari A C and Thomsen C 2008 Silicon nanowire optical Raman line shapes at cryogenic and elevated temperatures Phys. Status Solidi b 245 2090–3
[20] Faraci G, Gibilisco S, Russo P, Pennisi A R and La Rosa S 2006 Modified Raman confinement model for Si nanocrystals Phys. Rev. B 73 033307 [21] Miska P, Dossot M, Nguyen T D, Grun M, Rinnert H, Vergnat M and Humbert B 2010 Embedded silicon nanocrystals studied by photoluminescence and Raman spectroscopies: exciton and phonon confinement effects J. Phys. Chem. C 114 17344–9 [22] Beeman D, Tsu R and Thorpe M F 1985 Structural information from the Raman spectrum of amorphous silicon Phys. Rev. B 32 874 [23] Wu X L, Siu G G, Tong S, Liu X N, Yan F, Jiang S S, Zhang X K and Feng D 1996 Raman scattering of alternating nanocrystalline silicon/amorphous silicon multilayers Appl. Phys. Lett. 69 523–5 [24] Smit C, Swaaij R A C M M V, Donker H, Petit A M H N, Kessels W M M and Sanden M C M V D 2003 Determining the material structure of microcrystalline silicon from Raman spectra J. Appl. Phys. 94 3582–8 [25] Lu Z Q, Quinn T and Reehal H S 2005 Polarization-dependent Raman spectra of thin crystalline silicon films J. Appl. Phys. 97 033512 [26] Paillard V, Puech P, Laguna M A, Carles R, Kohn B and Huisken F 1999 Improved one-phonon confinement model for an accurate size determination of silicon nanocrystals J. Appl. Phys. 86 1921–4 [27] Kolb G, Salbert T and Abstreiter G 1991 Raman-microanalysis of strain and crystal orientation in laser-crystallized silicon Fresenius’ J. Anal. Chem. 341 166–70 [28] Hyun J and Lauhon L 2011 Spatially resolved plasmonically enhanced photocurrent from Au nanoparticles on a Si nanowire Nano Lett. at press
7
Search
Collections
Journals
About
Contact us
My IOPscience
Tailoring the crystal structure of individual silicon nanowires by polarized laser annealing
This article has been downloaded from IOPscience. Please scroll down to see the full text article. 2011 Nanotechnology 22 305709 (http://iopscience.iop.org/0957-4484/22/30/305709) View the table of contents for this issue, or go to the journal homepage for more
Download details: IP Address: 128.125.121.187 The article was downloaded on 01/07/2011 at 21:00
Please note that terms and conditions apply.
IOP PUBLISHING
NANOTECHNOLOGY
Nanotechnology 22 (2011) 305709 (7pp)
doi:10.1088/0957-4484/22/30/305709
Tailoring the crystal structure of individual silicon nanowires by polarized laser annealing Chia-Chi Chang1 , Haitian Chen2 , Chun-Chung Chen2, Wei-Hsuan Hung3 , I-Kai Hsu3 , Jesse Theiss2 , Chongwu Zhou1,2 and Stephen B Cronin1,2,4 1
Department of Physics, University of Southern California, Los Angeles, CA 90089, USA Department of Electrical Engineering, University of Southern California, Los Angeles, CA 90089, USA 3 Department of Materials Science, University of Southern California, Los Angeles, CA 90089, USA 2
E-mail: [email protected]
Received 20 April 2011, in final form 6 June 2011 Published 1 July 2011 Online at stacks.iop.org/Nano/22/305709 Abstract We study the effect of polarized laser annealing on the crystalline structure of individual crystalline–amorphous core–shell silicon nanowires (NWs) using Raman spectroscopy. The crystalline fraction of the annealed spot increases dramatically from 0 to 0.93 with increasing incident laser power. We observe Raman lineshape narrowing and frequency hardening upon laser annealing due to the growth of the crystalline core, which is confirmed by high resolution transmission electron microscopy (HRTEM). The anti-Stokes:Stokes Raman intensity ratio is used to determine the local heating temperature caused by the intense focused laser, which exhibits a strong polarization dependence in Si NWs. The most efficient annealing occurs when the laser polarization is aligned along the axis of the NWs, which results in an amorphous–crystalline interface less than 0.5 μm in length. This paper demonstrates a new approach to control the crystal structure of NWs on the sub-micron length scale. (Some figures in this article are in colour only in the electronic version)
at 450 ◦ C [14]. Cui et al developed a one-step growth method of core–shell NWs at higher temperature (485 ◦ C) on stainless steel substrates in which transition metals, like Ni and Fe, act as catalysts [11]. In this method, the diameter of the c-Si core shows no dependence on the growth time, however, the thickness of the a-Si shell increases linearly with growth time. Post-growth treatments to modify silicon crystal structure have also been investigated. Colli et al demonstrated that the amorphization due to ion implantations can be fully recovered by thermal annealing at 800 ◦ C [7]. Liu et al have studied the amorphization of c-Si NWs via dry oxygen annealing at 800 ◦ C [15]. During the oxidation process, lattice distortioninduced stress disintegrates the c-Si into small crystal grains embedded in a-Si. By increasing the oxidation time the NWs would be completely amorphized with a thick layer of oxide. In this paper, we use Raman spectroscopy and high resolution transmission electron microscopy (HRTEM) to
The ability to control and quantify the crystal structure of silicon nanowires (Si NWs) is crucial for obtaining a fundamental understanding of their optical [1–4] and electrical [5–7] properties, and for evaluating their potential applications [8–11]. A variety of growth approaches have been studied, and the crystallinity of NWs can be varied significantly by adjusting the growth conditions [5, 8, 12]. Crystalline– amorphous core–shell Si NWs have been used in solar cells [8], transistors [13, 14], and lithium ion batteries [11] due to their special chemical and electrical properties. Dong et al have grown core–shell Si NWs using a two-step chemical vapor deposition (CVD) process with a precursor of SiH4 . In this processes, Au catalyzed crystalline Si (c-Si) NWs were first grown at 435 ◦ C and then an amorphous Si (a-Si) was deposited 4 Address for correspondence: Department of Electrical Engineering, University of Southern California, Powell Hall of Engineering PHE 624, Los Angeles, CA 90089-0271, USA.
0957-4484/11/305709+07$33.00
1
© 2011 IOP Publishing Ltd Printed in the UK & the USA
Nanotechnology 22 (2011) 305709
C-C Chang et al
Figure 1. (a) SEM image of as-grown Si NWs. (b) TEM images of individual NWs. (c) Optical image of a silicon nanowire with focused laser spot. (d) AFM image of the NW with a diameter of 78 nm. (e) Raman spectra of the NW before and after annealing at the same spot.
study the crystallization process of core–shell Si NWs under laser irradiation in an argon environment. A continuous (CW) 532 nm wavelength laser is focused through a 100× objective lens (0.9 NA) and used as a local heat source to modify the crystallinity of core–shell Si NWs. In our Raman spectrometer, a holographic notch filter is used, which enables us to take both Stokes and anti-Stokes Raman spectra. A half-wavelength polarizer is placed before the notch filter to rotate the laser polarization. In situ Raman spectra are taken during the annealing process with the same incident laser, which give information about the heating temperature based on the anti-Stokes:Stokes Raman intensity ratio [16] and the temperature-induced Raman lineshape broadening and frequency downshifting [17–19]. Low laser power (0.5 mW) Raman spectra with an integration time of 60 s are taken before and after annealing to reveal changes in the crystallinity due to the local annealing process. A strong polarization dependence of this local heating is observed in the core–shell NWs. By using focused laser annealing, we are able to control the crystallinity of core–shell NWs by varying the laser power and/or laser polarization, and, thus, create amorphous–single crystalline junctions in individual NWs. Core–shell Si NWs were synthesized by the thermal CVD method. A 2 nm film of gold was first evaporated on Si(100) substrates and annealed at 530 ◦ C for 30 min to produce isolated gold nanoparticles, which serve as catalysts for the NW growth. The growth was then carried out at 530 ◦ C for 30 min while flowing 20 SCCM hydrogen (H2 ) and 111 SCCM silane (SiH4 ). The resulting nanowires have a diameter distribution extending from 40 to 100 nm based on TEM and atomic force microscopy (AFM) characterizations, as shown in figure 1. As-grown nanowires were sonicated in isopropyl alcohol and then transferred onto substrates (0.25 mm thick glass or 100 nm thick silicon nitride membranes). The Raman
spectrometer was calibrated with a single crystal (100) silicon substrate routinely before each experiment to ensure that the transverse optical (TO) phonon mode of silicon exhibits a Raman peak with a frequency of 520.5 cm−1 and a FWHM of 4.5 cm−1 . Figures 1(a) and (b) show electron microscope images of Si NWs grown in this study. Metal strips with grid markers were deposited on top of one of the NWs, as shown in figure 1(c). AFM was used to determine the diameter of this NW to be 78 nm, as shown in figure 1(d). Figure 1(e) shows the Raman spectra of an individual Si NW before and after laser annealing at 50 mW, while the laser polarization is aligned along the NW axis. In the following, we align the laser polarization along the axis of the NWs, unless stated otherwise. The laser annealing process is carried out in an argon environment at 1 atmosphere to prevent oxidation during annealing. Figure 2(a) shows the evolution of the Raman spectra taken after 30 s irradiations with different laser powers at the same focused laser spot. Before annealing, a broad peak centered at 480 cm−1 , the Raman signature of amorphous silicon, is observed in this NW. As the incident laser power is increased, this peak can be seen to upshift and become narrower due to the reduction in phonon confinement effects [1–4, 20, 21]. The raw spectra in figure 2(a) were fitted with two Lorentzian peaks, as shown in the inset of figure 2(b). The broad peak is attributed to amorphous Si, and the sharp peak is attributed to crystalline Si. The crystalline fraction can be obtained from the relative peak intensities by following the equation: X c = Ic /(Ic + γ Iα ); where Ic and Iα are the integrated Raman intensities of the crystalline and amorphous components, respectively. γ is the scattering ratio of crystalline and amorphous silicon, which ranges from 0.8 to 1 [22–25]. We use γ = 1 here. In figure 2(b), the crystallinity increases with increasing annealing power during 2
Nanotechnology 22 (2011) 305709
C-C Chang et al
Figure 2. (a) Raman spectra before and after laser annealing at different powers at the same laser spot. These spectra have been artificially offset for clarity. (b) Crystalline fraction X c , (c) Raman linewidth, and (d) Raman shift relative to single crystal bulk and estimated mean crystalline size after laser annealing. The inset in (c) shows the crystalline fraction plotted as a function of laser irradiation time at two constant laser powers.
the crystallization process. Also, the lineshape and frequency of the crystalline component becomes narrower and upshifts as the annealing power is increased. These observations indicate that the crystalline core of the NW grows, while the amorphous shell becomes thinner due to the crystallization process. Note that the crystalline Raman peak becomes visible after laser annealing at 20 mW but is downshifted from the single crystal value to 513 cm−1 and significantly broadened to 20 cm−1 . We attribute this to the effect of phonon quantum confinement because of the small crystalline core. As the crystalline core increases due to successive annealing, the crystalline peak blue shifts to 520 cm−1 , while the linewidth narrows to 7 cm−1 , which is closer to the values of single crystal bulk Si. The linewidth and Raman shift ω of the crystalline peak are plotted as a function of annealing power in figures 2(c) and (d), where the peak downshift is from 520.5 cm−1 . The crystalline size D can be calculated from the bond polarization model: ω = B(a/D)γ , where B = 2.0 cm−1 , γ = 1.44, and a is the silicon lattice constant (a = 0.543 nm) [26]. Various models [1–4, 20, 21] have been developed to obtain the mean crystal grain size from these Raman features. However, they are all in agreement when the crystal size is above 3 nm [20]. The mechanism of CW laser-induced crystallization is different from that of a pulsed laser. The crystallization takes place in the solid phase under CW laser irradiation. The annealing temperature depends on the laser power and the heat loss to the substrate. The crystalline core acts as a seed for crystallization and the growth of the crystalline core is expected to depend on the annealing time at a given temperature. However, the crystalline fraction of the annealed
NWs does not show any time dependence, as shown in the inset of figure 2(c). This is because the irradiated nanowire reaches thermal equilibrium on a time scale of microseconds, and the resulting changes in crystallinity take place far below our minimum time resolution. Figure 3(a) shows the Raman spectra of a core–shell Si NW before and after annealing with 50 mW. A Raman spatial map scan is used to resolve the crystallinity changes along the NW after the single spot laser annealing. The integrated crystalline peak intensity Ic plotted as a function of position gives a full width at half maximum of 680 nm by fitting with a Lorentzian function. This is slightly larger than the nominal laser spot size (500 nm) obtained with the 100× objective lens. We believe that the slightly wider profile of the crystalline peak intensity is due to the finite thermal gradient produced while annealing, which results in different levels of crystallization that have occurred near the annealing spot. The integrated amorphous peak intensity Iα follows the opposite trend of the crystalline peak, as expected in the crystallization process. Further spatial mapping of the Raman intensity is plotted in figure 3(c), which shows the modulations of both a-Si and c-Si peaks along the length of the NW. Nanowires were also deposited on 100 nm thick silicon nitride (Si x N y ) membranes (SPI, Inc.), enabling HRTEM imaging to be taken after laser annealing. Figure 4(a) shows a TEM image of a locally annealed NW. Although the same laser conditions (50 mW with polarization along NW axis) were used, the annealed region is twice as large as that of the NW deposited on the glass substrate. This is because the heat conduction through the Six N y membrane is far less than that 3
Nanotechnology 22 (2011) 305709
C-C Chang et al
the NW. However, in the annealed region (figure 4(d)), the crystalline core fills nearly the entire NW cross section, with a thin silicon oxide layer. The same uniform crystalline structure as in figure 4(d) is observed within 1 μm of the annealed region, indicated by a gray box in figure 4(a). The axial modulation in crystal structure is also reflected in the spatial Raman mapping (intensity, Raman shift, and linewidth), as shown in figures 4(f) and (g). The uniformity of Raman shift and linewidth within the annealed region agrees with the TEM inspection and further corroborates the high crystal quality. The Raman Stokes (S) and anti-Stokes (AS) spectra shown in figure 4(e) were taken during the laser annealing process. These Raman modes downshift and broaden to 496.6 and 18 cm−1 due to the high temperature reached [17–19]. The annealing temperature is estimated to be 1103 K, based ¯ 0 /kB T , where on the AS/S intensity ratio, IAS /IS = C e−hω C is a calibrated coefficient from room temperature data, h¯ ω0 = 64.6 meV is the phonon energy, and kB is Boltzmann’s constant. We also observe a strong polarization dependence of the laser-induced heating. The Raman intensity follows the expression I (ϕ) = c cos2 (ϕ), where ϕ is the angle between laser polarization and NW axis as shown in figure 5(a). The temperature profile obtained from the Raman AS/S intensity ratio over a wide range of polarization angles, in figure 5(b), shows a strong preferential heating effect when the laser polarization is aligned at ϕ = 0◦ and 180◦ . There is a maximum temperature difference of 200 K under 40 mW irradiation. In figure 5(c), the Raman linewidth oscillates between 5 and 8 cm−1 and the Raman shift varies from 520 to 517 cm−1 , which can also be understood by the preferential heating. The polarization dependence of the Raman intensity has been reported in crystalline silicon films [27, 25]. The intensity of Raman scattering depends on the orientation of the incident laser polarization relative to the crystal axes followed by the simplified expression I (θ ) = c cos2 (2θ ), where θ is the angle between laser polarization and crystal axes. However, we do not observe other oscillation patterns of Raman intensity rather than the one in figure 5(c), which implies that the former effect dominates in the NW system. The polarization dependence of laser heating is attributed to the different degrees of absorption of the polarized laser. A recent finite-difference time-domain (FDTD) study on light absorption in a 50 nm Si NW shows that the absorption per unit volume of transverse magnetic (TM) light is almost 20-fold larger than that of transverse electric (TE) light at a wavelength of 532 nm [28]. In figure 6, we demonstrate the effect of polarizationdependent annealing on a 12 μm-long core–shell NW with a fixed laser power of 50 mW at various laser polarization angles specified on the top of figure 6(a). Each annealing spot is separated by 1 μm along the wire axis to avoid proximity heating from the Gaussian laser beam. Prior to annealing, spatial Raman mapping was performed and correlated with the polarization angles to reveal the uniformity of crystallinity over the length of the NW. After preferential annealing, the same spatial Raman mapping was performed again to determine the crystallinity modulation along the NW, as plotted in figure 6(c).
Figure 3. (a) Raman spectra before and after focused laser annealing. (b) Integrated intensity of the crystalline peak ( Ic ) as a function of position, plotted together with the intensity of the amorphous peak ( Iα ). (c) Raman intensity mapping along the NW.
of the glass substrate, which also increases the amorphous– crystalline interface to 0.5 μm in length. The drawing in figure 4(b) illustrates the cross section at the boundary between the unannealed and annealed regions. The unannealed region remains as a core–shell structure, as seen in figure 4(c). Here, the crystalline structure can only been seen near the core of 4
Nanotechnology 22 (2011) 305709
C-C Chang et al
Figure 4. (a) TEM image of a locally annealed Si NW in which the annealed area is 1 μm indicated by the box. (b) Schematic diagram of the NW boundary between annealed and unannealed regions. Corresponding HRTEM images in (c) and (d) are indicated by the dashed boxes. (c) HRTEM image of the unannealed area near the core of the NW. (d) HRTEM image of the annealed segment near the NW surface. (e) Stokes and anti-Stokes Raman spectra taken during the laser annealing process. (f) Crystalline peak intensity, (g) Raman shift and linewidth plotted as a function of position along the NW.
Figure 5. (a) Experimental setup of polarization-dependent laser heating on crystalline NWs. (b) AS/S intensity ratio and corresponding temperature as a function of polarization angle. (c) Raman intensity, linewidth, and Raman shift plotted as a function of polarization angle with respect to the NW axis.
5
Nanotechnology 22 (2011) 305709
C-C Chang et al
Figure 6. (a) Schematic diagram of polarization-dependent laser-induced preferential annealing. Blue areas represent crystalline silicon. (b) Spatial Raman intensity mapping of a laser annealed NW at various polarization angles. (c) Crystalline fraction X c before and after annealing and calibrated annealing temperature plotted as a function of polarization angle and position.
Center funded by the US Department of Energy, Office of Science, Office of Basic Energy Sciences under Award Number DE-SC0001013. This research was also supported in part by DOE Award No. DE-FG02-07ER46376 (IK-H) and NSF Award No. CBET-08467265.
In figure 6(b), visible a-Si peaks are observed at integer position numbers, which are unannealed spots. The intensity modulation of a-Si peaks is merely due to the polarization angle effect described in the previous paragraph. At the annealed spots (half-integer position numbers), the a-Si peaks either diminish or become very weak. For these c-Si peaks, there are two factors responsible for the intensity modulation. Besides the polarization angle effect, the crystalline fractions after annealing are different, as depicted in figure 6(a). The drawing illustrates the crystal structure modulations in both axial and radial directions. Figure 6(c) also shows the corresponding annealing temperatures as a function of polarization angle. The improvement of crystallinity shows strong polarization dependence due to preferential heating. In conclusion, Raman spectroscopy was used to study the effect of laser annealing on crystalline–amorphous core– shell Si NWs grown by the CVD method. The crystalline fraction was found to increase gradually with increasing laser power based on the local Raman spectra. The Raman shift and linewidth show upshifting and sharpening, respectively, due to the reduction of phonon confinement effects, as the crystalline core of the NW increases. Strong preferential heating due to the laser polarization was utilized to control the crystallinity of individual NWs. Using the focused laser annealing, we are able to create an amorphous–crystalline interface less than 0.5 μm in length.
References [1] Li B, Yu D and Zhang S-L 1999 Raman spectral study of silicon nanowires Phys. Rev. B 59 1645 [2] Wang R-P, Zhou G-W, Liu Y-L, Pan S-H, Zhang H-Z, Yu D-P and Zhang Z 2000 Raman spectral study of silicon nanowires: high-order scattering and phonon confinement effects Phys. Rev. B 61 16827 [3] Piscanec S, Cantoro M, Ferrari A C, Zapien J A, Lifshitz Y, Lee S T, Hofmann S and Robertson J 2003 Raman spectroscopy of silicon nanowires Phys. Rev. B 68 241312 [4] Gupta R, Xiong Q, Adu C K, Kim U J and Eklund P C 2003 Laser-induced Fano resonance scattering in silicon nanowires Nano Lett. 3 627–31 [5] Gudiksen M S, Lauhon L J, Wang J, Smith D C and Lieber C M 2002 Growth of nanowire superlattice structures for nanoscale photonics and electronics Nature 415 617–20 [6] Cui Y and Lieber C M 2001 Functional nanoscale electronic devices assembled using silicon nanowire building blocks Science 291 851–3 [7] Colli A, Fasoli A, Ronning C, Pisana S, Piscanec S and Ferrari A C 2008 Ion beam doping of silicon nanowires Nano Lett. 8 2188–93 [8] Tian B, Zheng X, Kempa T J, Fang Y, Yu N, Yu G, Huang J and Lieber C M 2007 Coaxial silicon nanowires as solar cells and nanoelectronic power sources Nature 449 885–9 [9] Chan C K, Peng H, Liu G, McIlwrath K, Zhang X F, Huggins R A and Cui Y 2008 High-performance lithium battery anodes using silicon nanowires Nat. Nano 3 31–5
Acknowledgments C-C Chang, C-C Chen, and JT were supported as part of the Center for Energy Nanoscience, an Energy Frontier Research 6
Nanotechnology 22 (2011) 305709
C-C Chang et al
[10] Ruffo R, Hong S S, Chan C K, Huggins R A and Cui Y 2009 Impedance analysis of silicon nanowire lithium ion battery anodes J. Phys. Chem. C 113 11390–8 [11] Cui L-F, Ruffo R, Chan C K, Peng H and Cui Y 2009 Crystalline–amorphous core–shell silicon nanowires for high capacity and high current battery electrodes Nano Lett. 9 491–5 [12] Wacaser B A, Reuter M C, Khayyat M M, Wen C-Y, Haight R, Guha S and Ross F M 2009 Growth system, structure, and doping of aluminum-seeded epitaxial silicon nanowires Nano Lett. 9 3296–301 [13] Goldberger J, Hochbaum A I, Fan R and Yang P 2006 Silicon vertically integrated nanowire field effect transistors Nano Lett. 6 973–7 [14] Dong Y, Yu G, McAlpine M C, Lu W and Lieber C M 2008 Si/a-Si core/shell nanowires as nonvolatile crossbar switches Nano Lett. 8 386–91 [15] Liu L Z, Wu X L, Zhang Z Y, Li T H and Chu P K 2009 Raman investigation of oxidation mechanism of silicon nanowires Appl. Phys. Lett. 95 093109–3 [16] Scheel H, Reich S, Ferrari A C, Cantoro M, Colli A and Thomsen C 2006 Raman scattering on silicon nanowires: the thermal conductivity of the environment determines the optical phonon frequency Appl. Phys. Lett. 88 233114 [17] Su Z, Sha J, Pan G, Liu J, Yang D, Dickinson C and Zhou W 2006 Temperature-dependent Raman scattering of silicon nanowires J. Phys. Chem. B 110 1229–34 [18] Chen Y, Peng B and Wang B 2007 Raman spectra and temperature-dependent Raman scattering of silicon nanowires J. Phys. Chem. C 111 5855–8 [19] Scheel H, Khachadorian S, Cantoro M, Colli A, Ferrari A C and Thomsen C 2008 Silicon nanowire optical Raman line shapes at cryogenic and elevated temperatures Phys. Status Solidi b 245 2090–3
[20] Faraci G, Gibilisco S, Russo P, Pennisi A R and La Rosa S 2006 Modified Raman confinement model for Si nanocrystals Phys. Rev. B 73 033307 [21] Miska P, Dossot M, Nguyen T D, Grun M, Rinnert H, Vergnat M and Humbert B 2010 Embedded silicon nanocrystals studied by photoluminescence and Raman spectroscopies: exciton and phonon confinement effects J. Phys. Chem. C 114 17344–9 [22] Beeman D, Tsu R and Thorpe M F 1985 Structural information from the Raman spectrum of amorphous silicon Phys. Rev. B 32 874 [23] Wu X L, Siu G G, Tong S, Liu X N, Yan F, Jiang S S, Zhang X K and Feng D 1996 Raman scattering of alternating nanocrystalline silicon/amorphous silicon multilayers Appl. Phys. Lett. 69 523–5 [24] Smit C, Swaaij R A C M M V, Donker H, Petit A M H N, Kessels W M M and Sanden M C M V D 2003 Determining the material structure of microcrystalline silicon from Raman spectra J. Appl. Phys. 94 3582–8 [25] Lu Z Q, Quinn T and Reehal H S 2005 Polarization-dependent Raman spectra of thin crystalline silicon films J. Appl. Phys. 97 033512 [26] Paillard V, Puech P, Laguna M A, Carles R, Kohn B and Huisken F 1999 Improved one-phonon confinement model for an accurate size determination of silicon nanocrystals J. Appl. Phys. 86 1921–4 [27] Kolb G, Salbert T and Abstreiter G 1991 Raman-microanalysis of strain and crystal orientation in laser-crystallized silicon Fresenius’ J. Anal. Chem. 341 166–70 [28] Hyun J and Lauhon L 2011 Spatially resolved plasmonically enhanced photocurrent from Au nanoparticles on a Si nanowire Nano Lett. at press
7
