Fabrication of Crystalline Silicon Spheres by Selective Laser Heating

Mar 17, 2011 - materials from a cooling tower and successfully applied them in ... to-point resolution) observation, X-ray diffraction (XRD, Rigaku Ul...
0 downloads 0 Views 3MB Size
ARTICLE pubs.acs.org/Langmuir

Fabrication of Crystalline Silicon Spheres by Selective Laser Heating in Liquid Medium Xiangyou Li, Alexander Pyatenko, Yoshiki Shimizu, Hongqiang Wang, Kenji Koga, and Naoto Koshizaki* Nanosystem Research Institute, National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba Central 5, 1-1-1 Higashi, Tsukuba, Ibaraki 305-8565, Japan ABSTRACT: Micrometer and submicrometer crystalline silicon spheres were fabricated by selective laser heating of irregular silicon particles in liquid medium. TEM, SEM, XRD, and XPS characterized the structure and morphology of the prepared silicon spheres. The results suggested that they were spherical with a single crystalline structure. In this study, the formation mechanism of the spheres is analyzed, and the process parameters are optimized to obtain high-quality silicon spheres. A theoretical deduction regarding the relationship between critical laser energy density and particle size is also discussed, by which we can predict that larger spheres can be obtained at higher laser energy densities.

1. INTRODUCTION Micrometer or submicrometer spheres are important because they can be applied in many fields.1,2 High-quality microspheres can be used as optical devices (e.g., microlaser,3 micro-optical force sensor,4 liquid-crystal tunable filter,5 optical switching,6 displacement measurement,7 and Raman sources8). More recently, more and more people have become interested in nanosize-dependent cytotoxicity in vivo,9,10 making microspheres more attractive because they can be an alternative to nanomaterials in the medical field. Although the urgent requirement of microspheres has attracted many researchers’ interest, most of the research with spherical structures is based on noncrystalline materials such as glass (e.g., SiO2) or polymers [e.g., polystyrene (PS) and polymethylmethacrylate (PMMA)].11 Crystalline spheres have rarely been reported, possibly because the conventional heating preparation method generally induces nucleus formation and subsequent preferential crystal growth. Therefore, it is kinetically difficult to inhibit anisotropic crystal growth, making it difficult to fabricate crystalline spheres.12,13 Silicon is the popular and dominating material in the current optical and integrated-circuit areas. Therefore, fabricating silicon microspheres might result in new applications or a breakthrough in the current optoelectronic industry. Unfortunately, up to now, only a few studies have reported on the fabrication of silicon spheres. Minemoto et al. fabricated silicon spheres by dropping materials from a cooling tower and successfully applied them in solar cells with a semi-light-concentration system.14,15 Liu and co-workers developed a novel seeding crystallization technique by injecting silicon powder into molten silicon droplets as nuclei and stimulated solidification at low undercooling to form crystalline silicon spheres.16 However, spheres fabricated by dropping are often around the millimeter level, and a high temperature is necessary for silicon fusion. More recently, Lin and co-workers r 2011 American Chemical Society

applied a homogenized KrF excimer laser to fabricate silicon spheres on the silicon on insulator (SOI) platform and obtained 600 nm silicon spheres with an extremely smooth surface.17 However, it is necessary to use lithography technology to prepare the starting silicon rods, and the complex procedures and high cost restrict further application of this technology. Fenollosa et al. also demonstrated a method to obtain silicon microspheres by chemical vapor deposition techniques through decomposing disilane gas at high temperature.18 This method also needs complex procedures and expensive equipments. Here we present a very simple approach to fabricating micrometer and submicrometer silicon spheres by selective laser heating of irregular silicon particles in liquid medium. The prepared silicon spheres have a single crystalline structure with a smooth surface. Many researchers have fabricated nanosilicon particles in liquid medium;1922 however, to our knowledge, this is the first time that micrometer and submicrometer silicon spheres have been obtained by selective laser heating in liquid medium.

2. EXPERIMENTAL SECTION In a typical procedure, 0.02 g of micrometer sized silicon material (Kojundo Chemical Laboratory Co., Ltd., particle size 5 μm, purity >99%) was pulverized for 1 h by a mechanical milling machine (planetary ball mill, Fritsch Corp., type-07.301) and dispersed into 50 mL of solvent (e.g., ethanol and water). After ultrasonication, a 4 mL suspension was dropped into a glass cell for laser irradiation. The cell was covered by a quartz window. A Nd:YAG pulsed laser (Spectra-Physics, repetition rate 30 Hz, pulse width 8 ns, beam diameter 8 mm) with Received: January 19, 2011 Revised: March 1, 2011 Published: March 17, 2011 5076

dx.doi.org/10.1021/la200231f | Langmuir 2011, 27, 5076–5080

Langmuir

Figure 1. Schematic illustration of the experimental setup.

ARTICLE

Figure 3. XRD analysis of the particles before (black) and after (red) laser irradiation.

Figure 4. XPS depth profile analysis of Si 2p atomic concentration from surface (bottom) to inside (top) of the particles obtained in ethanol. Numbers at the right side are the depth from the surface calibrated by SiO2 etching rate.

3. RESULTS AND DISCUSSION

Figure 2. SEM of silicon particles before (a, b) and after (cf) laser irradiation (460 mJ/pulse 3 cm2, 30 min). Images a, c, and e are the morphologies without milling and b, d, and f are those with milling. Images e and f are magnified images of c and d, respectively. second harmonics (532 nm) was used to irradiate the suspension without focusing. During laser irradiation, a magnetic stirrer was used to prevent gravitational settling of the suspension. Figure 1 schematically depicts the experimental setup. After laser irradiation, the solution was centrifuged (6000 rpm, 30 min) and most of the supernatant was removed. Before further characterization, the sediment was washed with Milli-Q water. Several small droplets of the sediment-containing liquid was deposited onto a silicon and gold substrate for SEM (Hitachi S-4800 scanning electron microscope with 30 kV acceleration and 1 nm pointto-point resolution) observation, X-ray diffraction (XRD, Rigaku Ultima IV/PSK, Cu KR), and X-ray photoelectron spectroscopy (XPS, PHI Versa Probe) analysis after natural drying. A small droplet of the above solution was also deposited on a copper grid for high-resolution transmission electron microscopy (HRTEM, JEM-3000F) observation. The size of the particles in the colloidal solutions was measured by dynamic light scattering (DLS) with a particle size analyzer (Malvern, Zetasizer Nano-ZS).

3.1. Morphology. Figure 2 presents the typical morphology of starting materials and as-prepared silicon spheres. The starting silicon particles (Figure 2a) were mainly irregular, and the main part was large particles of several micrometers. After mechanical milling for 1 h, most of the very large particles were crushed into smaller and more uniform particles (Figure 2b). Without milling, only a few tiny irregular particles could be transformed into spheres by laser irradiation of the particles (Figure 2c). However, when the particles after milling were irradiated by laser, almost all of them became spherical (Figure 2d). This is because the particles without milling were too large and applied laser energy could not melt them entirely. However, when large particles were made smaller by milling, they could be melted entirely to form spheres. Figure 2e,f is the magnified images of Figure 2c,d, respectively. From this figure, it is easy to observe the differences between them. At the same time, we observe that only the surface of the large irregular particles was melted in Figure 2e. 3.2. Structures and Characterization. Figure 3 presents the XRD data of raw silicon particles and as-prepared silicon spheres. Almost no difference was found before and after laser irradiation, confirming that our technology changed just the morphology. Figure 4 depicts the XPS depth profiles of the silicon spheres, indicating that a thin oxidized layer formed on the surface. Comparison of O 1s XPS depth profiles of spheres obtained in water and in ethanol indicates that silicon spheres might be oxidized more easily in water, which can be proved by the increase of the oxygen atomic concentration (Figure 5). Therefore, ethanol is much better than water in this technology. 5077

dx.doi.org/10.1021/la200231f |Langmuir 2011, 27, 5076–5080

Langmuir A lattice plane can be clearly observed in the HR-TEM image, which also indicates that the structure of the prepared spheres is crystalline (Figure 6). However, this is the structure of a local area on a sphere, because we cannot obtain a lattice image of the whole large sphere. In order to judge whether it is singlecrystalline or polycrystalline, we characterized the whole sphere by SAED. Figure 7b is the SAED pattern obtained from the whole sphere shown in Figure 7a. This TED pattern exhibits a single regular spot array, indicating a single crystallographic orientation in a sphere, identified as the silicon [213] pattern.

Figure 5. XPS depth profile of O 1s atomic concentration of silicon spherical particles obtained in water and ethanol (460 mJ/pulse 3 cm2, 30 min). The depth was calibrated by the SiO2 etching rate.

Figure 6. HR-TEM of prepared silicon spheres obtained in ethanol (460 mJ/pulse 3 cm2, 30 min).

ARTICLE

Therefore, the prepared sphere was assumed to be a single crystalline structure. 3.3. Milling Effects. The milling technique is also very important for obtaining a large amount of silicon spheres. Large particles are crushed into small particles with lower heat capacity, resulting in complete melting by laser irradiation. The absorption increases overwhelmingly after milling. Figure 8 demonstrates an absorption increase of 3.6 times at 532 nm due to milling. The possible reason is that, when large particles are crushed into small particles by milling, fracture planes are induced, and the absorption areas increase correspondingly. Furthermore, Hwang et al.23 found that increased structure disorder in crystalline silicon predictably enhances optical absorption. Milling may also increase structural disorder by collision and crushing, which results in optical absorption enhancement and enables spheres to form more easily. Additionally, after laser irradiation, the absorption became weak, and this might arise from the recrystallization of the disordered structure and the oxidation of some small particles. 3.4. Process Parameters and Optimization. Figure 9 illustrates the sizes of the silicon particles or spheres determined by DLS measurement under different process conditions. Obviously, the peak position of particle size shifted to a smaller size after mechanical milling (red to green). The peak position of particle size after laser irradiation under the appropriate conditions (460 mJ/pulse 3 cm2, 30 min) is similar to that of particles before laser irradiation (green and blue), which proves that the particles melted to form spheres by themselves, not by merging with other particles. However, after 60 min of irradiation, the silicon spheres became much smaller (see blue to cyan), which may imply surface oxidation of the spheres and gradual etching by laser irradiation. Figure 10 presents SEM morphology under different laser irradiation conditions. Silicon spheres can be obtained only under appropriate process conditions (460 mJ/pulse 3 cm2, 30 min, Figure 10a). With lower laser energy density (Figure 10b) and shorter irradiation time (Figure 10c), most of the particles remain irregular. However, with longer irradiation time (Figure 10d), the amount of particles obviously decreases, perhaps due to oxidizing and gradual etching from the sphere surface after too long irradiation time. In summary, the optimum laser irradiation condition for silicon sphere fabrication is 330460 mJ/pulse 3 cm2 energy density for 3060 min. 3.5. Formation of Silicon Spheres. The mechanism of silicon sphere formation is quite intuitive. The starting materials include large and small irregular particles. When a pulsed laser is applied

Figure 7. TEM image of a larger silicon sphere (a) and the corresponding SAED patterns (b). Formation conditions: 460 mJ/pulse 3 cm2, 30 min. 5078

dx.doi.org/10.1021/la200231f |Langmuir 2011, 27, 5076–5080

Langmuir

ARTICLE

Figure 8. Absorption of silicon particles in ethanol before and after milling. Figure 10. SEM morphology under different process conditions: (a) 460 mJ/pulse 3 cm2, 30 min; (b) 262 mJ/pulse 3 cm2, 30 min; (c) 460 mJ/pulse 3 cm2, 10 min; and (d) 460 mJ/pulse 3 cm2, 180 min.

Figure 9. DLS measurement under different process conditions (460 mJ/pulse 3 cm2). The inset numbers are weighted mean and deviation of the particle size distribution.

to the colloidal solution, only the particles (not the solvent) are heated. This selective heating relies on the laser energy absorption of solid particles, as well as the low absorbance of the solvent medium. If the laser energy is sufficient, the particles melt and the shape of the particles changes, due to surface tension. If the energy is high enough to melt the particles completely, they become spherical. However, if the energy is insufficient to melt whole particles, the particles melt only on the surface, which results in surface smoothing without complete sphere formation. On the basis of the above discussion, silicon sphere formation is actually a process of heating and melting during pulse laser irradiation, and this process strongly depends on the optical absorption of the particles. To predict the critical conditions to form silicon spheres, laser selective heating in liquid was calculated. As it was shown in ref 24, the typical characteristic times of particle cooling process (both by irradiation and by boiling heat transfer) are about 104106 s, which is much shorter than the interval between two consecutive pulses (3  102 s) but much longer than the pulse duration (108 s). It permits us to neglect all the heat losses during the particle heatingmelting period and, on the other hand, to neglect the interpulse effect and to make all the calculations for one individual laser pulse.

Figure 11. Relationship between particle diameter and critical laser energy density based on heating and melting.

In such case, the laser energy absorbed by a particle is written as eq 1 Qabs ¼ Jσabs λ ðdp Þ

ð1Þ

where J is the laser fluence, dp is the diameter of the particle, and σabsλ is the particle absorption cross-section, which can be calculated by classical Mie theory using the complex refractive index of bulk silicon.2530 However, all the energy absorbed by the particle is consumed in heating and melting, which can be expressed as eq 2 Qabs ¼ Fp

πdp 3 s ½Cp ðTm  T0 Þ þ ΔHm  6

ð2Þ

where the density of silicon Fp is 2.32 g/cm3, the melting point Tm is 1685 K, and the ambient temperature T0 is 298.15 K. The heat capacities for solid silicon Cps and the enthalpy of melting ΔHm data were adopted from JANAF.29 The relationship between particle diameter and critical laser energy density for particle melting can be obtained by combining eqs 1 and 2 (Figure 11). The calculated results indicate that the critical laser energy density strongly depends on particle size, and 5079

dx.doi.org/10.1021/la200231f |Langmuir 2011, 27, 5076–5080

Langmuir with increasing laser fluence, the size of the formed spheres also increases. This result agrees with our experimental results. In our experiments, small particles formed more easily, and large particles are not spherical because they melt only on the surface. Our laser is restricted to a maximum fluence of only 460 mJ/pulse 3 cm2 without focusing; therefore, the maximum size of the spheres obtained is only 1000 nm at current maximum laser energy density. It is worth noting that some errors existed between experimental and calculated data. This is because our calculation is not a very exact calculation; there are some other factors not considered (e.g., solution, energy loss). However, the trend agrees with the experiment, and the data is on the same order of experimental results. Additionally, there is a bump in range of 300400 nm, and this might arise from the oscillations in absorbance as a function of particle diameter that are characteristic of Mie scattering. However, it is not difficult to predict that, with the development of laser technology, the laser energy density may be increased, and under larger energy, larger spheres can be obtained by this pulsed laser selective heating technology.

4. CONCLUSIONS In summary, micrometer and submicrometer (2001000 nm silicon spheres were fabricated by pulsed laser selective heating in liquid medium. Compared with other methods, this is a very simple way to prepare silicon spheres, which has not been reported before, to our knowledge. The silicon spheres have a single crystalline structure and a smooth surface. The processes to fabricate such silicon spheres were optimized, and the formation mechanism was discussed in this paper. Furthermore, the relationship between critical laser energy density and sphere size was calculated and simulated, allowing us to predict that larger spheres can be obtained at higher laser energy density. These silicon spheres can be possibly applied in many fields (e.g., solar cells, medicine optics, and measurement). ’ AUTHOR INFORMATION Corresponding Author

*Tel: (þ) 81-29-861-4879. Fax: (þ) 81-29-861-6355. E-mail: [email protected].

’ ACKNOWLEDGMENT This work was supported by the Japan Society for the Promotion of Science (JSPS: P09508) fellowship at the National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba, Japan.

ARTICLE

(8) Spillane, S. M.; Kippenberg, J. T.; Vahala, K. J. Nature 2002, 415, 621–623. (9) Nel, A. E.; M€adler, L.; Velegol, D.; Xia, T.; Hoek, E. M. V.; Somasundaran, P.; Klaessig, F.; Castranova, V.; Thompson, M. Nat. Mater. 2009, 8, 543–557. (10) Bhabra, G.; Sood, A.; Fisher, B.; Cartwright, L.; Saunders, M.; Evans, W. H.; Surprenant, A.; Lopez-Castejon, G.; Mann, S.; Davis, S. A.; Hails, L. A.; Ingham, E.; Verkade, P.; Lane, J.; Heesom, K.; Newson, R.; Case, C. P. Nat. Nanotechnol. 2009, 4, 876–883. (11) Im, S. H.; Jeong, U. Y.; Xia, Y. Nat. Mater. 2005, 4, 671–675. (12) Wang, H.; Pyatenko, A.; Kawaguchi, K.; Li, X.; SwiatkowskaWarkocka, Z.; Koshizaki, N. Angew. Chem., Int. Ed. 2010, 49, 6361–6364. (13) Ishikawa, Y.; Shimizu, Y.; Sasaki, T.; Koshizaki, N. Appl. Phys. Lett. 2007, 91, 161110.1–3. (14) Okamoto, C.; Tsujiya, K.; Minemoto, T.; Murozono, M.; Takakura, H.; Hamakawa, Y. Jpn. J. Appl. Phys. 2005, 44, 8351–8355. (15) Minenoto, T.; Okamoto, C.; Omae, S.; Murozono, M.; Takakura, H.; Hamakawa, Y. Jpn. J. Appl. Phys. 2005, 44, 4820–4824. (16) Liu, Z.; Nagai, T.; Masuda, A.; Kondo, M.; Sakai, K.; Asai, K. J. Appl. Phys. 2007, 101, 093505.1–5. (17) Hung, S.; Shiu, S.; Chao, C.; Lin, C. J. Vac. Sci. Technol. B 2009, 273, 1156–1160. (18) Fenollosa, R.; Meseguer, F.; Tymczenko, M. Adv. Mater. 2008, 20, 95–98. (19) Svrcek, V.; Sasaki, T.; Shimizu, Y.; Koshizaki, N. Appl. Phys. Lett. 2006, 89, 213113.1–8. (20) Qin, W.; Kulinich, S. A.; Yang, X.; Sun, J.; Du, X. J. Appl. Phys. 2009, 106, 114318.1–6. (21) Botti, S.; Coppola, R.; Gourbilleau, F.; Rizk, R. J. Appl. Phys. 2000, 88, 3396–3401. (22) Li, X.; He, Y.; Swihart, M. T. Langmuir 2004, 20 (11), 4720–4727. (23) Bondi, R. J.; Lee, S.; Hwang, G. S. Phys. Rev. B 2010, 82, 115214.1–12. (24) Pyatenko, A.; Yamaguchi, M.; Suzuki, M. J. Phys. Chem. C 2009, 113, 9078–9085. (25) Mie, G. Ann. Phys. 1908, 330, 377–445. (26) Bohren, C. F.; Huffman, D. Absorption and Scattering of Light by Small Particles; John Wiley: New York, 1983. (27) Kreibig, U.; Vollmer, M. Optical Properties of Metal Clusters, Springer Series in Material Science 25; Springer: Berlin, 1995. (28) Kelly, K. L.; Jensen, T.; Lazarides, A.; Schatz, G. C. In Metal Nanoparticles. Synthesis, Characterization and Application; Feldheim, D. L., Foss, C. A.; Jr., Eds.; Dekker: New York, 2002. (29) Edward Palik, D. Handbook of Optical Constants of Solids; Academic Press: New York,1988. (30) Thermodynamic database MALT group. Thermodynamic database MALT for Windows; Kagaku Gijutsu-Sha, 2005, CD-ROM available from http://www.kagaku.com/malt/index.html (accessed 12/01/ 2010).

’ REFERENCES (1) Li, Y.; Li, C.; Cho, S. O.; Duan, G.; Cai, W. Langmuir 2007, 23, 9802–9807. (2) Cai, M.; Painter, O.; Vahala, K. J. Opt. Lett. 2000, 25, 1430–1431. (3) Li, Y; Huang, X. J.; Heo, S. H.; Li, C; Choi, Y. K.; Cai, W.; Cho, S. O. Langmuir 2007, 23, 2169–2174. € ugen, (4) Ioppolo, T.; Kozhevnikov, M.; Stepaniuk, V.; Volkan Ot€ M.; Sheverev, V. Appl. Opt. 2008, 47, 3009–3014. (5) Gilardi, G.; Donisi, D.; Serpeng€uzel, A.; Beccherellil, R. Opt. Lett. 2009, 34, 3253–3255. (6) Tapalian, C.; Laine, J. P.; Lane, P. A. IEEE Photonics Technol. Lett. 2002, 14, 1118–1120. (7) Laine, J. P.; Tapalian, H. C.; Little, B. E.; Haus, H. A. Sens. Actuators 2001, A 93, 1–7. 5080

dx.doi.org/10.1021/la200231f |Langmuir 2011, 27, 5076–5080