Generation of Charged Nanoparticles and Their Deposition Behavior

Nov 7, 2012 - Seong-Han Park , Jin-Woo Park , Seung-Min Yang , Kwang-Ho Kim ... Sung-Soo Lee , Chan-Soo Kim , Tae-Sung Kim , and Nong-Moon Hwang...
0 downloads 0 Views 2MB Size
Article pubs.acs.org/JPCC

Generation of Charged Nanoparticles and Their Deposition Behavior under Alternating Electric Bias during Chemical Vapor Deposition of Silicon Woong-Kyu Youn,† Chan-Soo Kim,‡ Jae-Young Lee,† Sung-Soo Lee,† and Nong-Moon Hwang*,† †

Department of Materials Science and Engineering, Seoul National University, Daehak-dong, Gwanak-gu, Seoul 151-744, Republic of Korea ‡ Jeju Global Research Center, Korea Institute of Energy Research, Gimnyeong-ri, Gujwa-eup, Jeju-si, Jeju-do 695-971, Republic of Korea ABSTRACT: The generation of charged nanoparticles was confirmed under typical conditions of the thin film deposition during the silicon chemical vapor deposition (CVD) process. To exert the electric force on these charged nanoparticles, the alternating current (ac) bias was applied to the stainless substrate holder during CVD. The bias frequency and voltage significantly affected the microstructure evolution and the growth rate. As the frequency at ±100 V was increased from 0.2 to 0.5, 1, and 5 Hz, the mass of the film became respectively 1.1, 2.2, 3.0, and 3.5 times compared with that of the film deposited under the zero bias. As the bias voltage at 1 Hz was increased from 0 to ±50, ±100, ±150, and ±200 V, the mass was increased respectively by 1.8, 3, 4.5, and 8.5 times.

1. INTRODUCTION In the chemical vapor deposition (CVD) process, it has been believed that the reactant gases decompose mainly on the surface and produce atoms or molecules, which contribute to the growth of films or nanostructures.1 However, recently it has been shown that the reactant gases also decompose actively in the gas phase in many CVD systems, which is evidenced by the extensive generation of the gas phase nuclei.2−15 Using a differential mobility analyzer (DMA) or other particle detecting systems attached to a CVD reactor, Adachi et al.2,16−18 made an extensive study that a huge amount of SiO2 nanoparticles was generated in the gas phase almost unavoidably under the deposition condition of SiO2 films. The abundant generation of charged nanoparticles was confirmed during the syntheses of diamond films,3,4 zirconia films,5 copper films,6 silicon films,7,8 silicon nanowires,9 ZnO nanowires,10 carbon nanotubes,11,12 and GaN films or nanowires.13 Considering these results, the generation of charged nanoparticles is so general that it appears to be the rule rather than the exception in the CVD process. The reason why the generation of nanoparticles in the thermal CVD process has been unnoticed for so long would be that they are invisible to the naked eye because they are far smaller than the wavelength of visible light. In contrast with this, the nanoparticles in the plasma enhanced CVD (PECVD) process have been relatively well-known because they often grow to a visible size, which is called “dusty plasma”.19,20 It is well established that the plasma wall is negatively charged because electrons have much higher mobility than ions. Likewise, if gas phase nucleation occurs and the nuclei grow into nanoparticles, electrons will be attached to them and as a © 2012 American Chemical Society

result the nanoparticles become negatively charged. Then, the residence time of negatively charged nanoparticles increased compared with other neutral species because the plasma relatively abundant in positive ions attracts the negatively charged nanoparticles.21 Although charged nanoparticles suspended in the PECVD process had been believed to be harmful if incorporated into films, Cabarrocas22,23 suggested that these nanoparticles produced in the silane plasma improve the electrical property of the films if incorporated into films. He called such nanoparticle-incorporated films “polymorphous films”. The contribution of the nanoparticles to deposition has also been studied extensively by Ostrikov24,25 in the PECVD process. As to the possibility that nanoparticles can be a building block of epitaxial growth of films, Yoshida and his colleagues26−30 made extensive studies on the epitaxial growth of YBa2Cu3O7−x films with a building block of nanoparticles by the method called thermal plasma flash evaporation. Using this method, they could deposit high-quality epitaxial YBa2Cu3O7−x films with a growth rate as high as 16 nm/s.29 They called the related phenomenon “hot cluster epitaxy”. Using a microtrench fabricated on a Si wafer, they could estimate the size of the depositing nanoparticles to be about 0.3−10 nm.28 The fact that a significant amount of charged nanoparticles is formed during the CVD process has important implications since they could play a significant role in the microstructure evolution of films or nanostructures.14,15 Considering that most Received: October 30, 2012 Published: November 7, 2012 25157

dx.doi.org/10.1021/jp310705p | J. Phys. Chem. C 2012, 116, 25157−25163

The Journal of Physical Chemistry C

Article

3000F). To compare the growth rate, the mass change of films was measured by a microbalance (Denver Instrument TB215D). To examine the crystallinity of films, the Raman spectra (Jobin Yvon, LabRam HR Raman spectrometer) were analyzed with an Ar ion laser beam (514.5 nm) excitation and an incident power of 0.5 mW.

of nanoparticles are charged, applying the electric bias to the substrate would exert an electric force on these charged nanoparticles and direct their motion according to the given electric field. Motivated by these considerations, first we tried to confirm the generation of charged nanoparticles by measuring their size distribution using a DMA combined with a Faraday cup electrometer (FCE) in the silicon CVD process. Second, we examined the alternating current (ac) bias effect on the microstructure evolution and the growth rate. Applying the ac bias turned out to be a new powerful processing parameter in the thermal CVD process where charged nanoparticles are generated.

3. RESULTS AND DISCUSSION Figure 2 shows the size distribution and the number concentration of both positively and negatively charged silicon

2. EXPERIMENTAL PROCEDURES Silicon films were deposited on a quartz substrate for 2 h by conventional atmospheric-pressure CVD using a quartz tube reactor at a deposition temperature of 900 °C with silane (SiH4) diluted in helium (He), hydrogen (H2), and nitrogen (N2). Since quartz is insulating, the quartz substrate of 10 × 10 × 1 mm3 was on a stainless substrate holder plate of 12 × 12 × 6 mm3 to apply the bias. The substrate was placed at the center zone of the quartz tube. Another electrode of a stainless plate was placed above the substrate. The distance between the substrate holder and the upper electrode was fixed at 1 cm. The bias by the ac power supply was applied between the two electrodes with the upper electrode being grounded. The schematic of the experimental setup is shown in Figure 1. The flow rates of 10% SiH4−90% He, H2 (99.9999%), and N2 (99.9999%) were respectively 5, 50, and 1000 sccm.

Figure 2. Size distribution and the number concentration of negatively (open circle) and positively (closed circle) charged nanoparticles at a reactor temperature of 900 °C with gas flow rates of 5 sccm 10% SiH4−90% He, 50 sccm H2, and 1000 sccm N2.

nanoparticles generated during the CVD process, which was measured by the DMA−FCE system. Below the size of ∼30 nm, the number concentration of positively charged nanoparticles is higher than that of negatively charged ones, but from ∼30 to ∼75 nm, the number concentration of positively charged nanoparticles is lower than that of negatively charged ones. Above the size of ∼75 nm, the number concentration of charged nanoparticles was reversed again. However, the total number concentrations of positively and negatively charged nanoparticles are almost the same. As suggested by Hwang and Kim,14 the possible mechanism for the formation of charged nanoparticles is that nucleation takes place first in the gas phase and the neutral nuclei subsequently undergo surface ionization on any surface such as the quartz tube of the reactor. The positive and negative surface ionizations are described by the Saha−Langmuir equations32 as

Figure 1. Schematic of the experimental setup for the synthesis of silicon films with the ac bias applied to the substrate holder during the CVD process.

A DMA (TSI model 3081)31 combined with a FCE, which was designed to measure particle sizes in the range of 10−1000 nm, was connected to the CVD reactor for measurements of the size distribution of charged nanoparticles generated during CVD. The amount of charged nanoparticles, which were sizeclassified by DMA, was measured as a current on a FCE. Our measuring system did not use the charging system because nanoparticles are self-charged in the quartz reactor of the CVD system.9,10 The size distribution and number concentration of charged nanoparticles were measured in situ during the synthesis of silicon films. Details of the experimental setup for the DMA−FCE system connected to the CVD reactor can be found elsewhere.9−11 A sampling position for nanoparticles was just above the substrate. After measuring the size distribution of the positively and negatively charged silicon nanoparticles in the gas phase during CVD, we applied the electric ac bias to the stainless substrate holder. The microstructure was observed by the field-emission scanning electron microscopy (FESEM, JSM-7500F) and highresolution transmission electron microscopy (HRTEM, JEM-

g+ ⎛ IP − WF ⎞ n+ ⎟ = 0 exp⎜ − 0 ⎝ kT ⎠ n g

(1)

g ⎛ WF − EA ⎞ n ⎟ = 0 exp⎜ − 0 ⎝ ⎠ kT n g

(2)





where IP, EA, WF, and k are ionization potential, electron affinity of nanoparticles, work function of the quartz wall, and Boltzmann constant, respectively. n0, n+, and n− are the numbers of neutral, positive, and negative species, respectively. g0, g+, and g− indicate the statistical weight of neutral, positive, and negative species, respectively. Therefore, WF of the reactor wall and IP and EA of nanoparticles generated in the gas phase are the key factor in determining the polarity of charged 25158

dx.doi.org/10.1021/jp310705p | J. Phys. Chem. C 2012, 116, 25157−25163

The Journal of Physical Chemistry C

Article

Figure 3. FESEM images of silicon films deposited on a quartz substrate at (a) the zero bias and the ac bias voltage of ±100 V with frequencies of (b) 0.2, (c) 0.5, and (d) 5 Hz applied to the substrate holder for 2 h at a reactor temperature of 900 °C at a SiH4 flow rate of 5 sccm, H2 flow rate of 50 sccm, and N2 flow rate of 1000 sccm.

Figure 4. FESEM images of silicon films deposited on a quartz substrate at the ac bias voltages of (a) ±50, (b) ±100, (c) ±150, and (d) ±200 V with the frequency of 1 Hz applied to the substrate holder under the same processing conditions as Figure 3.

holder. For comparison, the silicon was deposited on the substrate without the bias under the same processing conditions. Figure 3 shows the FESEM images of the surface morphology of silicon films deposited on the quartz substrates for 2 h at the reactor temperature of 900 °C with gas flow rates of 5 sccm 10% SiH4−90% He, 50 sccm H2, and 1000 sccm N2.

nanoparticles. Since the nanoparticles generated in this experiment are sufficiently large, both IP and EA of nanoparticles approach the work function value of their bulk.33 To examine the frequency effect of the alternating bias on the deposition behavior, the ac bias voltage of ±100 V with frequencies of 0.2, 0.5, and 5 Hz was applied to the substrate 25159

dx.doi.org/10.1021/jp310705p | J. Phys. Chem. C 2012, 116, 25157−25163

The Journal of Physical Chemistry C

Article

Figure 5. Low-magnification FESEM images for silicon films deposited on a quartz substrate at the ac bias voltages of (a) ±50, (b) ±100, (c) ±150, and (d) ±200 V with the frequency of 1 Hz applied to the substrate holder under the same processing conditions as Figure 3.

at ±150 V. The microstructure deposited at ±200 V was much more porous than that deposited at ±150 V, whose aspect was shown more clearly in FESEM images of lower magnification. To have better understanding of the effect of the bias voltage on the microstructure evolution, the lower magnifications of Figure 4 are shown in Figure 5. The microstructure became more porous as the bias voltage increased. The film deposited at ±50 V is relatively dense as shown in Figure 5a. Although the porosity difference between Figures 4c,d appeared not to be appreciable, the microstructure of lower magnification showed a drastically different porosity as shown in Figures 5c,d. These results show that drastically different microstructures were evolved by changing the intensity of the alternating bias. To observe the magnified image of nanorods attached to nanoparticles in Figure 4d, the sample of Figure 4d was immersed in ethanol with ultrasonic treatment. The suspended particles, which had been captured on a TEM grid membrane of holey carbon, were observed by HRTEM. Figure 6a shows the low-magnification TEM morphology of the silicon nanoparticles with their surface covered with numerous nanorods. Individual nanorods were single crystalline as revealed by the HRTEM image of Figure 6c. Figure 6c shows an enlarged-lattice image of Figure 6b. The spacing between the parallel fringes of the crystalline was measured to be 0.31 nm. It is equal to the spacing of the {111} planes of crystalline silicon. These results imply that controlling the ac bias frequency or the magnitude of bias voltages may produce various microstructures. Besides, the ac bias technique can provide a promising method to produce highly porous microstructures. Charged nanoparticles are a kind of nanosized colloidal particles, which undergo self-assembly and produce highly ordered three-dimensional arrays. This phenomenon is called “colloidal crystallization”.36 In addition to the property of selfassembly, charged nanoparticles have the property of chargeenhanced diffusion, resulting in the epitaxial colloidal

Figure 3a is for the zero bias, and Figures 3b−d are for the bias voltage of ±100 V at frequencies of 0.2, 0.5, and 5 Hz, respectively. At the zero bias as shown in Figure 3a, the film shows appreciable roughness on the surface. At the frequency of 0.2 Hz, the surface is mainly covered with flakelike structures as shown in Figure 3b. At 0.5 Hz, a flakelike structure disappeared, and the surface became smoother than that of 0.2 Hz as shown in Figure 3c. At 5 Hz, the surface had a very porous structure as shown in Figure 3d. The surface morphology of the film deposited at the zero bias (Figure 3a) is similar to that of films deposited by PECVD34 and hotwire CVD (HWCVD).35 Care must be taken of when the ac bias is applied in PECVD and HWCVD processes, where the amount of negatively charged nanoparticles is much larger than that of positively charged ones. When the amount of positively and negatively charged nanoparticles is pronouncedly unbalanced, the unbalanced ac bias would be necessary to produce the result similar to Figure 3. These results indicate that the surface morphology of films can be changed considerably by changing the frequency of the alternating bias. To examine the effect of the bias voltage, the bias was applied to the stainless substrate holder at the biases of ±50, ±100, ±150, and ±200 V with a frequency of 1 Hz. Figure 4 shows the FESEM images of the surface morphology of silicon films deposited under the same processing condition of Figure 3. Under the bias of ±50 V as shown in Figure 4a, the film microstructure tended to have a flakelike structure. When the bias was increased to ±100 V, the microstructure tended to have round nodules as shown in Figure 4b. When the bias was increased to ±150 V, the microstructure became porous as shown in Figure 4c. The nanoparticles of ∼100 nm were aggregated as a chainlike structure with their surface covered with numerous nanorods. Finally, as the bias was increased to ±200 V, the nanoparticles deposited on the surface were ∼50 nm as shown in Figure 4d, being smaller than those deposited 25160

dx.doi.org/10.1021/jp310705p | J. Phys. Chem. C 2012, 116, 25157−25163

The Journal of Physical Chemistry C

Article

catalytic effect,38 and electrode materials.39 The bias frequency and the magnitude of bias voltages can be changed during deposition, producing more diverse microstructure. This technique could also produce films of a layered structure with each layer of different microstructures. For example, the first layer is dense, the second layer is porous, and the third layer is dense again. These results provide a significant implication in the CVD process, showing clearly that the electric bias, which can be either ac or dc, can be a new processing parameter in the deposition process, where the charged nanoparticles are generated. To examine the bias effect on the film growth rate, the mass increase in the substrate after deposition was measured using a microbalance with a scale in 10−5 g. Figure 7a shows the effect of the bias frequency on the mass change of silicon films grown on the substrate. As the frequency was increased, the mass continued to increase. As the frequency was increased from 0.2 to 0.5, 1, and 5 Hz, the mass of the film was increased by 1.1, 2.0, 3.0, and 3.5 times, respectively, compared with that of the film deposited under the zero bias. The slope of the mass increase with respect to bias frequency in Figure 7a tended to decrease with increasing frequency. Figure 7b shows the effect of the bias voltage on the mass change of silicon films. The effect of the mass increase by bias voltage is more pronounced than that by bias frequency. As the bias voltage was increased from 0 to ±50, ±100, ±150, and ±200 V, the mass was increased respectively by 1.8, 3, 4.5, and 8.5 times. Increasing the bias voltage applied to the substrate holder means the increase of the electric field between the two electrodes, which would increase the electric force to attract the charged nanoparticles toward the electrodes. Considering that both positively and negatively charged nanoparticles exist abundantly in the gas phase as shown in Figure 2 and that the charge would not build up on the quartz substrate because the ac bias is applied, it is expected that the deposition rate continued to increase with increasing bias voltage as shown in Figure 7b. The increased deposition rate by the ac bias indicates the increased production yield. Therefore, in the process where the production yield is important, the ac bias can be highly effective. For example, in the Siemens process, which produces polysilicon by decomposing trichlorosilane using CVD, the gas phase nuclei would have an adversary effect on the production yield. The ac bias would efficiently collect the charged nanoparticles, increasing the production yield.

Figure 6. TEM images of silicon nanoparticles with numerous nanorods deposited on a quartz substrate at the ac bias voltages of ±200 V: (a) a low-magnification TEM image, (b) a high-magnification TEM image, and (c) an enlarged HRTEM image from the squareenclosed area.

crystallization without voids. Therefore, the self-assembly of charged nanoparticles tends to produce a dense structure. The ac bias would prevent charged nanoparticles from undergoing the self-assembly and produce a porous structures. The structure grown by charged nanoparticles without the ac bias would resemble the dense deflocculation of colloidal particles whereas that with the ac bias would resemble the porous flocculation of them. The highly porous structure has a high surface area, which is favorable for a low dielectric constant,37

Figure 7. Mass changes of silicon films deposited on a quartz substrate for (a) the ac bias frequencies and (b) the voltages. 25161

dx.doi.org/10.1021/jp310705p | J. Phys. Chem. C 2012, 116, 25157−25163

The Journal of Physical Chemistry C

Article

Figure 8. Raman spectra of silicon films deposited on a quartz substrate for (a) the ac bias frequencies and (b) the voltages.

Figures 8a and 8b show the Raman spectra of the films deposited with different bias frequency and different bias voltage, respectively. To determine the volume fraction of crystalline silicon in the films, Xc, the Raman spectra in Figure 8a were deconvoluted into three parts of crystalline Ic at about 520 cm−1, intermediate Im at around 505−517 cm−1, and amorphous component Ia at around 480 cm−1.40 The volume fraction of the crystalline phase is determined by Xc = (Ic + Im)/ (Ic + Im + Ia). Xc was evaluated to be 0.50, 0.56, 0.60, 0.67, and 0.58 respectively at the zero bias, 0.2, 0.5, 1, and 5 Hz. As the frequency increased, the crystalline fraction was increased up to 1 Hz but was decreased when the frequency was 5 Hz. Figure 8b shows the Raman spectra for the films deposited at the bias voltages of 0, ±50, ±100, ±150, and ±200 V. From the Raman spectra, Xc was evaluated to be 0.50, 0.60, 0.67, 0.74, and 0.61 respectively at the biases of 0, ±50, ±100, ±150, and ±200 V. The crystalline fraction was increased with increasing bias voltage up to ±150 V but was decreased at ±200 V. The reason why the crystalline fraction increased with increasing bias frequency and voltage would be that the amount of deposition increased with increasing bias frequency and voltage. It was reported that the crystalline fraction evaluated from the Raman spectra tended to increase with increasing film thickness when the film thickness is below the penetration depth of Raman.35 The reason for crystallinity to decrease at the bias frequency of 5 Hz and the bias voltage of ±200 V seems to come from the fact that the films were more porous. It is known that Raman spectra of porous silicon films became broader than those of dense silicon films.41 The fact that charged nanoparticles tend to be generated in the gas phase and contribute to film growth during many CVD implies many potential applications. One example is the low temperature deposition of a crystalline phase, which turned out to be quite successful in the deposition of silicon nitride42 and silicon43 using HWCVD. In this case, crystalline nanoparticles are formed in the high temperature region near the hot wires, carried to the low temperature region, and deposited on the substrate at low temperature. Another example would be the acceleration of charged nanoparticles by electric field toward the substrate. This situation would be very similar to the ionized cluster beam deposition (ICBD),44 which is known to be very powerful in low temperature epitaxy. In addition to these examples, many applications would be possible based on the crystal growth by charged nanoparticles.

4. CONCLUSION Both positively and negatively charged nanoparticles were shown to be naturally generated in the gas phase during deposition of silicon by thermal CVD. The bias frequency as well as the bias voltage affected the microstructure and the growth rate of the deposited films. The bias on the substrate or the substrate holder can be a new processing parameter in the CVD process where charged nanoparticles are generated.



AUTHOR INFORMATION

Corresponding Author

*E-mail: e-mail:[email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was supported by the Brain Korea (BK21) program, Republic of Korea, and the Hi Seoul Science (Humanities) Fellowship from Seoul Scholarship Foundation. Technical help by M. J. Koo at Amherst College is greatly appreciated.



REFERENCES

(1) Venables, J. A.; Spiller, G. D. T.; Hanbucken, M. Rep. Prog. Phys. 1984, 47, 399. (2) Adachi, M.; Okuyama, K.; Tohge, N.; Shimada, M.; Sato, J.; Muroyama, M. Jpn. J. Appl. Phys. 1992, 31, L1439−L1442. (3) Jeon, J. D.; Park, C. J.; Kim, D. Y.; Hwang, N. M. J. Cryst. Growth 2000, 213, 79−82. (4) Ahn, H. S.; Park, H. M.; Kim, D. Y.; Hwang, N. M. J. Cryst. Growth 2002, 234, 399−403. (5) Jeon, I. D.; Gueroudji, L.; Kim, D. Y.; Hwang, N. M. J. Korean Ceram. Soc. 2001, 38, 218−224. (6) Jeon, I. D.; Barnes, M. C.; Kim, D. Y.; Hwang, N. M. J. Cryst. Growth 2003, 247, 623−630. (7) Cheong, W. S.; Hwang, N. M.; Yoon, D. Y. J. Cryst. Growth 1999, 204, 52−61. (8) Kim, C. S.; Youn, W. K.; Hwang, N. M. J. Appl. Phys. 2010, 108, 014313. (9) Kim, C. S.; Kwak, I. J.; Choi, K. J.; Park, J. G.; Hwang, N. M. J. Phys. Chem. C 2010, 114, 3390−3395. (10) Kim, C. S.; Chung, Y. B.; Youn, W. K.; Hwang, N. M. Aerosol Sci. Technol. 2009, 43, 120−125. (11) Kim, C. S.; Chung, Y. B.; Youn, W. K.; Hwang, N. M. Carbon 2009, 47, 2511−2518. (12) Lee, J. I.; Hwang, N. M. Carbon 2008, 46, 1588−1592.

25162

dx.doi.org/10.1021/jp310705p | J. Phys. Chem. C 2012, 116, 25157−25163

The Journal of Physical Chemistry C

Article

(13) Lee, S. S.; Kim, C. S.; Hwang, N. M. Aerosol Sci. Technol. 2012, 46, 1100−1108. (14) Hwang, N. M.; Kim, D. Y. Int. Mater. Rev. 2004, 49, 171−190. (15) Hwang, N. M.; Lee, D. K. J. Phys. D: Appl. Phys. 2010, 43, 483001. (16) Adachi, M.; Okuyama, K.; Tohge, N.; Shimada, M.; Satoh, J.; Muroyama, M. Jpn. J. Appl. Phys. 1993, 32, L748−L751. (17) Adachi, M.; Okuyama, K.; Tohge, N. J. Mater. Sci. 1995, 30, 932−937. (18) Adachi, M.; Okuyama, K.; Fujimoto, T.; Satoh, J.; Muroyama, M. Jpn. J. Appl. Phys. 1996, 35, 4438−4443. (19) Boufendi, L.; Plain, A.; Blondeau, J. P.; Bouchoule, A.; Laure, C.; Toogood, M. Appl. Phys. Lett. 1992, 60, 169−171. (20) Howling, A. A.; Sansonnens, L.; Dorier, J. L.; Hollenstein, C. J. Phys. D: Appl. Phys. 1993, 26, 1003−1006. (21) Fridman, A.; Kennedy, L. A. Plasma Physics and Engineering; Taylor and Francis: New York, 2004. (22) Cabarrocas, P. R. Curr. Opin. Solid State Mater. Sci. 2002, 6, 439−444. (23) Cabarrocas, P. R. J. Non-Cryst. Solids 2000, 31, 266−269. (24) Ostrikov, K. Rev. Mod. Phys. 2005, 77, 489−511. (25) Vladimirov, S. V.; Ostrikov, K. Phys. Rep. 2004, 393, 175−380. (26) Hayasaki, K.; Takamura, Y.; Yamaguchi, N.; Terashima, K.; Yoshida, T. J. Appl. Phys. 1997, 81, 1222. (27) Takamura, Y.; Hayasaki, K.; Terashima, K.; Yoshida, T. J. Vac. Sci. Technol. B 1997, 15, 558−565. (28) Terashima, K.; Yamaguchi, N.; Hattori, T.; Takamura, Y.; Yoshida, T. Pure Appl. Chem. 1998, 70, 1193−1197. (29) Takamura, Y.; Hayasaki, K.; Terashima, K.; Yoshida, T. J. Appl. Phys. 1998, 84, 5084−5088. (30) Yamaguchi, N.; Sasajima, Y.; Terashima, K.; Yoshida, T. Thin Solid Films 1999, 345, 34−37. (31) Knutson, E. O.; Whitby, K. T. J. Aerosol Sci. 1975, 6, 443−451. (32) Langmuir, I.; Kingdon, K. H. Proc. R. Soc. London, Ser. A 1925, 107, 61−79. (33) Seidl, M.; Meiwes-Broer, K. H.; Brack., M. J. Chem. Phys. 1991, 95, 1295−1303. (34) Cheng, Q.; Xu, S.; Ostrikov, K. J. Mater. Chem. 2009, 19, 5134− 5140. (35) Chung, Y. B.; Lee, D. K.; Lim, J. S.; Hwang, N. M. Sol. Energy Mater. Sol. Cells 2011, 95, 211−214. (36) Shaw, D. J. Introduction to Colloid and Surface Chemistry; Butterworths: London, UK, 1966. (37) Miller, R. D. Science 1999, 286, 421−423. (38) Leofanti, G.; Padovan, M.; Tozzola, G.; Venturelli, B. Catal. Today 1998, 41, 207−219. (39) Simon, P.; Gogotsi, Y. Nat. Mater. 2008, 7, 845−854. (40) Chung, Y. B.; Lee, D. K.; Kim, C. S.; Hwang, N. M. Vacuum 2009, 83, 1431−1434. (41) Kanemitsu, Y.; Uto, H.; Masumoto, Y. Phys. Rev. B 1993, 48, 2827−2830. (42) Kim, C. S.; Youn, W. K.; Lee, D. K.; Seol, K. S.; Hwang, N. M. J. Cryst. Growth 2009, 311, 3938−3942. (43) Lee, S. S.; Ko, M. S.; Kim, C. S.; Hwang, N. M. J. Cryst. Growth 2008, 310, 3659−3662. (44) Yamada, I.; Inokawa, H.; Takagi, T. J. Appl. Phys. 1984, 56, 2746−2750.

25163

dx.doi.org/10.1021/jp310705p | J. Phys. Chem. C 2012, 116, 25157−25163