14546
J. Phys. Chem. B 2006, 110, 14546-14549
ARTICLES Formation Mechanism of Si3N4 Nanowires via Carbothermal Reduction of Carbonaceous Silica Xerogels Feng Wang,†,‡ Guo-Qiang Jin,† and Xiang-Yun Guo*,† State Key Laboratory of Coal ConVersion, Institute of Coal Chemistry, Taiyuan 030001, China, and Graduate School of the Chinese Academy of Sciences, Beijing 100039, China ReceiVed: March 28, 2006; In Final Form: June 7, 2006
Si3N4 nanowires prepared from the carbothermal reduction of carbonaceous silica xerogels with metal salt additives usually contain a small amount of nanotubes. This paper is devoted to the investigation of the formation mechanism of the Si3N4 nanowires. As-prepared samples heated at 1300 °C for different reaction times (1, 5, 10, and 30 h) were characterized by X-ray diffraction, Fourier transform infrared spectroscopy, scanning electron microscopy, and energy-dispersive X-ray spectroscopy. The results showed that all the samples mainly consisted of nanowires, while their crystalline phases changed with the heating time. Nitrogendoped silicon oxide nanowires were first produced via the vapor-liquid-solid process and then underwent a stepwise surface nitrogenization to silicon nitride. The suggested mechanism can easily explain the existence of nanotubes in the Si3N4 nanowires.
Introduction Inorganic nanotubes and nanowires have recently received much attention because they usually exhibit unique optical, electrical, magnetic, and mechanical properties which are distinctive from those of their bulk materials.1-3 Among these materials, silicon nitride (Si3N4) nanowires are particularly noticeable owing to their excellent properties, such as wide band gap, high strength, high hardness, and good resistance to thermal shock and oxidation.4-6 As a result, various chemical and physical routes have been proposed for the synthesis of Si3N4 nanowires. These routes include chemical vapor deposition (CVD),7,8 carbothermal reduction of silica and carbon in nitrogen-contained ambience,9,10 nitrogenization of silicon powder,11,12 nitridation of Fe-Si catalysts,13 direct growth from silicon substrate,14 and solvothermal synthesis of Mg3N2 and SiCl4.15 In the meantime, the growth mechanism of Si3N4 nanowires has also been extensively discussed. Since Wagner and Ellis proposed the vapor-liquid-solid (VLS) mechanism,16 catalytic growth of most nanowires and whiskers has been explained by the mechanism. Si3N4 nanowires are not exceptional. However, there still exist some different descriptions in detail on the formation of Si3N4 nanowires prepared from different routes. Zhang et al. prepared Si3N4 nanowires by heating Si or Si/SiO2 powder in N2 or NH3 ambience with or without metal catalysts and suggested that the growth was a vapor-solid process.17 Chen et al. explained the growth mechanism of rodlike β-Si3N4 crystals from combustion synthesis by a synergy of VLS and vapor-solid.18 Kim et al. found that the Si3N4 nanowires grown directly from silicon substrate * Corresponding author. Phone: +86-351-4065282. Fax: +86-3514050320. E-mail:
[email protected]. † Institute of Coal Chemistry. ‡ Graduate School of the Chinese Academy of Sciences.
resulted from a combination of solid-liquid-solid and VLS.14 Yang et al. thought the growth of platelike and branched Si3N4 whiskers obtained from catalyst-assisted pyrolysis of polymeric precursors followed a solid-liquid-gas-solid mechanism.19 Lin et al. reported that the Si3N4 nanotubes prepared from a thermal heating CVD method were formed via the VLS growth.20 Although there are considerable disputes in the formation of Si3N4 nanowires, it is generally recognized that the nanowires are directly produced from liquid metallic droplets if metal catalysts are involved in the synthesis process. In our previous work, it was found that the Si3N4 nanowires prepared from the carbothermal reduction of a carbonaceous silica xerogel with Fe(NO3)3 additive usually contained a small amount of nanotubes.21 The existence of spherical metal particles demonstrated that the VLS process occurred in the preparation. This paper is devoted to the investigation of the formation process of the Si3N4 nanowires by analyzing changes of the morphologies, crystalline phases, and compositions of the samples obtained from different reaction times. The present studies suggested that silicon oxide nanowires containing a little nitrogen component were first produced via the VLS mechanism and then underwent a stepwise surface nitrogenization to silicon nitride. These results are helpful to understand the existence of nanotubes in the Si3N4 nanowires. Experimental Section The detailed preparation of the carbonaceous silica xerogel has been described elsewhere.21 The xerogel was placed in an alumina tubular furnace and heated to 1000 °C at a rate of 8 °C/min then to 1300 °C at a rate of 3 °C/min under a flowing nitrogen atmosphere (200 mL/min). When the temperature was maintained for different times (1, 5, 10, and 30 h), raw products were obtained. Before further characterizations, the raw products
10.1021/jp0619282 CCC: $33.50 © 2006 American Chemical Society Published on Web 07/12/2006
Formation Mechanism of Si3N4 Nanowires
Figure 1. SEM images of the samples prepared from different reaction times: (a) 1 h, (b) 5 h, (c) 10 h, and (d) 30 h.
were purified by air oxidation at 700 °C for 2 h to remove residual carbon and then by acid treatment with hydrochloric acid (HCl) for 72 h to remove other impurities. The purified samples were characterized by X-ray diffraction (XRD, Rigaku D-Max/RB, Cu KR radiation), scanning electron microscopy (SEM, LEO-438VP), and Fourier transform infrared spectroscopy (FTIR, DIGIL B/FTS 3000MX, KBr), respectively. The chemical composition of the droplets found at the nanowire tip was analyzed by energy-dispersive X-ray (EDX) spectroscopy attached to the SEM. Results and Discussion SEM images of the samples prepared at 1300 °C for different reaction times (1, 5, 10, and 30 h) are shown in Figure 1. From the figure, it can be found that all the four samples mainly consist of nanowires with a diameter of ∼100 nm and a length of tens of micrometers. Particulate and cottony substances are easily observed from the surfaces of the nanowires shown in Figure 1a-c, but are almost not found from those in Figure 1d. The nanowires in Figure 1d have a relatively smooth and clean surface. These SEM images indicate that prolonging the reaction time from 1 to 30 h does not make remarkable changes to the sample morphologies. From the as-prepared samples before the acid treatment, spherical metallic droplets can easily be found from nanowire tips. Figure 2 displays an SEM image of a droplet from the as-prepared sample after 10 h of heat treatment and the corresponding EDX spectrum. The EDX analysis shows that the droplet is mainly composed of Si, Fe, C, O, and N elements.
J. Phys. Chem. B, Vol. 110, No. 30, 2006 14547 The copper signal appearing in the EDX result originates from the copper grid for supporting samples. These results imply that the nanowires in the present work were produced via the VLS process.22,23 To investigate the differences between the four samples in crystalline phases, the XRD technique was employed. From the XRD patterns shown in Figure 3, the locations and relative intensities of the diffraction peaks change obviously with the reaction time. For the sample obtained from 1 h of reaction (Figure 3a), there presents a broad band between 15° and 30° of 2θ, which is usually assigned to amorphous silicon oxide (SiOx, x e 2).24 Additionally, small peaks at 2θ ) 20°, 26.4°, and 37.2° are characteristic diffraction peaks of Si2N2O. The result indicates that the 1 h sample mainly consists of amorphous silicon oxide and contains a small amount of Si2N2O species. When the reaction time is prolonged to 5 h, the XRD patterns (Figure 3b) show that the sample is predominantly composed of Si2N2O25,26 incorporated with a little R-Si3N4. This demonstrates that there occurs a transformation from amorphous silicon oxide to the Si2N2O phase. When the reaction time is further prolonged to 10 h, the XRD patterns (Figure 3c) show that the dominate phase in the sample has changed from Si2N2O to R-Si3N4. For the 30 h sample, the XRD patterns (Figure 3d) are in good agreement with those from the Si3N4 sample purified by hydrofluoric acid (HF),21 and this indicates that the sample consists of nearly pure R-Si3N4 phase. In addition, the peak at 2θ ) 45° results from iron silicides, which act as a catalyst during the formation of the nanowires. From the above results, it is concluded that the formation of the Si3N4 nanowires from the present synthesis route underwent a phase transformation of SiOx f Si2N2O f Si3N4. To further confirm the conclusion from the XRD results, FTIR spectroscopy was used to analyze the four samples obtained from different reaction times. The spectrum of the 1 h sample (Figure 4a) exhibits additional absorption peaks at 1005, 950, 900, and 540 cm-1 besides the characteristic absorption at 1100 cm-1 of the Si-O bond in silicon oxide.27 These additional peaks are generally assigned to the characteristic absorption of N-Si-O.28 The 5 h sample has a spectrum similar to that of the 1 h sample, as shown in Figure 4b. However, the characteristic absorption of N-Si-O becomes stronger, while the absorption of Si-O becomes weaker. With the extension of reaction time to 10 h, the N-Si-O peaks are still very strong, and meanwhile, there appear new absorption peaks at 850, 682, 600, and 492 cm-1, as shown in Figure 4c. The new peaks originate from the characteristic adsorption of R-Si3N4,29,30 and they imply that the sample after 10 h of reaction contains a
Figure 2. SEM image of a liquid droplet at the nanowire tip and corresponding EDX spectrum.
14548 J. Phys. Chem. B, Vol. 110, No. 30, 2006
Wang et al.
Figure 3. XRD patterns of samples prepared from different reaction times: (a) 1 h, (b) 5 h, (c) 10 h, and (d) 30 h. (f, Si3N4; [, Si2N2O; g, SiOx; ], iron silicides.)
considerable amount of Si3N4. For the forth sample, one almost cannot find the characteristic peaks resulting from silicon oxide or the N-Si-O composite. On the contrary, the absorption peaks of Si3N4 become very strong, as shown in Figure 4d. From the above analysis, the same conclusion can be obtained that the Si3N4 nanowires produced in the present experiments are converted from silicon oxide and then Si-N-O composite. From the above SEM, XRD, and FTIR results, it is suggested that the formation of Si3N4 nanowires was achieved by stepwise surface nitrogenization of silicon oxide nanowires. The process can be described as follows. At the reaction temperature (1300 °C), liquid Si-Fe-O droplets are first formed by the reaction of silica and iron:
SiO2 + Fe f Fe-Si-O
Figure 4. FTIR spectra of samples prepared from different times: (a) 1 h, (b) 5 h, (c) 10 h, and (d) 30 h.
Figure 5. SEM image of Si3N4 nanowires purified by air oxidation and HF + HCl treatment. The nanotubes (arrowed) show concentrically circular cross sections.
(liquid eutectic droplets) (1)
The metallic droplets act as a catalyst in the whole process. Gaseous SiO and CO are produced in the following manner: 9,10
SiO2 + C f SiO + CO
(2)
The liquid droplets continually absorb SiO, CO, N2, and O2 from the ambience and allow them to react in the droplets:
O2 f O(a) + O(a)
(3)
N2 f N(a) + N(a)
(4)
SiO + O(a) f SiOx
(x e 2)
CO + O(a) f CO2
(5) (6)
where (a) represents atomic species. After the concentrations of Si, O, C, and N in the droplets approach saturation, nitrogendoped silicon oxide (SiOx) nanowires precipitate from the droplets; meanwhile, CO2 is released. Dissociated nitrogen atoms can migrate to the nanowire surface and replace oxygen in silicon oxide at the high temperature:
SiOx + N(a) f Si2N2O
(7)
Si2N2O + N(a) f Si3N4
(8)
With the extension of the time of heat treatment in N2
atmosphere, the replacement will progress from the nanowire surface to the interior. Finally, Si3N4 nanowires are formed. Because the Si3N4 nanowires are formed by the surface nitrogenization of silicon oxide nanowires and the nitrogenization cannot be thoroughly completed sometimes, there possibly appear some coaxial nanocables with a nitrogen-doped silicon oxide or Si2N2O core and a Si3N4 sheath in as-prepared samples. After treatment in the mixture of hydrochloric acid (HCl) and hydrofluoric acid (HF), the core can be removed while the sheath can survive. Therefore, Si3N4 nanotubes can occasionally be found from the nanowire products prepared by the carbothermal reduction. Figure 5 is an SEM image of the Si3N4 nanowires after air oxidation and acid treatment. The figure shows the existence of a few nanotubes. Moreover, it is found that the nanotubes have a cross section (arrowed) with clear concentric circles, which possibly result from the stepwise nitrogenization. Summarily, the present work indicates that the formation of Si3N4 nanowires involves the VLS process. However, the Si3N4 nanowires are not directly produced from liquid metallic droplets but are formed by stepwise surface nitrogenization of silicon oxide nanowires, which directly precipitate from the droplets. The formation of Si3N4 nanowires can be sketched in Figure 6. It should be pointed out that the oxygen in reaction 3 may come from the impurities in the nitrogen ambience. The dissociation of N2 (reaction 4) can also directly occur on the nanowire surface; therefore, the suggested mechanism implies that Si3N4 nanotubes could be largely prepared by employing different metal catalysts in the carbothermal reduction. In fact, we have
Formation Mechanism of Si3N4 Nanowires
J. Phys. Chem. B, Vol. 110, No. 30, 2006 14549 References and Notes
Figure 6. Sketch of the growth mechanism of Si3N4 nanowires.
added La(NO3)3 in the sol-gel process and found that the amount of nanotubes in the final Si3N4 product has a remarkable increase (Figure S1 in the Supporting Information). Moreover, these nanotubes have a zigzag interior wall (Figure S2 in the Supporting Information), which is very different from other nanotubes grown via the VLS mechanism.31,32 The reason is that lanthanum species have less catalytic activity for N2 dissociation than iron, and thus reaction 4 mainly occurs on the nanowire surface. These results further support the suggested mechanism. Conclusion The carbothermal reduction of carbonaceous silica xerogels containing metallic additives in N2 ambience usually produces Si3N4 nanowires mixed with a small amount of nanotubes. Based on SEM, XRD, and FTIR results, we suggested a formation mechanism of the Si3N4 nanowires prepared from this way. Silica first reacted with metal and produced eutectic droplets, which absorbed vaporous SiO, CO, N2, and O2 from the ambience. After their concentrations in the droplets approached saturation, silicon oxide nanowires directly precipitated and then changed to silicon nitride by a surface-to-interior nitrogenization. The transformation of crystalline phases can simply be represented as SiOx f Si2N2O f Si3N4. The suggested mechanism reasonably interpreted the existence of nanotubes in nanowires. Acknowledgment. The work was financially supported by the National Natural Science Foundation of China under Grant No. 20471067. The authors thank Professors J. L. Yang and M. Li for their help in the FTIR characterization. Supporting Information Available: SEM and high-resolution TEM images of the lanthanum-assisted Si3N4 sample. This material is available free of charge via the Internet at http:// pubs.acs.org.
(1) Hu, J. Q.; Lu, Q. Y.; Tang, K. B.; Deng, B.; Jiang, R. R.; Qian, Y. T.; Yu, W. C.; Zhou, G. E.; Liu, X. M.; Wu, J. X. J. Phys. Chem. B 2000, 104, 5251. (2) Wong, E. W.; Sheehan, P. E.; Lieber, C. M. Science 1997, 177, 1971. (3) Xia, Y. N.; Yang, P. D. AdV. Mater. 2003, 15, 351. (4) Carduner, K. R.; Blackwell, C. S.; Hammond, W. B.; Reidinger, F.; Hatfield, G. R. J. Am. Chem. Soc. 1990, 112, 4676. (5) Xu, Y. J.; Cao, C. B.; Chen, Z.; Li, J.; Wang, F. C.; Cai, H. N. J. Phys. Chem. B 2006, 110, 3088. (6) Tang, C. C.; Bando, Y.; Sato, T.; Kurashima, K. J. Mater. Chem. 2002, 12, 1910. (7) Kijima, K.; Setaka, N.; Tannaka, H. J. Cryst. Growth 1974, 24/25, 183. (8) Motojima, S.; Yamana, T.; Araki, T.; Iwanage, H. J. Electrochem. Soc. 1995, 142, 3141. (9) Arik, H. J. Eur. Ceram. Soc. 2003, 23, 2005. (10) Wu, X. C.; Song, W. H.; Zhao, B.; Huang, W. D.; Pu, M. H.; Sun, Y. P.; Du, J. J. Solid State Commun. 2000, 115, 683. (11) Gopalakrishnan, P. S.; Lakshminarsimham, P. S. J. Mater. Sci. Lett. 1993, 12, 1422. (12) Inomata, Y.; Yamane, T. J. Cryst. Growth 1974, 21, 317. (13) Huo, K. F.; Ma, Y. W.; Hu, Y. M.; Fu, J. J.; Lu, B.; Lu, Y. N.; Hu, Z.; Chen, Y. Nanotechnology 2005, 16, 2282. (14) Kim, H. Y.; Park, J.; Yang, H. Chem. Phys. Lett. 2003, 372, 269. (15) Zou, G. F.; Hu, B.; Xiong, K.; Li, H.; Dong, C.; Liang, J. B.; Qian, Y. T. Appl. Phys. Lett. 2005, 86, 181901. (16) Wagner, R. S.; Ellis, W. C. Appl. Phys. Lett. 1964, 4, 89. (17) Zhang, Y. J.; Wang, N. L.; He, R. R.; Liu, J.; Zhang, X. Z.; Zhu, J. J. Cryst. Growth 2001, 233, 803. (18) Chen, D. Y.; Zhang, B. L.; Zhuang, H. R.; Li, W. L.; Xu, S. Y. Mater. Lett. 2002, 57, 399. (19) Yang, W. Y.; Xie, Z. P.; Li, J. J.; Miao, H. Z.; Zhang, L. G.; An, L. N. Solid State Commun. 2004, 132, 263. (20) Lin, F. H.; Hsu, C. K.; Tang, T. P.; Lee, J. S.; Lin, J. Y.; Kang, P. L. Mater. Chem. Phys. 2005, 93, 10. (21) Wang, F.; Jin, G. Q.; Guo, X. Y. Mater. Lett. 2006, 60, 330. (22) Zhang, R. Q.; Lifshitz, Y.; Lee, S. T. AdV. Mater. 2003, 15, 635. (23) Wang, D. W.; Dai, H. J. Angew. Chem., Int. Ed. 2002, 41, 4783. (24) Liu, Y. Z.; Zhou, Y. Q.; Shi, W. Q.; Zhao, L. L.; Sun, B. Y.; Ye, T. C. Mater. Lett. 2004, 58, 2397. (25) Han, W. Q.; Fan, S. S.; Li, Q. Q.; Gu, B. L. Appl. Phys. Lett. 1997, 71, 2271. (26) Gundiah, G.; Madhav, G. V.; Govindaraj, A.; Seikh, Md. M.; Rao, C. N. R. J. Mater. Chem. 2002, 12, 1606. (27) Luo, Y. M.; Zheng, Z. M.; Xie, Z. M.; Zhang, Z. J. Mater. Lett. 2004, 58, 2114. (28) Lu, J. G.; Rozgonyi, G.; Rand, J.; Jonczyk, R. J. Cryst. Growth 2004, 269, 599. (29) Theodossiu, E.; Bauman, H.; Matz, W.; Mu¨cklich, A. Phys. Status Solidi A 2002, 194, 47. (30) Mazdiyasni, K. S.; Cooke, C. M. J. Am. Ceram. Soc. 1973, 56, 628. (31) Bakkers, E. P. A. M.; Verheijen, M. A. J. Am. Chem. Soc. 2003, 125, 3440. (32) Tang, C. C.; Ding, X. X.; Huang, X. T.; Gan, Z. W.; Qi, S. R.; Liu, W.; Fan, S. S. Chem. Phys. Lett. 2002, 356, 254.
