solidi current topics in solid state physics Chemical vapour infiltration of nano-structured carbon in porous silicon

Transcription

1 solidi status physica pss c current topics in solid state physics Chemical vapour infiltration of nano-structured carbon in porous silicon G. Mattei1, V. Valentini1, and R. Polini2 1 2 C.N.R. Istituto dei Sistemi Complessi, ISC, Sezione di Montelibretti, Area della Ricerca di Roma 1, C.P. 10, Monterotondo Scalo (Roma), Italy Dipartimento di Scienze e Tecnologie Chimiche, Università di Roma Tor Vergata, Via della Ricerca Scientifica 1, Roma, Italy Received 17 March 2006, revised 29 November 2006, accepted 4 December 2006 Published online 9 May 2007 PACS Hk, Am, Ly, Rm, Bc, 81.15Gh For the first time the results of the chemical vapor infiltration (CVI) of nanostructured sp2 carbon in mesoporous silicon layers under process conditions normally used to grow diamond films by Hot Filament Chemical Vapour Deposition (HFCVD) are presented. The combined use of micro-raman spectroscopy and Field Emission Gun Scanning Electron Microscopy (FEG-SEM) clearly demonstrated that disordered graphitic carbon was infiltrated in the PS pores, thus permeating completely the PS layer. Such a nanostructured carbon infiltration provided new properties to the PS material, which are potentially of great relevance for opto-electronics and sensors applications. phys. stat. sol. (c) 4, No. 6, (2007) / DOI /pssc T N I R P E R

2 phys. stat. sol. (c) 4, No. 6, (2007) / DOI /pssc Chemical vapour infiltration of nano-structured carbon in porous silicon G. Mattei *, 1, V. Valentini 1, and R. Polini 2 1 C.N.R. Istituto dei Sistemi Complessi, ISC, Sezione di Montelibretti, Area della Ricerca di Roma 1, C.P. 10, Monterotondo Scalo (Roma), Italy 2 Dipartimento di Scienze e Tecnologie Chimiche, Università di Roma Tor Vergata, Via della Ricerca Scientifica 1, Roma, Italy Received 17 March 2006, revised 29 November 2006, accepted 4 December 2006 Published online 9 May 2007 PACS Hk, Am, Ly, Rm, Bc, 81.15Gh For the first time the results of the chemical vapor infiltration (CVI) of nanostructured sp 2 carbon in mesoporous silicon layers under process conditions normally used to grow diamond films by Hot Filament Chemical Vapour Deposition (HFCVD) are presented. The combined use of micro-raman spectroscopy and Field Emission Gun Scanning Electron Microscopy (FEG-SEM) clearly demonstrated that disordered graphitic carbon was infiltrated in the PS pores, thus permeating completely the PS layer. Such a nanostructured carbon infiltration provided new properties to the PS material, which are potentially of great relevance for opto-electronics and sensors applications. 1 Introduction Recently, porous silicon (PS) has been used as substrate for the nucleation and growth of diamond films and of carbon nanotubes by chemical vapor deposition (CVD). In particular, the structure of PS layers, patterned with suitable metal catalyst (e.g. Fe, Ni, etc.), has been used as template to grow up self oriented regular arrays of carbon nanotubes, in view of potential applications as electron field emission display, chemical sensors, etc. [1 3]. Besides, the typical microstructure of PS can provide a surface topology considered useful to grow up a diamond film [4] and a porosity suitable for porous diamond synthesis [5]. Several works have been published on this subject, and the optimal conditions to grow diamond films on PS by CVD have been defined. A general necessary prerequisite is the pretreatment of the PS top surface using a diamond powder suspension ( seeding ) prior to CVD [6]. Without this seeding pretreatment, only few diamond crystals nucleate with densities lower than 10 5 cm 2. It seems to be worth to investigate the effects on the PS of the CVD of carbon in these special preparation conditions. Indeed, in this paper, the results of the chemical vapour deposition of carbon in meso-porous silicon layers under process conditions normally used to grow diamond films by Hot Filament Chemical Vapour Deposition (HFCVD), but without diamond seeding, are presented and discussed. 2 Experimental Several PS samples were prepared by partial anodic dissolution of a (100) p + doped (resistivity 0.01 Ω cm) Si plate in a Teflon electrochemical cell containing a HF(45 wt.%):etoh = 1:2 solution. A cur- * Corresponding author: giorgio.mattei@isc.cnr.it

3 2050 G. Mattei et al.: CVI of nano-structured carbon in porous silicon rent density of 60 ma/cm 2 was used. The anodization times were 100 s and 400 s and allowed to prepare 3 and 12 µm thick PS layers, respectively. After preparation and rinsing, the PS samples were characterized by reflectance Fourier Transform Infrared (FTIR) spectroscopy that allowed to obtain the porosity ( 74 %) and to assess the thickness of the PS layers. Some of the prepared samples were cut in two halves and subsequently used without any diamond seeding as substrates for carbon deposition by Hot Filament Chemical Vapour Deposition (HFCVD). The gas phase (1.0 vol.% CH 4 /H 2 mixture, 4.8 kpa total pressure, 300 sccm total flow rate) was activated by a hot (2180 C) tantalum filament (0.3 mm in diameter) wound in a 1.4 mm internal diameter spiral and positioned 8 mm from the substrate. Process times were 3 and 14 h for samples with 3 and 12 µm thick PS layers, respectively. Substrate temperature was 700 C. The used deposition conditions, apart from the absence of diamond seeding, were those normally employed to deposit high-quality diamond films on silicon. The carbon deposited PS samples (PS/C) and the as prepared PS layers were characterized by Fourier transform infrared (FTIR) spectroscopy, by scanning electron microscopy (SEM, Leica Cambridge mod. 360 with LaB 6 source and LEO Supra 35 with Field Emission Gun) and by Raman spectroscopy. The micro- and macro-raman measurements were performed by a Dilor XY triple-spectrometer (laser wavelength nm). For the FTIR measurements a Biorad-Digilab FTS40A spectrometer was employed. 3 Results and discussion SEM analyses (Fig. 1) showed that the as prepared PS samples were meso-porous with a pore average diameter around 20 nm and with columnar pores developing perpendicularly to the top surface. After the carbon deposition process, all the PS/C samples preserved their columnar pores with dendritic structure, whereas on the top surface the average pore diameters increased up to nm, most likely due to Si etching performed by monohydrogen under typical diamond CVD process conditions [7]. 2.5µm top surface cross-section detail in the center of the cross-section 2.5µm Fig. 1 SEM images of the top-surface and the cross-section of a PS/C sample before (top) and after (bottom) 14 h CVD. The stars along the dotted line in panel (A) indicate the positions where the micro-raman spectra (laser spot diameter 1 µm) reported in Fig. 3 were collected. Moreover, the SEM analyses indicated that on the surface there were a few diamond micro-crystals and that the diamond nucleation was very low on the PS area and similar to that on the surrounding bare c-si wafer region. These findings confirm that i) the process conditions here employed were those typical of

4 phys. stat. sol. (c) 4, No. 6 (2007) 2051 diamond CVD and ii) the surface topology of meso-porous silicon has no effect on the enhancement of the heterogeneous nucleation of diamond from the gas phase. FTIR spectra (two examples are presented in Fig. 2) of as prepared PS samples were dominated by fringes due to the interference at the air/ps and PS/Si interfaces. Absorption bands due to the vibrations of the SiH x species, always present in freshly prepared PS layers (curves a), were clearly detectable. In the infrared spectra of the PS/C samples (curves b) the interference fringes were absent (sample A) or strongly modified (sample B). The SiH x bands disappeared whereas new features attributable to silicon oxide and to vibrations involving carbon bonds were detectable. a a Reflectance (a. u.) A SiH Reflectance (a. u.) b B SiH b C=C, C-O-C, Si-O-Si, C-C C=C, C-O-C, Si-O-Si, C-C Wave number (cm -1 ) Wave number (cm -1 ) Fig. 2 FTIR reflectance spectra of two PS samples (12 µm thick), before (a) and after (b) carbon CVD process (A: 14h continuous deposition; B: two steps deposition, 3 h and 11 h). On the flat c-si portion of the PS/C sample the only Raman signal present was that at 520 cm 1 due to the zone center optical phonon of crystalline Si itself. On the contrary, on the whole surface of the PS area the Raman spectra were dominated by two broad bands, D and G band, at 1349 cm 1 and 1604 cm 1 that are typical of sp 2 carbon or disordered graphite [8]. We recall that in graphite only one band due to the G mode is allowed in the Raman spectrum. The presence of disorder or of nanostructures generates the appearance of new features as the D band and a third component D (sometimes resolved in our spectra, as in Fig. 3A) on the high frequency side of G band. In order to better understand the transformations PS underwent after CVD, all the PS/C sample were cut in two parts and the Raman signal was collected on the PS cross-section by micro-raman technique along a line perpendicular to the surface. In all the cases, repeated analyses performed on several samples demonstrated that basically the same spectrum was measured on the top surface as well as on any point along the PS cross-sections, as illustrated in Fig. 3A. These unprecedented findings clearly indicated that the carbon deposit completely permeated all the pores in the PS/C layer. The chemical vapour infiltration (CVI) of carbon in the PS structure has to be attributed to a dramatic change of gas phase composition inside the pores with a relative deficiency of atomic hydrogen concentration and consequent formation of disordered sp² carbon [9]. Given the meso-porous nature of the infiltrated PS layers, confirmed by careful high resolution SEM investigations of samples cross sections (see Fig. 1 bottom), such sp² carbon material was clearly nanostructured. At this stage, the real nature of this nanosructured carbon cannot be confirmed by the Raman spectra that are compatible with different carbon species, for example multiwalls nanotubes (MWNTs) or carbon nanorods [10, 11]. Additional work is in progress, including TEM investigations, to clarify this point. The presence of the infiltrated carbon is expected to modify deeply all the properties of the PS matrix. Indeed, as indicated by the infrared measurements, the material was no longer transparent and the interference fringes were no longer present. Only in few cases, as in the example reported in Fig. 2B, the residual interference indicates the presence of a very thin PS top layer not completely filled by the car-

5 2052 G. Mattei et al.: CVI of nano-structured carbon in porous silicon bon deposit which grows from the bottom (PS/Si interface) to the top of the PS layer. The thickness of this non-infiltrated PS layer was evaluated to be around 0.9 µm. Intensity (a. u.) A D G D Intensity (a.u.) s a b B c Raman shift (cm -1 ) Raman shift (cm -1 ) Fig. 3 (A) A typical Raman spectrum (after smoothing and base line removing) collected from the surface of a PS/C sample. The arrows indicate the three bands present in the spectrum after a deconvolution with lorentzian components. (B) Micro-Raman spectra collected on the cross-section of a PS/C sample along the line indicated in Fig. 1(A) in the points indicated by the stars. Spectrum (s) is from the surface of the same sample. This effect was further demonstrated by the Raman measurements. In fact, in PS/C samples the signal due to the Si skeleton of the PS was strongly reduced, being a few percent of that of the as prepared PS. This result is compatible with a reduction of the optical penetration depth in the visible from microns to hundreds of nanometers, as expected when moving from a meso-porous Si to a massive carbon deposit. Raman Shift (cm -1 ) A Temperature ( C) Raman Intensity (a.u.) PS PS/C Wavenumber (cm -1 ) 6.6 KW/cm KW/cm KW/cm 2 57 KW/cm 2 Fig. 4 (A) Dependence of the frequency of the Si Raman band on the temperature for a PS sample. (B) Micro- Raman spectra of a PS and a PS/C sample using different laser power density. B Another material property that changed was the thermal conductivity, as proved by Raman measurements. In fact, the thermal conductivity of PS is quite low (up to two orders of magnitude lower than that of c-si) [12] and it is very easy, during the Raman measurements, to heat locally the sample, even using low laser powers [13]. The increase of the local temperature is manifested in the Raman spectrum of Si and PS by a shift to lower Raman frequency (red shift) and an asymmetrical broadening of the Si band. These changes are reversible with the temperature. Therefore, Raman spectra have been used to evaluate the thermal conductivity of the PS material [14]. As an example, the dependence of the frequency of the

6 phys. stat. sol. (c) 4, No. 6 (2007) 2053 Si Raman band on the local temperature (measured from the intensity ratio of the Stokes and anti-stokes spectra) for a PS sample is reported in Fig. 4A. Figure 4B compares the micro-raman spectra collected at various laser power densities of an as prepared PS sample and of a carbon infiltrated one, PS/C. As it is clear, the red shift was much bigger for the PS sample, being around 20 cm 1 for 6.6 kw/cm 2, with respect to the PS/C sample (only 3 cm 1 for 230 kw/cm 2 ). This demonstrates that the local temperature induced by laser heating was much higher for much lower power density on the as prepared sample. It is possible to infer from these results a significant increase of the thermal conductivity for the infiltrated PS/C samples attributable to the substitution of the empty pores with the nanostructured sp 2 carbon deposit. Also the electrical conductivity is expected to increase because of the presence of the sp 2 carbon in the pores. Indeed, preliminary measurements have confirmed this point. This last finding could be of great technological relevance to improve the quality of electric contacts in PS-base devices for optoelectronics and sensors applications. 4 Conclusions The results of carbon chemical vapor infiltration in PS layers have been presented for the first time and discussed. It has been shown that nanostructured sp² carbon permeated completely the pores of the material under typical process conditions usually employed to grow diamond from the gas phase. This new nanocomposite material, PS/C, exhibits peculiar properties with respect to the starting PS layer: it is strongly absorbing in the visible as well as in the infrared, and the thermal conductivity is much higher than for PS. Preliminary measurements indicated that also the electrical conductivity is greater than that of the as prepared PS; further work is in progress on this subject. These findings can be of great relevance for technological applications in various fields employing PS based devices. Acknowledgements The authors wish to thank Mr. G. Piciacchia (ISC-CNR, Roma) for his valuable technical assistance. References [1] S. Fan, M. G. Chaplin, N. R. Franklin, T. W. Tombler, A. M. Cassel, and H. Dai, Science 283, 512 (1999). [2] S. Fan, W. Liang, H. Dang, N. Franklin, T. Tombler, M. Chapline, and H. Dai, Physica E 8, 179 (2000). [3] D. Xu, G. Guo, L. Gui, and Y. Tang, Appl. Phys. Lett. 75(4), 481 (1999). [4] Z. Liu, B.Q. Zong, and Z. Lin, Thin Solid Films 254, 3 (1995). [5] N. G. Ferreira, A. F. Azevedo, A. F. Beloto, M. Amaral, F. A. Almeida, F. J. Oliveira, and R. F. Silva, Diamond Relat. Mater. 14, 441 (2005). [6] V. Baranauskas, A. C. Peterlevitz, D. C. Chang, and S. F. Durrant, Appl. Surf. Sci. 185, 108 (2001). [7] J. C. Arnault, S. Hubert, and F. Le Normand, J. Phys Chem. B 102, 4856 (1998). [8] J. Robertson, Mater. Sci. Eng. R 37, 129 (2002). [9] A. Glaser, S. M. Rosiwal, B. Freels, and R. F. Singer, Diamond Relat. Mater. 13, 834 (2004). [10] Y. S. Woo, D. Y. Jeon, I. T. Han, Y. J. Park, H. J. Kim, J. E. Jung, J. M. Kim, and N. S. Lee, J. Appl. Phys. 94(10), 6789 (2003). [11] G. Zou, J. Lu, D. Wang, L. Xu, and Y. Qian, Inorg. Chem. 43, 5432 (2004). [12] W. Lang, in: Properties of Porous Silicon, edited by L. Canham (INSPEC, London, 1997), p [13] E. D. Obraztsova, L. P. Avakyants, and G. B. Demidovich, J. Electron. Spectrosc. Relat. Phenom. 64/65, 587 (1993). [14] V. Lysenko, S. Perichon, B. Remaki, D. Barbier, and B. Champagnon, J. Appl. Phys. 86, 6841 (1999).