Structural properties of porous 6H silicon carbide

Share Embed


Descrição do Produto

solidi

status

physica

pss

Phys. Status Solidi C 8, No. 6, 1950–1953 (2011) / DOI 10.1002/pssc.201000222

c

www.pss-c.com

current topics in solid state physics

Structural properties of porous 6H silicon carbide

Pascal Newby*,1,2, Jean-Marie Bluet1, Vincent Aimez2, Luc G. Fréchette2, and Vladimir Lysenko1 1 2

Université de Lyon, Institut des Nanotechnologies de Lyon, INL-UMR5270, CNRS, INSA de Lyon, Villeurbanne 69621, France Centre de Recherche en Nanofabrication et Nanocaractérisation (CRN2), Université de Sherbrooke, Sherbrooke, Québec, Canada

Received 30 April 2010, accepted 21 June 2010 Published online 25 November 2010 Keywords porous, silicon carbide, anodisation, morphology, Raman * Corresponding author: e-mail [email protected], Phone: +33 4 72 43 87 32, Fax: +33 4 72 43 85 31

In this work we have studied the effect of current density and UV illumination on the morphology of porous SiC formed by electrochemical etching. Raman spectroscopy was also carried out, in order to correlate the porous SiC structure with the obtained spectra. We show that current density is important in controlling the type of structure, whereas UV illumination plays a role in pore nucleation at lower current densities. The shape of the Raman spectra, in particular the longitudinal optical (LO) peak, also depends strongly on the SiC nanoscale morphology.

2 µm

Example of “sinuous” morphology of porous SiC.

© 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1 Introduction Porous silicon carbide was first discovered by Shor et al. [1] during their work on electrochemical etching of SiC, as this material is very difficult to etch with traditional silicon chemical-etching methods. Following this work, porous SiC has been studied by several groups [2-5] and has appeared to be a useful material for several applications, for example: (i) substrates for epitaxy of monocrystalline SiC [6,7], (ii) in- and out-diffusion of dopants [8], (iii) fabrication of membranes for biomedical applications [9]. Previous work has shown that a wide variety of porous SiC morphologies can exist [5]. However, to our knowledge, a precise study of the effect of each fabrication parameter has not been reported yet, and understanding of the porosification mechanism is still quite limited. In this work, we present preliminary results of a systematic study of the effect of fabrication parameters (specifically current density and UV illumination) on porous SiC morphology at nanoscale level. In addition, Raman spectroscopy measurements were carried out to correlate the effect of nanostructure on the spectra.

2 Experiment 2.1 Sample fabrication 6H on-axis, highly doped (ρ = 0.062 Ω.cm) n-type SiC wafers were anodised starting from the polished Si face. The SiC samples were electrochemically porosified in a Teflon etch-cell, according to the anodisation process generally used for fabrication of porous silicon [10], using a 48% hydrofluoric acid, ethanol and deionised water (1:5:4) electrolyte. All samples were etched for 30 minutes using a pulsed current with a 50% duty cycle [11]. Two parameters were varied to study their effect on the porous SiC morphology: current density and UV illumination. Current density ranged between 3 and 200 mA/cm2, and a 365 nm mercury-vapour lamp was used for illumination. 2.2 Characterisation After anodisation, porous SiC samples were fractured and cross-sectional SEM imaging was carried out using a field emission gun (FEG) microscope (FEI inspect F) to study their nanostructure. Raman characterisation was carried out with a Thermo Scientific Raman DXR at a wavelength of 532 nm, on the surface or

© 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Contributed Article Phys. Status Solidi C 8, No. 6 (2011)

1951

cross-section of the porous SiC samples. The spectral baseline (due to fluorescence of the samples) was subtracted from all porous SiC Raman spectra presented in this paper. 3 Results and discussion 3.1 Effect of current density Current density was varied between 3 and 200 mA/cm2, under UV illumination, and four types of morphology were identified, consistent with those published earlier [5]. These are: (i) from 3 to 5 mA/cm2: low-porosity “dendritic” morphology (Fig. 1a); (ii) 5-20 mA/cm2: high-porosity “dendritic” (Fig. 1b); (iii) 15-50 mA/cm2: “sinuous” (Fig. 1c); iv) “columnar” (Fig. 1d). Sharp transitions between the different morphologies were observed, rather than a gradual progression, as illustrated in Fig. 2. Indeed, in the case of the high-porosity dendritic and sinuous structures, both exist as spatially separate zones within the same layer (Fig. 2a). As for the transition between sinuous and columnar morphologies (Fig. 2b), it occurs at the same depth throughout the layer. It is worth remarking that, within the range of current densities used, the columnar morphology has only been obtained in such bi-layers. When current density increases, the relative thickness of the columnar layer (in comparison to that of the sinuous layer) increases too, suggesting that the transition to a columnar structure is mass-transfer related.

Figure 2 SEM images showing co-existence of two different morphologies within the same porous SiC layer.

Figure 1 SEM images of the four porous SiC morphologies fabricated with different current densities: (a) low-porosity “dendritic”, (b) high- porosity “dendritic”, (c) “sinuous”, (d) “columnar”. www.pss-c.com

For this series of samples, all parameters were constant except for current density, so this parameter is essential for controlling morphology. The co-existence of the highporosity dendritic and sinuous morphologies has already been shown and explained by the coalescence of vacancies [12]. In this case, as well as for the transition between sinuous and columnar morphologies, we suggest that depletion of the electrolyte with depth of the porous layer or an increase with time of the electrolyte temperature may also play a role, as has already been demonstrated for porous silicon [11,13,14]. It can be observed that when current density increases, the preferred orientation of the SiC crystallites becomes increasingly vertical, which suggests an evolution of etching mechanism and a change in the rela-

© 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

solidi

status

physica

pss

c

1952

P. Newby et al.: Structural properties of porous 6H silicon carbide

tive importance of chemical and electrochemical etching steps. 3.2 Effect of UV illumination The role of this parameter was studied by making samples at current densities of 5, 50 and 200 mA/cm2, with and without UV illumination. At 5 mA/cm2 under illumination, a uniform layer composed of the low-porosity dendritic morphology is obtained (Fig. 3a), whereas a discontinuous layer made of small non-uniform porous zones is formed in the dark (Fig. 3b). At 50 mA/cm2 a homogeneous layer containing the sinuous structure is observed (Fig. 3c). The sample etched in the dark is continuous (Fig. 3d), but the structure is nonuniform and the shape of the etch-front suggests nonuniform nucleation (ie. an extension of the isolated porous areas visible in Fig. 3b). At 200 mA/cm2, there is much less difference between the illuminated and nonilluminated samples (Fig. 3e,f): the type of morphology and the thickness of the layers are similar, although the pores are not as straight in the sample anodised in the dark. With UV illumination

Without UV illumination

5 mA/cm2

From these results it can be deduced that UV illumination has a strong effect at low current densities, whereas for a high current density its influence is much less pronounced. The discontinuous layer observed at 5 mA/cm2 and the non-uniform etch-front obtained at 50 mA/cm2 suggest that UV illumination plays an important role in pore nucleation at lower current densities. The fact that there is little difference at 200 mA/cm2 may indicate that the holes are generated mainly by breakdown mechanisms, while at lower densities they are probably at least partially photo-generated. 3.3 Raman spectroscopy Raman spectra for each type of morphology, as well as for bulk 6H–SiC, are presented in Fig. 4. There are several differences between the bulk SiC spectrum and those of porous SiC: in the case of the porous samples, (i) the forbidden E1 transverse optical (TO) peak at 797 cm-1 is more intense, (ii) the longitudinal optical (LO) peak at 967 cm-1 shifts towards higher energy and its intensity decreases, (iii) a new relatively wide peak is visible around 945 cm-1, and becomes more intense for morphologies formed at higher current densities. The E2(6/6) peak at 767 cm-1 is less intense for the columnar morphology, but for this sample the spectrum had to be acquired on the sample cross-section.

LO mode

TO modes Columnar Sinuous

Intensity (a.u.)

Dendritic high porosity

Intensity (a.u.)

50 mA/cm

2

Dendritic low porosity E2(2/6)

200 mA/cm2

E2(6/6)

750

E1 Bulk SiC 800

850

900

-1

950 1000 1050

Raman shift (cm ) Figure 4 Raman spectra of bulk 6H-SiC and 4 porous SiC morphologies. Note: intensity scales for TO and LO modes are different. LO peak is normalised to corresponding TO E2(2/6) peak. Figure 3 Effect of UV illumination. Samples fabricated at current densities of 5 mA/cm2 (a,b), 50 mA/cm2 (c,d) and 200 mA/cm2 (e,f). Left side with illumination and right side without.

© 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

The E1 TO forbidden peak is slightly visible for bulk SiC, but only because the laser beam is not perfectly per-

www.pss-c.com

Contributed Article Phys. Status Solidi C 8, No. 6 (2011)

1953

pendicular to the sample due to the micro-Raman setup using a high numerical aperture objective. The higher intensity of the E1 TO peak in the spectra for porous SiC is due to the relaxation of selection rules caused by diffusion of light within the porous layer. The blueshift of the LO peak can be explained by plasmonphonon coupling (LOPC peak). This phenomenon is wellknown for bulk SiC [15]. Indeed, the plasmon frequency is given by:

ωp =

ne2 m*ε0ε∞

(1)

where n is the free-carrier density, e the electron charge, m* the effective mass and ε the dielectric constant. In bulk SiC the blue-shift of the peak is usually caused by an increase of doping level. However, in the case of porous SiC, none of these values should change from those of the initial substrate, except the dielectric constant. Indeed, by introducing pores, the effective dielectric constant decreases, thus increasing the plasmon frequency and causing a shift of the LOPC peak towards higher energies. The appearance of a new peak around 945 cm-1 has already been observed and attributed to the surface polariton mode [16]. This effect becomes more important as the surface to volume ratio increases and the size of nanocrystallites decreases, and therefore the peak becomes more intense for the morphologies created at higher current densities. 4 Conclusions We have shown initial results on the effects of current density and UV illumination on the morphology of porous SiC. Raman spectra for these morphologies have also been presented. Four different types of morphology have been observed; current density controlled the type of morphology formed, as all other parameters were constant. UV illumination seems to play an important role in pore nucleation at lower current densities, whereas at high current densities holes necessary for the electrochemical reaction are probably generated by breakdown mechanisms. Raman spectra shape shows a strong dependence on the porous SiC structure. The E1 TO peak is visible due to relaxation of selection rules caused by light diffusion within the porous layer. The LO peak shifts towards higher energies and its intensity decreases for morphologies formed at higher current densities because of plasmon-phonon coupling. Finally, a new peak appears at 945 cm-1, which corresponds to a surface polariton mode. The effect of other anodisation parameters will also be studied in the future. This includes electrolyte concentration, length of etch-stop time, substrate resistivity, anodisation on the C-face, crystalline orientation, and wavelength and intensity of UV illumination. This should lead to better

www.pss-c.com

control of the porous SiC structure, and also help understand the formation mechanism. Raman spectroscopy can be developed to be used as a fast and non-destructive method for characterising the nanostructure of porous SiC. Acknowledgements The authors wish to thank Prof. B. Champagnon and all the staff of the CECOMO measurement platform for their help in carrying out the Raman measurements. GM Canada, Nanoquébec and NSERC helped fund this project.

References [1] J. S. Shor, I. Grimberg, B.‐Z. Weiss, and A. D. Kurtz, Appl. Phys. Lett. 62, 2836 (1993). [2] A.O. Konstantinov, C.I. Harris, and E. Janzen, Appl. Phys. Lett. 65, 2699 (1994). [3] S. Zangooie, P.O.A. Persson, J.N. Hilfiker, L. Hultman, and H. Arwin, J. Appl. Phys. 87, 8497 (2000). [4] M. Mynbaeva, Mater. Res. Soc. Symp. Proc. 742, 309 (2003). [5] Y. Shishkin, Y. Ke, R.P. Devaty, and W.J. Choyke, Mater. Sci. Forum 483-485, 251 (2005). [6] J.E. Spanier, G.T. Dunne, L.B. Rowland, and I.P. Herman, Appl. Phys. Lett. 76, 3879 (2000). [7] M. Mynbaeva, S.E. Saddow, G. Melnychuk, I. Nikitina, M. Scheglov, A. Sitnikova, N. Kuznetsov, K. Mynbaev, and V. Dmitriev, Appl. Phys. Lett. 78, 117 (2001). [8] M. Mynbaeva, N. Kuznetsov, A. Lavrent'ev, K. Mynbaev, J.T. Wolan, B. Grayson, V. Ivantsov, A. Syrkin, A. Fomin, and S.E. Saddow, Mater. Sci. Forum 433-436, 657 (2003). [9] A.J. Rosenbloom, D.M. Sipe, Y. Shishkin, Y. Ke, R.P. Devaty, and W.J. Choyke, Biomed. Microdevices 6, 261 (2004). [10] L. Canham (ed.), Properties of Porous Silicon (Inspec/IEE, London, 1997). [11] X. Y. Hou, H. L. Fan, L. Xu, F. L. Zhang, M. Q. Li, M. R. Yu, and X. Wang, Appl. Phys. Lett. 68, 2323 (1996). [12] M.G. Mynbaeva, D.A. Bauman, and K.D. Mynbaev, Phys. Solid State 47, 1630 (2005). [13] S. Billat, M. Thoenissen, R. Arens-Fischer, M.G. Berger, M. Krueger, and H. Lueth, Thin Solid Films 297, 22 (1997). [14] P.Y.Y. Kan and T.G. Finstad, Thin Solid Films 515, 5241 (2007). [15] S. Nakashima and H. Harima, Phys. Status Solidi A 162, 39 (1997). [16] T.V. Torchynska, A.V. Hernandez, A.D. Cano, S. JimenezSandoval, S. Ostapenko, and M. Mynbaeva, J. Appl. Phys. 97, 33507 (2005).

© 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Lihat lebih banyak...

Comentários

Copyright © 2017 DADOSPDF Inc.