Brevier - Technische Keramik, Hrsg.: Informationszentrum Technische ... und Technische Kommission der Fachgruppe Technische Keramik, Fahner Verlag, Lauf.
Mater. Res. Soc. Symp. Proc. Vol. 1069 © 2008 Materials Research Society
1069-D01-03
Alternative Routes to Porous Silicon Carbide Bettina Friedel1, and Siegmund Greulich-Weber2 1 Physics, University of Cambridge, Cavendish Laboratory, JJ Thomson Avenue, Cambridge, CB30HE, United Kingdom 2 Physics, University of Paderborn, Warburger Strasse 100, Paderborn, Germany ABSTRACT A low-cost alternative route for large-scale fabrication of high purity porous silicon carbide is reported. This allows a three-dimensional arrangement of pores with adjustable pore diameters from several 10 nanometers to several microns. The growth of SiC is here based on a combined sol-gel and carbothermal reduction process. Therein tetraethoxysilane is used as the primary silicon and sucrose as the carbon source. We provide two different sol-gel based ways for preparation of porous SiC, obtaining either a regular porous or a random porous type. Regular porous SiC with monodisperse ordered spherical pores of predefined size is obtained via liquid infiltration of a removable opal matrix. Whereas random porous material with polydisperse pores of an adjustable size distribution range, but without order, can be achieved via free gas phase growth. This is performed by degradation of granulated sol-gel prepared material inside a sealed reaction chamber, resulting in a SiO/CO/SiC rich gas atmosphere, which causes SiC growth inside the granulate itself. For both types doping of the initially semi-insulating porous SiC is possible either during the sol-gel preparation or via the gas phase during the following annealing procedure. As probing dopants we have used P, N, B and Al, which are well known from 'conventional' SiC. Composition and structure of the obtained material was investigated using scanning electron microscopy, X-ray diffraction, nuclear magnetic resonance and Fourier transform infrared spectroscopy.
INTRODUCTION SiC is not only a powerful material for electronic and optoelectronic applications but also for advanced photonic applications such as photovoltaic devices, taking advantage of the large electronic bandgap. Due to its excellent properties in particular its hardness and its chemical inertness it is a challenge to use SiC devices also in harsh environment as e.g. filters or catalytic converters at high temperatures [1]. Main drawbacks using SiC for such applications are the expensive production and difficulties to process the material. Especially porous SiC is currently under discussion for various applications. For sophisticated applications porous SiC is usually fabricated destructive by labour intensive electro-chemical etching of electronic quality wafers [2]. Although notable progress has been achieved in the recent past, high quality porous structures as needed for photonic applications in silicon carbide have not been achieved so far. However, it depends on the intended application, which kind of porosity is sufficient. There are also methods starting with SiC powder [3] leading to ceramic porous SiC, which however, is inappropriate for electronic or photonic applications. We are reporting here on rather inexpensive constructive fabrication methods providing a large range of products of high purity porous SiC.
All methods to be presented are based on a sol-gel process combined with carbothermal reduction [4]. At around 1800°C we obtain 3C-SiC single crystal growth. Porous SiC can be grown on various substrates like commercial SiC wavers or sapphire and as free standing bulk material [5]. Pre-structured SiC with well defined monodisperse periodic pores is achieved by using a close packed two or three dimensional carbon spheres template, which is infiltrated with sol-gel SiC precursor [6,7]. The carbon template is removed afterwards, thus a SiC skeleton is left. The pores diameter can be adjusted between several 100nm and 2µm due to an appropriate choice of carbon spheres. The SiC finally consists of polycrystalline microcrystals surrounding the empty spheres, or in the case of thin SiC layers consisting of a monocrystalline 3C-SiC film. Using a modified sol-gel process we are able to convert any graphite part into 3C-SiC [8]. Also this material is consisting of 3C-SiC micro crystals, while the porosity depends on the graphite raw material. The main advantage of this process is that a given graphite part is preserved in shape and size after conversation of graphite into silicon carbide. Thus any porous structure easily prepared from carbon will be converted into SiC with our method.
EXPERIMENTAL DETAILS Base Material Synthesis. Precursor solution: Starting material either for template infiltration or for further processing towards the precursor granulate, is a sucrose containing silica sol. This was prepared via hydrolysis and condensation of tetraethoxysilane (TEOS) in ethanolic solution with deionized water in the presence of hydrochloric acid. A specific amount of sucrose, leading to a Si:C ratio of 1:4 was added via prior dissolution in deionized water. The obtained sol is either used immediately (coating, infiltration) or is kept for further processing. Carbon charged silica granulate: The sucrose silica sol is allowed to gel fast (depending on the HCl content), followed by a slow drying process under ambient conditions. Finally the solid carbohydrate charged silica gel pieces are annealed at 1000˚C under inert atmosphere (argon or nitrogen). The obtained granules are the basic material for all our gas phase supported SiC preparations. Porous Silicon Carbide Preparation. SiC with ordered uniform spherical pores: In this case porous SiC is grown by liquid solgel precursor infiltration in a removable template. Thereby a colloidal crystal of monodisperse spherical carbon particles is used as the template [6,7]. This regular structure is infiltrated by filling the voids between the spheres with the above described SiC precursor solution (see Fig 1a). To cure the liquid material, it is annealed at 1000˚C in argon gas atmosphere for a few seconds and cooled down to room temperature. Due to shrinking processes of the precursor material, this infiltration cycle has to be repeated until a sufficient high fill factor in the structure is achieved.
Figure 1: Schematical the preparation of uniform porous SiC, by (a) Infiltration of a colloidal crystal with liquid SiC precursor, (b) high temperature annealing of the composite in argon gas atmosphere at 1800°C, (c) removing of the colloidal crystal template, by burning in air and finally (d) the inverted structure of silicon carbide
Afterwards the obtained compound structure of spherical carbon particles and carbon rich silica glass is annealed to 1800˚C for a few minutes in argon gas atmosphere (see Fig. 1b) to convert the precursor to silicon carbide. Finally the spherical carbon particles are removed by heating the whole structure short time at 1000˚C in air (see Fig 1c), leaving a silicon carbide scaffold consisting of uniform spherical interconnected pores (see Fig 1d). Therein the final pore size can be adjusted by the initially diameter of the carbon spheres used to grow the template. These can be synthesized with a size from 100nm to 2µm. Free gas phase growth: The special composition of the solid SiC precursor, described above makes it possible to coat any kind of heat resistant (up to 1800˚C) substrate with a polycrystalline porous SiC film. Therefore the granulated precursor material and substrate are placed in a gradient furnace under static argon atmosphere. As soon as the precursor reaches 1700˚C, it starts to degrade, causing a SiC rich gas atmosphere, which causes the growth of joined SiC micro crystals on the slightly cooler substrate. Depending on time and the used amount of precursor, films with a thickness between 100µm and several millimetres can be grown. Especially thicker films, e.g. grown on graphite can be used freestanding after removal of the substrate. Conversion of carbonaceous parts in porous SiC: In this technique carbon bodies (e.g. from graphite or glassy carbon) are fully or partially converted into porous silicon carbide without deformation or shrinkage. This method is also gas phase based and though uses the same experimental setup and conditions as the previous one, just the substrate is replaced by a carbon body of any shape or size. Due to the nature of carbonaceous materials, the forthcoming process differs to free gas phase growth of films. Gaseous SiO and CO, created along with the SiC vapour during degradation of the precursor granulate, increase porosity of the carbon material due to oxidation. On the other hand by SiO interaction also first silicon carbide seed crystals are generated inside the pores. Further SiC vapour condensates on these SiC seeds, crystals grow inside the structure, covering or even closing the pores. Both of these processes continue until the precursor is used up. Depending on the size of the body to be converted, the exchanging process has to be repeated with renewed precursor. The porosity of the obtained SiC body depends mainly on the initial porosity of the carbon body. Shape and size of the body are not changed. Introduction of dopants in porous SiC. As probing dopants we have used phosphorous, nitrogen, boron and aluminium, which are the most wanted ones from the technological side in 'conventional' semiconductor SiC. In contrast to customary SiC, sol-gel SiC is not initially nitrogen doped. In contrast to commercially
available SiC here p-type doping and semi-insulating properties are possible without nitrogen donor compensation or activation of intrinsic defects, respectively. In all above mentioned processes foreign ions can be introduced in two different ways. The dopant can be added during SiC precursor preparation, by adding either itself or a soluble compound to the initial solution. Suitable candidates are nitrates (N doping), borates (B), phosphates (P) or chlorides (e.g. Al). During degradation of the precursor, during annealing at 1800°C, the foreign ions are automatically built in into the growing SiC crystal structure. The other more difficult way is providing the dopant via gas phase during the high temperature annealing process (e.g. nitrogen). Latter method is only suitable for selected elements. Characterization. The grown and the converted materials respectively have been verified as SiC by nuclear magnetic resonance spectroscopy (NMR), Fourier transform infrared spectroscopy (FTIR) and X-ray diffraction (XRD). The pore size and structure of the porous samples has been determined by scanning electron microscopy (SEM). The nature of introduced dopants could be observed via electron paramagnetic resonance spectroscopy (EPR). RESULTS AND DISCUSSION Uniform porous SiC. Although there are some techniques known to grow porous silicon carbide with ordered structures, our method is unique for the growth of SiC with monodisperse spherical pores. Therefore well ordered carbon opals (as shown in Fig. 2) are needed, their fabrication has been reported elsewhere $REF$. After repeated infiltration with the liquid precursor sol and the high temperature annealing step, the material is turned in a closed SiC-carbon compound structure. Thereby the SiC shows a strong tendency to grow monocrystalline nearly without crystal defects, although the carbon spheres imply an enormous intrusion in the growing area.
Figure 2: Colloidal crystal from monodisperse carbon spheres, prepared by a sound supported sedimentation process [7].
Figure 3: SEM image of an inverted colloidal crystal on a monocrystalline SiC substrate. Inlay shows magnification of a single pore.
Only when the surface is polished, it becomes possible to remove the remaining carbon spheres from the structure by annealing in oxygen gas atmosphere. An example of such a laid open structure, here prepared on a monocrystalline 6H-SiC substrate, is presented in Fig. 3. The inlay shows the magnification of a single pore, a slight tendency of the pores to form hexagonally shaped can be seen. Gas phase grown films. With this method porous polycrystalline SiC films have been grown on various substrates, as for example on silicon carbide, on sapphire or on glassy carbon. Thicknesses of the films could be obtained from few microns to several millimetres easily. The suspicion, silica (quartz or cristobalite) could be present in the grown films as well, caused by SiO gas in the reaction chamber, could be cleared out by characterization of the material with NMR and XRD.
Figure 4: 29Si-MAS-NMR spectrum of a freestanding mesoporous SiC layer (dark solid curve). No silica has been found (light grey line indicates the position of a usual SiO2 signal).
Figure 5: XRD of a porous SiC layer, grown on a glassy carbon substrate (solid black line). The simulated reflexes indicate the typical intensity distribution of a 3C-SiC powder diagram.
Even with very sensitive 29Si solid state NMR spectroscopy no silica traces, usually found at a chemical shift of -110ppm, could be detected (Fig. 4). Just the signal at -17ppm, typical for silicon carbide has been observed. XRD reflexes from porous films (see example Fig. 5, measured at a porous SiC layer grown on a glassy carbon substrate) show concerning the appearance angle good accordance to calculated 3C-SiC powder spectra [9]. Although the intensities are not comparable, because of preferred crystal orientation of the grown films, this shows clearly the exclusive growth of 3C-SiC in all samples. Have films been grown with an adequate thickness on a removable substrate, as for example graphite or glassy carbon, it is possible to achieve a freestanding porous film. Are 3 diFigure 6: Porous polycrystalline SiC shell of a mensional parts covered, even SiC shells of these bodies can be obtained. An example therefore is preformer graphite substrate.
sented in Fig. 6, showing the porous SiC shell of a former completely covered glassy carbon substrate, which has been burned off afterwards (by annealing in oxidative atmosphere). Converted carbon. To discover the properties of the conversion process and the obtained porous silicon carbide samples, a large range of probe bodies has been either completely or at least partially converted. This method, based on molecular exchange processes, offers several advantages. To achieve a requested silicon carbide body, a carbon copy is processed to the requested shape and dimensions and is easily converted into porous silicon carbide. No direct mechanical processing of SiC is necessary this way. Thereby the carbon part is transformed to SiC from the outside
Figure 7: Cross section of a half way converted sample disc.
Figure 8: SEM image showing the porous polycrystalline structure of the cross section of a converted sample
towards the center of the sample. That can be seen clearly, when looking at the cross section of an incompletely converted sample disc (Fig. 7), whose remaining carbon has been removed by annealing in air. The white center of the disc is very soft and shows the beginning conversion. Another advantage of our conversion method is the mostly higher density compared to the samples made from other cheap pressing methods and additionally much more clean. Doped porous SiC. The sol-gel derived samples have been prepared with several dopants of interest for electronic applications of SiC, as e.g. N, Al, P and B. Also the introduction of rare earth ions has been tested. The presence of foreign ions has been investigated with EPR, a very powerful tool for verification of dopants in semiconductors.
Figure 9: EPR spectrum of nitrogen doped porous 3C-SiC.
Is has to be mentioned, that pure undoped silicon carbide, obtained from our sol-gel based method, shows initially no EPR signal, different to most commercially available SiC. Thereby nearly all of the tested introduced dopants could be detected positively in our samples. One example is shown in Fig. 9, a free gas phase grown sample doped with nitrogen, which has been introduced by adding nitric acid to the initial precursor solution. The measurement has been performed at 11K, at a frequency of νEPR = 9.9416 GHz. The g-factor could be determined with g = 2.0050 and is therefore according very well to the value known from monocrystalline 3C-SiC. The 14N hyperfine interaction with 3.5MHz is not resolved.
CONCLUSIONS The different reported sol-gel based growth processes for porous SiC deliver very clean materials of selectable pore size and pore arrangements, which are therefore suitable for photonic as well as for electronic applications. The conversion of carbonaceous bodies into porous lightweight and stable silicon carbide, offers a useful easy method for the production of requested SiC parts, without having to process the hard SiC directly. Two different doping procedures allow nand p-type doping with shallow donors such as nitrogen and phosphorous and shallow acceptors such as boron and aluminium. Overall, our described methods are easy in production at notably low costs. Numerous applications of doped and undoped porous SiC are imaginable, for instance in photonic, electronic, but also for machine parts or filtering media. REFERENCES 1. J.M. Tulliani, L. Montanaro, T.J. Bell, V. Swain, J. Amer. Ceram. Soc. 1999, 82, 961. 2. Y. Shishkin, W.J. Choyke, R.P. Devaty, J. Appl. Phys. 2004, 96, 2311. 3. Brevier - Technische Keramik, Hrsg.: Informationszentrum Technische Keramik (IZTK) und Technische Kommission der Fachgruppe Technische Keramik, Fahner Verlag, Lauf 1998. 4. B. Friedel, S. Greulich-Weber, Materials Science Forum 2006, 527-529, 759. 5. B. Friedel, Thesis, University of Paderborn 2007. 6. B. Friedel, S. Greulich-Weber, Small 2006, 2, 859. 7. B. Friedel, S. Greulich-Weber, Mater. Res. Soc. Symp. Proc. 2007, 951, 0951-E06-27. 8. B. Friedel, S. Greulich-Weber, Patent pending DE102006055469.8. 9. K. Yvon, W. Jeitschko, E. Parthé, J. Appl. Cryst. 1977, 10, 73.
