EP2443063A1

EP2443063A1 - A method for the manufacture of a silicon polytype material

Info

Publication number: EP2443063A1
Application number: EP09776683A
Authority: EP
Inventors: Yan Yu; Lin Gu; Wilfried Sigle; Joachim Maier; Peter A. Van Aken
Original assignee: Max Planck Gesellschaft zur Foerderung der Wissenschaften eV
Current assignee: Max Planck Gesellschaft zur Foerderung der Wissenschaften eV
Priority date: 2009-06-04
Filing date: 2009-06-04
Publication date: 2012-04-25
Also published as: WO2010139346A1

Abstract

A method of manufacturing a macroporous silicon polytype material comprising the steps of a) mixing a nanopowder consisting of a silicon compound with a nanopowder consisting of a metal able to reduce the silicon compound to silicon to form a mixture, b) heating the mixture in a gas atmosphere consisting of an inert gas optionally with an proportion of a reducing gas for the silicon compound at an elevated temperature in the range from 500°C to 1000°C for a period of several hours to reduce the silicon compound to silicon, and c) subsequently treating the resultant product in an acid to remove any excess metal and metal compound resulting from step b) and leave a macroporous silicon polytype material. A macroporous silicon polytype material with an almost direct bandgap is also described and claimed.

Description

A Method For The Manufacture Of A Silicon Polytype Material

The present invention relates to a method for the manufacture of a silicon polytype material and to a silicon polytype material, e.g. a silicon polytype material made by the method.

Owing to its position in the periodic system, natural abundance and environmental benignity, silicon is a natural candidate for making technological use of the interaction between photons and electrons. However, the indirect bandgap transition characteristics of silicon often makes it an inefficient emitter of light and thus inappropriate for demanding optoelectronic applications. Substantial efforts have been made in the past to enhance its luminescence. These efforts include fabricating low-dimensional architectures, for example as described in the papers by Wilson, W. L., Szajowski, P. F. & Brus, L. E. "Quantum confinement in size-selected surface oxidized silicon nanocrystals" in Science 262, 1242-1244 (1993), by Lu, Z. H., Lockwood, D. J. 85 Baribeau, J. -M. "Quantum confinement and light emission in SiO₂/ Si superlattices" in Nature 378, 258-260 (1995), by Pavesi, L., Dal Negro, L., Mazzoleni, C, Franzό, G. 85 Priolo, F. "Optical gain in silicon nanocrystals" in Nature 408, 440-444 (2000), by Wai Lek Ng et al. "An efficient room-temperature silicon-based light-emitting diode" in Nature 410, 192-194 (2001) and by Leong, D., Harry, M., Reeson, K. J. K. 86 Homewood, P. "A silicon/ iron-disilicide light-emitting diode operating at a wavelength of 1.5 μm" in Nature 387, 686-688 (1997). The efforts referred to also include the concept of fabricating nearly lattice-matched new phases as described in the papers by Ding, Z. et al. "Electrochemistry and electrogenerated chemiluminescence from silicon nanocrystal quantum dots" in Science 296, 1293- 1297 (2002) and by Zhang, P. et al. "Com- putational design of direct-bandgap semiconductors that lattice-match silicon" in Nature 409, 69-71 (2001).

In addition prior art efforts to enhance luminescence include the use of nanoporous structures as described in the papers by Cullis, A. G. & Can- ham, L. T. "Visible light emission due to quantum size effects in highly porous crystalline silicon" in Nature 353, 335-338 (1991) and by Hirschman, K. D., Tsybeskov, L., Duttagupta, S. P. 8B Fauchet, P. M. "Silicon-based visible light-emitting devices integrated into microelectronic circuits" in Nature 384, 338-341 (1996).

However, fabrication of a complex highly porous system, such as is described in the last two named papers and in the further papers by Nolan, M., O'Callaghan, S., Fagas, G., Greer, J. C. 85 Frauenheim, T. "Silicon nanowire band gap modification" in Nano Lett. 7, 34-38 (2007) and by Yao, D., Zhang, G. 85 Li, B. "A universal expression of band gap for silicon nanowires of different cross-section geometries" in Nano Lett. 8, 4557- 4561 (2008) - offering benefits from quantum confinement effects - remains a challenging task. This also applies to the realisation of twinning superlattice structures to modulate the Si electronic structure. Twinning superlattice structures are described in detail with respect to silicon and other materials in the paper by Ikonic, Z., Srivastava, G. P. 8& Inkson, J. C. "Electronic properties of twin boundaries and twinning superlattices in diamond-type and zinc-blende-type semiconductors" in Phys Rev. B 48, 17181-17193 (1993) and with respect to germanium in the paper by Ikonic, Z., Srivastava, G. P. 8ε Inkson, J. C. "Direct optical transitions in indirect semiconductors: The case of Ge twinning superlattices" in Phys Rev. B 52, 1474-1476 (1995). The object of the present invention is to provide a method of manufacturing a silicon polytype material which has an almost direct bandgap as well as a silicon polytype material of this kind with the method being relatively simple, economical and capable of use on a large scale with a good yield and with the silicon polytype material having an almost direct bandgap and likewise being capable of being produced simply and in relatively large quantities at low cost.

In order to satisfy these objects there is provided a method of manufacturing a macroporous silicon polytype material comprising the steps of:

a) mixing a nanopowder consisting of a silicon compound with a nanopowder consisting of a metal able to reduce the silicon compound to silicon to form a mixture,

b) heating the mixture in a gas atmosphere consisting of an inert gas, optionally with an proportion of a reducing gas for the silicon compound, at an elevated temperature in the range from 500⁰C to 1000⁰C for a period of several hours to reduce the silicon compound to silicon, and

c) subsequently treating the resultant product in an acid to remove any excess metal and metal compound resulting from step b) and leave a macroporous silicon polytype material.

The silicon compound is preferably at least one of SiO2 and SiN. The nanopowder of the silicon compound typically has a diamond structure and is for example a silicon dioxide nanopowder supplied by Sigma- Aldrich 89552 St Germany under the product designation 637246 with 5 to 15nm BET and 99.5 % metals basis, although other silicon compound nanopowders can also be used, i.e. nanopowders with particle sizes in the range from 1 nm to 1 μm.

The metal nanopowder is preferably magnesium and in the experiments carried out so far magnesium powder obtained from the company Alfa Ae- sar GmbH 8B CO Kg 76057 Karlsruhe Germany with the product designation 010233L 14803 of -325 mesh size and 99.8% purity was used, although other magnesium or aluminium nanopowders, for example in the size range from 1 nm to 1 μm, could also be used.

The acid used for the step c) is preferably HCl, typically 10% HCl but could be another acid such as HNO3, typically 10% HNO3.

The method may also comprise the further step d) of treating the silicon polytype material of step c) with a further acid to remove a surface oxide layer. This step can, if required, be carried out at a later stage, for example when the silicon polytype material of step c) has acquired a native oxide layer or includes residual oxide material, for example immediately prior to use of the silicon polytype material in an electronic or photoluminescent application. HF is normally used to remove the oxide or nitride layer from the silicon polytype material.

It is particularly preferred when the temperature range for step b) is selected to lie in the range 700⁰C to 1000⁰C and most preferably in the range 800°C to 950°C.

The pressure of the gas atmosphere in the method step b) is conveniently a relatively low pressure, for example in the range from 0.5 bar to 10 bar, especially in the range from 0.8 bar to 1.2 bar. The inert gas atmosphere used for step b) preferably includes a further gas or gaseous compound adapted to reduce the silicon compound such as H2 or NH3.

Investigations have shown that the result of method step c) is a silicon polytype material which can be described as a 3-dimensional (3D) macro- porous 9 R Si polytype with native surface oxidation, which exhibits a different electronic structure than bulk diamond- structured Si. Unlike the latter one which has an indirect bandgap transition close to the zone boundary, the conduction band minimum in the material made by the method of the invention significantly shifts towards the T point, indicating less momentum transfer required to fulfil the bandgap transition. It is particularly beneficial that to obtain the desired morphology a relatively straightforward synthesis process by magnesiothermic reduction of silica nanoparticles at temperatures in the range indicated facilitates the formation of a 3D macroporous structure as well as the 9R Si polytype growth leading to the distinct electronic property.

It should be noted that the desired almost direct bandgap is achieved by producing the porous silicon material of 9R polytype and indeed with pores preferably in the range from 100 to 300nm, although pore sizes which are slightly larger or smaller are acceptable particularly since the pore size distribution typically approximates to a Gaussian distribution determined by the particle size of the magnesium or aluminium powder that is used.

It should also be noted that the heating step b) is strongly exothermic and can lead to local heating of the mixture. It is believed that the 9R polytype material actually only arises at temperatures above 650⁰C; however, due to local hot spots heating to 500⁰C can lead to the formation of some re- gions of 9R polytype material. It is preferable if the temperature can be maintained in the range from 800⁰C to 950⁰C since then a high yield of 9R polytype material results with little and preferably no material of bulk diamond structure. Heating to temperatures above 950⁰C is problematic because it is close to the melting point of silicon and can result in destruction of the desired 9R polytype material, particularly at any local hot spots that may occur due to the exothermic nature of the reaction. At this point a comparison can usefully be made with the method described in the paper by Bao, Z. et al. "Chemical reduction of three-dimensional silica micro- assemblies into microporous silicon replicas" in Nature 446, 172-175 (2007). There a process is described for the chemical reduction of three- dimensional silica micro-assemblies into microporous silicon replicas. More specifically silica diatom frustules or other silica structures of biological origin were converted into MgO /Si bearing replicas by the following reaction with gaseous Mg:

2Mg(g) + SiO₂(s)→2MgO(s) + Si(s)

The MgO /Si composites generated by this reaction retained the three dimensional cylindrical morphology and nanoscale features of the Aulaco- seira frustules. After immersion in a IM HCl solution for four hours the selective and complete dissolution of the magnesia took place and the resulting silicon based product retained the 3D morphology and nanoscale features of the Aulacoseira frustules.

Although the method proposed by Bao Z. et al. initially seems to resemble the method of the present invention there are actually many significant differences. First of all Bao et al. operate with different starting products, gaseous Mg instead of Mg nanopowder and biological silica structures rather than silica nanopowder with a structure of the bulk diamond type. Bao et al. conduct their process at a low temperature of 650⁰C, i.e. above the melting point of Mg which is significantly less than the preferred temperature range of 800 to 950⁰C for the present invention. Moreover, Bao et al. clearly state that they find no change in morphology whereas the method of the present invention results in a profound change in the morphology of the silica from that of a bulk diamond form to a twinned super- lattice form intermixed with a bulk diamond form. Moreover, the silicon polytype material resulting from the method of the present invention is macroporous, meaning that it has pores predominantly in the size range above 50nm, typically an average pore size of 200nm with a generally Gaussian distribution extending from about 100 to 300nm. However, pore sizes less than lOOnm are possible if the magnesium powder is made correspondingly smaller and could be beneficial. Also larger pore sizes up to lμm are considered to be practical and useful. In any event such pore sizes are significantly larger than the micropores of less than or equal to 2nm reported in the Bao et al. reference. The Bao et al. reference does not contain any information on the bandgap of the resulting microporous material.

In contrast measurements made on the silicon polytype material made by the method of the present invention have provided direct experimental evidences from valence electron energy-loss spectroscopy, carried out using the method described in the paper by Gu, L. et al. "Band-gap measurements of direct and indirect semiconductors using monochromated electrons" in Phys. Rev. B 75, 195214 (2007) and associated imaging techniques that the bandgap transition of this material occurs at 1.1 eV close to the F point, removing a major obstacle for the technological use in Si-based high-performance optoelectronic devices. The present invention will now be explained in more detail with reference to the accompanying drawings and to a method of making the macropor- ous silicon polytype material of the present invention. In the drawings there are shown:

Figs. Ia and Ib schematic diagrams of the structures offering possibilities for band structure modification of Si crystal. The Si atoms are shown as grey circles with Fig. Ia showing a surface oxidized porous structure with high surface to volume ratio (SVR); the black dots indicates oxygen atoms and with Fig. Ib showing a twinning superlattice structure with twinning planes marked by dotted lines. The crystal orientation is reversed at these planes,

Figs. 2a to 2c the morphology of the 3D macroporous Si polytype of the present invention with Fig. 2a showing a bright field (BF) micrograph showing the interconnected macroporous structure, Fig.2b showing a HRTEM micrograph with the corresponding SAD (selected-area electron diffraction) pattern shown in the inset revealing the polytype superstructure and with Fig. 2c showing a Wiener-filtered HRTEM micrograph magnified from the region marked in Fig. 2b together with the corresponding diffractogram demonstrating the 9R polytype structure. A schematic of the 9R stacking sequence is shown in the insert at the top right of Fig. 2c,

Figs. 3a to 3c structural and chemical investigation of the macroporous polytype Si before exposure to the HCl solution, with Fig. 3a showing a schematic illustration for the formation of the Si 3D structure with MgO is displayed as grey spheres and polyhedra, with Fig. 3b showing a BF micrograph and with Fig. 3c showing an elemental mapping of the same region as presented in Fig. 3b using the characteristic volume-plasmon energy-losses of MgO and Si to verify the macroporous polytype structure evolution, the cavities shown as dark patches originate from the void space occupied by the MgO products prior to the reaction with the HCl solution and the light grey areas show the Si material,

Figs. 4a to 4c diagrams confirming a bandgap transition of the natively surface-oxidized 3D macroporous 9R Si polytype with significantly less momentum transfer requirement in comparison to diamond- structured Si, with Fig. 4a showing a 9R-Si diffraction pattern in [110] orientation showing the two positions (dotted circles) at which DF VEELS spectra in Fig. 4c were recorded with the center of the collection aperture being located at 0.5 and 1.3 nπr¹ respectively with the aperture diameter of about 0.8 nrrr¹, circles with a grey outer boundary in indicate the regions within which spectral artefacts are likely to occur, with Fig 4b showing a diffraction pattern of diamond-structured Si in [110] orientation showing the two positions (dotted circles) at which DF VEELS spectra were recorded and with Fig. 4c showing spectra of 9R polytype Si (from positions 9R-Si-I and 9R-SΪ-2) and diamond- structured Si (from positions Si- 1 and Si- 2) obtained by DF VEELS. Both spectra of the macroporous 9R Si show a bandgap transition energy of about 1.1 eV. This is the case for diamond- structured Si only at large momentum transfer range (position 2), as expected for its band structure.

Figs 5a to 5c schematic band structures of unfolded (Fig. 5a), one time folded (Fig. 5b) and two times folded superlattice systems (Fig. 5c) with the dashed lines being drawn to facilitate comparison and with E_g and k stand for the bandgap energy and the required momentum transfer respectively,

Fig. 6 surface morphology of the natively surface-oxidized 3D macroporous 9R Si polytype, cavities with size distributions ranging from 100 to 300 nm were observed throughout the structure,

Fig. 7 chemical composition analyses of the 9R Si polytype after HCl solution treatments using EDX spectroscopy performed in image mode with an illuminated area diameter of about 1 μm, more than 93 at.% of Si and less than 7 at.% of oxygen were identified, indicating the presence of native surface oxides, the Cu and C signal come from the carbon lacey film supported by a Cu grid on which the studied material was deposited,

Fig. 8 a table showing the composition extracted from EDX analyses of the 9R Si polytype after HCl solution treatment, the uncertainty is the statistical error of the peak counts of the two species, Fig. 9 the microstructure of Si materials synthesized at 650⁰C, which is close to the melting point of metallic Mg, the other growth parameters were unchanged, the material contains mainly diamond- structured Si as revealed by the lattice fringes extending outside the 9R Si polytype,

Fig.10 a simulated map of the energy loss versus electron momentum of a diamond- structured Si specimen with a thickness of 30 nm and an incident electron energy of 200 keV, the Cerenkov losses associated with the surface modes occur mainly at a momentum transfer range below 40 pm ¹.

To achieve direct bandgap transitions in a nominally indirect bandgap semiconductor is of profound interest for both theoretical and practical aspects but remains a crucial challenge. It is known that the transition properties of indirect semiconductors can be modified by taking advantages of quantum confinement and surface termination effects. For example, a natively surface-oxidized porous structure, which is shown as a schematic in Fig. Ia, provides a high surface area to volume ratio (SVR) as well as a stable surface environment. The white area in the middle of Fig. Ia shows a pore of the porous structure. In addition, semiconductor polytypes, which can be viewed as a special class of twinning superlat- tices, offer the possibility to modulate band structures depending on the stacking sequences of the materials (Figs. 5a to 5c). Success in precisely controlling twinning superlattice growth achieved by the method of the present invention promises further access to direct transitions in indirect- bandgap semiconductors without complications such as the need for momentum-compensating phonons for instance. Relying on the periodic reversal of the crystal orientation (Fig. Ib), folding- mediated direct optical transitions are predicted, which differs significantly from the approaches to modulate the chemical composition and stacking sequences, e.g. in the Si/ Ge superlattice system. Since the material achieved by the present invention involves only pure silicon, it is fully compatible with existing Si- based semiconductor processing technologies, for example as described in the above mentioned Hirschman et al. reference. The twin boundary between the two crystal orientations provides a perfectly lattice-matched interface, which addresses the critical enquiries on the long-term stability due to the presence of excessive strain. The 3-dimensional (3D) macropor- ous structure enables further enhancement of the luminescence resulting from both quantum confinement and surface termination effects, thus the optical efficiency for room-temperature operation can be further improved.

Thus the silicon material of the present invention is an initially natively surface-oxidized 3D macroporous silicon polytype, namely the 9R structure, which possesses a bandgap transition with significantly less momentum transfer requirement in comparison to diamond- structured Si. The folding-mediated transition mechanism, quantum confinement and surface termination effects contribute conjunctively to its distinct electronic property. Evidence is available from the valence electron energy-loss spectroscopy measurements (VEELS) and associated imaging techniques in a transmission electron microscope (TEM), i.e. the SESAM microscope as described in the paper by Koch, C. T. et al. "SESAM: Exploring the frontiers of electron microscopy" in Microsc. Microanal. 12, 506-514 (2006), that the bandgap energy of the natively surface-oxidized macro- porous 9 R Si is close to the F point, indicating only a marginal momentum transfer being required for bandgap transitions. The common crystalline form of Si has a diamond structure with the space group Fd3 m (No. 227). Till now more than 10 Si polytypes have been observed since the discovery of high-pressure phases of Si as described in the papers by Wentorf, R. H. Jr. & Kasper, J. S. "Two new forms of silicon" in Science 139, 338-339 (1963) and by Kailer, A., Gogotsi, Y. G. & Nickel, K. G. "Phase transformation of silicon caused by contact loading" in J. Appl. Phys. 81, 3057-3063 (1997) and the references therein. Besides the 9R Si polytype, 9R-structured diamond and SiC polytypes, which also have the [21]3 stacking sequence, were reported before in the papers by Lifshitz, Y. et al. "Epitaxial diamond polytypes on silicon" in Nature 412, 404 (2001) and by Kaiser, U., Chuvilin, A., Brown, P. D. & Richter, W. "Origin of threefold periodicity in high-resolution transmission electron microscopy images of thin film cubic SiC" in Microsc. Microanal. 5, 420- 427 (1999). However, most of these structures were synthesized with high pressure at elevated temperatures. The recent report on chemical reduction of silica by Bao et al. as discussed above enables production of 3D microporous silicon replicas at lower temperature and pressure which, however, is still far from fabricating porous polytypes.

Thus the present invention has allowed the fabrication of 3D macroporous Si 9R polytypes by magnesio thermic reduction of silica nanoparticles at temperatures above 500⁰C and particularly with a good yield at temperatures above 650⁰C and especially in the range from 800 to 950°C.

More specifically Figs. 2a to 2c shows the microstructure of the natively surface-oxidized 3D macroporous 9R Si polytype. A bright-field (BF) transmission electron micrograph (Fig. 2a) demonstrates the interconnected macroporous structure with pore diameters ranging from 100 to 300 nm. This is confirmed by the scanning electron micrograph shown in Fig. 6. The chemical composition was measured by energy-dispersive X- ray spectroscopy (EDX) over a large region and more than 93 at.% of Si and less than 7 at.% of O after absorption corrections were detected (Fig. 7 and the table of Fig. 8) . Since a thin surface oxidation layer forms upon exposing fresh Si to ambient atmosphere, the noticeable concentration of O is indicative of the high SVR. Note that small composition variations are possible depending on different SVRs. The high-resolution transmission electron microscopy (HRTEM) micrograph (Fig. 2b) and selected-area electron diffraction (SAD) patterns (insert of Fig. 2b) reveal the 9R polytype superstructure. Compared to the diffraction pattern of diamond- structured Si, two extra diffraction spots are detected between the 000 and 009 reflections, the latter one overlapping with the diamond- structured Si {111} reflections. These extra spots correspond to the 9R polytype structure with the orientation relationship of [001]9R| < 1 1 l>si, where the subscript Si stands for diamond- structured Si. The Wiener- filtered HRTEM micrograph in Fig. 2c, magnified from the region marked in Fig. 2b, reveals the atomic ordering throughout the whole structure. A diffractogram, shown in the insert of Fig. 2c, obtained by fast Fourier transformation of the HRTEM image, confirms the 9R structure. The [21J3 stacking sequence is displayed in the schematic with the different atomic layers being indicated.

At the growth temperature of 850 ⁰C which is above the melting point of metallic Mg (about 650 ⁰C), the Mg reactants are essentially present in the liquid form. It is very likely that the solid-liquid interface facilitates the formation of planar stacking faults and twinning boundaries. This is confirmed by the fact that materials grown at lower temperature, e.g. 650 ⁰C, consist mainly of diamond- structured Si (see Fig. 9) due to the lower reaction temperature not suitable for vapor-liquid- solid growth. At higher tern- peratures especially at temperatures in the range from 800⁰C to 950⁰C there is a higher proportion of the twinning superlattice structure. Fig. 3a shows a schematic of the formation of the 3D macroporous polytype structure. In this schematic, alternative segments as displayed by different transparency resemble the 9R stacking sequence. Amid the bulk material, residues of the reaction are distributed, namely MgO in solid form, which was subsequently removed in a HCl solution leaving cavities of comparable sizes. To testify the structural evolution, a zero-loss-filtered BF micrograph and an elemental mapping of the same region using the characteristic volume-plasmon losses of MgO at 24 eV and Si at 17 eV were acquired prior to the exposure to the HCl solution. The BF micrograph shown in Fig. 3b demonstrates two sets of {111} planar stackings. The angle between the planes was measured to be about 107.5°, close to the theoretical angle of 109.5° between the (111) and (l ϊ ϊ) planes. The elemental mapping shown in Fig. 3c shows the MgO spheres and polyhedra embedded in the Si matrix of sizes comparable to the cavities as shown in Fig. 2a. This provides strong evidence that the macroporous structure is essentially deduced from the void space after removal of MgO by the HCl solution.

Valence electron energy-loss spectroscopy establishes a direct link between the local electronic structure and spectroscopic features with both high spatial and angular resolution. However, due to possible artefacts (see Fig. 10), understanding the indirect band transition characteristics of Si by VEELS analyses requires profound pre-knowledge. Both theory and experimental work by Kroger, E. "Berechnung der Energieverluste schneller Elektronen in dϋnnen Schichten mit Retardierung" (calculation of the energy losses of fast electrons in thin layers with retardation) in Z. Physik 216, 115-135 (1968) and by Chen, C. H., Silcox, J. & Vincent, R. "Electron-energy loss in silicon: Bulk and surface plasmons and Cerenkov radiation" in Phys. Rev. B 12, 64-71 (1975) revealed that these spectroscopic artefacts are only detected by electrons scattered at small angles. Thus, substantial suppression of the artefacts and direct interpretation of the Si VEELS spectra are possible by using the dark-field (DF) mode, where only electrons at large scattering angles are detected.

To reveal the bandgap transition properties of the macroporous 9R Si polytype, DF VEELS was performed using different momentum transfer ranges at positions 1 and 2 indicated by the dotted circles in the [HO]- oriented diffraction pattern (Fig. 4a) using a collection aperture diameter of about 0.8 nm ¹. Both positions, with the center of the collection aperture located at 0.5 and 1.3 nnr¹ respectively, are sufficiently far from the T point (> 100pm-¹) in order to avoid spectral artefacts (see Fig. 10). A similar diffraction pattern of diamond- structured Si in [110] orientation with two positions, from which the corresponding DF VEELS spectra (Fig. 4c) were acquired, is shown in Fig. 4b. The background of the zero-loss-peak tails were subtracted by a power-law fit which gives an uncertainty comparable to the experimental energy resolution on the order of 0.1 eV. A bandgap onset energy of about 1.1 eV, as revealed by the first intensity increase, was detected from the 9R Si for both positions 1 and 2. This band-gap energy is comparable to the indirect band-gap energy of bulk diamond- structured Si. This observation confirms that a bandgap of about 1.1 eV with significantly less momentum transfer requirement in comparison to diamond- structured Si is present in the material of the present invention. In contrast, DF VEELS acquired from diamond-structured Si at similar positions in reciprocal space, along F-X direction at 0.5 and 1.3 nm ¹ respectively, exhibits a very different spectral behavior. The indirect band- gap onset at 1.1 eV is only detected with fairly large momentum transfer at position 2, in accordance with its band structure for the indirect transi- tion close to the X point. At position 1 , interband transitions other than the bandgap transition at the X point dominate and the onset of interband transitions is shifted to higher energies.

A bandgap transition with significantly less momentum transfer requirement has been observed for the material of the invention in comparison to diamond-structured Si in a natively surface-oxidized 9R Si polytype with 3D macroporous structure. This distinct transition property arises from conjunctive effects of quantum confinement, native surface oxidation and band folding, which makes it a highly desired candidate material for further development in Si-based high-efficiency optoelectronic devices.

As a method for the preparation of the 3D macroporous silicon polytype s powder, we follow this reaction (g: gas, I: liquid, s: solid): 2Mg(g, I) + SiO₂ (s) → 2MgO(s) + Si(s) ( 1 )

1 g magnesium nanopowder (- 100 nm, 99.9%, Aldrich) was ground in the agate mortar with 1 g Siθ2 nanopowder (99.99%, Aldrich) to obtain a uniform precursor powder. The mixture (dark gray) with excessive Mg was then transferred into a tube furnace for heat treatment under an Ar (95 vol.%) / H₂ (5 vol.%) atmosphere at 850 ⁰C for 8 h. The heating rate was kept at 5 K/min. The as-grown powders (black) were first immersed in a IM HCl solution for 12 hours at room temperature to remove MgO, followed by etching off SiO₂ using the HF (20%) solution for pure silicon. The HF etching process was conducted in an argon-filled glove box, where both oxygen and H₂O levels were below 1 ppm.

The as-grown powder is etched with HCL to remove any Mg and Mg- compounds remaining in the as-grown powder, to further purify the as- grown powder. Following the first etching step the as-grown powder can be placed into a centrifuge with ethanol or, any other kind of solvent to separate the as- grown powder from the HCL. This process can be repeated for 3 to 5 times to make sure, that an HCL free powder is obtained.

After the as-grown powder has been etched with HCL and possibly been placed into a centrifuge to remove any remaining HCL in the as-grown powder, the as-grown powder can either be packaged for sale, or it can be further purified by being placed into an HF solution as described above. If the powder is intended for sale, it may not be etched using HF to prevent to an unnecessary loss of the overall yield of the twinning superlattice structure due to surface oxidation.

Following the etching with HF, the as-grown powder can again be placed into a centrifuge this time with de-ionized water or, any other kind of solvent to separate the as-grown powder from the HF. This process can also be repeated for 3 to 5 times to make sure, that an HF free powder is obtained.

This powder can then be transferred to its intended use which can be, for example, the embedding of the obtained nanocrystals into a matrix which can then provide a platform for potential optoelectronic applications, as is disclosed in the publication by Pavesi, L., Dal Negro, L., Mazzoleni, C, Franzό, G. & Priolo, F. Optical gain in silicon nanocrystals. Nature 408, 440-444 (2000).

If required, the as-grown powder can be placed into a vacuum oven (Ley- bold Hereaus) and dried, for example, at 80⁰C and at approximately 10 to 20 mbar to obtain a dried as-grown powder. The initial Mg and SiO2 powder size used was approximately 100 nm.

In a further embodiment in accordance with the invention 0.25 g of magnesium nanopowder (- 100 nm, 99.9%, Aldrich) was ground in the agate mortar together with 0.25 g SiO2 nanopowder (99.99%, Aldrich) to obtain a uniform precursor powder.

This mixture (dark gray) with excessive Mg was then transferred into a tube furnace for heat treatment under an Ar (95 vol.%) / H2 (5 vol.%) atmosphere at 650⁰C for 12 h. The heating rate was kept at 5 K/min.

The as-grown powders (black) were first immersed in a IM HCl solution for 12 hours at room temperature to remove MgO. Following the etching with HCL, the as-grown powders were placed into a centrifuge to separate them from any residual MgO and/ or Mg followed by the etching off SiO2 using the HF (20%) solution for pure silicon. The HF etching process was conducted in an argon-filled glove box, where both oxygen and H2O levels were below 1 ppm.

The resulting powder was then washed with de-ionized water or, any suitable liquid such as ethanol to separate the as-grown powders from the HF. This process can also be repeated for 3 to 5 times to make sure, that an HF free powder is obtained.

To verify the method in accordance with the invention, the surface morphology was investigated using a JEOL 6300F field-emission scanning electron microscope (JEOL, Tokyo, Japan) operated at 15 keV. TEM specimens were prepared by dispersing the materials in ethanol followed by sample collection using carbon lacey films. High-resolution transmis- sion electron microscopy (HRTEM) was performed using a JEOL 4000EX transmission electron microscope (JEOL, Tokyo, Japan) operated at 400 keV. The interpretable resolution defined by the contrast transfer function of the objective lens is 0.16 nm. Energy-dispersive X-ray spectroscopy (EDX) analysis was carried out using an EDAX system (EDAX, Mahwah, NJ, USA) attached to the Zeiss SESAM microscope (Carl Zeiss, Oberko- chen, Germany) operated at 200 keV. For EFTEM elemental mapping, an energy- selecting slit of 0.9 eV was used to obtain the material chemical information. For DF VEELS, spectra were acquired in diffraction mode with a spectrometer entrance aperture of about 0.8 nm"¹ in diameter for the selection of the momentum transfer range.

Thus in the studies to date a 3-dimensional macroporous 9R silicon poly type with native surface oxidation has been produced, which exhibits a different electronic structure than bulk diamond-structured Si. The conduction band minimum in this material significantly shifts towards the r point, indicating less momentum transfer required to fulfil the bandgap transition, which removes a major obstacle for the technological use in Si- based high-performance optoelectronic devices.

Schematic band structures of unfolded, one time folded and two times folded superlattice systems are shown in Figs 5a to 5c. Comparably, formation of twinning superlattices provides another method to achieve folding-mediated quasi-direct bandgap transtions in indirect semiconductors. Polytype superlattices, which can be viewed as a special class of twinning superlattices, possess potential band-folding capabilities. The band structures of polytype Si materials have been discussed previously, where the bottom of the conduction band clearly shifts from the M point towards the r point as the structure transforms from diamond structure to 6H polytype. By way of further background information the field-emission (scanning) transmission electron microscope SESAM (Carl Zeiss, Oberkochen, Germany), abbreviation of Sub-Electron-volt-Sub-Angstrom-Microscope, delivers outstanding stability and optical performance which allows valence electron energy-loss spectroscopy (VEELS) and valence energy-filtered transmission electron microscopy (EFTEM) analyses with both high energy and high spatial resolution. Owing to the electrostatic Ω-type monochro- mator and the improved stability, the energy resolution of the microscope defined by the full-width-at-half-maximum (FWHM) of the zero-loss peak is below 90 meV for routine applications. The in-column MANDOLINE energy filter provides high dispersion, high transmissivity and high isochro- maticity.

Claims

1. A method of manufacturing a macroporous silicon polytype material comprising the steps of:

b) heating the mixture in a gas atmosphere consisting of an inert gas optionally with an proportion of a reducing gas for the silicon compound at an elevated temperature in the range from 500⁰C to 1000⁰C for a period of several hours to reduce the silicon compound to silicon, and

2. A method in accordance with claim 1, wherein the silicon compound is at least one of Siθ2 and SiN.

3. A method in accordance with claim 1 or claim 2 and comprising the further step of treating the silicon polytype material of step c) with a further acid to remove a surface oxide layer.

4. A method in accordance with any one of the preceding claims, wherein the temperature range for step b) is selected to lie in the range 700⁰C to 1000⁰C and most preferably in the range 800⁰C to 950°C.

5. A method in accordance with any one of the preceding claims, wherein the pressure of the gas atmosphere in the method step b) is a relatively low pressure, for example in the range from 0.5 bar to 10 bar, especially in the range from 0.8 bar to 1.2 bar.

6. A method in accordance with any one of the preceding claims, wherein the inert gas atmosphere used for step b) includes a further gas or gaseous compound adapted to reduce the silicon compound such as H 2 or NH3.

7. A method in accordance with any one of the preceding claims in which the nanopowder of the silicon compound typically has a diamond structure.

8. A method in accordance with any one of the preceding claims in which the silicon compound is a nanopowder with powder sizes selected to lie in the range from 10,000nm to 50nm, preferably in the range from lOOOnm to 50nm and especially in the range from 500 to 50nm and most preferably in the range from 300 to lOOnm.

9. A method in accordance with any one of the preceding claims in which the metal nanopowder is at least one of magnesium and aluminium and is in the form of a nanopowder with the normal structure of the metal and with powder sizes selected to lie in the range from lOOOnm to 50nm, preferably in the range from 700nm to 50nm and especially in the range from 500 to 50nm and most preferably in the range from 300 to lOOnm.

10. A method in accordance with any one of the preceding claims in which the inert gas atmosphere used for step b) preferably includes a further gas or gaseous compound adapted to reduce the silicon compound such as H2 or NH3.

11. A method in accordance with any one of the preceding claims in which the acid used for the step c) is preferably HCl, typically 10%HCl but could be another acid such as HNO3, typically 10%HNO₃.

12. A method in accordance with any one of the preceding claims in which the product of step c) is treated in a further step d) with HF to remove a native oxide or nitride from the silicon poly type material.

13. A macroporous silicon polytype material made by a method in ac- ciordance with any one of the preceding claims.

14. A macroporous silicon 9R polytype material comprising regions with a diamond form and regions with a twinning superlattice structure and having an almost direct bandgap.

15. A macroporous silicon 9R polytype material in accordance with claim 14 and present in the form of single crystal material.