From sol–gel prepared porous silica to monolithic porous Mg₂Si/MgO composite materials

Powder X-ray diffractograms (PXRD) were recorded using a Bruker D8 diffractometer with a DaVinci Design and CuK_α radiation. Evaluation of crystallite sizes was done according to the Scherrer equation and Rietveld refinement was performed using TOPAS V4-2 software (Bruker).

Sample morphology was examined using a Zeiss Ultra Plus scanning electron microscope (SEM) operated at an accelerating voltage of 2 kV with an in-lens detector. Silica samples were sputtered with gold to provide an electrically conductive surface. An Oxford Instruments X-Max energy dispersive X-ray (EDX) detector was used for elemental analysis.

The microstructure of the samples was studied with transmission electron microscopy (TEM) using a TECNAI F20 field emission microscope operated at an accelerating voltage of 200 kV. Images were recorded with a Gatan Orius SC 600 charge-coupled device (CCD) camera.

Nitrogen sorption isotherms were recorded at 77 K using a sorption porosimeter (Micromeritics, ASAP 2420). Prior to the measurement, samples were degassed for 3 h at 100 °C in vacuum. The Brunauer-Emmett-Teller (BET) surface area was evaluated using adsorption data in a relative pressure range p/p₀ 0.05–0.25. The mesopore size distribution was calculated on the basis of the desorption branch using the Barrett–Joyner–Halenda (BJH) model.

Based on the experiments by Szczech et al. [18] with diatomaceous earth SiO₂, and also on our own previous experiments [20], we reacted hierarchically structured silica, consisting of a cellular network comprising struts with hexagonally ordered mesopores with a diameter of 8 nm and a pore wall thickness of 3–5 nm (see Supplementary Fig. S1 ESI), directly with four moles of Mg vapor at a temperature of 650 °C for 2 h following Eq. 3. The PXRD pattern in Fig. 1b confirms the complete conversion to a Mg₂Si/MgO composite with a ratio of 44:56 by weight (molar ratio Mg₂Si/MgO = 0.41). Starting with amorphous silica, indicated by a very broad peak in the PXRD diffraction pattern at 22° 2 Theta (Fig. 1a), a product with crystallite sizes of 73 ± 5 nm (Mg₂Si) and 21 ± 5 nm (MgO) was obtained according to the Scherrer equation. Mg vapor can easily diffuse into and through the macroporous network, followed by condensation and reaction with the SiO₂ on the struts (see Supplementary Fig. S1 ESI), which are typically of about 1 µm in length and 100 nm in diameter. This build-up of a pure crystalline material upon the magnesiothermic conversion is accompanied by the destruction of pores in the mesoscopic range. However, a perfect preservation of the macroporous network of Mg₂Si/MgO is observed (see Fig. 1b). Similar results were obtained recently for the reaction of Si with Mg to yield pure Mg₂Si [20].

It is very likely that the presence of MgO mitigates the locally released heat from the exothermic reaction helping to preserve the macro-morphology [23]. Elemental analysis by EDX confirmed a homogeneous distribution of Mg₂Si and MgO in the macroporous network (see Supplementary Information, Fig. S2) again supporting the contention of the benefit of the better diffusion in a highly porous precursor material. The loss of mesopores is indicated by a decrease of the specific surface area (SSA) from 700 m²g⁻¹ in silica to less than 50 m²g⁻¹ in the composite sample. This data proves that we can successfully prepare a monolithic, but nanosized, Mg₂Si/MgO composite material via a direct one-step magnesiothermic reaction. However, any adjustment of the ratio of the components to each other does not seem to be possible. Variations in the amounts of Mg for the magnesiothermic reaction would result in either silica and/or silicon containing samples or residual unreacted Mg.

In order to gain better control over the molar ratio between Mg₂Si and MgO in the final composite material, we developed a route based on two successive magnesiothermic reductions making use of the two parallel reactions as given in Eqs. 1 and 3. The aim is to first convert silica with a very low amount of Mg to a silica/silicon composite material, which is reacted in a second step with Mg to give a Mg₂Si/MgO composite (see Fig. 2).

In principle, the reaction of silica and Mg is performed as described before with the exception that the amount of Mg is now adjusted to values lower than four moles (thus sub-stoichiometric with respect to Equation 2).

\(x_{\mathrm{A}}={{0.75}}; x_{\mathrm{B}}={{0. 44}};x_{\mathrm{C}}={{0.20}}\)

This leads to an incomplete reaction resulting in a monolithic material consisting of silicon, MgO and remaining silica. The given amount of Mg, indicated by “x” in Eq. 4, is always adjusted to < 2. In this series, the values have been set to (A, B, and C) x_A = 0.75; x_B = 0.44, x_C = 0.20, and thus the amount of Mg determines the content of residual silica. MgO is removed by HCl etching leaving a macroporous silicon/silica network behind. Although the distribution of Si and SiO₂ in this network is unknown, we conclude that it is homogeneous since it is homogeneously distributed in the final product (see Supplementary Information; EDX data in Fig. S3). MgO removal is necessary to be able to adjust the desired ratios in the final product. The dried Si/SiO₂ composite monolith is then subjected to a second reaction with Mg. The amount of Mg is calculated on the basis of the calculated ratios of Si/SiO₂ (see Table 1) and the stoichiometry given in Eqs. 5a and b. While silica is now converted to magnesium silicide and MgO (Eq. 5b), silicon directly gives magnesium silicide according to Eq. 5a. In this step, Si and SiO₂ form the final macroporous Mg₂Si network with homogeneously distributed MgO crystallites. Depending on the amount of silica in this step, a number of different compositions (A2, B2, and C2) are accessible without further purification (see also photographs in Fig. 2).

The monolithic shape of the starting silica gel is preserved upon all processing steps to the final composite material without significant shrinkage. As expected, the color change from white to slight brownish and furthermore to bluish demonstrates the transformation from silica to silicon/silica and Mg₂Si/MgO. Along with the transformation from silica to the Mg₂Si/MgO composite materials, we obtain a slight decrease in density overall: 0.35 g cm⁻³ for silica and 0.10–0.20 g cm⁻³ for Mg₂Si/MgO (A2 0.10; B2 0.13; C2 0.20). Interestingly, the densities of the intermediate stage of Si and silica (after removal of MgO) are the lowest, with measured values of 0.067 g cm⁻³ for A1, 0.085 for B1, and 0.095 for C1. Considering the low densities of amorphous silica (~2.20 g cm⁻³), silicon (2.32 g cm⁻³) and Mg₂Si (1.99 g cm⁻³) compared to MgO (3.58 g cm⁻³), this is not unexpected.

To prove our concept we prepared three different samples with theoretical Mg₂Si/MgO ratios of 0.66, 1.14, and 2.50 indicated with A2, B2, and C2 (Table 2). For the first reaction, we reacted (2 – 2x) mole of Mg and 1 mole of SiO₂, with x = 0.20, 0.44, and 0.75 being the amount of remaining silica after the first step. A detailed overview of the calculated reaction compositions is given in Table 2. Since Mg powder contains traces of MgO impurities we take advantage of the boat-mesh set-up, which ensures spatial separation from the reactants, since MgO always remains beneath the mesh. In addition, the chosen Ar atmosphere avoids formation of MgO traces due to Mg oxidation. We investigated the intermediate SiO₂/Si composites by PXRD diffraction as shown in Fig. 3a after extraction of MgO. All three samples (A1, B1, and C1) show an amorphous signal with a broad reflection at ~22° 2 Theta indicating remaining silica accompanied by Bragg reflexes for crystalline silicon. Unfortunately, the amorphous character of silica made detailed calculations of the composition by Rietveld refinement impossible.

The three Mg₂Si/MgO composite samples were characterized by PXRD diffraction (see Fig. 3b). For all three samples only Mg₂Si and MgO are present (reference pdf files: Mg₂Si-PDF-87-0651 and MgO-PDF-35-0773). For comparison, we normalized the XRD patterns and calculated the ratios of Mg₂Si and MgO by Rietveld refinement. As shown in Table 2, the obtained values for the molar ratio between Mg₂Si and MgO match very well with the calculated contents of silica and Mg and show the great versatility and reliability of the method. Expectedly, the absolute experimental amounts obtained in the final product are lower than the theoretically calculated ones (c.f. Tables 2 and 3) due to material loss during washing, processing by use of the mesh or even handling. The crystallite sizes calculated by the Scherrer equation are in the range 193 to 269 nm (for Mg₂Si) and ~30 nm (for MgO) (see Table 2). The morphology of the obtained composite samples was investigated by SEM and TEM. Figure 4 shows the microstructure indicating in all cases a preservation of the macroporous cellular network. The TEM images in Fig. 4 reveal the fine dispersion of Mg₂Si and MgO crystallites in each strut with the absence of pores in the mesoscopic range, which proves the contention that Si and SiO₂ are also homogeneously distributed in the intermediate product. We examined the distribution of Mg₂Si and MgO also by EDX and obtained a homogeneous distribution in an area of several square µm of the sample (see Fig S3 ESI). The loss of mesopores during processing is also supported by nitrogen sorption analysis and very low SSAs ( < 50 m²g^-1) in comparison to the highly porous silica precursor (700 m²g⁻¹).