Inorganic monoliths hierarchically textured via concentrated direct emulsion and micellar templates

Abstract

Hierarchical inorganic porous monoliths have been prepared using concentrated emulsion and micellar templates. The texture of these monoliths has to be tuned varying either the pH of the continuous aqueous phase, the emulsification process or the oil volumic fraction. These materials show interconnected macroporosity associated with vermicular-type mesostructuration. They possess an average mesoporosity of 800 m² g⁻¹ associated with bulk density as low as 0.08 g cm⁻³, which is comparable to values obtained for silica aerogel.

---

Introduction

Mesostructuration of inorganic materials is today a rather well established method.¹ To achieve both a higher degree of control and a hierarchically organized structure, mineralization can be confined to macroscopic interfaces as demonstrated by the ever-growing field of bio-inspired materials.² In such a context, ordered macroporous inorganic materials have been generated using direct concentrated non aqueous³ or aqueous emulsions.⁴ These authors aim to associate a high degree of ordered macroporosity with inorganic walls as dense as possible, to achieve the maximal contrast in the refractive index between the matrix and the macropores, an important characteristic for reaching high-performance photo-band gap properties.⁵ More recently, porous silica has been obtained based on the use of emulsions stabilized by silica particles in the absence of surfactants.⁶ On the other hand, materials with an emphasis on heterogeneous catalysis or phase separation properties should provide a high internal surface area. With regard to this last issue, distinct silica beds have been obtained with hierarchical porosity using emulsion-polymer double templates⁷ and, more recently, our group has developed a new process to obtain macrocellular silica monoliths with a high degree of control over both cell sizes and morphologies.⁸ In this paper, we present a simple method to prepare, in a one step process, porous monoliths, (Fig. 1) hierarchically textured, with organized vermicular-type mesoporosity that makes use of a double template, i.e. direct emulsion at the macroscale and micellar templates at the mesoscale.

Fig. 1 Inorganic monolith of the 1Si-HIPE0.035 sample.

Experimental

Synthesis

Tetraethoxy-orthosilane (TEOS) and tetradecyltrimethylammonium bromide 98% (TTAB) were purchased from Fluka, HCl 37% and dodecane 99% were purchased from Prolabo. Procedures were based on the use of both micelles and direct emulsion templates. Typically 5 g of tetraethoxy-orthosilane (TEOS) was added to 16.5 g of tetradecyltrimethylammonium bromide (TTAB) aqueous solution at 35% in weight. The aqueous mixture was then brought to pH 0.5 (1.8 ml of HCl 37%) or 0.035 (5.84 ml of HCl 37%) leading, after the oil emulsification process, to the materials labelled as 1Si-HIPE0.5 and xSi-HIPE0.035 respectively. A detailed synthesis concerning the 1Si-HIPE0.035 is as follows and can be extended to all other materials described in this study: 5 g of TEOS was added to 16.5 g tetradecyltrimethylammonium bromide (TTAB) aqueous solution at 35% in weight. Then, 5.84 ml of HCl (37%) was added. This aqueous phase was left under stirring for approximately 10 minutes in order to perform TEOS hydrolysis, leading to a decrease in the aqueous phase turbidity. Then the emulsion was prepared by hand in a mortar by including 35 g of dodecane drop-by-drop in the as-prepared aqueous phase. This final emulsion was then transposed into a canister and left in this state for a one-week period. The resulting material was then washed and calcined as mentioned below in the text. Beyond the pH effect over both silica condensation kinetics and the mesostructuration process, we also intended to tune the macroscopic void space texture and connectivity, playing with either with the emulsification process (emulsion shear by hand for pH 0.035 or using an ultraturax apparatus for pH 0.5) or with the oil volumic fractions (ρ_o). In this last step, 35 g, 40 g, 45 g and 60 g of dodecane were then emulsified drop-by-drop by hand in a mortar within the pH 0.035 aqueous media described above, leading respectively to the materials labelled as 1Si-HIPE0.035, 2Si-HIPE0.035, 3Si-HIPE0.035 and 4Si-HIPE0.035. The corresponding oil volumic fractions are shown in Table 1. The use of a viscous dispersed phase in a high concentration state avoids droplet migration into more mobile regions during the sol-gel process. Contrary to other works,^3,4 such a characteristic allows the nonhomogeneous macropore structuration to be overcome, hence forming homogeneous polydisperse emulsions (i.e. without any macroscale phase separation between an oily monophasic system and the emulsion). These concentrated direct emulsions were left at ambient temperature for a 15 day period to complete the sol-gel process when a pH value of 0.5 is used; this time scale is diminished to one week for a pH value of 0.035. In a second step, the monoliths were washed, by immersion in an acetone/THF (1 : 1) mixture three times over a 24 hour period. Finally, to remove the organic supramolecular-type templates, the hybrid organic-inorganic materials were treated at 650 °C for a period of six hours. The heating speed was monitored at 2 °C/min with a first plateau at 200 °C for 2 hours. The cooling process was uncontrolled and directed by the oven inertia.

Table 1 Mercury intrusion porosimetry data and oil volumic fraction of the starting emulsions

Materials	1Si-HIPE0.5	1Si-HIPE0.035	2Si-HIPE0.035	3Si-HIPE0.035	4Si-HIPE0.035
Intrusion volume/cm^3 g^−1	2.8	10.9	12.5	4.85	3.2
Porosity (%)	70	91.5	92	83	77
Bulk density/g cm^−3	0.25	0.083	0.07	0.17	0.23
Skeletal density/g cm^−3	0.845	0.98	0.91	1.06	1.04
Oil volumic fraction	0.67	0.67	0.70	0.73	0.78

---

Characterization

TEM experiments were performed with a Jeol 2000 FX microscope (accelerating voltage of 200 kV). The samples were prepared as follows: silica scaffolds in a powder state were deposited on a copper grid coated with a Formvar/carbon membrane.

SEM observations were performed with a Jeol JSM-840A scanning electron microscope operating at 10 kV. The specimens were gold-coated or carbon-coated prior to examination. Surface areas and pore characteristics at the mesoscale were obtained with a Micromeritics ASAP 2010 employing the Brunauer–Emmett–Teller (BET) method.

Intrusion/extrusion mercury measurements were performed using a Micromeritics Autopore IV apparatus to determine the scaffolds’ macrocellular cell characteristics.

X-ray diffraction experiments were carried out on an 18 kW rotating anode X-ray source (Rigaku-200) using Ge (111) crystal as the monochromator. The scattered radiation was collected on a two dimensional detector (Imaging Plate system from Mar Research, Hamburg). The sample–detector distance was 500 mm.

Results and discussion

The final inorganic monolith-type materials (Fig. 1) depict typical polymerized high internal phase emulsion (poly-HIPE)-type interconnected macroporous textures with polydisperse cell and window sizes within the micrometer range (Fig. 2).⁹ At this point we would like to note that we call “cell windows” the holes that separate two adjacent macroscopic cells.

Fig. 2 SEM visualization of the inorganic monolith-type material macrostructure. a) 1Si-HIPE0.5, b) 1Si-HIPE0.035. The black arrow shows an example of the so called “external junction” and the white one shows a typical cell window.

Effect of pH and emulsification process. Characterization at the mesoscopic and macroscopic length scale

Depending both on the pH in use during the synthesis and the emulsification process we can observe that the macroscopic texture is drastically changed. A pH value of 0.5 offers complete mineralization of the continuous aqueous phase leading to well distinguished walls (Fig. 2a), whereas for a pH of 0.035, silica condensation seems to occur at the oil-water interface leading to a texture that resembles hollow spheres strongly packed together (Fig. 2b). Also, as discussed below, this demonstrates the fact that the mineralization process first starts at the oil-water interface and then extends up to the core of the continuous aqueous phase.

The second observation obtained from SEM experiments, is the high discrepancy between the monolith’s macrocellular size, when we compared the materials 1Si-HIPE0.5 and 1Si-HIPE0.035 (Fig. 2). 1Si-HIPE0.5 presents a polydisperse diameter distribution ranging from 0.5 to 5 microns. The mean diameter is around two microns. The diameter distribution of 1Si-HIPE0.035 is by far more polydisperse and spreads from 1 to 100 microns. These differences in texture and morphology come from three important factors. The first one is the difference between the methods used to emulsify the oil droplets, 1Si-HIPE0.5 has been obtained by emulsifying the dodecane with an ultraturax apparatus leading to much smaller and more monodisperse oily droplets compared to the Si-HIPE0.035 emulsions obtained by hand. The second feature, which enhances the first one, is the difference in the aqueous media viscosity that decreases from TTAB 35% at pH 0.5 to 0.035 thus minimizing the shear effect that directly tunes the size of the oil droplets. The third factor concerns the inorganic core shrinkage effect that increases when condensation occurs closer to the isoelectric point of silica, i.e. where the fractal character of the forming gel is high.¹⁰

In order to better quantify the macroscale void space distribution we did some mercury intrusion porosity measurements. At this point we have to specify that mercury intrusion porosimetry measures the size of the macroscopic pore “windows” between the emulsion templated cells and not the diameter of the cells themselves. Also, we have to note that all the materials described in this work depict a scaffold with mechanical strength high enough to endure mercury porosity measurements, which is an important property. The Si-HIPE0.5 monolith-type material is associated with an intrusion volume of 2.8 cm³ g⁻¹ and a porosity of 70% (bulk and skeletal density of 0.25 g cm⁻³ and 0.845 g cm⁻³ respectively), whereas 1Si-HIPE0.035 has a high intrusion volume of 10.9 cm³ g⁻¹ with a porosity calculated to be 91.5% (bulk and skeletal density of 0.083 g cm⁻³ and 0.98 g cm⁻³ respectively). Fig. 3 shows the macroscopic cell window distribution as measured by mercury intrusion porosimetry. The 1Si-HIPE0.5 average macroscopic window shows a mostly bimodal distribution centered at 200 nm (Fig. 3a). This bimodal feature is a characteristic commonly observed for the emulsions in use for this study.¹¹ The window size distribution of the 1Si-HIPE0.035 monolith is by far more polydisperse and spreads from 20 up to 4000 nm associated with two major contributions respectively centered at 20 nm and 1500 nm (Fig. 3b). SEM observations allow us a better understanding of the measured porosity distribution. Those macroscale connecting windows, induced by the shrinkage of the monoliths, are located in the thin films that separate two adjacent droplets. The diameters of the holes located in the thin films are around two microns in size for a droplet of 60 microns (see Fig. 7b). The size of the hole dealing with this internal cell junction (see white arrow in Fig. 2b) is thus roughly equal to the size of the droplets divided by 30. Assuming that the same proportionality of size is still correct for the smaller droplets of around one micron, the connecting window cells should be in the range of 1/30 of a micron in size. At this point, and taking into account the initial direct emulsion polydispersity for both materials, we can then understand the polydispersity of the macrocellular window cell size distribution. Note, moreover, that in the case of the xSi-HIPE0.035 materials, an additional external porosity may emerge from the hollow spheres’ aggregation process (see black arrow in Fig. 2b). On the contrary, the walls of the 1Si-HIPE0.5 material, completely mineralized, exclude this additional feature. For all the xSi-HIPE0.035 materials, the size involved in this external porosity is related to the size of the nodes where the Plateau borders meet. The SEM pictures show that the size involved in this porosity is about 2000 nm (see white arrow Fig 2b). These holes thus enhance the peak centered at 1500 nm (Fig. 3b). In order to better specify and verify this behavior, excluding the pH parameter, we did play with the oil volumic fractions of the starting emulsions to tune the average macrocellular size and thus the connecting window cell size. This study is discussed below in the text.

Fig. 3 Pore size distribution for monoliths as measured by mercury intrusion porosimetry. a) 1Si-HIPE0.5, b) 1Si-HIPE0.035.

Wall cells observed by TEM show vermicular type mesoporous texturation (Fig. 4a and 4b) that can be confirmed with small angle X-ray diffraction profiles (Fig. 4c and 4d). The scattering curves exhibit a large peak centered at 0.13 Å⁻¹ and 0.15 Å⁻¹ for the hybrid organic-inorganic 1Si-HIPE0.5 and 1Si-HIPE0.035 materials respectively that shift to 0.17 Å⁻¹ and 0.18 Å⁻¹ after the heat treatment. In both cases, this single peak is in good agreement with a mesoscale worm-like structure¹⁶ and the shift of the d-spacing, from 48 to 37 Å for 1Si-HIPE0.5 and from 43 to 35 Å for 1Si-HIPE0.035, induced by the calcination treatment is associated with the inorganic core shrinkage effect. It is worthwhile to note that TTAB aqueous solutions at both pH values of 0.5 and 0.035 depict the X-ray diffraction pattern signature of this mesoscopic organization, meaning that the worm-like structures observed in both cases within the solid state materials are induced by a direct morphosynthesis between the organic template and the mineralization process. Also, the decrease of the d-spacing between the lyotrope organic template at pH 0.5 (56 Å) and the hybrid material 1Si-HIPE0.5 (48 Å) is certainly due to the confinement effect provided by the condensation process. This phenomenon is also observed for the lyotropic TTAB solution at pH 0.035 showing a d-spacing of 54 Å that shifts to 43 Å for the hybrid 1Si-HIPE0.035 material. Those shrinkage and sintering effects described above are also observed for all the xSi-HIPE0.035 materials series whatever the oil volume fraction in use. (ESI Fig. S1†).

Fig. 4 a) 1Si-HIPE0.5 vermicular texture observed by TEM. The scale bar corresponds to 50 nm. b) 1Si-HIPE0.035 vermicular texture observed by TEM. The scale bar corresponds to 5 nm. c) 1Si-HIPE0.5 small angle X-ray diffraction profiles, (■) before calcination, (△) after calcination and (○) TTAB solution at a pH value of 0.5 d) 1Si-HIPE0.035 small angle X-ray diffraction profiles (■) before calcination, (△) after calcination and (○) TTAB solution at a pH value of 0.035.

For the 1Si-HIPE0.5 monolith, N₂ adsorption at 77 K rises rapidly in the low relative pressure range (0 to 0.3) followed by some small additional nitrogen adsorption at higher relative pressure values indicating the textural macroporosity (Fig. 5a). An hysteresis loop can be observed between the adsorption and desorption curves, above a relative pressure of 0.5 and up to 1.0 (Fig. 5). The specific surface area of the silica monoliths was calculated to be 832 m² g⁻¹ using the Brunauer–Emmett–Teller (BET) method. The porosity distribution provided by the original density functional theory depicts pore widths ranging from 12 to 40 Å describing both micro- and mesoporosity (Fig 5b).¹⁷ The BJH desorption cumulative surface area of pores taking into account only pore diameters larger than 35 Å and up to 500 Å reports a surface area of 64 m² g⁻¹. The N₂ adsorption/desorption curves obtained for the 1Si-HIPE0.035 depict approximately the same shapes as 1Si-HIPE0.5, nevertheless associated with a higher adsorbed volume (Fig. 6a). This difference is extended to the pore size distribution curve where the pore size distribution resembles that of 1Si-HIPE0.5 but again with higher differential surface area values (Fig. 6b). The BJH desorption cumulative surface area of pores, taking into account only pore diameters larger than 35 Å and up to 500 Å, is 70.5 m² g⁻¹. Concerning the xSi-HIPE0.035 series, where the only parameter involved is the oil volumic fraction, we do not observe drastic changes both in the N₂ adsorption/desorption curves (Fig. S2, ESI†) or in the porosity distribution curves (Fig. S3, ESI†), the overall nitrogen physisorption data are summarized on Table 2.

Fig. 5 a) Nitrogen adsorption and desorption isotherms of the 1Si-HIPE0.5 monolith-type material. ○ adsorption curve, + desorption curve. b) Pore-size distribution for silica foam as calculated from the original density functional theory model.

Fig. 6 a) Nitrogen adsorption and desorption isotherms of the 1Si-HIPE0.035 monolith-type material. ○ adsorption curve, + desorption curve. b) Pore-size distribution for silica foam as calculated from the original density functional theory model.

Effect of oil volumic fraction. Characterization at the macroscopic length scale

To accomplish this study we chose the more acidic pH condition in order to enhance the condensation kinetics. The results are depicted in Fig. 7. We can observe that whatever the oil volumic fraction conditions, the general texture resembles “aggregated hollow spheres”. For the 1Si-HIPE0.035 and 2Si-HIPE0.035 porous monoliths we observe almost the same macrocellular average cell size without any dramatic increase in the cell size windows, confirmed with mercury intrusion porosimetry experiments (Fig. 8). When we increase the starting emulsion oil volumic fraction, the macrocellular cell sizes diminish drastically (Fig. 8 c, e and g) . In fact, considering the rheology of the emulsion, it is well known that the viscosity of direct emulsions increases dramatically when the oil volumic fraction reaches values above 0.74.^12,13 For the starting emulsions of the materials 3Si-HIPE0.035 and 4Si-HIPE0.035, this phenomenon increases the stress applied to the oily droplets, which induces smaller macrocellular cells¹⁴ within the solid state replica. This rheological effect has a strong impact on the cell windows taking into account mercury intrusion porosimetry (Fig. 8). First, when the oil volumic fraction is the subject of a small increase from 1Si-HIPE0.035 (ρ_o = 0.67) to 2Si-HIPE0.035 (ρ_o = 0.70), both the largest cell junction and smallest cell junction sizes are increased from 1.4 to 1.6 µm and 21 to 45 nm respectively. For those two oil volumic fraction values, the increase in the viscosity is not high enough to overcome the increase in the droplet size and we observe a simple effect of geometry. When the oil volumic fraction is increased-up from 2Si-HIPE0.035 to 3Si-HIPE0.035 (ρ_o = 0.73) and 4Si-HIPE0.035 (ρ_o = 0.78) the largest window cell sizes decrease from 1.4 to 0.5 and 0.25 µm respectively whereas the average smallest cell junction size weakly increases from 25 to 50 nm for 1Si-HIPE0.035 and 2Si-HIPE0.035; this value of 50 nm is constant whatever the increase in the oil volumic fraction (Fig. 8 b, c and d) of the materials, 2Si-HIPE0.035, 3Si-HIPE0.035 and 4Si-HIPE0.035 respectively. The decrease of the largest window cell from 1500 to 400 nm is certainly related to the decrease of the macrocellular cell diameters. Indeed, decreasing the macrocellular cell sizes reduces the characteristic size of the external porosity while weakly increasing the size of the window cell junctions when going from an oil volumic fraction value of 0.67 to 0.70. The small increase of the smallest window cell, when the oil volumic fraction increases from 0.67 to 0.70, might be related to two antagonist effects, namely the small increase of the droplet sizes (Fig. 7 a and c), in this range of oil volumic fraction the viscosity is not yet increasing enough to induce a strong shear effect,^12–14 and the decrease of the Plateau border thickness. Indeed, when the oil fraction increases above 0.70 , the thickness of those Plateau borders decreases, they become thinner and weaker and allow the amount of macrocellular connecting-cells promoted during the shrinkage process to be enhanced.

Fig. 7 SEM visualization of the inorganic monolith-type material macrostructure. a) and b) 1Si-HIPE0.035, c) and d) 2Si-HIPE0.035, e) and f) 3Si-HIPE0.035, g) and h) 4Si-HIPE0.035.

Fig. 8 Pore size distribution for the xSi-HIPE0.035 monolith series as measured by mercury intrusion porosimetry. a) 1Si-HIPE0.035, b) 2Si-HIPE0.035, c) 3Si-HIPE0.035, d) 4Si-HIPE0.035.

Hypothesized condensation mechanism

At this point, it is important to note that the shrinkage of the monoliths is increasing when the starting emulsions’ oil volumic fractions are increasing, for the same TEOS concentration and pH. This effect is in direct relation to the fact that the condensation as previously mentioned certainly starts at the oil-water interface. In fact the oil-water interface might promote silica condensation by minimizing the nucleation enthalpy, acting as a defect. Also, it is well known that the oil-water interface of an emulsion is associated with a higher surfactant concentration than the core of the continuous aqueous phase.¹¹ In our case, this specific region of high surfactant concentration should enhance silica condensation. This consideration associated with the first hypothesis might explain why the silica condensation first starts at the oil-water interface and then extends up to the wall core. If we increase the condensation at the interface by minimizing the oil droplet diameter and so optimizing the droplet number we will increase the number of cell walls per unit of volume, leading both to materials with higher bulk densities (Table 1) and a higher shrinkage effect as, beyond the TEOS concentration and pH conditions, the oil-water interface should promote silica condensation. It is also worthwhile to note that for oil volumic fractions above 0.78, powder materials are obtained rather than monoliths. This feature might be related to the fact that the wall thicknesses decrease and do not have the expected mechanical strength to support the inorganic scaffolds. Also, as the shrinkage increases with increasing oil volumic fraction, the out-going oil flow induced by the shrinkage effect certainly damages the macrocellular walls thus inducing the collapse of the inorganic scaffold. Overall, with the process described above, densities as low as 0.08 g cm⁻³ can be reached, (see Table 2), which is comparable to values obtained for silica aerogel, 0.1 g cm⁻³.¹⁵

Table 2 Nitrogen physisorption data

Materials	1Si-HIPE0.5	1Si-HIPE0.035	2Si-HIPE0.035	3Si-HIPE0.035	4Si-HIPE0.035
BET surface m^2 g^−1	830	820	740	720	670
BJH surface m^2 g^−1 (desorption curve)	64	71	70	62	65

---

Conclusion

In conclusion, we describe a facile route to obtain monolith-type materials associated with both a hierarchically organized porosity and high internal surface area. pH conditions have been shown to be an important factor toward controlling the monolith final textures, also the oil-water interface seems to promote the inorganic condensation process. At the morphosynthesis stage, by increasing the oil volumic fraction of the starting emulsion above the typical value of 0.7, we increase both the emulsion viscosity and the shear strength, so minimizing both the macroscopic cell sizes and the external junction sizes while keeping the window cell diameters constant. For oil fractions above 0.78, powders are obtained rather than monolith-type materials. Beyond the control over the textural porosity we are working on controlling the elastomeric properties of the described materials. Also, this work is currently being extended to other oxides or mixed-oxide materials and will be published in due course.

References

1. C. T. Kresge, M. E. Leonowicz, W. J. Roth, J. C. Vartuli and J. S. Beck, Nature, 1992, 359, 710; J. S. Beck, J. C. Vartuli, W. J. Roth, M. E. Leonowicz, C. T. Kresge, K. D. Schmitt, C. T.-W. Chu, D. H. Olson, E. W. Sheppard, S. B. McCullen, J. B. Higgins and J. L. Schlenker, J. Am. Chem. Soc., 1992, 114, 10834.
2. S. Mann, Nature, 1988, 332, 119; D. D. Archibald and S. Mann, Nature, 1993, 364, 430; P. Feng, X. Bu, G. D. Stucky and D. J. Pine, J. Am. Chem. Soc., 2000, 5, 994; H. Yang, A. Kuperman, N. Coombs, S. Mamiche-Afara and G. A. Ozin, Nature, 1996, 379, 703.
3. A. Imhof and D. J. Pine, Nature, 1997, 389, 948.
4. G.-R. Yi and S. M. Yang, Chem. Mater., 1999, 11, 2322.
5. P. L. Flaugh, S. E. O’Donnel and S. A. Asher, Appl. Spectrosc., 1984, 38, 847.
6. B. P. Binks, Adv. Mater., 2002, 14, 1824.
7. H. Zhang, G. C. Hardy, M. J. Rosseinsky and A. I. Cooper, Adv. Mater., 2003, 15, 78.
8. F. Carn, A. Colin, M.-F. Achard, H. Deleuze and R. Backov, Adv. Mater., 2004, 6, 140.
9. H. Deleuze, R. Faivre and V. Heroguez, Chem. Commun., 2002, 2822.
10. C. J. Brinker and G. W. Scherer, in Sol-Gel Science: the Physics and Chemistry of Sol-Gel Processing, Academic Press, San Diego, 1990.
11. A. Adamson, in Physical Chemistry of Surfaces 5th edn., Wiley-Interscience, 1990.
12. S. A. Bagshaw, E. Prouzet and T. J. Pinnavaia, Science, 1995, 269, 865; E. Prouzet and T. J. Pinnavaia, Angew. Chem., Int. Ed., 1997, 36, 516; C. Boissière, A. Larbot and E. Prouzet, Chem. Mater., 2000, 12, 1937.
13. P. A. Webbs and C. Orr, Surface area and Pores structure by Gas Adsorption, in Analytical Methods in Fine Particles Technology, Micromeritics, Norcross, GA, 1997.
14. M. Mooney, J. Colloid Interface Sci., 1951, 6, 162.
15. T. G. Mason, J. Bibette and D. A. Weitz, J. Colloid Interface Sci., 1996, 179, 439.
16. M.-P. Aronson and M.-F. Petko, J. Colloid Interface Sci., 1993, 159, 134.
17. N. Hüsing and U. Schubert, Angew. Chem., Int. Ed., 1998, 37, 23.

---

Footnote

† Electronic supplementary information (ESI) available: XRD profiles, nitrogen physisorption data and pore size distribution calculated from density functional theory, for the xSi-HIPE0.035 series. See http://www.rsc.org/suppdata/jm/b4/b400984c/

---