1 Introduction

Ordered mesoporous materials represented by mesoporous silicas [1, 2] have attracted much attention because of their potential applications such as catalysts, adsorbents, separation media, and sensors. A number of studies have been carried out to establish synthetic methodologies to prepare novel mesoporous solids [3]. Template method is the most successful approach to prepare ordered nanoporous materials. The term “template” often indicates “soft template” such as surfactant micelles constructing ordered mesostructures through cooperative self-assembly of the micelles and inorganic precursors. Although it is advantageous that a variety of the mesoporous structures can be formed with a number of combinations of surfactants, inorganic precursors, and growth conditions, the resultant structures are strongly affected by the experimental conditions and the nature of chemical species in a complicated manner. It is often difficult to prepare mesoporous structures of desired compositions as expected. On the other hand, an ordered mesoporous structure is inversely replicated by using a mesoporous solid as a “hard template”, where the precursors are simply introduced in the template and solidified. The inverse replica of mesoporous silica as the hard template was first prepared by Ryoo et al. in 2000 [4]. Sucrose was infiltrated into a SBA-15 template to obtain an inverse carbon replica CMK-3 with an ordered 2D hexagonal structure of SBA-15 after chemically dissolving the SBA-15 template. Many ordered mesoporous carbon materials were prepared in similar approaches since then [5–11]. Extension of the hard template approaches to other components was successfully applied to ordered mesoporous materials composed of metals [10, 12–17], metal oxides [18–25], non-oxide ceramic components [26–33], and organic polymers [34]. The hard template approach is thus becoming attractive to prepare various mesoporous structures with a variety of compositions as summarized in recent reviews [35–39]. Differently from the soft template approach, the hard template approach should enable reciprocating replication retaining the nanopore ordering with interconversion between different compositions via a negative replica. Such interconversion needs adequate selection of the negative replica media and replication method. As carbon and silica offer a variety of different nanopore structures as representative nanoporous materials, they are predominantly selected as the primary sources of nanoporous hard templates. They are also suitable as the negative replica media, because they can be readily eliminated in different manners. Nanoporous carbon, which is eliminated on calcination, can be used as the template of nanoporous metal oxides, and nanoporous silica, which can be dissolved in a HF or NaOH solution, can be used as the template of nanoporous noble metals and partly metal oxides. Appropriate selection in the replication method should lead to improved precision in the interconversion. Especially, the negative replication method between carbon and silica should be important as a model study in the reciprocating replication.

Several attempts have been carried out to prepare mesoporous materials with carbon templates [40–46]. Firstly, two research groups reported liquid phase infiltration to obtain inverse silica replicas of CMK-3 by Schüth et al. and Kim et al. Schüth et al. prepared the inverse silica replica by immersing the template in solutions containing tetraethyl orthosilicate (TEOS) [40, 41]. Kim et al. used sodium metasilicate as the silica source [42, 43]. These methods are advantageous in simplicity of the procedure, whereas multiple infiltration steps are generally required to load sufficient amount of the silica sources. As another infiltration approach of chemical species in a mesoporous carbon, novel infiltration method of silica species via vapor phase was reported by Parmentier et al. [44, 45]. They prepared an inverse replica of CMK-3 by introducing TEOS vapor in the carbon template under a reduced pressure at 700 °C followed by subsequent pyrolysis of TEOS at 950 °C in vacuum and the combustion of the carbon at 700 °C in air. Compared with the liquid phase infiltration, larger amount of silica was loaded at the single deposition step. One of the disadvantages of this CVI technique is use of special vacuum-heating apparatus with high temperature heat treatment. Another problem is considerably large structural shrinkage in a series of heat treatment at high temperatures.

In this work, we report a simple infiltration procedure leading to well-ordered mesoporous silica replicas in a single step. Although the technique is based on vapor phase chemical transport of TEOS, the preparation conditions are milder and simpler compared with the reported CVI technique [44, 45] since no pyrolysis of TEOS is necessary. The reaction is carried out in an autoclave in the presence of small amount of water. The multi-step replication process was monitored by characterizing the replicated materials as well as intermediate composites.

2 Experimental

Starting from mesoporous silica SBA-15, an SBA-15/carbon composite and mesoporous carbon CMK-3 were prepared basically according to the reported procedures [2, 4]. A silica source was infiltrated via vapor phase in CMK-3 to form a CMK-3/silica composite. Finally, the carbon template was calcined by heating the composite in air to obtain a replica of the starting SBA-15 (inverse-replica of CMK-3). In this paper, the obtained SBA-15/carbon composite, CMK-3/silica composite, and a replica of the starting SBA-15 are denoted as composite-1, composite-2, and MSR-1, respectively.

2.1 Materials

A triblock copolymer surfactant Pluronic P123 (EO20PO70EO20, Mav = 5800, BASF) was purchased from Aldrich. Tetraethyl orthosilicate TEOS (95%) was obtained from Kanto Chemicals. Hydrochloric acid and furfuryl alcohol were purchased from Wako chemicals. Sulfuric acid was obtained from Nakarai. Purified water with Millipore Milli-Q Jr was used for the preparations.

2.2 Preparation of CMK-3

Mesoporous silica SBA-15 was prepared according to the procedure described in [2]. P123 and TEOS were used as the surfactant and silica source, respectively. Mesoporous carbon CMK-3 was prepared by a method based on [5], where furfuryl alcohol was used as the carbon source instead of sucrose.

2.3 Preparation of mesoporous silica replica

In a Teflon-lined autoclave, 0.1 g of CMK-3 was put on a stainless-steel dish separately from a mixture of TEOS and H2O (1.0 g TEOS :0.18 g H2O, H2O/TEOS = 2.0) at the bottom. The autoclave was heated under an autogenous pressure at a variable temperature x °C (x = 100–220, mainly x = 180) for 2 days and TEOS was introduced in the carbon template via vapor phase transport (VPT). The CMK-3/silica composite obtained at x °C is denoted as composite-2−x. The composites were calcined at 550 °C in air to remove the carbon template. The mesoporous silica replica obtained from composite-2−x is denoted as MSR-1−x.

2.4 Characterization

Small angle X-ray diffraction (SAXRD) experiments were carried out in the range of 0.5–3° using a Rigaku Rad-C diffractometer with an attachment for the small angle measurement. Thermogravimetry-differential thermal analyses (TG-DTA) were performed with a Rigaku Thermo plus2 analyzer at a heating rate of 10 °C/min under a flowing air at a rate of 100 sccm. Scanning electronic microscopy (SEM) images were obtained with Shimadzu SUPERSCAN at acceleration voltage of 10 kV. Nitrogen adsorption isotherms were measured at 77 K with Bel-Japan BELSORP 28A. The samples for the adsorption measurements were degassed at 400 °C for 6 h prior to the measurements. Pore size distributions were calculated using the Dollimore-Heal (DH) method from the adsorption branch of the isotherms [47].

3 Results and discussion

In the autoclaves just after the VPT experiments, black powder was always found on the stainless steel net in the Teflon liner. There was also small amount of liquid at the bottom of the Teflon liner. The 1H NMR of the remaining liquid confirmed the existence of ethanol and an absence of unreacted TEOS in the liquid. Adhesion of a small amount of white powder, which should be silica, was observed both on the top and bottom of the autoclave. The amount of the recovered silica powder was ca. 25 wt% of the quantity theoretically anticipated from that of the starting TEOS. Therefore, it was reasonable to assume that the rest of silica source was included in the black powder as a CMK-3/silica composite. These results indicate that the hydrolysis/condensation reaction of TEOS considerably proceeds after 48 h in spite of no addition of catalyst and the reaction took place preferentially in the carbon template.

Nitrogen adsorption isotherms of CMK-3 and composite-2-180 are shown in Fig. 1. Table 1 shows the pore properties calculated from the isotherms. The isotherm of CMK-3 is of type IV with a hysteresis loop around a relative pressure of 0.4, which is attributed to capillary condensation in the ordered mesopore. The isotherm of composite-2-180 indicates type I with little adsorption capacity. It means that the mesopore of CMK-3 was almost filled leaving little volume of micropore. Thus, it is confirmed that dense silica loading in CMK-3 mesopore is achieved.

Fig. 1
figure 1

The nitrogen adsorption–desorption isotherms of the CMK-3 and Composite-2-180

Full size image
Table 1 Pore properties of CMK-3 and Composite-2-180
Full size table

Figure 2 shows the TG curves for composite-1, CMK-3, and composite-2-180. CMK-3 showed a steep weight loss of nearly -100% between 450 and 660 °C, which is attributed to the combustion of the carbon, whereas the weight losses on the carbon combustion were ca. -40% for composite-1 and composite-2-180. The residues at 1000 °C were white powder, confirming that silica component was actually included in composite-2-180 as well as composite-1. In composite-2-180, a weight loss starts from 300 until 450 °C, which is probably due to the combustion of organic components included in the silica such as ethoxy groups. Actually, an exothermic peak is observed in the DTA curve of composite-2-180 around 420 °C. The carbon content in composite-2-180 was estimated to be 34.6 wt% according to TG. Since the residue is silica loaded in the carbon template, the SiO2/C mass ratio was 1.55 (Table 2). This value is comparable to that of the starting composite-1, which was evaluated to be 1.72.

Fig. 2
figure 2

TG-DTA curves of a composite-1, b CMK-3, and c composite-2-180

Full size image
Table 2 Estimated compositions of the composite-1 and composite-2−x
Full size table

Figure 3 shows SAXRD patterns of SBA-15, CMK-3, and MSR-1-180. All the patterns exhibit three distinctive diffraction peaks assigned to the 100, 110, and 200 reflections of a 2D hexagonal lattice (p6mm). The calculated lattice constants were 12.1, 10.0, and 9.8 nm from d100 for SBA-15, CMK-3, and MSR-1-180, respectively. The ratios of lattice shrinkage were 17% for CMK-3 from SBA-15 and 2% for MSR-1-180 from CMK-3, respectively. It should be noted that the lattice contraction of MSR-1-180 from CMK-3 was much smaller than that in the reported CVI method, which was estimated to be ca. 22% according to the XRD patterns in the literature [13]. This is probably due to milder heat treatment condition and denser silica loading in this method, contributing in sustention of the inverse-replica structure. The CMK-3/silica composite obtained by CVI reportedly contained 30 wt% of silica [13], while our composite contained more silica content, 53.7 wt% (Fig. 1).

Fig. 3
figure 3

SAXRD patterns for the SBA-15, CMK-3, and MSR-1-180

Full size image

Figure 4 shows nitrogen adsorption isotherms and the corresponding pore size distribution of SBA-15 and MSR-1-180. Table 3 shows the pore properties calculated from the isotherms. MSR-1-180 has lower specific surface area and pore volume than the parent SBA-15. The isotherm of MSR-1-180 is of type IV with a hysteresis loop around a relative pressure of 0.6. The pore-size distribution is narrow and is centered at 4.9 nm, which is smaller than 7.1 nm of SBA-15. The difference is mainly attributed to the structural shrinkage of CMK-3 during the replication of SBA-15 as shown in Fig. 3.

Fig. 4
figure 4

Nitrogen adsorption–desorption isotherms and the corresponding pore size distribution for the SBA-15 and MSR-1-180

Full size image
Table 3 Pore properties of SBA-15 and MSR-1−x
Full size table

The SEM images of SBA-15, CMK-3, and MSR-1-180 are shown in Fig. 5a, b, and c, respectively. All of them exhibit bead-like morphology, and neither apparent inter-grain deposit nor particles of other morphologies are observed even after the replication. It is concluded that macroscopical morphology is retained as well as the mesostructures.

Fig. 5
figure 5

The SEM images for the a SBA-15, b CMK-3, and c MSR-1-180

Full size image

Controlling the loading amount of silica in CMK-3 was considered to be a key factor in this technique. Therefore, other MSR samples were prepared by infiltrating at varied temperatures between 100 and 220 °C to examine influence of infiltration temperature on loading amount of silica. The SiO2/C mass ratio of MSR series was 0.95 for composite-2-100, 1.40 for composite-2-140, 1.55 for composite-2-180 and 1.48 for composite-2-220 as shown in Table 2. The silica content increases with temperature up to 180 °C.

Figure 6 shows the SAXRD patterns of the SBA-15, MSR-1-100, MSR-1-140, MSR-1-180, and MSR-1-220. All the patterns are indexed with 2D hexagonal lattice as the parent SBA-15. Although MSR-1-100 shows the lowest SiO2/C mass ratio of 0.95, the 2D hexagonal lattice is retained after the removal of carbon template. There is correlation between SiO2/C mass ratio and the lattice contraction. The contraction becomes larger as the SiO2/C mass ratio decreases (Table 2). MSR-1-100 with the lowest silica loading shows the largest lattice contraction by 7%. Since it was reported that the CMK-3/silica composites prepared in liquid phase did not preserve the structural regularity in [10] even with a similar or higher SiO2/C mass ratio compared to MSR-1-100, it proves that our approach enables effective silica loading.

Fig. 6
figure 6

The SAXRD patterns for the SBA-15, MSR-1-100, MSR-1-140, MSR-1-180, and MSR-1-220

Full size image

The nitrogen adsorption–desorption isotherms of MSR-1-100 compared with MSR-1-180 and the corresponding pore size distribution curves are shown in Fig. 7. Adsorption properties of all the MSR series samples are listed in Table 2. MSR-1-140 and MSR-1-220 shows similar properties to MSR-1-180, although they have lower specific surface areas and pore volumes than the parent SBA-15 as discussed above. MSR-1-100 has a larger pore volume with a broader bimodal pore size distribution than MSR-1-180, since MSR-1-100 has low density pore wall caused by incomplete infiltration in part of CMK-3 pores.

Fig. 7
figure 7

The nitrogen adsorption–desorption isotherms and the corresponding pore size distribution of the MSR-1-100 and MSR-1-180

Full size image

The MSR series samples show basically similar structural properties except MSR-1-100. Small dependence of the silica loading amount on the reaction temperature is also an advantage of this technique. The hydrolysis/condensation reaction can take place anywhere in liquid and gas phase in an autoclave or in carbon mesopore competitively in this technique. In our reaction conditions, it is considered that not all of TEOS exist in vapor phase and roughly only one-third of TEOS is estimated to exist in vapor phase at the boiling point when 1 g of TEOS was added to a Teflon-lined autoclave of 54 dm3 in volume as in a typical condition. The rest exists in the liquid phase at the bottom. The residual silica without the vaporization on the bottom was 25 wt% of input TEOS after VPT of TEOS at 180 °C. Accordingly we assumed that the vaporized TEOS molecules are transferred in vapor phase and preferentially adsorbed on CMK-3.

In order to confirm preferential adsorption in CMK-3, preparation of mesoporous silica replica composite-2 (cf. Sect. 2.3) was carried out without adding water. The obtained sample is denoted composite-2-NW. Figure 8 shows the TG/DTA curves of composite-2-NW compared with composite-2-180. The TGA curve of composite-2-NW shows a weight loss of 61% between 180 and 660 °C. Since white powder remained after heating until 1000 °C, it should be silica. The TG curve starts to decline at 180 °C much lower than that of composite-2-180 because adsorbed TEOS on CMK-3 vaporizes at the boiling point (168.5 °C). The amount of the vaporized TEOS is roughly estimated to be 13% of composite-2-NW at maximum. Higher exothermic peak for DTA curve of composite-2-NW compared with that of composite-2-180 is observed at around 400 and 500 °C, which is attributed to combustion of organic components such as ethoxy groups as mentioned above. Therefore, it is apparent that TEOS selectively adsorbs on CMK-3 even without water vapor. Without addition of water, the hydrolysis/condensation reaction of TEOS on heating in the autoclave poorly proceeds with small amount of water vapor, which is probably supplied from air or adsorbed on CMK-3, leading to non-vaporized silica species. The loading amount of silica in CMK-3 for composite-2-NW is somewhat smaller than composite-2-180. The SAXRD pattern for MSR sample obtained by calcining composite-2-NW at 550 °C exhibits a broad single reflection around 1.2° (Fig. 9). This clearly shows that the replication is possible in spite of lower structural ordering because the loading amount of silica is small and silica network is insufficiently formed without addition of water.

Fig. 8
figure 8

TG/DTA of composite-2-NW and composite-2-180

Full size image
Fig. 9
figure 9

SAXRD pattern for the composite-2-NW

Full size image

Also under the condition of coexistence of TEOS and water, as soon as preferential adsorption of TEOS vapor on CMK-3 occurs, partial pressure of TEOS in the autoclave should decrease. TEOS vapor is supplied from the liquid phase at the bottom of the autoclave to maintain the vapor pressure. Probably hydrolysis/condensation is accelerated within the nanopore of CMK-3, leading to formation of vacant space after desorption of ethanol. This available nanospace is ready for re-transportation of TEOS from the liquid phase, which continues until the nanopores of CMK-3 are sufficiently filled with silica. During this vapor phase transportation process, the competitive hydrolysis/condensation of TEOS should occur especially in CMK-3 and liquid phase at the bottom of the autoclave. Since vaporization and transportation of TEOS is slow when the temperature is low, most of TEOS is partially hydrolyzed in the liquid phase at the bottom before being transported to CMK-3. In higher temperature, TEOS vapor transported to CMK-3 increases because of higher vapor pressure, resulted in formation of composite-2 with higher loading amount of silica.

4 Conclusion

Ordered mesoporous silica replicas were obtained using CMK-3 as the hard template. The silica source was introduced by VPT at 180 °C for 2 days. The TG analyses of the CMK-3/silica composite showed that sufficient amount of silica was loaded in the mesopore of CMK-3. The inverse silica replica obtained by removal of CMK-3 has structural regularity of SBA-15 as confirmed by SAXRD, N2 adsorption, and SEM. Although the competitive hydrolysis/condensation of TEOS should occur in the nanopore of CMK-3, in the liquid phase at the bottom of the Teflon vessel, and on the Teflon wall during the vapor phase transportation process, it proved that the reaction took place mainly in CMK-3 to form the composite. The loading amount of silica in the MSR series gradually increased with temperature from 100 °C to 180 °C and 53.7% of silica at maximum was introduced in CMK-3 at 180 °C. This simple infiltration technique should be applicable to other types of nanoporous carbons with considerable advantages compared with the conventional liquid phase infiltration or CVI techniques.