Preparation of Novel Mesoporous Silica Using a Self-Assembled Graphene Oxide Template

Novel mesoporous silicas rolled with silica sheets with 2D regular spacing were prepared using a self-assembled graphene oxide (GO) template formed by mixing GO with Pluronic123 (P123). Self-assembled GO templated mesoporous silicas (SGT-PMS) showed well-developed X-ray diffraction peaks with d-spacings of 9.8–10.8 nm depending on the amount of GO, indicating mesoporous structures. The specific surface areas increased from 603.8 to 861.2 m2g−1 on adding GO. The pore size distribution was in the range 5.1–5.8 nm and pore volume in the range 0.80–0.99 m3g−1. The SEM images of SGT-PMS showed irregular elliptical particles with various sizes. TEM images showed that the cross section of SGT-PMS particles comprises a roll of silica sheets with 2D regular spacing. The pore walls of SGT-PMS are firmer and thicker than those for PMS without GO as indicated by the corresponding intensities of Q3 and Q4 signals. These results were explained well by the self-assembled GO templating mechanism.

In this study, novel PMS rolled with 2D silica sheets are successfully synthesized using a self-assembled GO template (SGT). When GO sheets are added to a Pluronic123 (P123) solution, hydrophilic groups of P123 molecules are adsorbed on the surface of GO sheets by interacting with π-electrons of GO sheet, resulting in more hydrophobic GO sheets. Such hydrophobic GO sheets form self-assembled GO aggregates, which form a roll stacked with GO sheets due to the neutral interaction between surface alkyl groups. Tetraethyl orthosilicate (TEOS) is easily introduced into the hydrophobic galleries between GO sheets thereby composing self-assembled GO aggregates. The gelation of TEOS in the gallery produces rolls sandwiched with silica sheets between GO sheets. Their calcination results in a roll of two-dimensional silica sheets with very regular spacing.

Materials and methods
Preparation of GO. GO was prepared according to the modified Hummer's method 29,43,44 . Briefly, in a 2-L three-necked round-bottom flask, commercial graphite powder (Aldrich, 5 g) and NaNO 3 (3.75 g) were added to concentrated H 2 SO 4 (350 mL). This mixture was stirred in an ice-water bath, and 20 g of KMnO 4 was slowly added over 1 h, followed by continuous stirring for 2 h in the ice-water bath. After the mixture was stirred vigorously for 2 days at room temperature, 700 mL of a 5 wt.% H 2 SO 4 aqueous solution was added over 1 h with stirring at a constant temperature of 98 °C. The resultant mixture was further stirred for 2 h at 98 °C. After the temperature was reduced to 60 °C, 15 mL of H 2 O 2 (30 wt.% aqueous solution) was added, and the mixture was stirred for 2 h at room temperature. To remove extraneous oxidation products and other inorganic impurities, the resultant mixture was purified by repeating the following procedure 20 times: centrifugation, removal of the supernatant liquid, dispersion of the solid using vigorous stirring, and ultrasonication for 1 h at a power of 150 W. The resultant solid was recovered by centrifugation, washed with deionized water and ethanol (EtOH) until H + was removed, and then dried in air at 40 °C.

Synthesis of SGT-PMS-x.
The weight ratios of the reactant solutions are listed in Table 1. Briefly, 1.0 g of P123 (M av = 5800, Aldrich) was dissolved in a mixture of 30 mL of 2-N HCl and 8.0 mL of distilled water at 38 °C. Then a certain amount of as-prepared GO (weighing 0, 1.5, 2.2 or 3.0 mg) was added and homogeneously dispersed under sonication. For each solution, 2.15 g of TEOS (Aldrich) was added with vigorous stirring for 5 min, resulting in gel type precipitation. Here, SGT was also prepared by homogeneously dispersing under sonication after adding 3.0 mg of as-prepared GO to 30 ml of above P123 solution without TEOS. The resultant solutions were aged for 5 h at 80 °C. As-synthesized solid samples were recovered by filtration and washed with deionized water, and then dried for 3 h at 80 °C. The dried samples were labelled as SGT and as-synthesized SGT-PMS-x (x = 1.5, 2.2, 3.0 mg) and x represented the different weights (mg) of GO added. Finally, dried powders of as-synthesized SGT-PMS-x were calcined in a furnace for 5 h at 600 °C in air, which removed the P123 and GO templates. The final samples were also labelled as SGT-PMS-x (x = 1.5, 2.2, 3.0) and x represented different weights (mg) of GO added. For comparison, the ordered mesoporous silica prepared without GO was labelled as PMS-0.
Characterization. The powder X-ray diffraction (XRD) patterns of the samples were recorded on a Bruker D8-Advance X-ray powder diffractometer using Cu Kα radiation (λ = 0.1542 nm) with scattering angles (2θ) of 1°-10°, operating at 40 keV, with a cathode current of 20 mA. Fourier transform infrared (FT-IR) were recorded using Attenuated Total Reflectance (ATR) sampling accessory on the Nicolet iS50 FT-IR spectrometer. X-ray photoelectron spectra (XPS) were recorded using a VersaProbe, Ulvac-PHI, with Al Ka excitation radiation (hv = 1486.6 eV). The pressure in the analyser was maintained at approximately 6.7 × 10 −7 Pa. XPS data was processed using a DS 300 data system.
Scanning electron micrographs were obtained using a JEOL JSM-840A scanning electron microscope (SEM). The transmission electron micrographs (TEM) were obtained with a JEOL JEM-200 CX transmission electron microscope operated at 200 kV, using a thin-section technique. The powder samples were embedded in epoxy resin and then sectioned with ultra-microtome. Atomic force microscopy (AFM) images were obtained using an AutoProbe CP/MT scanning probe microscope (XE-100(PSIA)). Imaging was performed in non-contact mode using a V-shaped Ultralever probe B (Park Scientific Instruments, B-doped Si with a frequency ƒ c = 78.6 kHz, spring constant k = 2.0-3.8 N m −1 , and nominal tip radius = 10 nm). All images were collected under ambient conditions at 50% relative humidity and 23 °C with a scanning raster rate of 1 Hz. Samples for AFM were prepared by depositing dispersions of GO in EtOH on a freshly cleaved mica surface (Ted Pella Inc. Prod No. 50) and allowing them to air-dry.
Solid-state 29 Si MAS NMR spectra was recorded on a Bruker CXP-100 spectrometer at a resonance frequency of 19.89 MHz with a 45° pulse and a recycle delay of 7 s. Raman spectra were obtained using a Jobin Yvon/ Horiba LabRAM spectrometer equipped with an integral microscope (Olympus BX 41). A 514.5-nm Ar-laser was used as an excitation source. Samples were sonicated in EtOH and three drops were placed on a glass slide www.nature.com/scientificreports www.nature.com/scientificreports/ for observation. The samples were viewed using green and red laser apparatuses with maximum magnifications of 50× and 100×, respectively. N 2 adsorption isotherms were obtained at 77 K using a nitrogen sorption instrument (Micromeritics ASAP 2020). Pore size distributions were calculated by the Barrett-Joyner-Halenda (BJH) method using the adsorption branches of the isotherms.  Fig. 1(b,c) show AFM images indicating GO particles with a thickness of ~1 nm and length of ~0.5 μm, corresponding with previous reports 45,46 . However, the GO prepared in this work comprises single or multiple sheets, because AFM images can be discriminated well for the single GO sheets. Figure 2 shows FT-IR peaks for the P123, GO and SGT. P123 peaks [ Fig. 2(a)] arises at 800~1200 cm −1 (-C-C-), 2850~2900 cm −1 (-C-H) and 3200~3600 cm −1 (-O-H) and GO peaks [ Fig. 2(b)] at 800~1200 cm −1 (-C-C-), 1750 cm −1 (-C=C-). Whereas, the characteristic peaks of P123 and GO largely decrease or nearly disappear in SGT peaks [ Fig. 2(c)], indicating that P123 interacts with GO. When GO sheets are added to a Pluronic123 (P123) solution, hydrophilic groups of P123 molecules are adsorbed on the surface of GO sheets by interacting with π-electrons of GO sheet, resulting in more hydrophobic GO sheets. Such hydrophobic GO sheets form self-assembled GO aggregates, which form a roll stacked with GO sheets owing to the neutral interaction between the surface alkyl groups. In particular, the absence of -OH peak in SGT indicates that OH groups of P123 interact with π-electrons of GO sheet.   Fig. 4(a-d), the removal of the GO and P123 templates by calcination resulted in the decrease of d-spacing; d-spacings of PMS-0, SGT-PMS-1.5, SGT-PMS-2.2, and SGT-PMS-3.0 were now 9.8, 9.4, 10.2, and 10.8 nm, respectively. This is attributed to the contraction of gallery owing to the removal of template. Figure 5 shows XPS spectra for the as-synthesized SGT-PMS-3.0 and SGT-PMS-3.0. The absence of carbon peak indicates that organic templates such as GO and P123 are removed completely by the calcination. Figure 6 shows SEM images of PMS-0 and SGT-PMS-x(x = 1.5, 2.2, 3.0). PMS-0 shows a regular particle spherical morphology [ Fig. 6(a)]. However, SGT-PMS-x samples have particle morphologies different from that of PMS-0, exhibiting irregular and elliptical large particles with various sizes and shapes [ Fig. 6(b-d)]. Here, small spherical particles [ Fig. 6(b)], which are caused by the PMS-0 phase, disappeared on increasing the amount of GO added [ Fig. 6(d)]. This is attributed to the decrease in the P123 micelles owing to the adsorption of P123 molecules on the GO surface in accordance with the increase in the amount of added GO.

Results and discussion
The solid-state 29 Si MAS-NMR spectra of PMS-0 and SGT-PMS-3.0, as shown in Fig. 7(a,b), indicate that the pore wall structure of SGT-PMS-3.0 is fairly different from that of PMS-0. Here, SGT-PMS-3.0 is shown as  www.nature.com/scientificreports www.nature.com/scientificreports/ a representative of SGT-PMS-x. In general, Si atoms of the silicate network exhibit signals (Q 2 , Q 3 , and Q 4 ) in the range of ~ 90 to 115 ppm. Q 2 (HO) 2 Si(OSi) 2 (near 90 ppm), Q 3 HOSi(OSi) 3 (near 100 ppm) and Q 4 Si(OSi) 4 (110 to 1154 ppm) signals were observed due to the diverse environments of silicon. In particular, the intensity of Q 3 and Q 4 signals in the SGT-PMS-3.0 is more than double that of those for PMS-0. As the Q 3 and Q 4 signals are highly reflective of the degree of the silicate network development, this indicates that the pore walls in SGT-PMS-3.0 are well developed and firmer and thicker than those in PMS-0. Figure 8 shows the Raman spectra of as-synthesized PMS-0 and SGT-PMS-x samples. Raman spectroscopy is employed to characterize the bonding, ordering, and crystallite size in carbon materials. A Raman band arises at ~1582 cm −1 (the G-band) from the in-plane phonon modes of graphene, indicating sp 2 bonding. The D-band at ~1353 cm −1 corresponds to disorder in the graphene layers caused by the presence of sp 3 bonding 47 . The peaks of GO observed at ~1350 and ~1590 cm −1 correspond to the D and G bands, respectively 48 . However, in SGT-PMS-x, the D band occurs at 1315 cm −1 , which is attributed to the influence of P123 adsorbed on GO surface. However, the G and D bands disappeared after calcination (Fig. 9), indicating that the GO and P123  www.nature.com/scientificreports www.nature.com/scientificreports/ templates were removed during calcination. XPS peaks (Fig. 5) show that the GO and P123 templates are removed during calcination. Figure 10 shows the N 2 adsorption isotherms of PMS-0 and SGT-PMS-x samples. A typical steep increase with mesopore filling at a relative pressure of 0.6-0.8 is observed. The specific surface areas obtained using the Brunauer-Emmett-Teller (BET) equation are listed in Table 2, with considerably different values of 603.8, 713.6, 861.2, and 840.8 m 2 g −1 for PMS-0, SGT-PMS-1.5, SGT-PMS-2.2, and SGT-PMS-3.0, respectively. The relatively high specific surface areas of PMS-0 and SGT-PMS-x samples indicate that PMS-0 and SGT-PMS-x are silica microparticles with microporous and mesoporous structures. In particular, the increase in surface area with the addition of GO is attributed to the increase in micropores due to the development of a pore wall. In porous materials formed using several templates, specific surface area consists of micropores developed due to the pore walls and mesopores formed by burn-off of templates. Figure 11 shows the BJH pore-size distribution of PMS-0 and SGT-PMS-x samples; pore size of PMS-0, SGT-PMS-1.5, SGT-PMS-2.2 and SGT-PMS-3.0 is 5.8, 5.6, 5.3, and 5.1 nm, respectively (Table 2). This is against the trend of d-spacing. In general, an increase in d-spacing results in an increase of pore size. However, if silicate sheets are thickened further by gelation using SGT, gallery height can be reduced relatively (because the gallery height = d-spacing -thickness of silica sheet), resulting in the decrease  www.nature.com/scientificreports www.nature.com/scientificreports/ in pore size accompanied by the increase in surface area. This was already explained by the increase of the intensities of Q 3 and Q 4 peaks with the addition of GO. The increase of silica sheet thickness and the decrease of gallery height can bring out the increase of surface area, allowing the development of micropores. Therefore, the increase   www.nature.com/scientificreports www.nature.com/scientificreports/ in surface area accompanied by the decrease of pore size are attributed to the increase in silica sheet thickness by the formation of SGT-PMS.
TEM images, as shown in Fig. 12, show the cross-section structures of SGT-PMS particles which cannot be confirmed by XRD peaks, SEM images, and 29 Si MAS-NMR spectra. These prove that large irregular particles, as shown in SEM, are rolled in silicate sheets stacked with regular spacing. The rolls with a layered structure are also accompanied by spherical particles with nanosized hexagonal pores. Here, the nanosized hexagonal pores originated from PMS-0, whereas the rolls with layered structure originated from SGT-PMS. These results indicate that SGT-PMS is a novel mesoporous silica phase formed by SGT.  www.nature.com/scientificreports www.nature.com/scientificreports/ The schematic in Fig. 13 shows the SGT mechanism as a representation of the effect of the addition of GO on the physical properties such as particle morphology, regular pore structure, and surface area of SGT-PMS-x. When GO sheets are added to the P123 solution, hydrophilic head groups of P123 molecules are adsorbed on the surface of GO sheets by the molecular interaction between OH groups of P123 and π-electrons of GO sheet, resulting in hydrophobic GO sheets surrounded by alkyl chains. Such hydrophobic GO sheets form SGT like a roll stacked with GO sheets owing to neutral interaction between the alkyl chains. TEOS can be introduced into the hydrophobic galleries between GO sheets of SGT. The gelation of TEOS in the gallery of SGT results in two-dimensional thick silica sheets. The removal of the GO and P123 templates from as-synthesized samples leave only rolls comprising stacked silica sheets with regular spacing. Here, when GO is added in a small amount, SGT are accompanied by P123 micelle templates, resulting in a mixture of PMS-0 and SGT-PMS phases. The increase in the added amount of GO results in the increase in SGT-PMS-x particle size, thus leading to an increase in SGT size. Here, the galleries between silicate sheets are very regular because they are made of SGT supported by wide hydrophobic GO sheets. In future studies, we expect that mesoporous materials with various metal oxides can be produced by self-assembled GO templated gelation.

conclusion
Novel mesoporous silicas were prepared successfully by the gelation of TEOS using a self-assembled GO template. SGT-PMS-x showed well-developed XRD peaks with d-spacings of 9.8-10.8 nm depending upon the amount of GO, thereby indicating mesoporous structures. The specific surface areas were increased from 603.8 to 861.2 m 2 g −1 , depending upon the amount of GO. The pore size distribution remained in the range 5.1-5.8 nm and the pore volume was in the range 0.80-0.99 m 3 g −1 . However, the pore size distribution remained in the range 5.1-5.8 nm. In particular, SEM images of SGT-PMS-x showed irregular and elliptical large particles with various sizes, which are fairly different from that of PMS-0. TEM images indicate that SGT-PMS particles comprises silicate sheets with regular spacing, which are fairly different from that of PMS-0 without GO. These results were well explained by the self-assembled GO templating mechanism. The above results indicate that mesoporous materials with various metal oxides can be produced by using self-assembled GO templates.