Segregation of confined ionic liquids inducing the formation of super-micropores in the silica matrix

Abstract

In the present work, we prepared a super-microporous silica gel via a nonhydrolytic methanol–formic-acid solvolytic sol–gel process and using a hydrophobic short-chain ionic liquid (IL), 1-butyl-3-methylimidazolium hexafluorophosphate, as the template solvent and tetraethyl orthosilicate (TEOS) as the precursor. Low-temperature solvent extraction (65–70 °C) and subsequent vacuum drying were adopted to remove the confined IL. Transmission electron microscopy (TEM), nitrogen sorption (BET), and small/wide-angle X-ray scattering (S/WAXS) experiments were carried out to explore the structures of the prepared porous silica gels and the corresponding formation mechanism. The results from BET and TEM revealed that the synthesized silica gels with higher IL/TEOS molar ratio of ≥2.0 formed fractal matrix with embedded disordered worm-like super-micropore (pore size of 1.1–1.2 nm and surface areas of 690–800 m² g⁻¹). Further, S/WAXS analysis revealed the formation of super-micropore in the silica matrix strongly depended on the formation of highly-branched fractal silica framework (mass-fractal dimension of silica matrix D_f > 2.8) and the segregation of the confined IL, which was basically driven by the strong IL/pore wall repulsive interactions when excess amount of IL was used here.

---

1 Introduction

Super-microporous materials with pore sizes between 1 and 2 nm, which are intermediate between the pore size gap of microporous zeolites (pore size < 1 nm) and mesoporous materials (such as porous SiO₂, TiO₂, cellulose, etc. and with pore size > 2 nm), have attracted considerable concern because of various applications such as size- and shape-selective catalysts, active fillers of nanofiltration membrane allowing water and monovalent salts to pass, low-k dielectric materials, low-refractive materials, and thermal insulators, etc.^1–10 Porous materials with the micropore size and uniform pore distribution are also confirmed to enhance the mechanical strength and prevent the contamination of undesired chemical species deposited on them.^3,4,8–10 Up to now, the technologies developed for preparing such super-microporous materials are concentrated at sol–gel routes associated with the use of various structure-directing agents as templates such as surfactants, block-copolymers, bolaamphiphiles, and long-chain ionic liquids.^1–7 Utilizing structure-directing agents as templates aims to prepare ordered porous materials with uniform pore distribution, whereas high-temperature calcination (>400 °C) is unavoidably selected to remove template.^1–7 Generally, the calcination would cause the collapse of silica framework, thereby resulting in the formation of denser silica (pore size < 1 nm).^8–10 For low-refractive materials, such a denser silica would induce the promotion of density and refractive index, consequently especially undesirable.^8,9

The preparation of porous materials recently tends to use ionic liquids (ILs) as template solvents mainly considering that ILs have been set as the green and recyclable solvents and exhibited diverse properties such as negligible vapour pressure at room temperature, excellent thermal stability, high ionic conductivity, wide electrochemical windows, and the behaviours of both lyotropic and thermotropic liquid crystals.^11–20 ILs are defined as salts with melting temperature below 100 °C, which typically contain bulky cations, small anions, and asymmetric ionic pairs in their chemical structures, thus having ability to be self-dissociated in the absence of solvents.^21,22 Bulk ILs would segregate nonpolar alkyl tails into charged matrix, leading to the formation of nanoscale domains. Such nanodomains will make unique nanoscale structural heterogeneities, thereby resulting in a scattering peak observable in the small/wide-angle X-ray scattering (S/WAXS) or neutron scattering profiles.²³ Previous results also presented that the formation of structural heterogeneity was probably derived by the electrostatic interactions between the charged groups and the van der Waal forces between the alkyl chains. The nanodomain size depends on the length of alkyl tail, symmetry in cation structure, and temperature.^24,25 The presence of nanodomain would also make the ILs to be viscous fluids compared to the traditional organic solvents.²⁶

Thus far, the literatures^11–20 regarding the preparation of porous materials with ILs as template solvents are mostly concentrated on the sol–gel synthesis of “mesoporous” materials since Dai et al.¹¹ first prepared a silica aerogel monolith using an ionic liquid, 1-ethyl-3-methylimidazolium bis[(trifluoromethyl)sulfony]amide, as the solvent and tetramethyl orthosilicate (TMOS) as the sol–gel precursor via a nonhydrolytic formic acid solvolytic method. The silica with embedded ILs thus bear one category of hybrid materials, i.e., ionogels,^14,21 which are extensively available for supercapacitors, fuel cells, drug deliveries, biosensors, catalysis, and miscellaneous applications, as reviewed in the current papers.^22,26 In contrast, extremely few works presented the preparation of super-microporous materials using ILs.^1,2,16,20 The pioneering studies were published by Zhou and Antonietti et al.,^1,2 who first successfully manufactured the highly ordered monolithic super-microporous lamellar silica with long-chain IL, 1-alkyl-3-methylimidazolium chloride (named as [C_nmim][Cl], where the carbon atom number of alkyl chain n ≥ 10), whose pore sizes were ca. 1.2–1.5 nm and surface areas exceeded 1300 m² g⁻¹. The high-temperature calcination (550 °C) was selected to remove the IL. Later, Zhang et al.¹⁶ reported that the prepared porous silica using short-chain IL (BMImN(CN)₂) exhibited the type I sorption isotherms, indicative of the formation of microporous silica, but dramatic reduction of surface area (∼20 m² g⁻¹) occurred after the calcination (∼300 °C).

With the above-described, the scientific and industrial interests are emerging: (1) considering the production cost, if the super-microporous material can be prepared through using short-chain IL as solvent and subsequent low-temperature heat treatment to remove the IL; (2) what the formation mechanism of such synthesized super-microporous material could be. For first, the present work will attempt to adopt a nonhydrolytic sol–gel process to synthesize super-microporous silica using a hydrophobic short-chain IL, 1-butyl-3-methylimidazolium hexafluorophosphate ([C₄mim][PF₆]), as the template solvent and tetraethyl orthosilicate (TEOS) as the precursor. Methanol and formic acid are used as the co-solvent and gelating agent, respectively. The embedded IL was removed via solvent extraction (65–70 °C) and subsequent vacuum drying at room temperature, in place of high-temperature calcination. Recently, it is reported that the reduced precursor/solvolytic reagent (like TEOS/formic acid) molar ratio would influence the properties of the resultant silica gels.²⁷ Similar to the sol–gel process by the present work, earlier work also presented the mesoporous silica could be obtained by the formulation with TEOS/formic acid/[C₄mim][PF₆] molar ratio of 1/8/0.4 (cf.ref. 28). Different from that, in this work, the sol–gel reaction was carried by fixing methanol/formic acid/TEOS/IL molar ratio of 2/4/1/x, x = 0.5–3.0. The results from the nitrogen gas sorption, and transmission electron microscopy (TEM) will verify that the super-microporous silica with disordered worm-like pore (pore sizes are ca. 1.1–1.2 nm) could be synthesized at x ≥ 2.0. For second, S/WAXS experiments associated with scanning electron microscopy (SEM) were further used to explore the formation of super-micropore in the silica framework.

2 Experimental

2.1 Materials, and preparations of porous silica gels

Tetraethyl orthosilicate (TEOS, 98%) and formic acid (98%) were acquired from Acros Organics. The methanol (99.9%) and 1-butyl-3-methylimidazolium hexafluorophosphate ([C₄mim][PF₆], ≥97%) were obtained from Echo and Sigma-Aldrich, respectively. The sol–gel reaction was carried by fixing methanol/formic acid/TEOS molar ratio of 2/4/1 and changing molar ratio of IL to TEOS, denoted by n_IL/n_TEOS. For exemplified specimen with n_IL/n_TEOS = 1, IL (15.82 g) was mixed with methanol (3.60 g) and formic acid (10.25 g) under mild magnetic stirring. Subsequently, TEOS (11.63 g) was added dropwise at room temperature. After the solution was magnetically stirred for ca. 6 hours, a transparent sol was obtained. The sol was stood in a closed flask at room temperature (RT) for 12 hours for gelation and then in an open flask at RT for 5 days for aging. Subsequently, the embedded ionic liquid was removed from the wet silica gel by extracting the sample with methanol in an extraction and stripping apparatus at 65–70 °C for 5 days. Finally, the monolith of wet silica gel was dried under a high vacuum for 8 hours and the dry porous gel was obtained. Through thermogravimetric analyses (TGA), the measured TGA thermogram of the synthesized dry silica gel confirmed the confined IL had been removed (see Fig. S1a in ESI†). The same process was continuously used to prepare the specimens with other n_IL/n_TEOS ratios of 0.5–3.0. Finally, all the prepared specimens were ground into powders for the subsequent characterizations.

2.2 Characterizations

The nitrogen gas sorption of the dried porous gel was obtained by a gas adsorption analyzer (Micromeritics ASAP 2020). Prior to degassing, the samples were pre-treated by an isothermal annealing at 90 °C for 1 hour, followed by heating to 250 °C. Finally, the samples were degassed at 250 °C under 2 µm_Hg pressure till they were completely free of moisture and adsorbed gases.

The solid matrix structures of the synthesized gels were probed by a Bruker Nanostar SAXS instrument at room temperature. The SAXS apparatus consisted of a Kristalloflex K760 1.5 kW X-ray generator (operated at 40 kV and 35 mA) and cross-coupled Göbel mirrors for Cu Kα radiation (λ = 1.54 Å), which resulted in a parallel beam of approximately 0.05 mm² in a cross section at the sample position. The scattering intensity was detected by a Siemens multiwire-type area detector with a 1024 × 1024 resolution mode. The ground powders were introduced into the sample cells composed of two Kapton windows for SAXS measurements. All the intensity data were corrected for empty beam scattering, background and the sensitivity of each pixel on the area detector. The mesoscopic structural heterogeneities of bulk IL and the confined IL in the porous silica matrix were examined by high-resolution WAXS conducted at Endstation BL17A1 of National Synchrotron Radiation Research Center, Hsin-Chu, Taiwan. The wavelength of incident beam was 0.13344 nm and the configuration covered a measurable q range from 1.0 to 30.0 nm⁻¹. Finally, the intensity was output as a function of the scattering vector, q = (4π/λ)sin(θ/2) (θ = scattering angle).

Transmission electron microscopy (TEM) was utilized to examine the real-space morphologies of the synthesized dry gels. The TEM samples were prepared by diluting the ground dry silica powder with ethanol under an ultrasonic environment. Then, a droplet of the suspension was deposited onto a copper grid and dried for 1 min before use. Finally, the ultrathin samples were examined by a JEOL JEM-2100F field-emission TEM operated at 200 kV.

The surface topography image of the synthesized dry gel was obtained with scanning electron microscopy (SEM) using a Hitachi S-4200 field-emission instrument with 15 kV. For sample preparation, the ground silica powder was directly mounted on a stub of metal with copper conductive tape, sputter-coated with a thin layer of gold, and then observed in the microscope.

3 Results and discussion

Fig. 1a shows the measured nitrogen gas sorption isotherms. Three distinct isotherms were clearly distinguished with increasing compositions. For n_IL/n_TEOS ≤ 1.0, the type IV isotherms with hysteresis loops were observed, indicating the formation of mesoporous structures.^29,30 Note that two distinct shapes of hysteresis loops on both type IV isotherms were also expressed, in which for n_IL/n_TEOS = 0.5, type H₂ hysteresis loop meaning the formation of open or closed pores with various radiuses was shown, and another type H₃ hysteresis loop regarding open slit-shaped capillary pores was exhibited for n_IL/n_TEOS = 1.0, respectively.³⁰ The BJH (Barrett–Joyner–Halenda) pore distributions in Fig. 1b showed the main peaks were centered at 9.15 nm and 24.18 nm, verifying the mesopores formed in the silica gels. For n_IL/n_TEOS ≥ 2.0, the sorption isotherms behaved like type I, closely associating with the formation of micropores. For microporous materials, NLDFT method was more properly used to calculate pore distributions.³¹ As shown in Fig. 1c are the calculated NLDFT pore distributions. We can see the main peaks of n_IL/n_TEOS = 2.0 and 3.0 were centered at 1.17 and 1.06 nm, respectively, unambiguously demonstrating the formation of super-micropores (pore size = 1–2 nm). However, Fig. 1c also shows the minor peaks were positioned at ca. ≥ 2.0 nm, implying a small fraction of mesopores possibly appearing in the synthesized silica gels. The sorption isotherm of n_IL/n_TEOS = 1.5 should be the combination of type I with type IV isotherms, as evidenced the broad peak (52.9 nm) and super-micropore peak (ca. 1.28 nm) in Fig. 1b and c, respectively. Such a hysteresis loop exhibited by n_IL/n_TEOS = 1.5 was attributed as a H₄-like, which was closely linking to the formation of narrow slit-like nanopore.³⁰ The structural parameters from nitrogen sorption data are further tabulated in Table 1. Generally, the BET (Brunauer–Emmett–Teller) surface areas (S_BET) measured here decreased with increasing compositions, but still high S_BET (>1000 m² g⁻¹) were available for low compositions. The silica gels with n_IL/n_TEOS = 2.0 and 3.0 exhibited fairly high S_micro/S_BET = 0.97 and 0.94, signifying the measured surface area most contributed by the micropore areas; besides, S_micro/S_BET of 0.54 for n_IL/n_TEOS = 1.5 also implied the near a half of surface area contributed by the micropore area.

Fig. 1 (a) Nitrogen gas sorption isotherms with various n_IL/n_TEOS ratios. The corresponding distributions of pore diameters were calculated from desorption branch of the isotherms with different methods: (b) BJH method for n_IL/n_TEOS = 0.5–1.5 and (c) NLDFT method for n_IL/n_TEOS = 1.5–3.0.

Table 1 Structural parameters from nitrogen gas sorption experiments^a

n IL/nTEOS	Pore type	S BET (m^2 g^−1)	S micro (m^2 g^−1)	V p (cm^3 g^−1)	D p (nm)
a BET surface area (SBET) was estimated by BET method. Micropore area (Smicro) was calculated by t-plot method. Pore diameters (Dp) were directly read the top values of the primary peaks shown in the pore distribution curves in the panel (b) of Fig. 1 for nIL/nTEOS = 0.5–1.0 and in the panel (c) of Fig. 1 for nIL/nTEOS = 1.5–3.0. The number in parentheses was read from BJH pore distribution (Fig. 1c). Vp means the pore volume.
0.5	H2	1043	—	1.72	9.15
1.0	H3	1133	77	3.42	24.18
1.5	H4-like	952	511	1.14	1.28 (52.94)
2.0	Micropore	806	779	0.42	1.17
3.0	Micropore	690	651	0.30	1.06

---

TEM and SAXS were subsequently carried to examine the structures of the synthesized silica gels. As displayed in Fig. 2a and b are the representative TEM images of the synthesized dry silica gels with n_IL/n_TEOS = 1.0 and 2.0, respectively. From Fig. 2a, the silica with n_IL/n_TEOS = 1.0 displayed the typical fractal silica framework formed by mesoporous silica gels, which consisted of the aggregates of the primary silica nanoparticles. By contrast, a disordered worm-like pore morphology was observed in Fig. 2b, manifesting the micropores appearing in the dry silica gel with n_IL/n_TEOS = 2.0, consistent with the BET results. Such a worm-like morphology had also be observed in the other system with organic solvents as well;^8,9 however, this is the first time to be visible in the silica gels synthesized by using short-chain IL as solvent.

Fig. 2 Representative TEM images showing the morphologies of the synthesized dry silica gels with n_IL/n_TEOS = (a) 1.0 and (b) 2.0.

Fig. 3a illustrates the SAXS profiles of wet silica gels (with confined IL) as a function of n_IL/n_TEOS ratios. As seen in Fig. 3a, all the scattering intensities I(q) followed power law decay on q as I(q) ∼ q^−D_f, indicating the formation of a fractal silica matrix, where D_f was mass-fractal dimension of silica matrix (1 < D_f < 3, D_f = 1 and 3 referred to the formation of open and compact fractals, respectively). However, all the I(q) also displayed another scattering dependence of I(q) ∼ q⁰, leading to a plateau appearing at high-q region. Such a plateau would extend to a broader q range with increasing n_IL/n_TEOS. This unique relation of I(q) ∼ q⁰ does not mean any physical meaning and mainly arises from the scattering contribution from polarity peak of confined IL (see WAXS patterns in Fig. 6b). This extra scattering contribution overlaps the scattering intensity contributed from silica matrix, thus resulting in the formation of high-q plateau. Besides, this overlapping will also strongly influence the quantitative analysis of fractal silica matrix of wet silica gels synthesized here.

Fig. 3 SAXS profiles as a function of n_IL/n_TEOS ratios: (a) wet silica gels (with confined IL), and (b) the corresponding dry silica gels (IL-removed). The red solid line in panel (b) means the simulation from the scattering function of fractal object. All the profiles are plotted in a log–log scale.

The panel b of Fig. 3 shows the SAXS profiles of dry silica gels after removing IL through solvent extraction and subsequent vacuum drying. It is certain that large-scale collapse of silica matrix did not occur during the IL-removing process, as evidenced the similar power law dependences of I(q) ∼ q^−D_f for n_IL/n_TEOS ≤ 1.5 (the D_f values were ca. 2.5, similar to the ones of wet silica gels). For n_IL/n_TEOS ≥ 2.0, however, hierarchical fractal structures were formed, as measured two power law dependences of I(q) ∼ q^−D_f at low-q region and I(q) ∼ q^−D_p at high-q region. The intermediate-q region was a transition region between two fractal structures. Low-q exponent (D_f ≈ 2.8) was attributed to the mass-fractal dimension of silica matrix, whereas high-q exponent (D_p ≈ 2.9) corresponded to the tiny mass-fractal particle bound by a surface fractal (D_p = 3 + D_m − D_s, D_m and D_s mean the mass and surface fractal dimension of the constitute particle. 1 < D_m < 3, D_m approaching 3 means a sphere with an uniform density; D_s always lies in between 2 and 3 and can be used to examine the surface roughness of the primary silica particle. D_s = 2 and 3 signal a perfect smooth and an extreme rough surface forming on particle, respectively). It is, consequently, suggested that in this sense the synthesized silica matrix basically formed a mass-fractal framework, which was constituted of the aggregation of tiny mass-fractal particle bound by fractal surface. A monotonous scattering curve measured with single power law dependence for n_IL/n_TEOS ≤ 1.5 was possibly ascribed as that the D_f should be very close to D_p.

Next, we would like to further quantitatively analyse the structural characteristics via fitting scattering intensities with structure and form factor equations of fractal object. The scattering intensity (I(q)) from fractal aggregate formed by clustering from constitute particle described as the following formula, viz.

I(q) = AS(q)P(q) (1)

A is a constant with value of ϕΔρ²V_p (ϕ is the particle volume fraction, Δρ² is the contrast factor, V_p is the particle volume), P(q) is the form factor, and S(q) is the structure factor. S(q) is usually given by³²

where D_f is the mass-fractal dimension of silica matrix, ξ_f is the correlation length that can be used as a measure for the size of the fractal network aggregate, and r₀ is the effective radius of the building block. ξ_f and r₀ may be considered as the maximum and the minimum length scale, respectively, between which the fractal object exhibits the feature of self-similarity. That is to say, the scattering intensity of the fractal object shows the power-law dependence of I(q) ∼ q^−D_f in the 1/ξ_f < q < 1/r₀.³²Γ(x) is the gamma function. Because a non-particular constitute particle is characterized in this work, P(q) is expressed as following equation:^33,34here ξ_p is a measure of the linear size of the constitute particle, which is proportional to the radius of gyration R_g,p, and has a relation of ξ_p² = 2R_g,p²/[D_p(D_p + 1)], D_p is the fractal dimension of the constitute particle.^32,33 Because the constitute particle may have rough surface and mass-fractal body, D_p = 3 + D_m − D_s (D_m and D_s mean the mass and surface fractal dimension of the constitute particle). If the constitute particle is the spherical particle containing uniform density, then D_m = 3, and thus D_p = 6 − D_s (i.e., Bale and Schmidt scattering power law I(q) ∼ q^−(6−D_s)).³⁵

Finally on the basis of above mention, if we want to use formula (1) to fit the measured scattering intensity, the fitted parameters could be set forth as A, D_f, ξ_f, r₀, R_g,p, D_p. Generally, D_f can be read directly from the slope. The red solid lines depicted in Fig. 3b were the fitted results using formula (1) and Table 2 listed the fitted structural parameters (D_f, ξ_f, r₀, R_g,p, D_p). The formula (1)–(3) had been well-analysed the silica aerogel, as recently published by Vollet et al.'s work.³⁶ From Table 2, the fitted R_g,p values expressed that the constitute particle was so tiny that its scattering contribution from form factor became very small. The fitted r₀ values reflected that the mass- and surface-fractal effects of constitute particle becoming significant should be at q > 1/r₀ (e.g., 1/r₀ ∼ 0.99 nm⁻¹ for n_IL/n_TEOS = 0.5); it was, however, hard to distinguish the real tuning point between power law regions of silica matrix and constitute particles because the D_f and D_p showed very close values (Table 2). Furthermore, the q range depicting self-similarity of fractal silica aggregate, 1/ξ_f < q < 1/r₀, also tended to increase with increasing compositions, but over-estimated ξ_f values for n_IL/n_TEOS ≥ 1.5 might appear as a result of the simulated 1/ξ_f exceeding the lower limit of accessible q range. Because the used model cannot allow us to further precisely resolve D_s or D_m, we here selected nitrogen gas sorption isotherm as another independent method to estimate D_s.

Table 2 Structural parameters acquired from SAXS scattering intensities and nitrogen adsorption isotherms^a

n IL/nTEOS	ξ f (nm)	r o (nm)	D f	R g,p (nm)	D p	D s	D m
a ξ f is the correlation length; r0 is the effective radius of the building block; Df is the mass-fractal dimension of silica matrix; Rg,p refers to the radius of gyration of the constitute particle; Dp is the fractal dimension of the constitute particle, Dp = 3 + Dm − Ds, Dm and Ds mean the mass and surface fractal dimension of the constitute particle, respectively.
0.5	2.87	1.01	2.51	1.38	2.70	2.45	2.15
1.0	7.03	1.07	2.62	1.34	2.95	2.55	2.50
1.5	33.23	1.14	2.52	1.11	2.97	2.31	2.28
2.0	31.05	6.05	2.83	1.45	2.98	∼3.00	∼3.00
3.0	34.35	2.38	2.95	0.42	2.98	∼3.00	∼3.00

---

The data of nitrogen adsorption isotherm was used to estimate D_s, which could be independently resolved through Frenkel–Halsey–Hill equation, i.e. FHH scaling law. FHH scaling law normally applies to fit nitrogen gas adsorption isotherm data and has formula as defined^37–39

where Q is the quantity adsorbed at the relative pressure P/P₀, Q_m is the quantity adsorbed of a monolayer, C is a constant. −1/m value is the slope of the linear region when the measured data of nitrogen adsorption isotherm is plotted in ln(Q/Q_m) vs. ln(ln(P₀/P)). Whereas because of the different stages occurring upon adsorption, two distinct dependences of 1/m on D_s can thus be distinguished by two different interface interactions. For the van der Waals adsorption regime in the early stage adsorption, the attractive van der Waals forces predominate the film/gas interface between the nitrogen and solid adsorbate, 1/m < 1/3, and thus^38,39

By contrast, for capillary condensation adsorption regime, multilayer adsorption occurs and the film/gas interface is controlled by liquid/gas surface tension. In this stage, 1/m > 1/3, and therefore^38,39

Fig. 4 represents the FHH plot of the dry silica gels with n_IL/n_TEOS = 0.5–1.5. From Fig. 4, the solid lines were well fitted to the adsorption data. For n_IL/n_TEOS = 0.5 and 1.0, the fitted 1/m values were equal to 0.547 and 0.450, respectively, larger than 1/3 and unambiguously exhibiting the occurrence of capillary condensation. By contrast, the 1/m of n_IL/n_TEOS = 1.5 was ca. 0.231 (<1/3), revealing that attractive van der Waals forces predominated adsorption process. Finally, the estimated D_s values for n_IL/n_TEOS = 0.5–1.5 were 2.45, 2.55, and 2.31, respectively; accordingly, the resolved D_m values were ca. 2.15, 2.50 and 2.28 for n_IL/n_TEOS = 0.5–1.5 via D_p = 3 + D_m − D_s (Table 2). Moreover, it is necessary to emphasize here that we cannot use FHH scaling law to fit nitrogen adsorption isotherm data for n_IL/n_TEOS ≥ 2.0, which will give an unreasonable D_s values exceeding 2 < D_s < 3. From the fitted D_p ∼ 2.98 for n_IL/n_TEOS ≥ 2.0 and the constraints of 1 < D_m < 3 as well as 2 < D_s < 3, we can reasonably infer that for n_IL/n_TEOS ≥ 2.0, both the most plausible D_m and D_s are very close to 3.

Fig. 4 FHH plots with various n_IL/n_TEOS ratios. Solid lines were depicted by the modeling of FHH scaling law.

Through TEM and SAXS analysis, we had analysed the silica matrix structures for all the synthesized silica gels. From Table 2, the resolved D_s and D_m values demonstrated that the constitute silica nanoparticle indeed belonged to a non-particular fractal object bound by a rough surface fractal, and thus our assumption regarding the use of form factor model of a mass-fractal particle bound by a surface fractal was reasonable. The constitute silica nanoparticles basically showed tiny sizes (1.11–1.45 nm), a more tiny size of ca. 0.42 nm was even obtained by silica with n_IL/n_TEOS = 3.0. All silica gels prepared here also exhibited highly-branched fractal silica framework, as obtained D_f > 2.5, whereas more highly-branched fractals (D_f = 2.83–2.95) were acquired with n_IL/n_TEOS ≥ 2.0. So high structural compactness certainly reduced the voids in the silica framework, thereby leading to the super-micropore obtained. As for the dry silica gels having n_IL/n_TEOS = 0.5–1.5, the mesoporous nature of pores increases with increasing the amount of IL in silica gels while further increasing the amount of IL gives no any signature of mesoporosity. We can also explain these findings through the measured D_f values. As shown in Fig. 3b and Table 1, the results had verified that the silica matrix basically formed a fractal aggregate. Also, the measured mass-fractal dimension of silica matrix D_f = 2.5–2.6 for n_IL/n_TEOS ≤ 1.5 and D_f = 2.8–2.9 for n_IL/n_TEOS ≥ 2.0, respectively. Based on the definition of D_f, 1 < D_f < 3, D_f ∼ 1 and 3 referred to the formation of open rod-like and compact sphere-like fractals, respectively. Besides, the measured D_f values indeed exhibited that the fractal obtained with n_IL/n_TEOS ≥ 2.0 are more compact than the one with n_IL/n_TEOS ≤ 1.5. Accordingly, the size of void (or pore) embedded in the silica matrix with n_IL/n_TEOS ≥ 2.0 is fairly smaller than the one with n_IL/n_TEOS ≤ 1.5, thereby leading to the formation of mesopore and super-micropore in the silica matrix with different molar ratios, as measured results by BET. The study by Shimizu et al.⁹ also developed a template-free catalytic sol–gel process using a nonionic hydroxyacetone (HA) catalyst as well as short-chain monohydric alcohol, thereby successfully producing the microporous silica. Through SAXS, they verified that the micropore formation mainly resulted from the aggregation among weakly-branched polymer-like silica chains (D_f ∼ 2.0). The weakly-branched polymer-like silica chains first formed in the sol–gel solution, then polymer–polymer aggregation occurred after removing solvent and the residual organic compounds.⁹ Different from that case, our case reflected that a more highly-branched fractal silica framework would directly form in the presence of IL, and IL-removing did not cause large-scale silica framework collapse.

Up to now, the relevant results and discussion are concentrated on exploring the morphologies of pore and silica matrix. It would be interested in understanding the formation mechanism of such the super-micropores embedded in the silica matrix. Our previous studies⁴⁰ on the [C₆mim][PF₆]-based silica gels had revealed that the IL/pore wall repulsive interactions would cause the segregation of confined [C₆mim][PF₆] from wet silica gels with high compositions (i.e., n_IL/n_TEOS = 2.0), thereby leading to narrow H₄-like mesopores forming in the silica matrix. In that case, a large amount of –OH groups would form on the pore wall and give rise to the perturbation of the P–F vibration peak of confined [C₆mim][PF₆] visible in the measured IR spectra. This perturbation was induced by the repulsive interactions existing between silanol and anionic [PF₆]⁻ groups. As a result of the similarity in the chemical structure between [C₄mim][PF₆] and [C₆mim][PF₆], similar repulsive interactions are also ascribed as the principal driving force to segregate confined [C₄mim][PF₆] outside the wet silica gels, and thus possibly lead to the formation of micropores in the silica matrix. If it is truth, we would observe a “Low-q” scattering peak (which links to the nanoscale structural heterogeneity of confined IL) shifting to lower q in the WAXS pattern of confined [C₄mim][PF₆].

Fig. 5 shows the measured WAXS pattern of bulk [C₄mim][PF₆]. From Fig. 5, the diffraction pattern clearly exhibited three peaks, involving a low-q shoulder (q_polar = 4.3 nm⁻¹), an intermediate-q peak (q_charge = 9.8 nm⁻¹), and a high-q peak (q_adj = 14.2 nm⁻¹). Such peaks had been reported by the previous literatures.^41–43 However, a more detailed theoretical explanation to such peaks was recently presented by Margulis et al.,²⁵ who summarized the diffraction behaviours of various ILs by computational simulations and suggested that these three peaks exactly corresponded to the polarity alternation, charge alternation, and adjacency correlation, respectively.²⁵ Polarity alternation (viz. polar–apolar alternation) is connected to nanoscale structural heterogeneity, which typically results from non-polar nanodomains embedded in the charged matrix. Through Bragg's relation (d = 2π/q), the characteristic length of nanodomain (d) was estimated as ca. 1.46 nm. Adjacency correlation is ascribed as interactions between neighboring atoms. Charge alternation is attributed to the correlations between charged groups. For the sake of subsequent discussion, these peaks are hereafter called as polarity peak, charge peak, and adjacency peak, respectively.

Fig. 5 WAXS pattern of bulk [C₄mim][PF₆]. The logarithmic scale was adopted on y-axis for peak clarity.

Fig. 6a shows the WAXS patterns of the confined [C₄mim][PF₆] in the wet silica gels with various compositions. The plot of the peak positions vs. compositions is displayed in Fig. 6b. From Fig. 6a, all the diffraction patterns of charge peaks and adjacency peaks presented composition-independent, whose peak positions (q^C) and characteristic lengths (d^C) were very close to the ones of bulk IL (cf. the panels b and c of Fig. 6). This indicates that the IL/wall interactions do not dramatically disturb the inter-atom correlations between neighboring atoms and electrostatic interactions between the charged groups. By contrast, significant variations of polarity peaks corresponding distinct composition regions were clearly seen, which displayed the approximately same tendency as our previous studied wet silica gels with IL [C₆mim][PF₆].⁴⁰ If scrutinizing the polarity peak positions in the panels a and b in Fig. 6, we could find that for n_IL/n_TEOS ≤ 0.5, there was no polarity peak visible, signifying that the nanoscale structural heterogeneity formed by confined IL was destructured because the IL was confined in the narrow pore with rough surface on the pore wall, as explained in the previous paper.⁴⁴ For intermediate-n_IL/n_TEOS region (1.0 ≤ n_IL/n_TEOS ≤ 2.0), the polarity peak gradually emerged and obviously shifted from q^C_polar = 4.87 to 4.37 nm⁻¹, indicative of the formation of close packing of confined IL than bulk IL. Whereas the polarity peaks would level off at q^C_polar = 4.37 nm⁻¹ for n_IL/n_TEOS > 2.0, almost equal to the q_polar of bulk IL (see the comparison of dot line and solid line in Fig. 6a), thus unambiguously indicating that a large amount of IL was segregated outside of wet silica gel. The further detailed information could be seen more clearly in the panel c in Fig. 6, which expressed the dependence of the calculated characteristic lengths (d^C, C means the “confined”) vs. compositions.

Fig. 6 (a) WAXS patterns of the wet silica gels (with confined IL) as a function of compositions; (b) and (c) showing the corresponding composition-variable peak positions (q^C) and characteristic distances (d^C), respectively. “C” means the “confined” and the dot line in the panel (a) refers to the bulk [C₄mim][PF₆]. The d^C values are calculated based on the Bragg's relation (d^C = 2π/q^C).

According to the above S/WAXS analysis, it is certainly ascertained that with high n_IL/n_TEOS ≥ 2.0, the obtained super-microporous silica gels embedded disordered worm-like micropores could be attributed to the formation of highly-branched fractal silica framework (D_f > 2.8) and the segregation of the confined IL, which was basically driven by the strong IL/pore wall repulsive interactions when excess amount of IL was used here. Accordingly, we presented a formation mechanism regarding the super-microporous silica with hydrophobic short-chain [C₄mim][PF₆] as solvent. As shown in the Scheme 1, when excess amount of [C₄mim][PF₆] as solvent was used to synthesize the porous silica gels with high compositions (n_IL/n_TEOS ≥ 2.0), the sol–gel reaction initially formed many tiny silica sol nanoparticles (stage a, sol formation), and then these tiny silica sols would gradually aggregate into a weakly-branched silica framework (stage b). Later, as illustrated in stage c of Scheme 1, with gelation continuously undergoing, the weakly-branched silica framework further grew into a highly-branched silica framework. Synchronously, because excess amount of hydrophobic [C₄mim][PF₆] was confined into narrow hydrophilic nanopores, the IL/pore wall repulsive interactions were strong enough to segregate the large amount of confined IL in the highly-branched silica framework out of the wet silica gel. These segregated IL would form a continuous phase to disperse and isolate these wet silica gel micelles including highly-branched silica framework and the residual components (TEOS, methanol, formic acid, a small amount of confined ILs, and other sol–gel by-products). Later, further gelation and aging reaction would continuously be undergone in these micelles and create a more highly-branched fractal silica framework (D_f > 2.8). Finally, after removing ILs and other sol–gel residuals, super-microporous silica spheres with micro-scale sizes (cf.Fig. 7) were acquired in the powder form (stage d).

Scheme 1 A schematic picture illustrating the postulated formation mechanism of the super-micropores formed in the silica matrix with different stages: (a) sol formation; (b) formation of weakly-branched silica framework; (c) formation of highly-branched fractal silica framework and segregation of confined ILs; finally, (d) the removal of IL and the formation of dry super-microporous silica.

Fig. 7 Representative SEM image showing the synthesized dry silica gel with n_IL/n_TEOS = 2.0.

4 Conclusions

Porous silica gels with various pore sizes had been synthesized via a nonhydrolytic methanol–formic-acid solvolytic sol–gel process and using hydrophobic [C₄mim][PF₆] as template solvent as well as TEOS as precursor. The embedded IL in the wet silica gel was removed via solvent extraction (65–70 °C) and subsequent vacuum drying at room temperature rather than high-temperature calcination. Various pore sizes and surface areas could be easily altered by mediating molar ratio n_IL/n_TEOS. With n_IL/n_TEOS ≤ 1.0, the mesoporous silica gels with pore size = 9.0–24.0 nm and surface area > 1000 m² g⁻¹ could be prepared. Then, the silica gels with super-micropore size = 1.1–1.2 nm and surface area = 690–800 m² g⁻¹ could be synthesized with high n_IL/n_TEOS ≥ 2.0. The silica gel containing micropore and mesopore was co-existed at n_IL/n_TEOS = 1.5. The TEM and S/WAXS results demonstrated that the obtained super-microporous silica gel embedded disordered worm-like micropores could be attributed to the formation of highly-branched fractal silica framework (D_f > 2.8) and the segregation of the confined IL, which was basically driven by the strong IL/pore wall repulsive interactions when excess amount of IL was used here.

Acknowledgements

We gratefully acknowledge financial support from Industrial Technology Research Institute, Taiwan. We also deeply thank Prof. Hsin-Lung Chen and Dr Kuei-Yu Kao from Department of Chemical Engineering at National Tsing Hua University for helping data acquisition of SAXS and high-resolution WAXS conducted at Endstation BL17A1 of National Synchrotron Radiation Research Center, Hsin-Chu, Taiwan.

References

1. Y. Zhou and M. Antonietti, Adv. Mater., 2003, 15, 1452.
2. Y. Zhou and M. Antonietti, Chem. Mater., 2004, 16, 544.
3. R. Ryoo, I. Park, S. Jun, C. W. Lee, M. Kruk and M. Jaroniec, J. Am. Chem. Soc., 2001, 123, 1650.
4. M. Kruk, T. Asefa, M. Jaroniec and G. A. Ozin, J. Am. Chem. Soc., 2002, 124, 6383.
5. Y. S. Lin, H. P. Lin and C. Y. Mou, Microporous Mesoporous Mater., 2004, 76, 203.
6. R. Wang, S. Han, W. Hou, L. Sun, J. Zhao and Y. Wang, J. Phys. Chem. C, 2007, 111, 10955.
7. A. K. L. Yuen, F. Heinroth, A. J. Ward, A. F. Masters and T. Maschmeyer, Microporous Mesoporous Mater., 2012, 148, 62.
8. W. Shimizu and Y. Murakami, ACS Appl. Mater. Interfaces, 2010, 2, 3128.
9. W. Shimizu, J. Hokka, T. Sato, H. Usami and Y. Murakami, J. Phys. Chem. B, 2011, 115, 9369.
10. W. Shimizu, S. Nakamura, T. Sato and Y. Murakami, Langmuir, 2012, 28, 12245.
11. S. Dai, Y. H. Ju, H. J. Gao, J. S. Lin, S. J. Pennycook and C. E. Barnes, Chem. Commun., 2000, 243.
12. Y. Zhou, J. H. Schattka and M. Antonietti, Nano Lett., 2004, 4, 477.
13. Y. Liu, M. Wang, Z. Li, H. Liu, P. He and J. Li, Langmuir, 2005, 21, 1618.
14. M.-A. Néouze, J. Le Bideau, P. Gaveau, S. Bellayer and A. Vioux, Chem. Mater., 2006, 18, 3931.
15. T. Wang, H. Kaper, M. Antonietti and B. Smarsly, Langmuir, 2007, 23, 1489.
16. J. Zhang, Y. Ma, F. Shi, L. Liu and Y. Deng, Microporous Mesoporous Mater., 2009, 119, 97.
17. K. M. S. Meera, R. M. Sankar, S. N. Jaisankar and A. B. Mandal, Colloids Surf., B, 2011, 86, 292.
18. L. Viau, M. Néouze, C. Biolley, S. Volland, D. Brevet, P. Gaveau, P. Dieudonné, A. Galarneau and A. Vioux, Chem. Mater., 2012, 24, 3128.
19. C. M. Wu, S. Y. Lin and H. L. Chen, Microporous Mesoporous Mater., 2012, 156, 189.
20. H. Sanaeishoar, M. Sabbaghan and F. Mohave, Microporous Mesoporous Mater., 2015, 217, 219.
21. J. Le Bideau, L. Viau and A. Vious, Chem. Soc. Rev., 2011, 40, 907.
22. M. P. Singh, R. K. Singh and S. Chandra, Prog. Mater. Sci., 2014, 64, 73.
23. O. Russina, A. Triolo, L. Gontrani and R. Caminiti, J. Phys. Chem. Lett., 2012, 3, 27.
24. W. Zheng, A. Mohammed, L. G. Hines Jr, D. Xiao, O. J. Martinez, R. A. Bartsch, S. L. Simon, O. Russina, A. Triolo and E. L. Quitevis, J. Phys. Chem. B, 2011, 115, 6572.
25. J. C. Araque, J. J. Hettige and C. J. Margulis, J. Phys. Chem. B, 2015, 119, 12727.
26. R. Hayes, G. G. Warr and R. Atkin, Chem. Rev., 2015, 115, 6357.
27. A. K. Gupta, Y. L. Verma, M. P. Singh and R. K. Singh, J. Solid State Chem., 2016, 242, 29.
28. Y. L. Verma, M. P. Singh and R. K. Singh, J. Nanomater., 2012, 2012, 570719.
29. K. S. W. Sing, D. H. Everett, R. A. W. Haul, L. Moscou, R. A. Pierotti, J. Rouquerol and T. Siemieniewska, Pure Appl. Chem., 1985, 57, 603.
30. J. Rouquerol, D. Avnir, C. W. Fairbridge, D. H. Everett, J. H. Haynes, N. Pericone, J. D. F. Ramsay, K. S. W. Sing and K. K. Unger, Pure Appl. Chem., 1994, 66, 1739.
31. J. Landers, G. Y. Gor and A. V. Neimark, Colloids Surf., A, 2013, 437, 3.
32. T. Freltoft, J. K. Kjems and S. K. Sinha, Phys. Rev. B: Condens. Matter Mater. Phys., 1986, 33, 269.
33. P. Wong and Q. Cao, Phys. Rev. B: Condens. Matter Mater. Phys., 1992, 45, 7627.
34. H. Wu, J. Xie, M. Lattuada and M. Morbidelli, Langmuir, 2005, 21, 3291.
35. H. D. Bale and P. W. Schmidt, Phys. Rev. Lett., 1984, 53, 596.
36. A. P. Perissinotto, C. M. Awano, D. A. Donatti, F. S. de Vicente and D. R. Vollet, Langmuir, 2015, 31, 562.
37. B. Sahouli, S. Blacher and F. Brouers, Langmuir, 1996, 12, 2872.
38. A. B. Jarzębski, J. Lorenc and L. Pająk, Langmuir, 1997, 13, 1280.
39. M. J. Watt-Smith, K. J. Edler and S. P. Rigby, Langmuir, 2005, 21, 2281.
40. C. M. Wu and S. Y. Lin, J. Phys. Chem. C, 2015, 119, 12335.
41. A. Triolo, A. Mandanici, O. Russina, V. Rodriguez-Mora, M. Cutroni, C. Hardacre, M. Nieuwenhuyzen, H. Bleif, L. Keller and M. A. Ramos, J. Phys. Chem. B, 2006, 110, 21357.
42. A. Triolo, O. Russina, B. Fazio, R. Triolo and E. D. Cola, Chem. Phys. Lett., 2008, 457, 362.
43. C. Hardacre, J. D. Holbrey, C. L. Mullan, T. G. A. Youngs and D. T. Bowron, J. Chem. Phys., 2010, 133, 074510.
44. C. M. Wu, S. Y. Lin, K. Y. Kao and H. L. Chen, J. Phys. Chem. C, 2014, 118, 17764.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra19932a

---