Synthesis of a bilayered SDS/ionic liquid stabilized ferrofluid and magnetic cubic mesoporous silica using a long chain ionic liquid

Abstract

Bilayered sodium dodecyl sulfate (SDS)/ionic liquids (alkyl imidazolium type ionic liquid C_nmim⁺, n = 10, 12, 14, and 16) were first employed to modify the surface of Fe₃O₄ nanoparticles for the formation of stable aqueous ferrofluids. Subsequently, in the synthesized bilayered SDS/C₁₆mim⁺-stabilized ferrofluid (Fe₃O₄/SDS/C₁₆mim⁺), magnetic mesoporous silica with a cubic (Ia3d) mesostructure (Fe₃O₄@MCM-48) has been synthesized using tetraethylorthosilicate as the silicon source and the long chain C₁₆mim⁺ as a template again. The research results showed that the combination of SDS and long chain C_nmim⁺ (such as C₁₆mim⁺) is superior to that of SDS and short chain C_nmim⁺ (such as C₁₀mim⁺) for the production of stable ferrofluids in water. Moreover, Fe₃O₄@MCM-48 (0.1 g) with Fe₃O₄ particles embedded in the network of cubic mesoporous silica showed decolorization rates of up to ∼90% in 25 min for four dye (methylene blue, rhodamine B, malachite green, and methyl violet) solutions (50 mL, 100 mg L⁻¹). In addition, the malachite green-adsorbed Fe₃O₄@MCM-48 can be separated and recovered under an external magnetic field. This method in which the long chain ionic liquid C₁₆mim⁺ can be manipulated in turn not only as the modifier but also as the mesoporous templating agent provides important guidelines for the multifunctional applications of C₁₆mim⁺ in the synthesis of various nanostructured materials.

---

1. Introduction

Surfactant-stabilized magnetic particles have been extensively studied for the stabilization of magnetic fluid suspensions by steric repulsion of surfactants.¹ The electrostatic repulsion occurring in the magnetic particles can be also produced by the introduction of a surfactant-coated bilayer as proposed by Shimoiizaka et al. in the 1980s.² The assembly of bilayered surfactants onto the magnetic particles was achieved by first coating the surface of the preformed magnetic particles with the primary surfactant, followed by coating the surface of the primary surfactant-coated particles with a second surfactant. For example, Wooding et al. employed saturated and unsaturated fatty acids as primary and secondary surfactants to construct stable magnetic fluids in water.³ Based on these results, Laibinis and Hatton group further demonstrated and quantified surfactant bilayered structures using iron oxide nanoparticles.⁴ Subsequently, various surfactant bilayered structures were developed to stabilize the magnetic fluids, such as some combinations of sodium oleate and PEG-type,⁵ oleic acid and polyisobutylene succinimide,⁶ oleic acid and SDS,⁷ and so on. Despite this demonstrated progress, exploiting new species for surfactant bilayered coating and enhancing the stabilization of aqueous magnetic fluids continue to be a research issue.

Recently, the application of ionic liquids has opened a new route in the fields of synthesis and creation of new and interesting materials.^8–10 Particularly, long-chain ionic liquids with imidazolium head groups and long alkyl chains (more than 10 carbon atoms) can show both the surface activity of common surfactants and the special properties of the short-chain ionic liquids such as the properties of low melting temperature, strong solubility, increasing polarity in water, and so on. Numerous studies were performed on the self-organization properties of long-chain ionic liquids in aqueous solutions.^11–13 For example, Zheng and his group contributed much effort to study the micelle properties of ionic liquids with long alkyl chains in water via monitoring their conductivity characterization, fluorescence properties, surface tension changes, etc.^14–16 These results showed that the activity of long-chain ionic liquids is slightly higher than that of cationic surfactants with a quaternary ammonium type structure in aqueous solutions. Besides, long-chain ionic liquids were proved to exhibit the properties of both thermotropic and lyotropic liquid crystals.^17,18 The various features of long-chain ionic liquids have also been utilized to establish various composite nanomaterials.¹⁹ For instance, 1-hexadecyl-3-methylimidazole ions (abbr. C_nmim⁺, n refers to the number of carbon atoms in the alkyl chain on the imidazolium group) were reported to exhibit unique templating behavior in the preparation of silica with a hexagonal mesoporous structure (MCM-41-type).²⁰ Zhou et al. synthesized supermicroporous lamellar silica using the long-chain C₁₆mim⁺ as the template.²¹ After that, it was found that silica with cubic and hexagonal mesoporous structures can be also prepared using the long-chain C₁₆mim⁺ as the template in our earlier research work.²² More recently, we have reported self-assembly bilayer structures of the C_nmim⁺/C_nmim⁺ onto magnetic Fe₃O₄ nanoparticles. Moreover, we found that in the bilayer C₁₆mimCl⁺-based ferrofluid (Fe₃O₄/C₁₆mim⁺/C₁₆mim⁺), hexagonal magnetic mesoporous silica with Fe₃O₄ NPs embedded into the porous silica framework (Fe₃O₄@MCM-41 type) can be prepared using the C₁₆mim⁺ as the template again.²³ However, unexpectedly, although we have made many attempts to investigate the effects of varied synthesized proportions and parameters on the formation of ordered cubic magnetic mesoporous silica (Fe₃O₄@MCM-48 type) in the ferrofluid (Fe₃O₄/C₁₆mim⁺/C₁₆mim⁺), we still cannot obtain the cubic magnetic mesoporous silica Fe₃O₄@MCM-48. This may be due to the narrow phase area of the cubic phase in the liquid crystal phase diagram, or the mismatching ratio of the Fe₃O₄/C₁₆mim⁺/C₁₆mim⁺ particles, silicon precursor (TEOS) and the C₁₆mim⁺ template agent.

In this study, first, we prepared a self-assembly bilayer of SDS/C_nmim⁺ (n = 10, 12, 14, and 16) to modify magnetic Fe₃O₄ NPs for the production of stable ferrofluids. Second, based on the bilayered SDS/C₁₆mim⁺-based ferrofluid, magnetic mesoporous silica with a cubic mesoporous structure and Fe₃O₄ NPs embedded into the cubic mesoporous network was synthesized using TEOS as a silicon source and the C₁₆mim⁺ as a mesoporous templating agent. Furthermore, the adsorption properties of the obtained cubic magnetic mesoporous silica (Fe₃O₄@MCM-48) were also investigated.

In our design, the primary goal is to construct a bimolecular modification layer SDS/C_nmim⁺ onto Fe₃O₄ NPs. It can be expected that the cationic imidazole head-group on the Fe₃O₄ NPs surface extends to the solution, which may lead to a high charge density and strong hydrophilicity on the particle surface. This special surface structure should be able to form a stable aqueous ferrofluid. Our second goal is to confirm the magical templating behavior of the C₁₆mim⁺ in the synthesis of cubic mesoporous silica with magnetism in the obtained Fe₃O₄/SDS/C₁₆mim⁺ ferrofluid. In order to prove the adsorption performance of this magnetic cubic mesoporous silica, four dyes, methylene blue, rhodamine B, malachite green, and methyl violet, were selected as adsorption models. Furthermore, the adsorbed amount of the magnetic cubic mesoporous silica Fe₃O₄@MCM-48 for malachite green solutions with different initial concentrations was investigated, also including its magnetic separation properties and reuse efficiency. We expect that the long-chain ionic liquid C_nmim⁺ should be manipulated in a synthetic system in turn not only as the outer layer modifier of Fe₃O₄/SDS/C_nmim⁺ particles but also as the mesoporous templating agent in the formation of magnetic cubic mesoporous silica.

2. Experimental section

2.1 Materials

1-Chlorodecane, 1-chlorododecane, 1-chlorotetradecane, 1-chlorodexadecane, 1-chlorooctadecane and 1-methylimidazole (analytical reagent, AR) were purchased from Acrös Organics. Sodium dodecyl sulfate (SDS) (AR) was purchased from Tianjin Zhiyuan Chemical Reagent Co., Ltd China. FeCl₂·4H₂O and FeCl₃·6H₂O (AR) were obtained from Chengdu Jinshan Chemical Reagent Co., Ltd, China. Tetraethylorthosilicate (TEOS) (AR) was purchased from Shanghai Aladdin Bio-Chem Technology Co., Ltd, China. All the chemicals were used as received.

2.2 Preparation of SDS-coated Fe₃O₄ particles

Sodium dodecyl sulfate (SDS)-coated Fe₃O₄ particles were first synthesized by the chemical coprecipitation of Fe²⁺ and Fe³⁺ from ammonia water in a molar ratio of 1/2 M with a little amount of SDS, followed by a modifying process of SDS. In a typical preparation step, SDS (0.10 g), FeCl₂·4H₂O (0.86 g), and FeCl₃·6H₂O (2.35 g) were dissolved in deionized water (50 mL) by mechanical stirring under N₂ purging. After mixing for 5 minutes, the temperature of the mixture was increased to 353 K, followed by adjusting the pH value of the mixture to 8 with ammonia (25% w/w). Then, SDS was further added in five 0.1 g amounts over 5 min until the amount of total SDS reached 0.6 g. After 30 minutes, the reaction mixture was cooled slowly. Through a magnetic decantation method, the resultant reaction solid can be obtained from the solution, followed by washing using deionized water and ethanol. This washing-magnetic decantation steps were repeated three times. The resultant solid product is denoted as Fe₃O₄/SDS.

2.3 Preparation of bilayered SDS/C_nmim⁺-stabilized ferrofluids

Long-chain alkyl imidazolium ionic liquids C_nmim⁺, (n = 10, 12, 14, 16, and 18, respectively) were synthesized and identified according to the earlier report.²³ In order to synthesize stable aqueous ferrofluids, the Fe₃O₄/SDS nanoparticles were further coated with a series of C_nmim⁺. In a typical synthesis procedure of Fe₃O₄/SDS/C₁₆mim⁺ ferrofluid, C₁₆mim⁺ solution (1 mL, 40 g L⁻¹) was added to 20 mL of deionized water. Then, the pH value of the solution was adjusted to 3 by dropping a certain amount of dilute hydrochloric acid, followed by adding Fe₃O₄/SDS (0.5 g) to the solution under ultrasound. After mixing for 15 minutes, the C₁₆mim⁺ solution with a concentration of 40 g L⁻¹ was added drop by drop to the mixture again in 1 mL amounts over 5 min with mechanical stirring. This process was continued until no obvious precipitate and phase separation phenomenon were observed when a small permanent magnet was placed at the bottom of the beaker containing the mixture for 10 min. The resultant solution is denoted as the Fe₃O₄/SDS/C₁₆mim⁺ ferrofluid. The preparation procedures of other SDS/C_nmim⁺ bilayered structures onto the Fe₃O₄ particles were accomplished by repeating the above operation steps using the corresponding combination of SDS and C_nmim⁺ (n = 10, 12, and 14, respectively) as inner and outer layer modifiers, respectively. For subsequent detection, a small portion of dried Fe₃O₄/SDS/C_nmim⁺ particles were separated out of the obtained ferrofluids using the decantation method with a small permanent magnet placed at the bottom of a beaker to provide a magnetic field of about 0.3 T over three weeks.

2.4 Preparation of magnetic cubic mesoporous silica

In the obtained Fe₃O₄/SDS/C₁₆mim⁺ ferrofluid, magnetic cubic mesoporous silica was prepared using the hydrothermal synthesis method in alkaline medium using C₁₆mim⁺ as the templating agent and TEOS as the silicon source. Typically, NaOH and C₁₆mim⁺ were dispersed in the Fe₃O₄/SDS/C₁₆mim⁺ ferrofluid with mechanical stirring. Then, TEOS was added in drops to the mixture. The starting molar composition was: 1.0 TEOS:xFe₃O₄/SDS/C₁₆mim⁺:0.2 C₁₆mim⁺:0.5 NaOH:70 H₂O (x = 0.050, 0.075, 0.10, respectively). After 1 hour of the reaction, the reaction mixture was transferred to a sealed Teflon-lined autoclave. The hydrothermal reaction in the autoclave was carried out for 3 days in an oven at 373 K. The resultant white powder was further filtered and washed and then calcined at 823 K for 5 h to remove the C₁₆mim⁺ templating agent. The final product is denoted as Fe₃O₄@MCM-48.

2.5 Characterization

Transmission electron microscopy (TEM) was carried out on a JEM-2100 electron microscope with an acceleration voltage of 200 kV. The samples were prepared by dipping a carbon-coated copper TEM grid (300 meshes) into the sample dispersions and dried at room temperature. The structural analysis of the samples was performed on an X-ray diffractometer (TTR III powder X-ray diffractometer) with Cu Kα (λ = 0.1542 nm) radiation. The XRD was operated at 40 KV and 30 mA over a 2θ range of 1–8° and 20–80°, respectively. The electrophoresis values of the colloidal suspensions were measured using an electrophoresis apparatus (Nanjing Sangli DYL-3). The magnetization experiments of the samples were conducted on a vibrating sample magnetometer (VSM 7407). The thermal properties of samples were investigated using thermogravimetric analysis (TGA) and differential thermal analysis (DTA), respectively, on a ZRY-1P thermal analysis apparatus. The N₂ adsorption–desorption isotherms for the magnetic cubic mesoporous silica were obtained at 77 k on a Tristar 3000 automated gas adsorption analyser. Before the sorption experiments, a Micromeritics VacPrep 061 degasser was used to remove the gas of the samples overnight at 423 K under 100 μTorr pressure. The BJH (Barrett–Joyner–Halenda) method was employed to analyze the obtained data. The BET value was determined at a range of P/P₀ = 0.05–0.35, and pore size distribution curves were obtained from the desorption branch of the isotherm. The Fourier transform infrared (FT-IR) spectra of samples were characterized on a TENSOR 27 FTIR spectrometer. The contents of iron in samples were determined by a well-known o-phenanthroline spectrophotometer, and then, the content of the modified double-layer was obtained by subtracting the corresponding amount of iron oxide from the total amount of the sample.

2.6 Adsorption experiment of magnetic cubic mesoporous silica

Four dye solutions, methylene blue, rhodamine B, malachite green, and methyl violet solutions, were selected as absorbed objects to study the adsorption and separation properties of the magnetic cubic mesoporous Fe₃O₄@MCM-48. In a typical adsorption procedure, the synthesized Fe₃O₄@MCM-48 (0.1 g) was added to 50 mL of the dye aqueous solutions (100 mg L⁻¹). At several fixed time intervals, the resultant sediments were separated from the adsorbed suspensions using a dumping method under magnetic induction, and then, the absorption spectra of the remaining solutions were recorded on an UV-vis spectrograph (SHIMADZU UV-1780) at 664, 554, 617, and 580 nm for methylene blue, rhodamine B, malachite green, and methyl violet solutions, respectively. The decolorization rates of the resultant solutions were calculated according to the following equation:

Decolorization rate:

where A₀ is the initial absorbance value of the solution, A_t is the solution absorbance value after the adsorption.

Moreover, the adsorption and separation characteristics of Fe₃O₄@MCM-48 in the malachite green solutions with different initial concentrations of 25, 50, 100, 200, and 300 mg L⁻¹ were also investigated, respectively. The concentrations of the malachite green solutions after adsorption at several fixed time intervals can be given according to the standard working curves of malachite green solutions. The adsorption quantity of Fe₃O₄@MCM-48, q_t, (mg g⁻¹) was determined from the following equation:

where C₀ and C_t (mg L⁻¹) are the concentrations of malachite green in the solution (mg L⁻¹) at initial time and time t, respectively. V is the volume of the solution (L), and m is the adding amount of Fe₃O₄@MCM-48 (g).

The Freundlich isotherm model is the most important adsorption model for multilayered adsorbent surfaces. The linear form of the Freundlich isotherm model can be represented as

where q_e (mg g⁻¹) represents the adsorption quantity of the adsorbent at equilibrium, C_e is the equilibrium concentration of malachite green solution (mg L⁻¹), K_f and 1/n are characteristic constants representing the adsorption capacity and intensity of the adsorption system, respectively.

In addition, the malachite green-adsorbed Fe₃O₄@MCM-48 can be calcined continuously at 823 K for 5 h to remove the adsorbed malachite green. By tracing the UV-vis absorption spectra at different cycle times, the recycling rate of Fe₃O₄@MCM-48 can be investigated. In contrast, the adsorption experiment of malachite green on the pure cubic mesoporous silica (MCM-48) prepared using the same C₁₆mim⁺ templating agent,²² without adding Fe₃O₄ NPs, was carried out. At the same time, the adsorption experiment of malachite green on the bare Fe₃O₄ NPs prepared by the same chemical coprecipitation method without adding C₁₆mim⁺ as the modifier was also performed.

3. Results and discussion

3.1 Synthesis and structure of bilayer SDS/C_nmim⁺ on Fe₃O₄ NPs

Our strategy to prepare bilayer SDS/C_nmim⁺ on Fe₃O₄ NPs includes first synthesizing Fe₃O₄ NPs and then immobilizing the bilayer SDS/C_nmim⁺ on Fe₃O₄ NPs. The well-established chemical coprecipitation technique of Fe²⁺ and Fe³⁺ in a molar ratio of 1 : 2 from ammonia water was used to synthesize the Fe₃O₄ NPs.²⁴ By adding NH₄OH (25% w/w), the pH of the reaction system was adjusted to 8, which led to the formation of positively charged Fe₃O₄ particles.^25,26 Subsequently, the obtained Fe₃O₄ NPs were coated with SDS physisorbed to the surface of Fe₃O₄ NPs by electrostatic action between the positively charged Fe₃O₄ particle surface and the negatively charged SDS head-group, as shown in Fig. 1. It was easy to observe that the SDS-coated Fe₃O₄ particles (Fe₃O₄/SDS) settled down from the solution within a few minutes (Fig. S1A, ESI†). The reason of this unstable appearance may be due to the alkyl chain of SDS on the Fe₃O₄ particle surface extending to the water. The black Fe₃O₄/SDS particles can be attracted by placing an experimental small magnet next to the beaker (Fig. S1B, ESI†). This simple experiment demonstrated that the Fe₃O₄/SDS particles are magnetic. Fig. S2(a) and (b) (ESI†) show the wide-angle XRD patterns of the pure Fe₃O₄ particles and the Fe₃O₄/SDS particles, respectively. All characteristic diffraction peak positions are consistent with those of the crystalline Fe₃O₄ with the face-centered cubic (fcc) structure according to JCPDS card, No. 00-001-1111. The peak intensity of Fe₃O₄/SDS particles is slightly lower than that of pure Fe₃O₄ particles. The average grain sizes were calculated from the Scherrer formula to be 9.52 nm for Fe₃O₄/SDS particles and 10.20 nm for pure Fe₃O₄ particles, respectively. These results indicated that SDS modification onto Fe₃O₄ particles limited the growth of Fe₃O₄ particles in part.

Fig. 1 Schematic illustration of the surface modification procedure of Fe₃O₄/SDS/C₁₆mim⁺ particles.

In order to synthesize stable aqueous ferrofluids, the Fe₃O₄/SDS particles were coated with C_nmim⁺ (n is 10, 12, 14, 16, respectively) as outer layer. It was found that the C_nmim⁺-coated Fe₃O₄/SDS particles (Fe₃O₄/SDS/C_nmim⁺) exhibited a remarkable stability in the aqueous colloidal suspensions. No sediment was found for more than 6 months (e.g. for Fe₃O₄/SDS/C₁₆mim⁺, Fig. S3, ESI†). Approximately +35 mV of Zeta potential (ζ) value was shown for the Fe₃O₄/SDS/C₁₆mim⁺ colloidal suspension in the corresponding electrophoresis experiment. These results suggest a hypothetical structure of the Fe₃O₄/SDS/C_nmim⁺ particles, that is, the positively charged imidazole head-groups in the outer layer of the particles extend to the aqueous solution, while the alkyl chain of imidazole is towards the Fe₃O₄/SDS particles or partially inserted into the alkyl chain of SDS coated on the particles due to hydrophobic interaction, as shown in Fig. 1. The hypothetical structure that the positively charged imidazole head-groups on the outer layer of the particles provided a very high charge density and hydrophilic particle surface, resulting in the formation of an aqueous stable ferrofluid.

Fig. 2(A) and (B) show the typical TEM image of the Fe₃O₄/SDS/C₁₆mim⁺ particles and its corresponding particle size histogram, respectively. It is clear that the Fe₃O₄/SDS/C₁₆mim⁺ particles exhibited the characteristics of polydispersed spherical particles with a clear interface and an average diameter of approximately 10.15 nm (Fig. 2(B)). Electron diffraction in the corresponding region is shown in the insets in Fig. 2(A). The diffraction image consists of a series of rings that can be indexed to a magnetite structure, which shows the same result as that of the wide-angle XRD pattern shown in Fig. S2 (ESI†). Fig. 2(C) shows the high-resolution TEM (HRTEM) image of a randomly oriented nanocrystallite, which is approximately 10 nm in diameter and possesses clearly resolved lattice fringes, indicating the high crystallinity of the sample particle. It should be noted that no double-layered structure of SDS/C₁₆mim⁺ was observed in the TEM image, which may be because of the low resolution of the molecular arrangement of double-layered SDS/C₁₆mim⁺ onto the Fe₃O₄ particles.

Fig. 2 TEM image of (A) Fe₃O₄/SDS/C₁₆mim⁺ particles and electron diffraction image of a large zone in the inset, (B) corresponding particle size histogram and (C) HRTEM image of a unique single-crystal particle.

Fig. 3 shows the magnetization curves of pure Fe₃O₄, Fe₃O₄/SDS, and Fe₃O₄/SDS/C₁₀mim⁺ particles at room temperature, respectively. These curves exhibited typical magnetization “S-shaped” curves, and no obvious hysteresis loops were observed, that is, there were no significant reduced remanence and intrinsic coercivity in these samples, indicating the superparamagnetic feature of these samples. The saturation magnetizations (M_s) of pure Fe₃O₄, Fe₃O₄/SDS, and Fe₃O₄/SDS/C₁₀mim⁺ particles were determined to be 62.61, 60.10 and 53.90 emu g⁻¹, respectively. It is mentionable that the gradual reducing change of M_s(Fe₃O₄) > M_s(Fe₃O₄/SDS) > M_s(Fe₃O₄/SDS/C₁₀mim⁺) can be observed with the increase of the number of modification layers. This phenomenon may be due to a separation of dipole coupling, which originated from the nonmagnetic species (such as SDS and C_nmim⁺) on the Fe₃O₄ particle surface.^27,28

Fig. 3 Magnetization curves of (a) pure Fe₃O₄ particles, (b) Fe₃O₄/SDS particles, and (c) Fe₃O₄/SDS/C₁₀mim⁺ particles.

Fig. 4 exhibits the typical thermogravimetric analysis (TG) (left axis) and differential thermal analysis (DTA) (right axis) curves of the Fe₃O₄/SDS/C₁₆mim⁺ particles, respectively. For the TG curve, when the temperature was below 120 °C, a small thermal weightlessness of about 2.64% can be observed, which should be due to the removal of adsorbed water from the sample. A more significant percentage weight loss of ∼10.24% appeared between 210 and 480 °C, which can be attributed to the combustion removal of the bilayered SDS/C₁₆mim⁺ molecules. The DTA curve of the sample exhibited an intense exothermic peak at 321 °C and a weak exothermic peak at 389 °C, which may be due to the thermal decomposition of the outer C₁₆mim⁺ and the inner layer SDS, respectively. These results can be easily explained by the differences in molecular weights and boiling point between C₁₆mim⁺ and SDS molecules. Furthermore, these two exothermic peaks appeared at different temperature positions, which may be due to the difference between the two interactions, such as hydrophobic interaction between SDS and C₁₆mim⁺ and electrostatic attraction between SDS and Fe₃O₄ particles. These results revealed the coexistence of two different kinds of molecules (SDS and C₁₆mim⁺) onto Fe₃O₄ particles. Similar to our results, the obvious difference in the weight loss between the bilayer fatty acid surfactants coated on Fe₃O₄ particles has also been reported by Shen et al.⁴

Fig. 4 TG and DTA curves of Fe₃O₄/SDS/C₁₆mim⁺ particles.

The multiple coverage parameters for inner layer SDS and outer layer C_nmim⁺ onto Fe₃O₄ NPs can be further calculated. In this calculation, the content of iron in samples was determined by well-known o-phenanthroline spectrophotometry. Then, the content of the modifier (SDS and C_nmim⁺) was obtained by subtracting the corresponding amount of iron oxide from the total amount of the sample. In addition, the Fe₃O₄ particle was regarded as a sphere with an average diameter of 10 nm according to the results of TEM (Fig. 2). These multiple coverage parameters of the bilayer SDS/C_nmim⁺ assembled onto Fe₃O₄ NPs are listed in Table 1. Two characteristics should be noted. It is first clear that the weight percents of the Fe₃O₄/SDS/C_nmim⁺ (n = 10, 12, 14, 16) particles are larger than those of the Fe₃O₄/SDS particles. The difference can be attributed to the outer layer C_nmim⁺. For example, the weight percents of the bilayer SDS/C_nmim⁺ were estimated onto Fe₃O₄ particles to be 10.46 wt% (SDS/C₁₀mim⁺), 12.50 wt% (SDS/C₁₂mim⁺), 17.42 wt% (SDS/C₁₄mim⁺) and 21.98 wt% (SDS/C₁₆mim⁺), respectively, corresponding to the monolayer SDS of 6.01 wt% (third column in Table 1). This result proved again the existence of the bilayer SDS/C_nmim⁺ molecular structures onto the Fe₃O₄ NPs. Secondly, it was found that with the increasing n value (the number of carbon atoms in the alkyl chain of C_nmim⁺) from 10 to 16, the weight percent of outer layer C_nmim⁺ increased from 4.45 wt% to 15.97 wt% (fourth column in Table 1), and the number of outer layer C_nmim⁺ on each Fe₃O₄/SDS particle increased from 316 to 1006 (sixth column in Table 1) (in the experiment, the modifying amount of adding outer C_nmim⁺ was the same). This C₁₆mim⁺ molecule number of 1006 on each Fe₃O₄/SDS particle is close to that of the saturated fatty acid modifiers (such as myristic acid) coated onto Fe₃O₄ particles with a densely packed structure pattern.³ In the experiment, we found that the long-chain C₁₆mim⁺ displayed a stronger assembly tendency (it was easier to form stable ferrofluid) than that of the short-chain ionic liquid (e.g. C₁₀mim⁺). The reason is that during the modification processes of outer layer C_nmim⁺, there is a competition of two interactions. One is a hydrophilic interaction between imidazole head-groups of C_nmim⁺ and water molecules. Another is a hydrophobic interaction between hydrocarbon tails of outer layer C_nmim⁺ and inner layer SDS. It is well known that positively charged ionic liquids, such as imidazolium salts, have strong hydrophilicity.¹⁹ The shorter the hydrocarbon chain of the C_nmim⁺, the stronger the hydrophilic tendency of the C_nmim⁺. When the short-chain C_nmim⁺ (such as C₁₀mim⁺) was employed as the outer layer modifier, the short-chain C₁₀mim⁺ might tend to stay in water because of its strong hydrophilic interaction with water molecules, which causes only a small amount of C₁₀mim⁺ to adhere to the particles, forming a loose arrangement in the particle surface. In the experiments, it can be observed that when n ≤ 10, the SDS/C_nmim⁺-coated Fe₃O₄ particle exhibited an unstable ferrofluid state, that is, the Fe₃O₄/SDS/C₁₀mim⁺ particles settled down within minutes. However, when the n value was increased to 18, the solubility of this C₁₈mim⁺ in water deceased markedly and an insoluble floating substance (C₁₈mim⁺) similar to sponge appeared, which indicated that it was impossible to produce an effective bilayer SDS/C₁₈mim⁺ structure due to the strong hydrophobicity of the C₁₈mim⁺. When n was 16, the hydrophobic and hydrophilic interactions of the C₁₆mim⁺ were found to be matching, resulting in a significant increase in the assembly capability. Thus, we believe that the Fe₃O₄/SDS/C₁₆mim⁺ particles can provide a strongly hydrophilic and highly charge density surface because of a closely packed structure of C₁₆mim⁺ molecules exposing their imidazole headgroups to the aqueous solution. The special particle surface structure can prevent the particles not only to agglomerate together but also to oxidize between the Fe₃O₄/SDS/C₁₆mim⁺ NPs due to electrostatic and steric repulsions, leading to the production of stable aqueous ferrofluids.

Table 1 Coverage parameters of the bilayer SDS/C_nmim⁺ onto Fe₃O₄ particles

Sample	Wt%^a (Fe3O4)	Wt% (modifier onto Fe3O4 NPs)	Wt% (outer layer modifier)	Wtout^b (outer layer Cnmim^+ on per g Fe3O4) (g)	N out (No. of molecules of outer layer Cnmim^+ on each particle)
a The mass percentage of Fe3O4 determined by the analysis of iron in samples using o-phenanthroline spectrophotometry. b Wtout = wt%(out)/wt%(Fe3O4), where the masspercentage of Fe3O4 was calculated by subtracting the masspercentage of Cnmim^+ and corresponding adsorbed water. c N out = N1/N2, where N1 = Wtout × NA/Mout is the number of Cnmim^+ molecules per g Fe3O4, with NA being the Avogadro constant and Mout being the molar mass of outer layer Cnmim^+, and N2 = V1/V2 is the number of Fe3O4 particles per g Fe3O4, with V1 = 1/ρ and V2 = 4πR^3/3 being the volume per g Fe3O4 and the volume per particle, respectively, where ρ is the density of Fe3O4 (5.18 g cm^−3) and R is the radius of the Fe3O4 particle.
Fe3O4/SDS	93.99	6.01	6.01	—	—
Fe3O4/SDS/C10mim^+	89.54	10.46	4.45	0.05	316
Fe3O4/SDS/C12mim^+	87.50	12.50	6.49	0.07	401
Fe3O4/SDS/C14mim^+	82.58	17.42	11.41	0.14	730
Fe3O4/SDS/C16mim^+	78.02	21.98	15.97	0.21	1006

---

3.2 Characterization of magnetic cubic mesoporous silica

The schematic diagram of the synthesized procedure of the cubic magnetic mesoporous silica is shown in Fig. 5. In this preparation procedure, Fe₃O₄/SDS/C₁₆mim⁺ particles, TEOS and C₁₆mim⁺ template were used as precursors, and subsequently, the procedure included a hydrothermal treatment and finally the calcination step.

Fig. 5 Schematic illustration of the synthesized procedure of magnetic cubic mesoporous silica prepared using C₁₆mim⁺ as the template in the Fe₃O₄/SDS/C₁₆mim⁺ ferrofluid.

Three samples were synthesized with initial molar ratios of n(Fe₃O₄/SDS/C₁₆mim⁺)/n(TEOS) of 0.050, 0.075 and 0.10, respectively, using C₁₆mim⁺ as the templating agent and TEOS as the silica source in the aqueous Fe₃O₄/SDS/C₁₆mim⁺ ferrofluid. The low-angle XRD patterns of the calcined samples (Fig. 6) show well-resolved reflections with d₂₁₁-spacings of 3.91, 3.79, and 3.87 nm, indicating a cubic gyroid phase (Ia3d) structure, which is consistent with the low-angle XRD patterns of the siliceous MCM-48 prepared using a quaternary ammonium type surfactant as the template.²⁹ The unit-cell dimensions a₀ of the cubic mesoporous structures were calculated to be 9.58, 9.29, and 9.48 nm for the three samples with the initial molar ratio n(Fe₃O₄/SDS/C₁₆mim⁺)/n(TEOS) of 0.050, 0.075 and 0.10, respectively (the unit-cell dimension a₀ assuming space group Ia3d was calculated according to the XRD data using a₀ = d₂₁₁√6). The unit-cell dimension a₀ values of the three samples are slightly larger than those of pure cubic mesoporous silica prepared with the same C₁₆mim⁺ as the template in our earlier work,²² which may be due to the introduction of the magnetic particles. The inset in Fig. 6 is an enlarged view of Fig. 6(b) in the range of 2θ = 3.5–5° (multiple peaks), further confirming the existence of the cubic phase structure. Compared with the low-angle XRD patterns of the pure cubic mesoporous silica prepared using C₁₆mim⁺ as the template,²² we found that the presence of the Fe₃O₄/SDS/C₁₆mim⁺ particles in the defined concentration range did not destroy the cubic gyroid phase structure of silica. If we compare the distinguishability of the diffraction peaks shown in Fig. 6(b) and (c), we can find that the resolution of the peaks in Fig. 6(b) is higher than that in Fig. 6(c), revealing that with an increasing initial molar ratio n(Fe₃O₄/SDS/C₁₆mim⁺)/n(TEOS) from 0.075 to 0.100, the order degree of the cubic magnetic mesopores would decrease. This result may be due to excessive addition of Fe₃O₄/SDS/C₁₆mim⁺ particles, which leads to the decrease of the matching degree of charge density between the silicon species and C₁₆mim⁺ template. This phenomenon can be observed when the guest species was loaded into the ordered mesoporous structure and the distinguishability of their XRD lines would be reduced.³⁰

Fig. 6 Low-angle XRD patterns of the three samples prepared with the initial molar ratio n(Fe₃O₄/SDS/C₁₆mim⁺)/n(TEOS) of (a) 0.050, (b) 0.075 and (c) 0.10 using C₁₆mim⁺ as the templating agent in the aqueous Fe₃O₄/SDS/C₁₆mim⁺ ferrofluid.

The TEM images of the calcined sample prepared with the initial molar ratio n(Fe₃O₄/SDS/C₁₆mim⁺)/n(TEOS) of 0.075 using C₁₆mim⁺ as the template are shown in Fig. 7. These TEM images taken along the [111] (Fig. 7(A)), [311] (Fig. 7(B)) and [110] (Fig. 7(C)) zone axes revealed that the sample exhibited an ordered mesoporous structure with cubic symmetry, which is in agreement with the TEM analysis on the pure cubic mesoporous silica (MCM-48) templated by the quaternary ammonium ion surfactant agent.^31,32 Simultaneously, in the TEM images (Fig. 7(A)–(C)), some dark quasi spherical particles can be observed. The dark particles should be Fe₃O₄ NPs, evidenced by some well-resolved Bragg peaks of XRD shown in Fig. S4A (ESI†), which was indexed to a crystalline ferromagnetic structure according to JCPDS card no. 28-0491. From the TEM images, we also found that the dark magnetite particles were randomly embedded into the channel network of the cubic ordered mesoporous silica. Therefore, these TEM images revealed the coexistence of the arrangement of cubic mesopore channels and quasi spherical Fe₃O₄ NPs. Fig. S4B (ESI†) further displays a magnetization “S” curve of this sample. No obvious hysteresis loop was observed, that is, there were no remarkable remanence and coercivity in the sample, demonstrating the superparamagnetic properties of the magnetic cubic mesoporous silica.

Fig. 7 TEM images of the sample prepared with the initial molar ratio n(Fe₃O₄/SDS/C₁₆mim⁺)/n(TEOS) of 0.075 using C₁₆mim⁺ as the templating agent in the Fe₃O₄/SDS/C₁₆mim⁺ aqueous ferrofluid, and the images were taken along (A) [111], (B) [311], and (C) [110] axes’ directions.

A nitrogen physisorption isotherm (Fig. 8(A)) was recorded to estimate the textural properties of this calcined sample (n(Fe₃O₄/SDS/C₁₆mim⁺)/n(TEOS) of 0.075). The isotherm is of type IV with a distinct hysteresis loop in the relative pressure 0.35–0.70, indicating a narrow pore size distribution in the mesoporous range.³³ The pore size distribution (Fig. 8(B)) is mainly located around 2.3–4 nm, and the average pore size calculated from the BJH model using the desorption branch was determined to be about 2.35 nm, which is consistent with the observation of the TEM images (Fig. 7). The BTE surface area and pore volume for the sample were determined to be 952 m² g⁻¹ and 0.66 cm³ g⁻¹, respectively. It can be found that this BET surface area of the magnetic cubic mesoporous sample is lower than that of the pure cubic mesoporous silica synthesized using C₁₆mim⁺ as the template (∼1250–1300 m² g⁻¹),²² which may be attributed to the introduction of the magnetic particles into the cubic mesoporous silica channel framework.

Fig. 8 (A) N₂ adsorption–desorption isotherm of the sample prepared with the initial molar ratio n(Fe₃O₄/SDS/C₁₆mim⁺)/n(TEOS) of 0.075 using C₁₆mim⁺ as the templating agent in the Fe₃O₄/SDS/C₁₆mim⁺ aqueous ferrofluid, and (B) its BJH pore size distribution calculated from the desorption branch using the BJH method.

In the classical preparation of mesoporous silica materials, surfactant chemistry is the key to the synthesis of solid mesophases.³⁴ It was reported that the synthesis of the cubic mesoporous phase (Ia3d) is more difficult than that of the hexagonal mesoporous phase. This reason may be due to the narrow phase area of the cubic phase in the liquid crystal phase diagram and the relatively harsh synthesis conditions for the synthesis of the cubic phase.³⁵ At the same time, the introduction of the modified Fe₃O₄ NPs would interfere with the synthesis of the composite mesoporous phase and increase the difficulty of the synthesis of the cubic phase. In our early experiments, we found that magnetic Fe₃O₄@MCM-41 with a coexistence structure of a hexagonal mesoporous silica array and Fe₃O₄ NPs can be obtained in a magnetic Fe₃O₄/C₁₆mim⁺/C₁₆mim⁺ fluid.²³ However, in the bilayer C₁₆mim⁺-based magnetic fluid, although we have made many attempts to study the effects of n(Fe₃O₄/C₁₆mim⁺/C₁₆mim⁺)/n(TEOS), n(TEOS)/n(C₁₆mim⁺) and n(TEOS)/n(H₂O) on the formation of the magnetic cubic mesoporous Fe₃O₄@MCM-48, we still cannot obtain Fe₃O₄@MCM-48. In this article, SDS and C₁₆mim⁺ were used as the modifiers for the inner and outer layers to construct modified bilayers onto the Fe₃O₄ particles, respectively. We found that in the Fe₃O₄/SDS/C₁₆mim⁺ ferrofluid, cubic magnetic mesoporous Fe₃O₄@MCM-48 can be formed. The results may be explained by two aspects. One is that the anion headgroup of SDS is smaller than the cationic imidazole headgroup of C₁₆mim⁺. When SDS is used as the modified inner layer of the modified double layer on the surface of Fe₃O₄ particles, smaller Fe₃O₄/SDS/C₁₆mim⁺ particles with higher radii of curvature can be obtained compared with the Fe₃O₄/C₁₆mim⁺/C₁₆mim⁺ particles. The existence of such small particles has little interference on the production of magnetic cubic mesoporous Fe₃O₄@MCM-48. Another reason may be due to the use of rational proportions of the n(Fe₃O₄/SDS/C₁₆mim⁺)/n(TEOS), n(TEOS)/n(C₁₆mim⁺) and n(TEOS)/n(H₂O) in the initial synthesis compositions. Based on our earlier research, we found that in the concentrated initial mixture solution, it is advantageous to synthesize cubic mesoporous silica by using C₁₆mim⁺ as the template.²² Therefore, on the other hand, the addition of the magnetic particles may increase the concentration of the initial mixture component, leading to the formation of the magnetic cubic mesoporous phase. The result also shows that it is possible to prepare the magnetic cubic mesoporous silica by using C₁₆mim⁺ as the template in the aqueous Fe₃O₄/SDS/C₁₆mim⁺ ferrofluid.

3.3 Adsorption properties of magnetic cubic mesoporous Fe₃O₄@MCM-48

To assess the adsorption ability of the synthesized Fe₃O₄@MCM-48, four dyes, methylene blue, rhodamine B, malachite green, and methyl violet, were selected as adsorption models, respectively. Fig. 9 shows the decolorization rates of the four dye solutions (100 mg L⁻¹, 50 mL) after adsorption using the synthesized Fe₃O₄@MCM-48 (0.1 g) as the adsorbent at different time intervals. As shown in Fig. 9, within 25 min, the decolorization rates of the four dye solutions increased rapidly and reached a plateau (∼90%), indicating the strong adsorption capacity of Fe₃O₄@MCM-48. Fig. 10 further shows decolorization rates of malachite green solutions and adsorbed amount of Fe₃O₄@MCM-48. Obviously, when the initial concentration of malachite green solution (50 mL) is less than 100 mg L⁻¹ (Fig. 10(A)), the decolorization rate of the malachite green solution increased faster in the first 5 min, and then the value reached a platform (95%) in the next 15 min, indicating that the adsorption equilibrium was reached. However, when the initial concentrations of the malachite green solutions were increased to 200 or 400 mg L⁻¹, the decolorization rates decreased to 84% or 66% at 15 min, respectively (Fig. 10(A)). These results indicated that there is a saturation value for the adsorption of Fe₃O₄@MCM-48. The adsorption capacity of Fe₃O₄@MCM-48 at equilibrium (30 min) was found to be 11.5, 15.0, 36.5, 85.5 and 161 mg g⁻¹ for the initial malachite green concentrations of 25, 50, 100, 200 and 400 mg L⁻¹, respectively (Fig. 10(B)). If we compare the adsorption capacities of the Fe₃O₄@MCM-48 and pure mesoporous silica MCM-48 prepared by the same C₁₆mim⁺ template at equilibrium (30 min),²² it was found that the adsorption quantity of Fe₃O₄@MCM-48 (161 mg g⁻¹) was lower than pure MCM-48 (376 mg g⁻¹). This phenomenon may be due to the fact that the BET surface area of Fe₃O₄@MCM-48 (952 m² g⁻¹) is lower than that of the pure mesoporous silica MCM-48 (∼1250–1300 m² g⁻¹).²² However, Fe₃O₄@MCM-48 was easily recovered from the adsorbed solution by magnetic separation technology, while the pure cubic mesoporous silica MCM-48 can only be retrieved by centrifugation. Another notable problem was that the bare Fe₃O₄ particles had an adsorption quantity of 10.5 mg g⁻¹, which is lower than that of Fe₃O₄@MCM-48. The reason is probably that Fe₃O₄@MCM-48 had adequate mesoporous channels and highly adsorption capacity compared with the bare Fe₃O₄ particles.

Fig. 9 Decolorization rates of (a) methylene blue, (b) rhodamine B, (c) malachite green, and (d) methyl violet solutions after adsorption using Fe₃O₄/MCM-48 as the adsorbent and magnetic separation at different time intervals. The initial concentrations of the four dye solutions were all 100 mg L⁻¹.

Fig. 10 (A) Decolorization rates of malachite green solutions and (B) adsorbed quantity of Fe₃O₄@MCM-48 at different time intervals. The initial concentrations of malachite green solutions are (a) 25, (b) 50, (c) 100, (d) 200 and (e) 400 mg L⁻¹, respectively.

For rough surfaces, the Freundlich isotherm model is considered as the most important multilayer adsorption model. For the adsorption of malachite green solutions at different initial concentrations (25, 50, 100, 200 and 400 mg L⁻¹), the equilibrium adsorption quantity (q_e) of Fe₃O₄@MCM-48 and the equilibrium concentration (C_e) of malachite green solutions at 30 min can be treated according to the Freundlich isotherm model. A well-defined linear relation was obtained (see Fig. S5, ESI†). From the linear relation, we can calculate that values of K_F and 1/n are 18 mg g⁻¹ and 0.53, respectively. This result (especially n is greater than 1) indicated that the synthesized Fe₃O₄@MCM-48 possessed spacious porous channels and special high surface area, resulting in relatively strong adsorption capacity. Therefore, in some liquid-phase adsorption processes, the synthesized Fe₃O₄@MCM-48 can be potentially used as an adsorbent. The synthesized Fe₃O₄@MCM-48 has generous surface silicon hydroxyl groups with negative charge in the mesoporous channels, which can adsorb positively charged cationic dyes, for example, methylene blue, malachite green, rhodamine B and methyl violet. Therefore, we believe that the physical adsorption caused by the electrostatic interaction between the cationic dyes and the silicon hydroxyls of negatively charged mesoporous channels may be the main reason, that is, the adsorption is more beneficial to cationic dyes.

If a small laboratory magnet was placed next to the small beaker containing the mixture, the dye-adsorbed Fe₃O₄@MCM-48 was recovered simply by decanting clear solution or removing clear solution using a pipette. From one side, this simple experiment also showed that Fe₃O₄@MCM-48 is magnetic. In other words, Fe₃O₄@MCM-48 can be considered as a magnetic adsorbent to remove some positively charged cationic dyes from the corresponding liquid-phase. We can also find that after the adsorption process, Fe₃O₄@MCM-48 was recovered by magnetic separation, dried at room temperature and calcined at 823K for 2 h to remove the adsorbed dyes. The reclaimed Fe₃O₄@MCM-48 can be used as an adsorbent again. It was found that after three successive cycle uses of Fe₃O₄@MCM-48, the decolorization rates of malachite green solutions decreased to 94, 91, and 85 mg L⁻¹, respectively. It is worth noting that the decolorization rate was decreased to about 74% after the fourth cycle, which suggests that its adsorption performance was reduced. This result may be attributed to the partial collapse of the cubic mesoporous channels of Fe₃O₄@MCM-48 due to the multiple recycled adsorptions and calcinations.

4. Conclusions

Stable aqueous ferrofluids were first synthesized by using bilayer SDS/ionic liquid (alkyl imidazolium ionic liquid C_nmim⁺, n = 10, 12, 14, and 16, respectively) as modifiers onto Fe₃O₄ NPs. It was found that the long chain ionic liquid (such as C₁₆mim⁺) is superior to the short chain ionic liquid (such as C₁₀mim⁺) in the construction of bilayer modifiers. The SDS/C₁₆mim⁺-based Fe₃O₄ particles can offer the hydrophilic and positively charged outer surface, resulting in the formation of a stable aqueous ferrofluid. Furthermore, in the aqueous SDS/C₁₆mim⁺-based ferrofluid, the magnetic cubic mesoporous silica framework (Fe₃O₄@MCM-48) can be synthesized by using C₁₆mim⁺ as the template again. The synthesized Fe₃O₄@MCM-48 with a specific surface area of up to 952 m² g⁻¹ showed high decolorization rates of up to ∼90% in 25 min for four dye solutions (methylene blue, rhodamine B, malachite green, and methyl violet solutions). In addition, it was found that the malachite green-adsorbed Fe₃O₄@MCM-48 can be separated and recovered from the corresponding solution under an external magnetic field. It was found that for successive three cycles, Fe₃O₄@MCM-48 enabled the decolorization rates of the malachite green solution to reach 94, 91, and 85 mg L⁻¹, respectively. The application of C₁₆mim⁺ as the modifier of the outer layer of Fe₃O₄/SDS/C₁₆mim⁺ particles and cubic mesoporous templating agent in turn may provide some information for the multifunctional applications of C₁₆mim⁺ in the synthesis of various nanomaterials.

Author contributions

Jing Shen: resources, funding acquisition, data analysis, and writing-original draft. Dongxiu Wang: investigation, data analysis and resources. Li Yang: investigation and resources. Tong Wang: funding acquisition. Qianmo Yang: investigation. Jun Wang: funding acquisition, writing – review and editing, and supervision.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

The authors are grateful to the National Natural Science Foundation of China (No. 21564018, 21063017, 21363029 and 22165034).

Notes and references

1. B. M. Berkovsky, V. F. Medvedev and M. S. Karkov, Magnetic Fluids: Engineering Applications, Oxford University Press, New York, 1993.
2. J. Shimoiizaka, K. Nakatsuka, T. Fujita and A. Kounosu, IEEE Trans. Magn., 1980, 16, 368–371.
3. A. Wooding, M. Kilner and D. B. Lambrick, J. Colloid Interface Sci., 1991, 144, 236–242.
4. L. F. Shen, P. E. Laibinis and T. A. Hatton, Langmuir, 1999, 15, 447–453.
5. R. Y. Hong, S. Z. Zhang, Y. P. Han, H. Z. Li, J. Ding and Y. Zheng, Powder Technol., 2006, 170, 1–11.
6. B. Bateer, Y. Qu, X. Meng, C. Tian, S. Du, R. Wang, K. Pan and H. Fu, J. Magn. Magn. Mater., 2013, 332, 151–156.
7. M. Soleymani and M. Edrissi, J. Dispersion Sci. Technol., 2016, 37, 693–698.
8. B. Xin and J. Hao, Chem. Soc. Rev., 2014, 43, 7171–7187.
9. X. Zhao, L. Guo, T. Xu, H. Wang, R. Zheng and Z. Jiang, New J. Chem., 2022, 46, 15901–15910.
10. Y. Tian, C. Xing, W. Wang, S. Zhang and Y. Zhang, New J. Chem., 2022, 46, 8855–8862.
11. M. Blesic, M. H. Marques, N. V. Plechkova, K. R. Seddon, L. P. N. Rebelo and A. Lopes, Green Chem., 2007, 9, 481–490.
12. J. Łuczak, J. Hupka, J. Thöming and C. Jungnickel, Colloids Surf., A, 2008, 329, 125–133.
13. H. Cao, Y. Hu, W. Xu, Y. Wang and X. Guo, J. Mol. Liq., 2020, 319, 114354.
14. B. Dong, N. Li, L. Zheng, L. Yu and T. Inoue, Langmuir, 2007, 23, 4178–4182.
15. T. Inoue, H. Ebina, B. Dong and L. Zheng, J. Colloid Interface Sci., 2007, 314, 236–241.
16. B. Dong, X. Zhao, L. Zheng, J. Zhang, N. Li and T. Inoue, Colloids Surf., A, 2008, 317, 666–672.
17. T. Bleasdale, G. Tiddy and E. Wyn-Jones, J. Phys. Chem., 1991, 14, 5385–5386.
18. F. Neve, O. Francescangeli and A. Crispini, Inorg. Chim. Acta, 2002, 338, 51–58.
19. M. Antonietti, D. B. Kuang, B. M. Smarsly and Z. Yong, Angew. Chem., Int. Ed., 2004, 43, 4988–4992.
20. C. J. Adams, A. E. Bradley and K. R. Seddon, Aust. J. Chem., 2001, 54, 679–681.
21. Y. Zhou and M. Antonietti, Adv. Mater., 2003, 15, 1452–1455.
22. T. Wang, H. Kaper, M. Antonietti and B. M. Smarsly, Langmuir, 2007, 23, 1489–1495.
23. J. Shen, W. He and T. Wang, RSC Adv., 2019, 9, 3504–3513.
24. R. Massart and V. Cabuil, J. Chem. Phys., 1987, 84, 967–973.
25. S. E. Khalafalla and G. W. Reimers, IEEE Trans. Magn., 1980, 16, 178–183.
26. E. Dubois, J. Chevalet and R. Massart, J. Mol. Liq., 1999, 83, 243–254.
27. J. Wang, Q. Chen, C. Zeng and B. Hou, Adv. Mater., 2004, 16, 137–140.
28. D. Wang, C. Cao, S. Xue and H. Zhu, J. Cryst. Growth, 2005, 277, 238–245.
29. Q. Huo, D. I. Margolese and G. D. Stucky, Chem. Mater., 1996, 8, 1147–1160.
30. C. K. Krishnan, T. Hayashi and M. Ogura, Adv. Mater., 2008, 20, 2131–2136.
31. V. Alfredsson and M. W. Anderson, Chem. Mater., 1996, 8, 1141–1146.
32. A. A. Romero, M. D. Alba, W. Zhou and J. Klinowski, J. Phys. Chem., 1997, 101, 5294–5300.
33. K. S. W. Sing, D. H. Everett, R. A. W. Haul, L. Moscou, R. A. Pierotti, J. Rouquérol and T. Siemieniewska, Pure Appl. Chem., 1985, 57, 603–619.
34. J. N. Israelachvili, D. J. Mitchell and B. W. Ninham, J. Chem. Soc., Faraday Trans., 1976, 272, 1525–1568.
35. J. C. Vartuli, K. D. Schmitt, C. T. Kresge, W. J. Roth, M. E. Leonowicz, S. B. McCullen, S. D. Hellring, J. S. Beck, J. L. Schlenker, D. H. Olson and E. W. Sheppard, Chem. Mater., 1994, 6, 2317–2326.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d2nj04958a

---