Synthesis and Characterization of Hematite, Magnetite and Maghemite Supported on Silica Gel

Abstract

A new method for obtaining nanosized particles of iron oxides using porous silica gel is proposed. In situ magnetometry was used to study the reduction of hematite deposited on silica gel during the thermolysis of glucose. The formed magnetite and maghemite obtained by subsequent oxidation of the magnetite were studied using X-ray diffraction and Mossbauer spectroscopy. It was shown that both the size of the oxide particles and the phase composition significantly depended on the porous structure of the silica gel. In particular, the formation of superparamagnetic maghemite particles on silica gels with pore sizes of 30, 15 and 10 nm was demonstrated.

1. Introduction

Materials containing nanosized particles of iron oxides are widely used in various fields, from catalysis to medicine [1,2,3,4]. First of all, nanosized particles of iron-containing substances are of interest for fundamental scientific research due to the differences in the physical properties of bulk and nanosized materials [5,6]. For example, synthetic nanosized magnetite has been proposed as a potential reactant for the purification of polluted groundwater and as an effective sorbent of metalloid pollutants due to its high surface area [7,8].

The targeted synthesis of preparations containing pure single-phase nanosized particles of iron with a monodisperse size distribution encounters significant difficulties due to the diversity of iron chemistry, which manifests itself in the stability of two oxidation states and rich crystal chemistry. All this causes problems with the coexistence of oxides of different compositions in one preparation, for example, such as magnetite (Fe₃O₄), maghemite (γ-Fe₂O₃), hematite (α-Fe₂O₃) and wustite (FeO) [9,10]. This problem is exacerbated by the tendency of nanosized iron oxide particles to aggregate into large agglomerates. This is especially important when working with magnetic particles such as Fe₃O₄ and γ-Fe₂O₃ [11]. To prevent particle aggregation, synthesis methods are used in carrier matrices, in particular, on silica gel. Silica was shown to be the best support in terms of both activity and selectivity in various catalytic reactions and, in particular, in Fisher–Tropsch synthesis [12,13,14,15,16].

Most often, to obtain such nanosized iron-containing preparations in a silica gel matrix, the sol–gel synthesis technique is used [9]. However, as a rule, as a result of using the sol–gel method, the synthesized nanoparticles of iron oxides are encapsulated in the bulk of the silica gel. In this case, the surfaces of the iron oxide nanoparticles are inevitably covered with silica shells and have limited access to the adsorbate. Such composites are not effective for catalysis due to the limited number of active sites accessible to reagents on the surface. In addition, it is necessary to take into account the possibility of the interaction of iron oxides with the silica gel material, especially at low iron concentrations [5].

Nanoparticles of iron oxides on silica gel can be prepared by impregnating the carrier with a solution of iron nitrate followed by thermolysis of the nitrate. This method is often used to prepare iron-supported catalysts [1,7]. Sonochemical synthesis of silica-supported iron oxide nanostructures was used by the authors of one work [17]. Preparation of magnetite on silica is a more difficult task, since the direct reduction of supported hematite inevitably leads to a mixture of oxides: Fe₂O₃, Fe₃O₄, FeO. To reduce hematite, it is possible to use hydrocarbons, as well as various carbohydrates. In this case, it is possible to channel the reaction towards the exclusive formation of magnetite [18]. The synthesis of ultrafine particles of magnetite using polysaccharides as a reducing agent was proposed in [19]. Silica-supported maghemite was prepared through the pyrolysis of a ternary composite of poly (vinyl alcohol) (PVA), iron (III) hydroxide and silica gel, and was characterized using X-ray diffraction and infrared spectroscopic and BET surface area measurements [20,21]. The synthesis of γ-Fe₂O₃ particles and the investigation of their properties are of fundamental importance for the development of science and technology. The preparation of pure γ-Fe₂O₃ nanophase presents some difficulties arising from the different metal oxidation states, which can lead to the presence of various oxides. The method of controlled oxidation of magnetite is usually used to obtain magnetite [11].

The aim of the present study is to demonstrate a technique for the selective synthesis of magnetite nanoparticles by means of the reduction of hematite supported on silica gel with pores of various sizes using glucose pyrolysis products, which makes it possible to synthesize nanosized maghemite particles of various specified sizes. The physicochemical properties of all of the synthesized composites were confirmed using Mossbauer spectroscopy, X-ray phase analysis and magnetic methods. Since the composites synthesized according to the method proposed in this work are supposed to be used as catalysts in the future, we refer to them as catalysts in what follows.

2. Materials and Methods

2.1. Materials

Iron (III) nitrate nonahydrate (Sigma-Aldrich, St Louis, MO, USA) (Fe(NO₃)₃·9H₂O) was used as a precursor. CARIACT silica gel with grades of Q-30, Q-15 and Q-10 (Q-X, where X = 30, 15, 10 nm as the average pore size in silica gel, with surface areas of 100, 200 and 300 m²/g, respectively) manufactured by Fuji Silysia Chemical, Ltd. (Nagoya, Japan) was used as a support. The nominal content of iron was 20 mass% for Q-30 and 15 mass% for Q-15 and Q-10.

Catalysts were prepared via incipient wetness impregnation of the support with aqueous solutions of hydrous iron nitrate. After impregnation, the catalysts were dried in a rotary evaporator at 80 °C for 2 h. Then, they were calcined in air at 450 °C for 4 h with temperature ramping of 10 °C/min. These samples were denoted as Fe₂O₃/Q-30, Fe₂O₃/Q-15 and Fe₂O₃/Q-10.

Obtained Fe₂O₃/Q samples (calcined at 450 °C) were additionally impregnated with an aqueous glucose solution at room temperature and dried in a rotary evaporator at 80 °C. The iron/glucose molar ratio was 10/1.

These samples were denoted as G/Fe₂O₃/Q-30, G/Fe₂O₃/Q-15 and G/Fe₂O₃/Q-10.

Subsequently, some of the catalysts were studied using the in situ magnetic method, and some were subjected to calcination in an argon current at a heating rate of 10 °C/min to a temperature of 450 °C for 4 h. The catalysts obtained from calcination in argon were designated as Fe₃O₄/Q-30, Fe₃O₄/Q-15 and Fe₃O₄/Q-10. Some of the catalysts obtained in this way were subjected to oxidation in an air current at 450 °C. The resulting oxidation catalysts were designated as γ-Fe₂O₃/Q-30, γ-Fe₂O₃/Q-15 and γ-Fe₂O₃/Q-10. A block diagram of the sequence of operations in the synthesis of catalysts is presented below.

Fe(NO₃)₃·9H₂O + SiO₂ → (80 °C)

→ Fe(NO₃)₃/SiO₂ → (450 °C air)

→ α−Fe₂O₃/SiO₂ → (C₆H₁₂O₆ 80 °C)

→ G/α−Fe₂O₃/SiO₂ → (450 °C Ar)

→ Fe₃O₄/SiO₂ → (450 °C air) →

γ−Fe₂O₃/SiO₂

2.2. X-ray Diffraction Study

The X-ray diffraction experiment was performed on a PANalytical Empyrean diffractometer. The immobile sample, in the form of a powder layer of 1 mm thick and about 20 mm in diameter on a non-reflective holder, was placed in the center of the vertical plane of the goniometer. The experiment was carried out in “Θ–Θ” geometry with synchronous movement of the PIXcel-3D detector and the ceramic X-ray tube along the goniometric circle. Diffraction patterns were measured on Cu Kα radiation in the scanning range for angle 2Θ of 5–100 degrees.

Mathematical processing of diffraction patterns, quantitative phase analysis of the samples (according to the Chung method) and particle size estimations using the Scherer method were carried out using the HighScore Plus program (version 4.9) and the JSCD file cabinet.

2.3. Mossbauer Spectroscopy

Mossbauer absorption spectra were obtained on an MS1104EM Express Mossbauer spectrometer (Cordon GmbH, Rostov-on-Don, Russia) in the constant acceleration mode with the triangular velocity reference signal. The 57Co nuclei in a metal rhodium matrix (RITVERC GmbH, St. Petersburg, Russia) with 50 mCi activity were used as the γ-radiation source. The spectra were obtained at room temperature (23 ± 3 °C). Mathematical processing was carried out for spectra using UNIVEM (Rostov-on-Don State University, Russia) software with a high resolution (1024 points). Experimental data were fitted using a superposition of symmetrical doublets and sextets with fixed line widths and intensity ratios. The chemical shifts were determined relative to α-Fe.

2.4. Magnetic Studies

Magnetic studies of the samples were carried out on a vibration magnetometer using the in situ regime. The test sample (10 mg) was placed in a measuring cell of the vibrating magnetometer [22], which is a flow-through microreactor with an internal volume of 0.3 cm³. The test sample was clamped between two membranes of porous quartz. In a nonisothermal experiment, the sample was heated (10 °C/min) to a predetermined temperature in the Ar or air at a magnetic field strength of 3 kOe with a flow rate of 10 cm³/min, while a change in magnetization was recorded continuously at a frequency of 1 Hz.

3. Results

Catalysts Fe₂O₃/Q-30, Fe₂O₃/Q-15 and Fe₂O₃/Q-10 did not reveal any changes in magnetization in an argon current up to the temperature of T = 600 °C. However, the exposure of catalysts G/Fe₂O₃/Q-30, G/Fe₂O₃/Q-15 and G/Fe₂O₃/Q-10 under identical conditions showed an increase in magnetization starting from 270 °C. Figure 1 shows the dependence of magnetization on temperature for the catalysts during heating in argon. The growth of magnetization is due to pyrolysis of glucose. During pyrolysis, the emitted CO reduces hematite particles to magnetite through the following reaction:

3Fe₂O₃ + CO = 2Fe₃O₄ + CO₂

At temperatures above 500 °C, glucose is completely transformed into a mixture of CO, CO₂ and H₂O.

The drop in magnetization to zero at T = 570–580 °C corresponds to the Curie temperature of magnetite. After cooling, the dependences of magnetization on the field were measured, and the results are shown in Figure 2.

Note the absence of remanent magnetization and coercive force in the case of catalysts Fe₃O₄/Q-15 and Fe₃O₄/Q-10. This indicates the presence of superparamagnetism in these catalysts at room T. The presence of superparamagnetic particles allowed us to estimate the size of the particles using the Langevin equation. In this way, it was possible to estimate the upper and lower limits of Fe₃O₄ particle size, d, using the low-field (d_LF) and high-field field (d_HF) approximations of the Langevin function [23], as follows:

$$d_{LF}^3=\frac{18k}{\pi Is}\left(\frac{\sigma }{{\sigma }_s}\right)/\left(\frac{H}{T_m}\right)$$

(1)

$$d_{HF}^3=\frac{6k}{\pi I_s}{\left[1-\left(\frac{\sigma }{{\sigma }_s}\right)\right]}^{-1}/\left(\frac{H}{T_m}\right)$$

(2)

where k is the Boltzmann constant and I_s is the spontaneous magnetization per unit volume for the bulk Fe₃O₄ saturation magnetization of the test sample at 93 K.

The saturation magnetization (I_s) was estimated by plotting emu/g versus l/H and extrapolating to zero, as shown in Figure 3.

The saturation magnetization of bulk magnetite (I_s) was assumed to be 70 emu/g [24,25,26,27].

Table 1. The upper and lower limits of Fe₃O₄ particle size (d), estimated for each case.

Medium, nm	d (max)	d (min)	Catalysts
5.7	7 nm	4.5 nm	Fe3O4/Q-15
5.4	7 nm	3.8 nm	Fe3O4/Q-10

shows the values of max. and min. particle size that were calculated using equations 1 and 2, as well as the average size of catalysts Fe₃O₄/Q-15 and Fe₃O₄/Q-10.

After oxidation in air at 450 °C, catalysts Fe₃O₄/Q-30, Fe₃O₄/Q-15 and Fe₃O₄/Q-10 acquired a red-orange color and remained magnetized at room temperature. We assume that magnetite oxidation results in the formation of maghemite.

If the system contains no multidomain particles, the particle size distribution can be analyzed using the following equation [23]:

γ = 1 − 2Jr/Js

(3)

where γ is the proportion of superparamagnetic particles and Jr is the saturation remanent magnetization.

According to the XRD data (given below), the maghemite particles in the catalysts were smaller than the single-domain size of 40 nm [25].

More detailed information about the composition of the systems under study was provided with X-ray phase analysis. The results of the X-ray phase analysis for all of the investigated objects are given below.

The X-ray diffraction patterns of all the samples at the first stage of synthesis (Figure 4, Figure 5 and Figure 6) were a superposition of reflections from two phases: a single, very wide and intense reflex of SiO₂ gel (111) and a system of lines of nanosized α-Fe₂O₃ phase.

On the diffraction patterns obtained, at the stages of transition from hematite to magnetite and, further, to maghemite, the intensity of the reflections of the forming Fe₃O₄ and γ-Fe₂O₃ phases increased, and the reflections themselves became narrower in comparison with the reflections of γ-Fe₂O₃. This indicates an increase in the size of the obtained Fe₃O₄ and γ-Fe₂O₃ nanoparticles within the pore size throughout synthesis.

The mean particle sizes in the silica gel pores that were determined using the Scherrer method are given in

Table 2. Iron oxide particle sizes found from Scherrer reflex width analysis.

Particle Size, nm			Catalysts
γ-Fe2O3	Fe3O4	α-Fe2O3
		22	α-Fe2O3/Q-30
		13	α-Fe2O3/Q-15
		5	α-Fe2O3/Q-10
	24		Fe3O4/Q-30
	14		Fe3O4/Q-15
	7		Fe3O4/Q-10
25			γ-Fe2O3/Q-30
14			γ-Fe2O3/Q-15
8			γ-Fe2O3/Q-10

. The average particle sizes given in the table agree well with the size estimates obtained from magnetic measurements. The exception is catalyst γ-Fe₂O₃/Q-30, because the particle size exceeded single-domain range and could not be determined using the magnetic method.

The phase transformations of nanoparticles occurring at all the stages of synthesis were investigated in more detail using Mossbauer spectroscopy.

To estimate the sizes of the obtained nanoparticles, spectra were measured at room temperature. Mathematical analysis of Mossbauer spectra of the samples obtained at all stages of synthesis allowed us to give a qualitative picture of the phase transformations taking place, to estimate the sizes of the obtained nanoparticles of iron oxides and to determine their quantitative characteristics. Particle size distributions for all investigated samples were also obtained from the mathematical processing of experimental Mossbauer spectra.

All the calculated parameters of the Mossbauer spectra are given in

Table 3. Mossbauer spectra parameters (T = 300 K).

S, %	Heff, kOe	Γexp, mm/s	Δ, mm/s	δ, mm/s	Phase	№	Sample
48		0.45	0.65	0.35	α-Fe2O3 d~6 nm	1	α-Fe2O3/Q-10
52		0.65	0.94	0.33	α-FeOOH(Si) d~2 nm	2
29	400	2	0	0.27	Fe3O4 d~9 nm	1	Fe3O4/Q-10
55		0.62	0.76	0.35	Fe3O4 d~7 nm	2
15		0.68	1.14	0.31	γ-Fe2O3 d~5 nm	3
43	452	2	0	0.27	γ-Fe2O3 d~9 nm	1	γ-Fe2O3/Q-10
44		0.65	0.76	0.34	γ-Fe2O3 d~7 nm	2
13		0.68	1.1	0.31	γ-Fe2O3 d~5 nm	3
76		0.47	0.69	0.34	α-Fe2O3 d~11 nm	1	α-Fe2O3/Q-15
24		0.67	1.1	0.3	α-FeOOH(Si) d~2 nm	2
33	440	1.2	0.5	0.35	Fe3O4 d~14 nm	1	Fe3O4/Q-15
51		0.76	0.92	0.37	Fe3O4 d~8 nm	2
16		0.6	2.8	1.14	γ-Fe2SiO4 d~9 nm	3
59	451	2	0	0.27	γ-Fe2O3 d~14 nm	1	γ-Fe2O3/Q-15
33		0.78	0.84	0.34	γ-Fe2O3 d~11 nm	2
8		0.68	1.1	0.31	γ-Fe2O3 d~8 nm	3
53	507	0.36	−0.22	0.37	α-Fe2O3 d~17 nm	1	α-Fe2O3/Q-30
29		0.45	0.66	0.34	α-Fe2O3 d~12 nm	2
18		0.68	1.1	0.31	α-FeOOH(Si) d~2 nm	3
49	452	0.6	0.02	0.64	Fe3O4 d~29 nm	1	Fe3O4/Q-30
	483	0.46	0	0.26	Fe3O4 d~29 nm
13	420	1	−0.03	0.45	Fe3O4 d~15 nm	2
17		0.91	1.06	0.36	Fe3O4 d~8 nm	3
21	485	0.36	0	0.32	γ-Fe2O3 d~27 nm	4
33	497	0.45	−0.02	0.34	γ-Fe2O3 d~29 nm	1	γ-Fe2O3/Q-30
21	478	0.45	0	0.3	γ-Fe2O3 d~20 nm	2
36	439	1.5	0	0.46	γ-Fe2O3 d~15 nm	3
10		0.6	0.85	0.36	γ-Fe2O3 d~9 nm	4

.

The Mossbauer spectrum of the α-Fe₂O₃/Q10 sample that was obtained at room temperature was a broadened paramagnetic doublet, characteristic of superparamagnetic compounds (Figure 7a).

Mathematical decomposition of the spectrum in Figure 7a revealed that it consisted of two components: a doublet of superparamagnetic α-Fe₂O₃ with particle sizes smaller than 10 nm [28,29] and a doublet of superparamagnetic α-FeOOH with particle sizes smaller than 5 nm [29]. To more precisely define both compounds’ particle sizes, we measured the α-Fe₂O₃/Q10 sample at 77.5 K. The α-FeOOH component of the spectrum was still a superparamagnetic doublet at this temperature, which indicates the very small sizes of its nanoparticles (2–3 nm). The α-Fe₂O₃ component showed hyperfine magnetic splitting, corresponding to nanoparticles of 5–6 nm.

α-Fe₂O₃ nanoparticles of 5 nm and less in size have a number of atoms on the surface comparable to their number in volume. The reactivity of the surface atoms is thus so high that the atoms of the Fe-O chain at the particle surface interact with the OH-Si groups at the SiO₂-gel surface. As a result, nanolayers of goethite doped with silicon are formed on the surface of iron oxide nanoparticles, forming α-FeOOH(Si). The intensities of the α-FeOOH(Si) and α-Fe₂O₃ subspectra were 52 and 48%, respectively, which is consistent with the comparable number of atoms on the surface and in the volume of nanoparticles of ~5 nm and less.

At the second stage of the synthesis of Fe₃O₄/Q-10 nanosized particles, according to the X-ray data, the average particle size increased to 7 nm. In the Mossbauer spectrum in Figure 7b, this increase was manifested in the appearance of the Fe₃O₄ component with a hyperfine magnetic field of the relaxation type (H = 400 kOe). The superparamagnetic doublet of the Fe₃O₄ oxide formed the largest part of the spectrum (55%). In addition, the superparamagnetic component attributed to γ-Fe₂O₃ (15%) appeared. Apparently, it was due to the transformation of part of the α-FeOOH(Si) nanoparticles into γ-Fe₂O₃, which occurs during thermolysis of glucose at temperatures above 400 °C [30]. The peak on the thermomagnetic curve of the G/Fe₂O₃/Q-10 sample at 520 °C (Figure 1), as well as the Curie temperature shift beyond 600 °C, was due to this transformation. At the third stage of the synthesis of γ-Fe₂O₃/Q-10 nanoparticles, the average size of the formed γ-Fe₂O₃ nanosized particles further increased to 8 nm (Table 2). Analysis of the Mossbauer spectrum parameters showed that only γ-Fe₂O₃ nanoparticles with a narrow size distribution of 6–9 nm were formed in SiO₂/Q-10 pores (Figure 8c).

In the SiO₂/Q-15 pores at all stages of γ-Fe₂O₃/Q15 synthesis, the nanoparticle sizes remained smaller than the pore size (Table 2). The sample formed at the first stage contained 76% α-Fe2O3 with particles of ~11 nm in size and 24% α-FeOOH(Si) with particles of ~2 nm in size. The Mossbauer spectra of catalysts α-Fe₂O₃/Q15 (a), Fe₃O₄/Q-15 (b) and γ-Fe₂O₃/Q-15 (c) are shown in Figure 9.

There were some features of phase transformations in the process of synthesis, however. The amount of α-FeOOH phase formed at the first stage of synthesis was two times less than in the case of α-Fe₂O₃/Q10 due to an increase in the average particle size to 7 nm and a reduction in the pore curvature. At the second stage, Fe₃O₄ particles were formed with most-probable sizes of around 8 and 14 nm. The contributions of these phases were 51 and 33%, respectively. Larger magnetite particles appeared in the spectrum as a relaxation-type component with a hyperfine magnetic field (H = 440 kOe). In addition, the second stage of Fe₃O₄/Q15 synthesis produced γ-Fe₂SiO₄ nanoparticles [31]. At the third stage of synthesis, only γ-Fe₂O₃ nanoparticles with a size distribution of 8–14 nm were formed in the Q-15 pores (Figure 10c). The main part was γ-Fe2O3 particles with a size of ~14 nm, and their share was 59%. The proportions of particles with sizes of ~11 nm and ~8 nm were 33 and 8%, respectively. We can see that for γ-Fe2O3/Q-15, the relative abundance of particles increased in direct proportion to their most probable size.

A large number of antiferromagnetic α-Fe₂O₃ nanosized particles were initially formed in the SiO₂/Q-30 pores (Figure 11a), whose sizes approached the size of the single-domain state (the effective magnetic field parameter in the Mossbauer spectrum was 507 kOe). In addition, a number of smaller α-Fe₂O₃ nanoparticles were formed. About 20% of the spectrum was α-FeOOH(Si) phase, which was 2–3 nm in size. It should be noted that the Mossbauer parameters of the α–FeOOH(Si) phase formed at the first stage of synthesis were identical for all pore sizes. This is proof of the formation of an identical thin shell around the α-Fe₂O₃ particles in all the pores. It was this phase that was transformed into γ-Fe₂O₃ at the second stage of synthesis (Figure 11b and Figure 12b). All α-Fe₂O₃ nanoparticles turned into Fe₃O₄. At the third stage of synthesis (Figure 11c), all the nanoparticles were transformed into γ-Fe₂O₃ with a broad nanoparticle size distribution of 9–29 nm (Figure 12c).

The average sizes of the nanoparticles of the obtained catalysts, estimated from the size distributions of the nanoparticles (Figure 12), were d = 7 nm for catalyst γ-Fe₂O₃/Q-10, d = 11 nm for γ-Fe₂O₃/Q-15, d = 25 nm for γ-Fe₂O₃/Q-30.

The dependences of magnetization on the magnetic field and pore sizes of catalysts γ-Fe₂O₃/Q-30, γ-Fe₂O₃/Q-15 and γ-Fe₂O₃/Q-10 were investigated by obtaining magnetization curves (Figure 13).

When a magnetic field is applied, particles with different relaxation times of magnetic moment (i.e., with different sizes) reach saturation magnetization at different values of the applied field, which leads to features in the γ-Fe₂O₃ magnetization curve.

The features of the curve are most clearly seen on the line related to γ-Fe₂O₃/Q-30, where the particle size distribution is the widest.

4. Conclusions

Based on the results of the present study, it can be concluded that the proposed chemical method for the synthesis of nanosized particles of iron oxides based on thermal reduction/oxidation in the presence of glucose is effective and promising for the synthesis of nanocatalysts with controlled particle sizes and their composition. Obviously, the carbon monoxide formed during glucose thermolysis contributes to the reduction of nanosized hematite particles on the surface of porous silica gel with the formation of nanosized magnetite particles. The subsequent oxidation of magnetite nanoparticles supported on silica gel for all catalysts leads to the formation of a single-phase system of nanosized maghemite particles. It has been shown that both the phase composition and particle size of iron oxides formed at all stages of catalyst synthesis depend on the pore sizes in the silica gel structure.

γ-Fe₂O₃/Q nanocatalysts were obtained in pores of all sizes, while the sizes of the resulting nanoparticles were limited by the pore size. As the pore size increased, the size distribution of the nanoparticles expanded, which corresponds to a wider range of relaxation times of the superparamagnetic nanosized particles’ magnetic moments, which, in turn, affected the values of the magnetization of the obtained nanosized catalysts (Figure 13). We hope to more carefully study these magnetization curves after removing the temperature dependences of all the obtained nanocatalysts’ Mossbauer spectra in the temperature range of 77.5–300 K. It will allow us to determine the relaxation times of magnetic moments for nanocatalyst particles of different sizes and the temperatures that block their magnetic state. The values of these parameters will make it possible to explain all the features of the magnetization curves.