Modified Fe₃O₄ Magnetite Core@Shell Type Nanomaterials for Highly-Responsive LPG Sensing: A Comparative Analysis

The bare Fe₃O₄ magnetic nanoparticles (MNPs) were prepared by the coprecipitation of Fe (II, III) salts in the base solution. Briefly, 29.15 g of FeCl₃6H₂O and 10.7 g of FeCl₂4H₂O were dissolved in 1 L deionized H₂O, then added to 190 ml of 10% solution of NaOH (pH = 10) at 60 °C with vigorous stirring at 600 rpm. The resulting Fe₃O₄ nanoparticles were washed three times with deionized water to remove impurities from the synthesis. The molar ratio of Fe(III) and Fe(II) salts is equal to 2:1. This ratio corresponds to the ratio of Fe(III): Fe(II) ions in the magnetite Fe₃O₄. According to the stoichiometry, the amount of sodium hydroxide is 0.4 mol.

$$\begin{eqnarray*}&&2{{\rm{FeCl}}}_{3}+{{\rm{FeCl}}}_{2}+8{\rm{NaOH}}\to {{\rm{Fe}}}_{3}{{\rm{O}}}_{4}+8{\rm{NaCl}}+4{{\rm{H}}}_{2}{\rm{O}}\end{eqnarray*}$$

To carry out the reaction, a 1.25 molar excess of NaOH is taken in order to increase the quantitative yield of the reaction.

A 1:4 suspension of activated carbon (Fe₃O₄:AC, where AC is activated carbon) was added to the precipitate of the resulting Fe₃O₄ nanoparticles. The resulting suspension was shaken on a rotary shaker (200 rpm) at 25 °C for 24 h (Fig. 1c).

Zoom In Zoom Out Reset image size

9 g of Fe₃O₄, 57.6 ml of TEOS, and 14.4 ml of APTES were mixed in 1 L of distilled H₂O and the mixture was shaken on a stirrer (1000 rpm, 60 min), pH = 10. The solution was washed with distilled water (pH = 7) and centrifuged (15 min 3000 rpm). The sample was dried in an oven (T = 50 °С, 24 h) (Fig. 1a).

9 g of Fe₃O₄, 57.6 ml of TEOS and 14.4 ml of APTES, and 0.3 g of HA were mixed in 1 l of distilled H₂O, 1 g of HA was added and shaken on a stirrer (1000 rpm, 45 min). The solution was washed with distilled water (pH = 7) and centrifuged (15 min 3000 rpm). The sample was dried in an oven (T = 50 °С, 24 h) (Fig. 1b).

The phase composition and primary particle size of the samples were determined by X-ray phase analysis (XPA) on a Philips X-pert diffractometer (Philips Analytical, Eindhoven, The Netherlands, CrKα radiation, λ = 2.29106 Å). The experimental data were smoothed with the well-known algorithm by A. Savitzky and M.J.E. Golay [described in Ref. 37, corrected by Steinier, Termonia, and Deltour in Ref. 38 was used. The measurements were performed at room temperature in the angular range of 10° < 2Θ < 110° with 0.025° resolution and exposure 1 s.

Quantitative analysis was performed by refinement of the total multiphase spectrum method (the Rietveld method) with a fundamental parameters approach, ³⁹ using the Math!3 software. A fixed Kα emission profile of seven lines was adopted plus an intensity-refined Kβ emission profile. Particle size was determined using the Scherrer equation using full width at half maximum (FWHM) ⁴⁰ using Match!3 software. The surface morphology of all three samples was analyzed by scanning electron microscopy (JEOL JSM-6490 LV).

The diffraction patterns of the samples MTA and MTAH contain six intense and broadened diffraction reflections, slightly differing in position and varying greatly in width and height (Fig. 2) with Miller indexes: (202, ∼45°), (311, ∼54°), (400, ∼66°), (422, ∼84°), (511, ∼90°), (404, ∼101°). The diffractogram of the magnetite-activated carbon sample contains the same peaks that differ significantly in width from the samples of magnetite-silica and magnetite-silica in the humic acid shell. In this case, the MAC sample contains only two broad groups of peaks: (311, ∼54.1°) and (440, ∼101.8°), which probably indicates a strong oxidation of magnetite to another species of iron oxide—maghemite during functionalization or its low concentration.

Zoom In Zoom Out Reset image size

The quantitative analysis by the Rietveld method shows all diffractograms can be interpreted as a face-centred cubic (Fd3m) lattice. The lattice parameter a for the magnetic nanoparticles has been calculated. Table I shows the unit cell parameters of each sample, the compounds, and χ², which is used as a Rietveld error indices and can be determined from the expected and weighted profile R factors χ² = (R_WP/R_exp)² ratio. ⁴¹ According to, ⁴¹ the value of χ² close to 1 is suggested to be satisfactory. The fit for all samples is excellent, both from the χ² and from visual inspection of measured and calculated patterns, and the residue (Fig. 1).

The preparation and functionalization of magnetite NPs could lead to their partial oxidation, which can be determined by comparing the diffraction patterns for the obtained samples with the reference ones. The lattice parameters for modified magnetite nanoparticles are smaller than the same for bare magnetite [JCPDS-ICDD 19-629] but larger than for maghemite [JCPDS-ICDD 39–1346]. This phenomenon is explained by the partial oxidation of iron ions (II) during functionalization due to an additional oxidative effect on Fe₃O₄ by oxidizing Fe²⁺ to Fe³⁺ by quinoid groups ⁴² and phenolic moieties ⁴³ of humic acids, for example, during the synthesis in situ ("the one pot"), which correlates with the results of electron-paramagnetic resonance spectroscopy, ⁴³ showed a direct relationship between the concentration of free radicals in HS and the amount of Fe (III) reduced by them. ⁴⁴ Functionalization with activated carbon to a greater extent leads to the oxidation of magnetite nanoparticles due to the oxidation by the OH-group of activated carbon.

The change of the lattice parameters indicates the formation of nonstoichiometric magnetite Fe_3−δ O₄ with Fe²⁺/Fe³⁺ ratio from 0 for Fe₂O₃ to ½ for Fe₃O₄ due to partial oxidation of Fe²⁺ during drying, functionalization, and oxidation. According to Gorski, ⁴⁵ stoichiometry can easily be converted to the following relationship.

$$\begin{eqnarray}&&x=\displaystyle \frac{F{e}^{2+}}{F{e}^{3+}}=\displaystyle \frac{1-3\delta }{2+2\delta },\end{eqnarray} \tag{ 1 }$$

The value of the unit cell parameter obtained by Rietveld's refinement and expression (1) makes it easy to calculate the compound Fe_3−δ O₄ (Table I), which shows a decrease of the content of stoichiometric magnetite to Fe_2.86O_4, Fe_2.85O_4, and Fe_2.73O₄ after silica, , and HA and activated carbon functionalization, respectively.

The coherent-scattering region size was derived from powder XRD data by Scherrer's method (Table I). Functionalization with silica and humic acids leads to a decrease in particle size, in contrast to functionalization of only silica due to the presence of two types of ligands in the nucleation stage.

The surface morphology of all three samples was analyzed by scanning electron microscopy (JEOL JSM-6490 LV). It is clearly visible from Figs. 3a–3d that the surface morphology of all three core–shell type magnetite samples were different and their porosity and particle sizes are also different. Figure 3a shows the SEM micrograph of MAC in which very pure porosity was observed. The particles have a bigger grain size along with irregular grain morphology. Figure 3b displays the SEM micrograph of MTA in which better porosity than MAC was observed, however, the particles have bigger grain sizes along with irregular grain morphology. Figures 3c and 3d demonstrate the surface morphology of sample MTAH. In which higher porosity is clearly visible. Also, particles are nearly spherical in shape with a minimum particle size of ∼74.44 nm, probably due to the encapsulation of magnetite nanoparticles in the shell of humic acids and silicaat the nucleation stage, which prevents the growth or aggregation of nanoparticles. By visualizing the surface morphology of all three samples, it is very easy to predict that sample MTAH will show higher reactivity to exposed gas because of its smaller particle size (hence higher surface-to-volume ratio) and larger pore size.

Zoom In Zoom Out Reset image size

The FTIR spectra of MTAH, MTA and MAC nanoparticles are compared in Fig. 4. FTIR spectrum of MTAH and MTA exhibits infrared absorption bands at 500 cm⁻¹, which can be assigned to the Fe–O–Si stretching mode. ⁴⁶ The three common bonds between 500 and 760 cm⁻¹ observed on MTAH and MTA can be ascribed to the stretching and deformation vibrations of Si–O, reflecting the coating of silica on the magnetite surface. Surface modification with TEOS and APTES is a complex process as it involves many different intermediates. ⁴⁷ The FTIR spectra showed the wide and intensive band at ∼1100 cm⁻¹ of MTA and MTAH which can be attributed to a superposition of the silanol (Si–O–H) and siloxane (Si–O–Si) groups, confirming the adsorption of APTES on MNP surface. At the same time, the Si–OH group band appeared at 960 cm⁻¹ as a shoulder at the Si–O–Si skeletal band and was slightly weaker in intensity than ever found in ordinary Stober silica particles, which showed a well-defined peak of Si groups –OH. ⁴⁸ The common band of MTA and MTAH at 1543 cm⁻¹ was assigned to the δNH groups of −NH2, both of which can be attributed to being present in the APTES structure. Also, the peak at 1640 cm⁻¹ for both MNPs was assigned to adsorbed water, and the broad peak centred at about 3400 cm⁻¹ was assigned to νO–H for adsorbed water or νN–H of the amino group.

Zoom In Zoom Out Reset image size

In the spectra of the obtained MAC nanoparticles, a peak of the characteristic vibration band of the carbonyl group (νСО = 1725–1680 cm^–1) is observed. At the same time, bands corresponding to symmetric (νsCOO = 1420–1360 cm^–1) and asymmetric (ν_asCOO = 1610–1550 cm^–1) vibrations of carboxylate ions appear in the spectra (Fig. 3). The interaction of iron ions is reflected in the bands for carbonyl groups: aliphatic (1100 cm^–1) and aromatic (1300 cm^–1, shoulder of 1100 cm⁻¹) present in phenolic and quinoid compounds in activated carbons.

Band gap analysis through UV-Visible spectroscopy is shown in Fig. 5. The band gap analysis is very important in the case of sensors. ^1,2 This is well known that when the band gap of material is increased, this shows the reduction in the size of particles. As the particle size reduces, the corresponding surface-to-volume ratio increases, which increases surface reactivity. ^1,2 This can be visualized from Figs. 5b–5d that band gap increases from 2.80 eV to 2.92 eV and then 4.08 eV for MAC, MTA, and MTAH samples. From SEM images in Fig. 3, it can be seen that MTAH is in nano-dimension, which can be used as a sensing material. This is the reason for high sensitivity of MTAH than the other two samples, which is discussed in detail later.

Zoom In Zoom Out Reset image size

It is known that the band gap of semiconductors often depends on the particle size. And in our case, indeed, with a decrease in the particle size of magnetite-based composites in the MAC-MTA-MTAH series, in accordance with the SEM data, the value of the band gap increases.