Influence of SiO2 Embedding on the Structure, Morphology, Thermal, and Magnetic Properties of Co0.4Zn0.4Ni0.2Fe2O4 Particles

(Co0.4Zn0.4Ni0.2Fe2O4)α(SiO2)(100−α) samples obtained by embedding Co0.4Zn0.4Ni0.2Fe2O4 nanoparticles in SiO2 in various proportions were synthesized by sol-gel process and characterized using thermal analysis, Fourier-transform infrared spectroscopy, X-ray diffraction, transmission electron microscopy, inductively coupled plasma optical emission spectrometry, and magnetic measurements. Poorly crystalline Co–Zn–Ni ferrite at low annealing temperatures (500 °C) and highly crystalline Co–Zn–Ni ferrite together with traces of crystalline Fe2SiO4 (800 °C) and SiO2 (tridymite and cristobalite) (1200 °C) were obtained. At 1200 °C, large spherical particles with size increasing with the ferrite content (36–120 nm) were obtained. Specific surface area increased with the SiO2 content and decreased with the annealing temperature above 500 °C. Magnetic properties were enhanced with the increase in ferrite content and annealing temperature.


Introduction
Spinel ferrite nanoparticles are widely studied due to their outstanding electrical and magnetic properties, high thermal and chemical stability, and applicability in different areas such as electronic, microwave, and communication devices, information storage systems, ferrofluid technology, solid oxide fuel cell, gas sensors, magnetocaloric refrigeration, and medical diagnosis [1][2][3][4]. The physicochemical properties of ferrites are determined by the preparation method, heat treatment, and chemical composition, as well as the type, stoichiometric ratio, and distribution of cations [5][6][7][8][9]. The particle morphology and surface coating may also influence the magnetic behavior of ferrites having the same compositions [10].
The synthesis route is a key factor in high-purity spinel ferrite nanoparticles' preparation [5,9,11]. The most common ways to synthesize nanostructured ferrites are sol-gel, solid-phase, hydrothermal, coprecipitation, auto combustion, sonochemical, microwave refluxing, etc. [7,10,11]. The solid-state synthesis produces nanoparticles with high yields and well-controllable grain size [3]. At the same time, the conventional ceramic method produces particles in the micrometer range that tend to agglomerate due to slow reaction kinetics [11]. Generally, wet chemical synthesis methods such as hydrothermal, sol-gel, and auto combustion are used to produce high-purity crystalline ferrite nanoparticles at low annealing temperatures [2,3,11]. The sol-gel process allows the easy, low-cost production of ferrite nanocomposites with controlled structure and properties [12]. Moreover, the sol-gel process may produce nanocomposite materials comprising highly dispersed magnetic ferrite nanoparticles [13][14][15][16]. This method consists of incorporating metal nitrates in tetraethyl
The thermal behavior of samples was studied by thermogravimetry (TG) and differential thermal analysis (DTA) by using a Q600 SDT (TA Instruments, New Castle, DE, USA) thermal analyzer, in air, up to 1200 • C, with a 10 • C/minheating rate using an SDT Q600 thermogravimeter and alumina standards. A D8 Advance (Bruker, Karlsruhe, Germany) diffractometer equipped with a LynxEye linear detector was used for the investigation of crystalline phases, using the CuKα radiation (λ = 1.54060 Å) in the 2θ range 10-80 • . The FT-IR spectra of the samples were recorded using a Spectrum BX II (Perkin Elmer, Waltham, MA, USA) Fourier-transform infrared (FT-IR) spectrometer, while the composition of Ni-Zn-Co ferrites was confirmed by Perkin Elmer Optima 5300 DV (Norwalk, CT, USA) inductively coupled plasma optical emission spectrometry (ICP-OES) after aqua regia digestion using a Speedwave Xpert (Berghof, Germany) microwave digestion system. N 2 adsorption-desorption isotherms were recorded at −196 • C by a Sorptomatic 1990 (Thermo Fisher Scientific, Waltham, MA, USA) instrument, which was used for calculation of the specific surface area (SSA) using the Brunauer-Emmett-Teller (BET) model. The particle morphology was investigated using a transmission electron microscope (TEM, HD-2700, Hitachi, Tokyo, Japan) and a digital image recording system on samples deposited on carbon-coated copper grids. The average particle size was estimated from TEM measurements using the UTHSCSA ImageTool image software for over 100 nanoparticles in each sample. The hysteresis loops were recorded in magnetic fields between −2 to 2 T, at room temperature, and magnetization versus applied field was measured on samples embedded in an epoxy matrix by a 7400 vibrating-sample magnetometer (VSM, Lake Shore Cryotronics, Westerville, OH, USA). The magnetic measurement uncertainty was 10%.
The thermal behavior of samples was studied by thermogravimetry (TG) and differential thermal analysis (DTA) by using a Q600 SDT (TA Instruments, New Castle, DE, USA) thermal analyzer, in air, up to 1200 °C, with a 10 °C/minheating rate using an SDT Q600 thermogravimeter and alumina standards. A D8 Advance (Bruker, Karlsruhe, Germany) diffractometer equipped with a LynxEye linear detector was used for the investigation of crystalline phases, using the CuKα radiation (λ = 1.54060 Å) in the 2 range 10-80°. The FT-IR spectra of the samples were recorded using a Spectrum BX II (Perkin Elmer, Waltham, MA, USA) Fourier-transform infrared (FT-IR) spectrometer, while the composition of Ni-Zn-Co ferrites was confirmed by Perkin Elmer Optima 5300 DV (Norwalk, CT, USA) inductively coupled plasma optical emission spectrometry (ICP-OES) after aqua regia digestion using a Speedwave Xpert (Berghof, Germany) microwave digestion system. N2 adsorption-desorption isotherms were recorded at −196 °C by a Sorptomatic 1990 (Thermo Fisher Scientific, Waltham, MA, USA) instrument, which was used for calculation of the specific surface area (SSA) using the Brunauer-Emmett-Teller (BET) model. The particle morphology was investigated using a transmission electron microscope (TEM, HD-2700, Hitachi, Tokyo, Japan) and a digital image recording system on samples deposited on carbon-coated copper grids. The average particle size was estimated from TEM measurements using the UTHSCSA ImageTool image software for over 100 nanoparticles in each sample. The hysteresis loops were recorded in magnetic fields between −2 to 2 T, at room temperature, and magnetization versus applied field was measured on samples embedded in an epoxy matrix by a 7400 vibrating-sample magnetometer (VSM, Lake Shore Cryotronics, Westerville, OH, USA). The magnetic measurement uncertainty was 10%.

Results and Discussion
The TG ( Figure 1a) and DTA curves ( Figure 1b indicates the maximum of the exothermic and endothermic effects, respectively) of sample α = 0% dried at 40 °C show two weak endothermic effects at 64 and 173 °C attributed to the loss of water from TEOS and an intense exothermic effect at 300 °C ascribed to 1,3PD decomposition [12]. These two processes result in a mass loss of 63.8% [12,15,16].  while those at 321-325 • C are attributed to Fe malonates' decomposition [12,[14][15][16]. The temperature corresponding to the formation of divalent metal (Co, Ni, Zn) malonates slightly increases, whereas that of trivalent metal (Fe) malonates slightly decreases [12,[14][15][16]. The transformations of the SiO 2 matrix during the thermal process make it challenging to delimitate the effects ascribed to malonate precursors' formation and decomposition [12,[14][15][16].
Except for sample α = 0%, the FT-IR spectra of samples dried at 40 • C ( Figure 2a) show a band at around 1380 cm −1 , characteristic of nitrates. This band is missing for the samples heated at 300 • C, confirming the metal malonates' formation and nitrates' decomposition up to 300 • C [12,15,16]. For samples dried at 40 • C, the band at 1590-1620 cm −1 is specific to O-H vibrations in 1,3PD and adsorbed molecular water, and the bands at around 2950 and 2870 cm −1 are specific to stretching vibration of C-H in the methylene groups of 1,3PD. This band does not appear in the FT-IR spectra of samples annealed at high temperatures, indicating the precursor's decomposition [15,16,21].
Except for sample α = 0%, the FT-IR spectra of samples dried at 40 °C ( Figure 2a) show a band at around 1380 cm −1 , characteristic of nitrates. This band is missing for the samples heated at 300 °C, confirming the metal malonates' formation and nitrates' decomposition up to 300 °C [12,15,16]. For samples dried at 40 °C, the band at 1590-1620 cm −1 is specific to O-H vibrations in 1,3PD and adsorbed molecular water, and the bands at around 2950 and 2870 cm −1 are specific to stretching vibration of C-H in the methylene groups of 1,3PD. This band does not appear in the FT-IR spectra of samples annealed at high temperatures, indicating the precursor's decomposition [15,16,21]. In the FT-IR spectra of sample α = 25-100% heated at 300 • C (Figure 2b), the band at around 1620 cm −1 characteristic to C=O of COO-groups' vibration indicates the formation of malonate-metal complexes [15,16,21].  [12,15,21]. The bands at 3420-3410 cm −1 are attributed to the vibration of O-H and hydrogen bonds in 1,3PD (40 • C) and metal malonates (300 • C) [15,16].
The XRD patterns of samples α = 100% annealed at 500, 800, and 1200 • C ( Figure 3) display no impurities or unreacted Fe, Ni, Co, and Zn oxides, the broadening of diffraction peaks being ascribed to ultrafine Co 0.4 Zn 0.4 Ni 0.2 Fe 2 O 4 particles [12,15,16]. The intensity of diffraction peaks matching to ferrites increases at high annealing temperatures indicating a high degree of crystallinity, high crystal nucleation (owing to the small growth rate and homogenous distribution), and large crystallites (owing to the coalescence process) [6,8,[14][15][16]22]. The degree of crystallinity (DC) was determined as the ratio between the area under all diffraction peaks and the total area under the amorphous halo and diffraction peaks [14][15][16]. The intensity of the main diffraction peak of cubic spinel ferrite at (311) plane was considered as a measure of the degree of crystallinity [14][15][16].
By annealing at 500 • C, the samples α = 25, 50 and 75% display single-phase Co 0.4 Zn 0.4 Ni 0.2 Fe 2 O 4 , but less crystallized than the sample α = 100%, the degree of crystallinity increases with the ferrite content embedded in the SiO 2 matrix. With sample α = 75% annealed at 800 • C, single phase Co 0. 4   The formation of Fe 2 SiO 4 appears due to the incomplete reduction of Fe 3+ to Fe 2+ , which further reacts with the SiO 2 matrix, forming Fe 2 SiO 4 [16]. Nanomaterials 2023, 13,    At all annealing temperatures, for samples with no ferrite content (α = 0%), no crystalline phases are observed and the halo between 16 and 30 • (2θ) matches the amorphous SiO 2 matrix. A possible explanation for the absence of crystalline phases corresponding to the SiO 2 could be the difficult diffusion of oxygen within the pores of the silica matrix [14][15][16]. Usually, the amorphous phase content is proportional to the area under the diffraction halo, but not all the amorphous phases produce diffraction halos due to the lack of a significant local order. The area under the amorphous halo and the total area of the diffraction peaks were used to explore the evolution of amorphous and crystalline phases [14][15][16]. Similar behavior is observed for the sample with high SiO 2 matrix content (α = 25%) annealed at 500 • C. Consequently, low annealing temperature and high SiO 2 content led to highly amorphous content. However, not all the amorphous phases imply diffraction halos owing to the absence of a large local order [14,21].
The average crystallite size (D XRD ) was calculated using the Scherrer formula (Equation (1)): where λ is the wavelength of CuK α radiation (1.5406 Å), β is the broadening of full width at half-maximum intensity (FWHM), and θ is the Bragg angle ( • ) [12][13][14][15][16]. The average crystallite size increases with the annealing temperature and Co 0.4 Zn 0.4 Ni 0.2 Fe 2 O 4 content by the grain growth blocking effect of the SiO 2 matrix (see Table 1) [14][15][16]. The samples with low ferrite content comprise both amorphous and crystalline phases. The changes in crystallite size may be associated with the influence of the SiO 2 matrix on the grain growth and lattice strains, in such a way that SiO 2 content increase, while the annealing temperature reduces the grain growth [23][24][25]. The largest crystallite size was obtained for non-embedded Co 0.4 Zn 0.4 Ni 0.2 Fe 2 O 4 (α = 100%), following the assumption that the SiO 2 matrix contributes to the reduction of crystallite size. One plausible explanation could be the improvement of the crystal-nuclei coalescence process, which occurs at high annealing temperatures (1200 • C). In addition, the annealing temperature reduces lattice strains and defects [14][15][16].
The elemental composition is confirmed by the Co/Zn/Ni/Fe molar ratio using the MW/ICP-OES analysis ( Table 1). The best experimental and theoretical data correlation is observed for the samples annealed at 1200 • C. For samples α = 25 and 50% annealed at 800 • C, the higher Fe content confirms the presence of Fe 2 SiO 4 as a secondary phase observed by XRD.
The SSA decreases with the increase of α, by the increase of D XRD (Table 1). For SiO 2 (α = 0%) and ferrite (α = 100%) samples, the SSA does not depend on the annealing temperature, while for samples α = 25, 50, and 75% annealed at 500 • C, an increase of SSA value was observed. A possible explanation could be the better organization and crystallization of samples annealed at 500 • C than at 300 • C. The pore size distribution (Figure 4c,d) shows that all samples contain different-sized pores up to 550 Å. The pores are generally under 100 Å in samples annealed at 500 • C and up to 200 Å in samples annealed at 300 • C, respectively. These results follow the variation of SSA described above.  The TEM image of the SiO2 matrix (α = 0%) consists of a dark area, without any possibility of identifying the matrix network, whereas those of samples α = 25-100% annealed at 500 and 800 °C are blurry, with low contrast, due to the small size poorly crystalline Co0.4Zn0.4Ni0.2Fe2O4 particles (~1 nm). For samples α = 25-100% annealed at 1200 °C ( Figure  5), the higher ferrite content embedded in the SiO2 matrix results in large spherical particles. The increase in particle size from 34 to 122 nm (Table 1, Figure 6) with the ferrite content could be the outcome of different reaction kinetics, variation of the particle growth rate, or crystalline clusters formation [14][15][16]28]. The different particle arrangement could be due to the solid bodies formed by well-faceted grains, while the particle agglomeration could be a consequence of small particle size, inter-particle interactions, interfacial surface tensions, and strong intermolecular friction produced during the conversion of thermal energy into internal heat energy [14][15][16]. The porous surface formed by the gases generated during the thermal decomposition also favors the particle's agglomeration [14][15][16].  The TEM image of the SiO 2 matrix (α = 0%) consists of a dark area, without any possibility of identifying the matrix network, whereas those of samples α = 25-100% annealed at 500 and 800 • C are blurry, with low contrast, due to the small size poorly crystalline Co 0.4 Zn 0.4 Ni 0.2 Fe 2 O 4 particles (~1 nm). For samples α = 25-100% annealed at 1200 • C (Figure 5), the higher ferrite content embedded in the SiO 2 matrix results in large spherical particles. The increase in particle size from 34 to 122 nm (Table 1, Figure 6) with the ferrite content could be the outcome of different reaction kinetics, variation of the particle growth rate, or crystalline clusters formation [14][15][16]28]. The different particle arrangement could be due to the solid bodies formed by well-faceted grains, while the particle agglomeration could be a consequence of small particle size, inter-particle interactions, interfacial surface tensions, and strong intermolecular friction produced during the conversion of thermal energy into internal heat energy [14][15][16]. The porous surface formed by the gases generated during the thermal decomposition also favors the particle's agglomeration [14][15][16]. The average crystallite sizes are consistent with the particle sizes determined from TEM, the differences being attributed to the interference in the diffraction patterns introduced by the amorphous SiO2 and large-size nanoparticles [12][13][14][15][16]. The particle size determined via TEM is generally larger than the crystallite size estimated by XRD, considering that a particle typically consists of several crystallites. The crystallite size can be calculated by analyzing the broadening of diffraction peaks without considering the effects of other factors that contribute to the diffraction peak width (i.e., instrumental contribution, temperature, microstrain, etc.) [29]. Moreover, even if they are few in number, the large nanoparticles significantly contribute to the diffraction patterns since they comprise a large fraction of atoms. The interference of the amorphous SiO2 with particle size lower than that of the embedded ferrite crystallites should also be considered [14][15][16]. The average crystallite sizes are consistent with the particle sizes determined from TEM, the differences being attributed to the interference in the diffraction patterns introduced by the amorphous SiO 2 and large-size nanoparticles [12][13][14][15][16]. The particle size determined via TEM is generally larger than the crystallite size estimated by XRD, considering that a particle typically consists of several crystallites. The crystallite size can be calculated by analyzing the broadening of diffraction peaks without considering the effects of other factors that contribute to the diffraction peak width (i.e., instrumental contribution, temperature, microstrain, etc.) [29]. Moreover, even if they are few in number, the large nanoparticles significantly contribute to the diffraction patterns since they comprise a large fraction of atoms. The interference of the amorphous SiO 2 with particle size lower than that of the embedded ferrite crystallites should also be considered [14][15][16].
The SiO 2 matrix (α = 0%) displays a diamagnetic behavior (Figure 7      coupling [31,32]. Generally, H c depends on crystallite sizes, magnetocrystalline anisotropy, domain walls and M S [3]. In the case of small single magnetic domain particles, they change the magnetization by spin rotation [34]. At high H c values, the thermal energy cannot induce the magnetization fluctuations for change of the magnetization [3]. Consequently, besides the particles' shape and the density of disordered surface spins, H c has an important contribution to the magnetic order inside a monodomain particle. For large particles, the density of surface spins is low and results in higher magnetization [26]. The magnetocrystalline anisotropy constant (K) was calculated using Equation (3): where M S is the saturation magnetization, µ 0 is vacuum permeability (µ 0 = 1.256 × 10 −6 N/A 2 ), and H c is the coercivity field (T) [35].
The K values ( (α = 100%). The strain on the ferrite nanoparticle surface induced by the SiO 2 matrix hinders the rotation of the magnetic moments from the particle-matrix interface [32]. As can be seen from Table 2, all the main magnetic parameters were strongly affected by the SiO 2 embedding. M S , M R , and K depreciate substantially with the increase of SiO 2 content due to the non-magnetic nature of the matrix. The H c value of the samples annealed at 1200 • C has a different trend; it is enhanced, suggesting the pinning of the magnetic moments from the surface of the particles, which is induced by the strain of the SiO 2 layer.
The chemical composition, crystallographic structure, particle size, atomic packing density, and internal defects highly influence the ferrites' magnetic properties [20]. Previous studies reported a decrease in the magnetocrystalline anisotropy of cubic ferrites when they are doped with Zn 2+ ions [36]. Ni-Co ferrites have large K due to the Co 2+ ions preference for the octahedral (B) sites [19].
The magnetization derivative curves (dM/d(µ 0 H)) vs. applied magnetic field are shown in the insets of Figure 8. The presence of a single peak indicates a single magnetic phase for samples α = 100%, considering that in a pure magnetic sample, the peak occurs at the nominal coercive field, suggesting crystalline samples with a single magnetic phase [14,16,31]. The sharp peaks indicate high magnetic purity, whereas the broad peaks suggest wide particle size distributions. The magnetization derivative for samples α = 25-75% annealed at 800 and 1200 • C indicate that two magnetic phases (from the triple Co-Zn-Ni ferrite) are magnetically coupled inside of the particle along their magnetic moments; the crystalline phases of Fe 2 SiO 4 (a typical paramagnet at room temperature [34,37]), cristobalite and tridymite identified by XRD do not display magnetic properties, the hard magnetic phase being dominant (since it has a larger H c ). If not instrumental, the peaks' asymmetry reveals the presence of two magnetic phases inside the particles, one forming from a solid solution of two ferrites and another being the third ferrite.
In our previous studies on Co-Ni, Co-Zn and Zn-Ni ferrites, only a single magnetic phase was obtained [14][15][16]. For samples annealed at 1200 • C, the hysteresis loops are, generally, broader, and the (dM/d(µ 0 H)) vs. µ 0 H curves are narrower and sharper than in samples annealed at 800 • C. The peak heights and their horizontal shifts are related to the strength of the magnetic phases [14,26,30]. For samples annealed at 800 • C, the broader peaks suggest a large particle size distribution associated with a large H c .
The structure and the magnetic properties of (Co 0.4 Zn 0.4 Ni 0.2 Fe 2 O 4 ) α (SiO 2 ) 100−α are highly influenced by the SiO 2 content and the annealing temperature. Our future studies intend to identify the metallic ion which does not couple with the other two metallic ions of the mixed ferrite leading to a second magnetic phase. Thus, by Co 0.4 Zn 0.4 Ni 0.2 Fe 2 O 4 nanoparticles, the magnetic behavior can be easily tuned. Furthermore, combining the best magnetic properties and morphological configuration may be of interest for several technical applications.

Conclusions
The influence of SiO 2 embedding on the structure, morphology, thermal, and magnetic properties of Co 0.4 Zn 0.4 Ni 0.2 Fe 2 O 4 particles obtained by the sol-gel process was investigated. Fe, Co, Zn, and Ni malonates formed in two stages, as indicated on the DTA curve. At 1200 • C, the XRD and FT-IR results supported the formation of a single-phase spinel structure and SiO 2 matrix. Highly crystalline single-phase ferrite starting from 300 • C for non-embedded Co 0.4 Zn 0.4 Ni 0.2 Fe 2 O 4 (α = 100%) and an amorphous halo without any crystalline phases for SiO 2 (α = 0%) sample were remarked. For samples α = 25-75%, the single crystalline phase Co-Ni-Zn ferrite at 500 • C was accompanied by Fe 2 SiO 4 and quartz at 800 • C, while at 1200 • C the major cristobalite phase was accompanied by Co-Ni-Zn ferrite, tridymite, and quartz. The crystallite size increased with ferrite content in the SiO 2 matrix, namely: 14.5-29.6 nm (500 • C), 26.3-52.4 nm (800 • C), and 33.3-118 nm (1200 • C), respectively. TEM images confirmed that the particles are in the nanometer range. The SSA gradually increased with the SiO 2 content and decreased with the annealing temperature above 500 • C. The main magnetic parameters increased with the Co 0.4 Zn 0.4 Ni 0.2 Fe 2 O 4 content: M S from 7.3 emu/g to 90.1 emu/g, M R from 1.2 to 13.7 emu/g, and K from 0.119·10 −3 to 0.934·10 −3 erg/cm 3 (at 1200 • C). H c increased with Co 0.4 Zn 0.4 Ni 0.2 Fe 2 O 4 content from 260 to 320 (at 800 • C) and decreased from 485 to 165 Oe (at 1200 • C). The M S and M R increased with the annealing temperature. As expected, the non-embedded Co 0.4 Zn 0.4 Ni 0.2 Fe 2 O 4 (α = 100%) was ferromagnetic with high M S , while the SiO 2 matrix (α = 0%) was diamagnetic with a small ferromagnetic fraction. Co 0.4 Zn 0.4 Ni 0.2 Fe 2 O 4 nonembedded into the SiO 2 matrix displays the behavior of a single magnetic phase, while the Co 0.4 Zn 0.4 Ni 0.2 Fe 2 O 4 embedded in the SiO 2 matrix shows two magnetic phases, the solid solution of two ferrites, and the third ferrite. Moreover, when the ferrite is embedded in the SiO 2 matrix, the particle sizes decreased and the main magnetic parameters depreciated.