Spinel Iron Oxide by the Co-Precipitation Method: Effect of the Reaction Atmosphere

Abstract

Synthesis atmosphere (i.e., air and nitrogen) effects on the physical properties and formation mechanism of spinel iron oxide nanoparticles prepared via the co-precipitation method have been investigated using a multi-technique approach. The obtained magnetic nanoparticles (MNPs) were characterized using the X-ray diffraction, transmission electron microscopy (TEM), SQUID magnetometry, Mössbauer spectroscopy and X-ray absorption near-edge Structure spectroscopy techniques. The synthesis procedure leads to the formation of a spinel structure with an average crystallite size of 9.0(9) nm. The morphology of the particles synthetized under an inert atmosphere was quasi-spherical, while the nanoparticles prepared in air present a faceted shape. The small differences observed in morphological properties are explained by the influence of the reaction atmosphere on the formation mechanism of the MNPs. The magnetic characterization indicates that both samples exhibit superparamagnetic behavior at 300 K. The investigation by means of the Langevin approach at 300 K also leads to equal values for the mean size of the magnetic cores (D_m). Additionally, the analysis of the Mössbauer spectra revealed the lack of spin disorder for both samples, resulting in a high saturation magnetization. The fit of XANES spectrum suggests that about 2/3 of the iron ions reside in a local environment close to that of γ-Fe₂O₃ and about 1/3 close to that of Fe₃O₄ for the sample synthetized in inert atmosphere.

1. Introduction

Spinel ferrite magnetic nanoparticles (MNPs) are attracting increasing interest due to the numerous areas of application that they have potential uses in, including biomedicine [1,2,3], magnetic resonance imaging (MRI) contrasting agents [4], drug delivery [5,6], and wastewater treatment [7]. A characteristic property exploited in all these applications is their superparamagnetism (SPM) [8]. Typically, decreasing particle size until a critical value r_c, induces the transition of the material from a magnetic multi-domain state (MD) to a magnetic single domain state (SD) [9]. On decreasing the particle size, the energy barrier, ΔE (i.e., the energy required for the magnetic moment to flip between two minima) value becomes comparable to or smaller than that of thermal fluctuations (k_BT) giving rise then to the so called superparamagnetic (SPM) regime in the system. Among the spinel ferrites, the most commonly used are magnetite (Fe₃O₄) and maghemite (γ-Fe₂O₃) due to their interesting properties such as high saturation magnetization, low toxicity and chemical stability [10]. However, the morpho-structural features and the magnetic properties of such materials are strongly dependent on the synthesis method used. Accordingly, several synthesis methods have been reported in the literature such as sol-gel [11], thermal decomposition [12,13], microemulsion [14], and co-precipitation [15,16,17,18]. Among them, co-precipitation is one of the most used methods due to the favorable synthetic conditions (i.e., high reaction yield, safe and inexpensive precursors and solvent). The effect of several synthesis parameters (e.g., type of precipitating agent, stoichiometric ratio, temperature, etc.) on the chemical composition, morphology, size distribution and magnetic properties of MNPs have been investigated [19,20,21,22,23,24], but few, and often contradictory works, on the effect of reaction atmosphere have been carried out. It has been reported that a normal reaction atmosphere (i.e., air), leads to higher saturation magnetization (M_s) and narrower size distribution [25,26]. Other studies indicate that a higher M_s and a sharper size distribution were obtained using a controlled reaction atmosphere (i.e., inert precipitation medium) [27,28]. Accordingly, this paper focuses on the effect of reaction atmosphere on the morpho-structural and magnetic properties of spinel iron oxide nanoparticles. To clarify the contradictory data in the literature, a multi-technique approach has been used to investigate the effect of reaction atmosphere on the morpho-structural and magnetic properties of the materials. Spinel iron oxide nanoparticles have been prepared by a co-precipitation approach using air and an inert (i.e., nitrogen) reaction atmosphere. The thorough characterization of materials just shows small differences in morphological properties that can been explained by the influence of the reaction atmosphere on the formation mechanism of the nanoparticles.

2. Materials and Methods

Spinel iron oxide nanoparticles were synthesized by the co-precipitation method using an aqueous solution of Fe²⁺ and Fe³⁺ with the stoichiometric ratio 1:2, by alkalization with 30% ammonia solution. The synthesis was done according to the following operating procedure: at 60 °C, a mixture of 6.21 mmol FeCl₃.6H₂O (>99%, Sigma-Aldrich, Schnelldorf, Germany) and 3.15 mmol FeCl₂.4H₂O (>99%, Sigma-Aldrich, Schnelldorf Germany) was dissolved in 100 mL 0.01 M HCl solution (Analysis iso, 37%, Carlo Erba, Paris, France.) previously purified by alternating vacuum and nitrogen (N₂) gas cycles three times. The N₂ atmosphere was maintained and the solution was alkalinized with 30% ammonium hydroxide (ACS, 30%, Carlo Erba, Paris, France). After two hours, the resulting precipitate was washed with distilled water in several cycles assisted by centrifugation, then dried overnight in the oven at 60 °C. Finally, the obtained powder was labeled as FO−CA. To verify the effect of reaction atmosphere, a second sample, labeled as FO−NA, was prepared in air, leaving the other experimental conditions unaltered. It is worth mentioning that these experiments have been repeated more than one time, (see supporting information for further details).

The crystalline structure was characterized by X-ray diffraction (XRD) measurements, using a TT 3003 diffractometer (Seifert, Ahrensburg, Germany) equipped with a secondary graphite monochromator, employing CuKα radiation (λ = 1.5418 Å). Data were collected in the 10°−70° 2ϑ range with a step size of 0.04° and counting 4 sec per step. Transmission electron microscopy (TEM) analysis was carried out using a CM200 microscope (Philips, Amsterdam, The Netherlands.) operating at 200 kV and equipped with a LaB6 filament. The samples in form of powder were dispersed in ethanol using ultrasonication. A drop of suspension was deposited on a commercial TEM grid covered with a carbon thin film and maintained in air until complete evaporation of ethanol. Frequency counts of particles size are fitted with a log-normal function [29] using ImageJ software [30]:

$$\mathrm{P}=\frac{1}{\mathrm{D}\mathsf{\sigma }\sqrt{2\mathsf{\pi }}}\exp \left[-\frac{{\ln }^2\left(\raisebox{1ex}{$\mathrm{D}$}\!\left/ \!\raisebox{-1ex}{$\langle {\mathrm{D}}_{\mathrm{TEM}}\rangle $}\right.\right)}{2{\mathsf{\sigma }}^2}\right]$$

(1)

where σ is the standard deviation and D_TEM is the median of the log-normal distribution.

Mössbauer spectra were recorded in transmission geometry using a ⁵⁷Co/Rh γ-ray source mounted on an electromagnetic driving unit undergoing a triangular velocity form. Experiments were carried out at 12 K using a bath cryostat. The quantity of powder is optimized at about 5 mg of Fe/cm². The fitting procedure was done using the home-made unpublished MOSFIT software involving magnetic sextets and quadrupolar components with Lorentzian lines. DC magnetization measurements were performed with a superconducting quantum interference device (SQUID) magnetometer (H_max = 5.5 T; T = 5–400 K). The sample in solid form was fixed using epoxy resin in polycarbonate capsule to prevent any movement of the powder during the measurement. All the magnetic measurements were normalized by the real mass of the sample. Saturation magnetization value (M_S) was determined from the portion at high field of M=(H) curves using as a formula:

$$\mathrm{M}={\mathrm{M}}_{\mathrm{S}}\left(1-\frac{\mathrm{A}}{H}-\frac{\mathrm{B}}{{\mathrm{H}}^2}\right)$$

(2)

Fe K-edge (7112 eV) XAFS measurements were carried out at the XAFS beamline of the ELETTRA synchrotron [31]. X-ray absorption spectra were measured in transmission geometry at room temperature, measuring the X-ray intensities before (I_o) and transmitted (I_t) through the sample using two gas optimized ionization chambers. An Fe metal foil is placed after the I_t detector and a third ionization chamber measured the transmitted X-ray flux (I_ref) which is used to monitor the X-ray energy calibration during the experiment. XAFS samples were prepared by mixing a small amount of MNPs powder with polyvinylpyrrolidone (PVP) polymer in approximately 1/10 weight concentration. About 100 mg of the mixtures were compressed in thin solid pellets suitable to be mounted on the sample holder by Kapton tape. To improve the data statistics and reduce artefacts due to X-ray beam drifts, at least three scans were collected for each sample and averaged after checking for the correct energy scale. In addition, Fe-K edge spectra from commercial reference compounds were measured, in particular pure Fe metal foil, FeO, Fe₃O₄ and γ-Fe₂O₃ oxide powders, in order to be used as standard references for Fe⁰, Fe²⁺ and Fe³⁺ oxidation states. The total absorption signal through the sample was calculated as a(E) = ln (I_o/I_t) [32]. The raw absorption spectra a(E), of NP and reference samples were processed for pre-edge background subtraction and post-edge jump normalization accordingly to standard XAFS data analysis procedures and analyzed in the XANES region using a Linear Combination Fit procedure using the Athena program (from the Demeter software package [33] (see the Supporting Information for further details).

3. Results and Discussions

3.1. Morpho-Structural Properties

XRD patterns (Figure 1) of both samples were indexed according to the spinel iron oxide (FO) structure (JCPDS card No.75-449). No extra phase has been detected and an average crystallite size of 9.0(9)nm was obtained for both samples by the Williamson-Hall method [34] (

Table 1. Crystallites size from X-ray Diffraction (D_XRD), particle diameter from TEM (D_TEM) and magnetic diameter (D_m) for FO-CA and FO-NA nanoparticles, uncertainties on the last digit are given in parentheses.

Sample	DXRD (nm)	DTEM (nm)	Dm(nm)
FO_CA	9.0(9)	9.0(9)	11(1)
FO_NA	9.0(9)	11(1)	12(1)

).

To investigate the magnetic structure of the sample Mössbauer spectroscopy under intense magnetic field (8T), applied parallel to the γ –beam has been recorded at low temperature (12 K) (Figure 2). When Mössbauer spectra of ferrites are recorded under an intense magnetic field, the magnetic splitting of the A-site Fe³⁺ ions increases and the B-site splitting decreases. Therefore, the applied field makes it possible to separate the A and B patterns and to obtain information about the inversion degree. For a thin sample, where thickness effects are negligible and where the direction of the hyperfine field is at random with respect to the gamma ray direction, the relative area of the Zeeman sextet are in the ratio:

$$3:2\mathrm{p}:1:1:2\mathrm{p}:3$$

(3)

where p is dependent on the canting angle ϑ:

$$\mathrm{p}=\frac{2{\sin }^2\mathsf{\theta }}{1+{\cos }^2\mathsf{\theta }}$$

(4)

By normalizing the total area to 1, the area of lines 2–5 (A_2,5) is equal to:

$${\mathrm{A}}_{2,5}=\frac{1}{2}{\sin }^2\mathsf{\theta }$$

(5)

In case of the absence of the canting effect, lines 2–5 disappear [35,36].

Fitting parameters are reported in

Table 2. Summary of obtained values. of hyperfine parameters (Isomer Shift, Quadrupolar Shift, Effective and Hyperfine fields, Angle $\mathsf{\theta }$, and weight) obtained at 12 K under external field of 8T.

Sample	Sites	$\mathit{\delta }$ (mm.S−1)	$2\mathit{\epsilon }$ (mm.S−1)	${\mathit{B}}_{\mathit{e}\mathit{f}\mathit{f}}$ (T)	${\mathit{B}}_{\mathit{h}\mathit{y}\mathit{p}}$ (T)	$\mathit{\theta }$ (°)	%
FO−CA	A	0.36	−0.01	60.5	52.8	16	38
	B	0.50	−0.00	46.2	53.8	20	62
FO−NA	A	0.35	−0.03	60.4	52.8	16	39
	B	0.51	−0.03	46.1	53.9	15	61

. Spectra of both samples clearly show the inverse spinel structure of the particles. The high oxidation degree of magnetite was confirmed by the low isomer shifts (δ) values [37]. Indeed, the mean isomer, which probes the electronic density at the ⁵⁷Fe nuclei, that is the valence state, are consistent with the presence of pure Fe³⁺ ions for both samples. The measurement at 77 K (see supporting information for further details) confirm this scenario with an isomer shift (0.45 mm/s) is about compatible with the presence of maghemite. X-ray Absorption Near-Edge Structure (XANES) spectroscopy measurements have also been performed on the FO−CA sample further confirming the presence of mainly maghemite (see supporting information for further details). The second and fifth lines have a non-zero intensity, indicating the presence of a non-collinear (canted) structure for iron magnetic moments with respect to the applied field. The canting angle(Ɵ) of iron spins in tetrahedral and octahedral sites for FO−CA and FO−NA is relatively small suggesting the high crystallinity of the particles.

Sample morphology has been investigated by transmission electron microscopy. Figure 3a,b show bright field TEM images showing general views of FO−CA and FO−NA, respectively. Samples contain aggregates of nanoparticles whose dimensions have been deduced by TEM images and statistically analyzed. A log-normal size distribution was obtained for both FO−CA and FO−NA, (see supporting information for further details) with mean size equal, within experimental error, for both samples, (Table 1). To further investigate the crystallographic structure, selected area electron diffraction (SAED) measurements have been performed on FO−CA and FO−NA. SAED patterns are composed of diffraction rings having the same diameters and intensities for both samples (see supporting information for further details). The interplanar distances associated with the diffraction rings reveal that they are both composed of polycrystalline iron oxide with a spinel structure, confirming XRD results. High−resolution TEM (HR−TEM) observations allowed us to better investigate the particles morphology. Figure 3c,d show HR−TEM images of FO−CA and FO−NA particles, respectively. Looking at the atomic planes, it is evident that FO−CA particles have a quasi-spherical shape while the FO−NA ones are generally faceted. Generally speaking, in colloidal synthesis, the formation of crystals is discussed in terms of nucleation and growth theory where chemical species are supposed to assemble directly from solution [38]. Accordingly, the best way to control the particle size and shape, is to separate these two steps (i.e., the nucleation step should be avoided during the growth of the nuclei [39]). Unfortunately in the co-precipitation method, it is not possible to separate the nucleation and the growth steps [40,41]. In typical co-precipitation synthesis, the pH of ferrous and ferric solution is increased via the addition of the base. Depending on the method of base addition [38] (e.g., continuous or abrupt addition) different formation pathways of magnetite nanoparticles can be observed [42,43]. In the investigated synthesis procedure, the base was added in abrupt way (i.e., pH quickly increases from 2 to 9.2), resulting in significant pH gradient leading to the formation of lepidocrocite(γ-FeOOH) at pH 7−8 [44,45,46,47],then transformed into spinel iron oxide at higher pH. Moreover, the presence of oxygen in a reaction medium is a key factor in promoting the oxidation of Fe²⁺ to Fe³⁺ [48,49]. In our case, the produced nanoparticles under different synthesis conditions(O₂/N₂) show different particles shape. The surface structure passes from quasi-spherical in FO−CA sample to faceted for FO−NA sample. For the latter, the excess of O₂ (air) creates higher Fe²⁺/Fe³⁺ oxidation rate, results in transforming most of the Fe(OH)₂ to γ-FeOOH then through a topotactic process transforms into magnetite [47,50]. This formation path way leads to irregularly shaped particles as observed in TEM image. In case of FO−CA sample, the synthesis under inert atmosphere lead to slow oxidation rate until the complete mixing of the solution (i.e., homogeneity in the pH of the solution) then direct transformation to magnetite involving the remaining Fe²⁺ in solution resulting in more uniformed and spherical particles shape [51].

3.2. Magnetic Properties

The temperature dependence of magnetization has been recorded according to ZFC−FC protocols at low magnetic field. The measurement was carried out as follows: first the sample was cooled down from room temperature to 5 K, then a static magnetic field of 2.5 mT was applied and M_ZFC was measured during the warming up of the sample. Once the desired temperature (300 K) was reached, the applied magnetic field was maintained, and M_FC was measured during the cooling down to 5 K. ZFC−FC magnetization curves of FO−NA and FO−CA samples (Figure 4), indicate a typical behavior for an assembly of interacting single domain particles. The temperature corresponding to the maximum in ZFC curve, T_max, is directly proportional to the average blocking temperature, and it was equal for both samples within experimental error (

Table 3. Temperature Corresponding to the Maximum in ZFC Curve (T_max), Irreversibility Temperature (T_irr), and Blocking Temperature (T_B-TRM) From TRM measurement and the blocking temperature from Hansen and Mørup approach (T_B.H.M) for FO−CA and FO−NA samples, uncertainties on the last digit are given in parentheses.

Sample	Tmax(K)	TB-TRM (K)	Tirr(K) (3%)	TB.H.M (k)
FO−CA	203(20)	104(10)	270(27)	125(12)
FO−NA	202(20)	100(10)	273(27)	124(12)

). Moreover, an irreversible magnetic behavior between ZFC and FC curves was observed; the related temperature (T_irr) can be ascribed to the blocking temperature of the biggest particles [45] (T_irr was estimated from the difference between M_FC and M_ZFC when it becomes lower than 3%). By decreasing the temperature, the FC curves present a plateau-like shape, indicating the presence of strong interparticle interactions as confirmed by the relatively high values of T_max [52].

Figure 5a,b show the temperature dependence of thermoremanent magnetization (TRM) and its derivative. In the TRM protocol the sample was cooled down to 5 K in magnetic field of 2.5 mT, then the field was turned off and the remanent magnetization was measured during the warming. For non-interacting nanoparticles with uniaxial anisotropy, the derivative of M_TRM is related to the anisotropy energy barrier distribution [35]:

$$\mathrm{f}\left(\Delta {\mathrm{E}}_{\mathrm{a}}\right)\propto -\frac{{\mathrm{dM}}_{\mathrm{TRM}}}{\mathrm{dT}}$$

(6)

Due to the presence of interparticle interactions, the derivative of M_TRM can only be considered as a rough estimation of the $\Delta$E_a distribution. The blocking temperature can be defined as the temperature for which the relaxation time is equal to the measuring time of the experimental technique. In a real system of nanoparticles, where a finite size distribution always exists, T_B is often defined as the temperature at which 50% of the particles overcome their anisotropy energy barriers. The obtained values of T_B, T_max and T_irr are equal within the experimental error confirming that no significant difference in magnetization dynamics is present between the two samples. The distribution of anisotropy energy also shows a sharp peak at low temperature (i.e., T_Low ~25 K), already observed in maghemite samples and can be ascribed to surface effects [9].

The phenomenological approach of Hansen and Mørup (H.M) has also been applied as a second estimation of the distribution of energy barriers in small particles from ZFC-FC magnetization curves using the equation [53]:

$$\langle {\mathrm{T}}_{\mathrm{B}.\mathrm{H}.\mathrm{M}}\rangle ={\mathrm{T}}_{\max }[1.792+0.186\ln (\raisebox{1ex}{${\mathrm{T}}_{\mathrm{irr}}$}\!\left/ \!\raisebox{-1ex}{${\mathrm{T}}_{\max }$}\right.-0.918)]-1+0.0039{\mathrm{T}}_{\mathrm{irr}}$$

(7)

Similarly to the previous results, samples also show equal T_B values according to the H.M. approach (Table 3).

The dependence of magnetization on an external magnetic field has been investigated at 300 K and 5 K. At 300 K, the results show superparamagnetic behavior for both samples (Figure 6a). Typically, for a material with superparamagnetic behavior, magnetization curves can also be used in order to extract the mean size of the magnetic core (D_m) (see supporting information for further details) using the Langevin function [54]. A small deviation of the fit from the experimental data has been observed, which can be related to the non-negligible interparticle interactions within the samples [55]. However, the value of D_m is in good agreement with experimental D_TEM and D_XRD. At 5 K (Figure 6b), the hysteresis loops show low values of reduced remanence magnetization (M_r/M_s) and coercive field (H_C) (

Table 4. Saturation magnetization (Ms) at 5 K and 300 K, reduced remanence magnetization (Mr/Ms) and coercive field (Hc) for FO−CA and FO−NA samples, uncertainties on the last digit are given in parentheses.

Sample	Ms (Am2Kg−1) at 300K	Ms (Am2Kg−1) at 5 K	Mr/Ms at 5 K	Hc(mT)
FO−CA	70.9(7)	88(3)	0.25	28.7(3)
FO−NA	69.4(1)	77(3)	0.25	28.1(3)

) in agreement with that is reported in literature [9]. The obtained high M_s value (close to the bulk one Ms = 92 Am²Kg⁻¹) indicates high crystallinity and magnetic ordering of the particles [56].

To evaluate the type and the strength of interparticle interactions, δm(H) plots at 5 K (Figure 7) have been investigated using direct current demagnetization (DCD) and isothermal remnant magnetization (IRM) protocols (see supporting information for further details). Generally, a positive peak in the δm vs-µ₀H curve is an indication of magnetizing interactions(e.g., exchange interactions) among nanoparticles while a negative peak suggests the existence of demagnetizing interactions (e.g., dipole-dipole interactions) [57]. δm plots confirming the presence of strong dipolar interactions, with equal strength in both samples. Since no significant difference between the two samples can be detected in terms of magnetic measurements, this gives us more evidence that the reaction atmosphere does not affect the magnetic properties of FO−CA and FO−NA samples in our case.

The more explanations could be found in Figures S1–S6, and Table S1.

4. Conclusions

The effect of atmosphere on the magnetic nanoparticles synthesized by the co-precipitation chemical method was investigated using a multi-technique approach. XRD and TEM measurements indicate the high crystallinity of FO−CA and FO−NA nanoparticles. However, TEM analysis highlights that the FO−CA sample shows more quasi-spherical particles, while FO−NA synthesized in air presents a relatively more residue of lepidocrocite in the final product. While no significant variation was observed for the magnetic properties (i.e., field and temperature dependence of magnetization, strength of interparticle interactions) in our study. The magnetic response of the materials is compatible with the presence of highly crystalline particles (i.e., of the low magnetic disorder) which was also confirmed by the Mössbauer spectrometry through the absence of spin canting effect.