Magnetic analysis of commercial hematite, magnetite, and their mixtures Open Access

Iron oxides are common compounds in laboratories and in nature, and hold widespread interest in sciences such as mineralogy, biology, geochemistry, and materials engineering.^1–4 The most naturally abundant iron oxides are hematite (α-Fe₂O₃) and magnetite (Fe₃O₄), and their magnetic behaviors have been of great interest owing to numerous applications. Magnetite is a ferrimagnetic phase with high magnetization (∼92 Am²/kg) and low coercivity (10-40 mT) while hematite is a canted antiferromagnetic phase having a small magnetization (∼0.4 Am²/kg) and high coercivity (∼100-400 mT).^1,5–9 Since oxidation-reduction reactions often occur in iron oxides, mixtures of magnetite-hematite are common, and magnetic interactions between discrete particles in a dense compact introduce additional complexity. It is reported that magnetizations of different phases are linearly additive in non-interacting mixtures, while interacting assemblages can introduce nonlinearities.^5,10–13

In this study, three commercial α-Fe₂O₃ powders are examined. These powders are characterized for microstructure, phase analysis, and magnetic properties. Secondly, the effects of various hematite-magnetite mixtures on magnetic hysteresis loops, FORC diagrams, IRM curves, and temperature-dependent magnetization are investigated. Mixtures containing components with distinct magnetic behaviors can potentially be characterized by hysteresis loops with “wasp-waisted” features.^5,14 Since FORC diagrams and IRM curves reveal the coercive field (H_C) distribution, they can provide insight into identifying individual magnetic phases with contrasting H_C within a mixture. Thermo-magnetic curves can also be used to identify individual phases if their magnetic transition temperature ranges do not overlap. We seek to investigate which method is more suitable to study the magnetic signals of hematite and magnetite when they coexist.

Three iron (III) oxide powders were investigated as hematite samples: Fisher 99.9% (LOT# 710181), Alfa Aesar 98% (LOT# E16Z021), and Sigma Aldrich 99.998% (LOT# JV04524HV). The Fe₃O₄ powder used as the magnetite precursor for the hematite (H) –magnetite (M) mixtures is iron (II,III) oxide Alfa Aesar 97% (LOT# X30B083). It was mixed with the Fisher hematite powder in various proportions, 99wt%H/1wt%M (99H1M), 95wt%H/5wt%M (95H5M), and 90wt%H/10wt%M (90H10M). The powders were weighed to 0.0001 g and mixed thoroughly using a vortex mixer and agate mortar and pestle.

XRD scans were obtained using an X’Pert Pro MPD (PANalytical, Netherlands) with a Co Kα X-ray (λ = 0.1789010 nm) at 40 kV and 40 mA. Secondary electron micrographs were obtained with a XL-30 field emission SEM (FEI, Hillsboro, OR), using accelerating voltage 5-10 kV, to analyze the particle morphology and size. A VSM (PMC3900, Lakeshore Cryotronics, Westerville, OH) with maximum applied field of 1.8 T was employed to measure the magnetic properties. The software FORCinel (V2.05 in IGOR Pro6, WaveMetrics, Portland, OR)¹⁵ was used to process the raw FORC data and obtain diagrams.

XRD (Figure 1a) shows single-phase hematite with hexagonal structure and $R3c¯$ space group for Fisher and Aldrich powders, but the Alfa sample shows additional minor impurity peaks which are difficult to identify due to low intensity. Mean crystallite sizes were calculated from peak broadening using the Scherrer equation¹⁶ (Table I). SEM images (Figure 1b) reveal relatively uniform and spherical particles for all powders. Mean particle sizes by image analysis (ImageJ¹⁷) are also listed in Table I.

Despite the same hematite phase detected by XRD, different magnetic behaviors are obtained as shown in Figure 2. Fisher shows the highest coercivity whereas Aldrich presents a straight thin loop. The Alfa sample, however, shows a wasp-waisted loop (necking in the middle section), which can be caused by either multiple magnetic phases with varying coercivities or multiple particle size distributions.^5,14,18,19 Considering the uniform particle size distribution observed in SEM micrographs of Alfa sample, and the presence of another phase detected by XRD, the observed behavior is likely from an Fe-containing impurity phase which likely has higher magnetization and lower coercivity than hematite. The FORC diagram of the Alfa sample (Figure 2b) confirms the simultaneous presence of a multidomain (MD) lower-H_C component (diverging away at the origin) and a single-domain (SD) higher-H_C component (small vertical spread along the H_u=0 axis). The latter is most likely from poorly interacting hematite particles. The FORC diagram of the Fisher sample also displays a SD behavior with center of the contours close to its bulk hysteresis loop coercivity (∼150 mT) indicative of fine hematite particles. Table I lists magnetic parameters derived from hysteresis loops for the hematite powders. The low coercivity and paramagnetic-like behavior of the Sigma-Aldrich sample can be attributed to its larger particle size and small crystallite size compared to the other two (each particle consists of many nano crystallites – nanostructured particles). It is known that increased particle size results in transitioning from SD to MD behavior which leads to low coercivity.

The Fisher hematite and Alfa Aesar magnetite powders were mixed and characterized magnetically. Particle size analysis of the magnetite performed by SEM (not shown) gives a mean size 158 ± 74 nm. The hysteresis behaviors of the mixtures (Figure 3a) reveal that magnetite controls the shape of the magnetic hysteresis loops, even in concentrations as low as 1 wt% (see supplementary material). A wasp-waisted loop is not observed for the mixtures with 5 and 10 wt% magnetite. Thus, major hysteresis loops are incapable of showing a lower-magnetization phase (hematite), when magnetite is present >5 wt%. Simple calculations reveal that saturation magnetization of these mixtures follows linear additivity to a good approximation, while coercivity is not linearly additive. Carter-Stiglitz et al.¹¹ have discussed that magnetization parameters are linearly dependent on the mixing ratio in non-interacting mixtures, whereas coercivity does not behave linearly. It is shown by others^12,13 that magnetization parameters of mixtures containing hard magnetic materials with small spontaneous magnetization (e.g., hematite and goethite) can be described by linear additivity of end-members because there are virtually no magnetostatic interactions. Mixtures of high-M_S soft magnetic materials (e.g., magnetite and maghemite), however, exhibit non-linear behavior due to magnetic interactions. For the present case, containing a low-M_S hard phase and a high-M_S soft phase, linear-additivity of magnetization is still followed to a reasonable extent since there are only minor interactions.

In FORC diagrams of the mixtures, measured with a field increment of 10 mT, magnetite exhibits a MD behavior and hematite a SD behavior (Figure 3b). The mixtures clearly show a progressive shift from an overall MD to SD pattern. However, for the sample with 10 wt% magnetite (90H10M), the contribution from SD hematite is muted due to magnetite. This could be true for any assemblage of a high-M_S component along with a low-M_S phase. Although Muxworthy et al.¹² have reported that FORC method is suitable for identifying hard magnetic minerals such as goethite and hematite in the presence of magnetite, Carvallo et al.¹³ found later that the hematite contribution cannot be detected by FORC when magnetite is present >12 wt%.

Thermomagnetic curves of normalized magnetization of the mixtures at 1 T from room temperature to 800°C were also obtained (Fig. S1 of the supplementary material). The mixtures with 5 and 10 wt% magnetite present similar thermomagnetic behavior, showing only the magnetite T_C. The 99H1M sample does show two transition temperatures attributed to magnetite and hematite. Thus, it is not possible to identify the presence of hematite from Magnetization-Temperature curves when magnetite concentration exceeds 5 wt%.

The IRM demagnetization curves (Figure 4a) were measured by first exposing the samples to a positive saturating field of 1.8 T, then applying stepwise increasing negative fields (backfield) parallel to the magnetization field up to 0.6 T, and measuring remanence at each step. 100 data points were measured for each sample. The IRM gradient curves were obtained by plotting the absolute value of first derivative, subsequently smoothed using a spline function. The gradient curves on a log₁₀ field scale were decomposed into cumulative log-Gaussian (CLG) distributions with zero skewness (see supplementary material). Each peak represents a magnetic component with a given coercivity distribution.²⁰ The majority of the hematite remanence is carried by peak 1 while magnetite has two major CLG coercivity distribution peaks (peaks 2 and 3). Linear-additivity was examined by analyzing height and area under the main hematite peak and the main magnetite peak (Figure 4b). The results based on the magnetite fit peak show that the IRM is linearly additive within a reasonable error, part of which could be caused by mixing and fitting errors. However, the hematite main peak does not obey linear additivity. This is because magnetite’s remanence mostly comes from its high magnetization which has been shown to be linearly additive. Hematite’s remanence mostly originates from its higher coercivity which is not linearly additive. Although magnetite fit peaks become dominant in 90H10M samples, the hematite contribution is still clear in IRM. The hematite contribution in hysteresis loop, FORC, and M-T curve of the same sample was muted by the presence of magnetite. Frank et al.²¹ have also shown that hematite can be visible in IRM curves when magnetite is present as much as 20 wt% in a hematite-magnetite mixture since remanent magnetization is a function of both coercivity and magnetization. Hematite (with intrinsically high coercivity) can maintain its contribution to remanence up to higher fractions of present magnetite. The usual SD state of hematite also magnifies its remanent signal. Therefore, in the case of magnetite-hematite mixtures, IRM measurements can provide a better method to identify hematite than the usual major hysteresis loops, FORCs, and M-T curves.

Three commercial hematite powders were characterized to assess differences which may affect their use as iron precursors. The hematites have varying particle and crystallite sizes and hence different magnetic properties. Additionally, mixtures of hematite-magnetite powders were studied. The magnetic signature of hematite, in major hysteresis loops, FORC diagrams, and thermo-magnetic curves, can be easily muted when magnetite is ≥5 wt%. IRM demagnetization curves, however, were able to identify hematite when magnetite coexisted in higher concentrations, since hematite can maintain its contribution to remanence more than spontaneous magnetization. The distribution peak related to magnetite in the IRM curves of mixtures is linearly additive whereas that of hematite is not because magnetite’s remanence is mostly controlled by its magnetization which obeys a linear additivity in non-interacting assemblages. These results can be valid for any mixture of a low-M_S hard material (e.g., hematite and goethite) and a high-M_S soft material (e.g., magnetite and maghemite).

See supplementary material for obtained magnetization-temperature curves as well as hysteresis parameters of the mixtures and the IRM fitting parameters.

This work was supported by Washington State University start-up funds.