Zeta-Fe₂O₃ – A new stable polymorph in iron(III) oxide family

Abstract

Iron(III) oxide shows a polymorphism, characteristic of existence of phases with the same chemical composition but distinct crystal structures and, hence, physical properties. Four crystalline phases of iron(III) oxide have previously been identified: α-Fe₂O₃ (hematite), β-Fe₂O₃, γ-Fe₂O₃ (maghemite) and ε-Fe₂O₃. All four iron(III) oxide phases easily undergo various phase transformations in response to heating or pressure treatment, usually forming hexagonal α-Fe₂O₃, which is the most thermodynamically stable Fe₂O₃ polymorph under ambient conditions. Here, from synchrotron X-ray diffraction experiments, we report the formation of a new iron(III) oxide polymorph that we have termed ζ-Fe₂O₃ and which evolved during pressure treatment of cubic β-Fe₂O₃ ( space group) at pressures above 30 GPa. Importantly, ζ-Fe₂O₃ is maintained after pressure release and represents the first monoclinic Fe₂O₃ polymorph (I2/a space group) that is stable at atmospheric pressure and room temperature. ζ-Fe₂O₃ behaves as an antiferromagnet with a Néel transition temperature of ~69 K. The complex mechanism of pressure-induced transformation of β-Fe₂O₃, involving also the formation of Rh₂O₃-II-type Fe₂O₃ and post-perovskite-Fe₂O₃ structure, is suggested and discussed with respect to a bimodal size distribution of precursor nanoparticles.

Introduction

Iron(III) oxide is a polymorphic compound, i.e., it can exist in two or more solid phases that are isochemical but have distinct crystal structures and thus different physical properties. Under ambient conditions, four different crystalline polymorphs of iron(III) oxide have been discovered and characterized in details^1,2,3,4,5: (i) α-Fe₂O₃, mineralogically known as hematite, which has a rhombohedrally centred hexagonal crystal structure ( space group with a = 5.034 Å and c = 13.752 Å); (ii) β-Fe₂O₃, which has a cubic body-centred crystal structure of bixbyite type ( space group with a = 9.393 Å); (iii) γ-Fe₂O₃, mineralogically known as maghemite, which has a cubic crystal structure of inverse spinel type ( space group with a = 8.351 Å and vacancies disordered over the octahedral cation sites in the crystal lattice); and (iv) ε-Fe₂O₃, which has an orthorhombic crystal structure (Pna2₁ space group with a = 5.072 Å, b = 8.736 Å and c = 9.418 Å). While α-Fe₂O₃ and γ-Fe₂O₃ are naturally abundant and can be prepared in diverse morphological forms with different sizes by various optimized synthetic routes, β-Fe₂O₃ and ε-Fe₂O₃ are rarely observed in nature and their stability is dependent on the nanodimensional character of their particles^1,2,3,4,5,6.

Due to their different physical properties, which arise from the differences in their crystal structures, all of the iron(III) oxide polymorphs have found applications in nanotechnology or show considerable promise in such applications. For instance, thin nanocrystalline films of α-Fe₂O₃ serve as very efficient electrodes in the photo-assisted electrolysis of water for hydrogen production in solar cells^7,8,9 and α -Fe₂O₃ nanoparticles are effective catalysts for various processes of heterogeneous catalysis^10,11,12. β-Fe₂O₃ was recently used as a chloroform sensor¹³ and identified as a suitable candidate for the preparation of anodes in lithium-ion batteries¹⁴. Ferrimagnetic and superparamagnetic nanoparticles of γ-Fe₂O₃, the most widely used iron(III) oxide polymorph, have been identified as a possible functional medium for magnetocaloric refrigeration¹⁵. They also have diverse biomedical applications because they are biocompatible and biodegradable while exhibiting useful magnetic properties; they have been used as MRI contrast agents, carriers for targeted drug delivery, heating units in magnetically-induced cancer therapy (i.e., hyperthermia) and sensors of various biologically important molecules^16,17,18,19. The most recently identified iron(III) oxide polymorph, ε-Fe₂O₃, shows the highest coercivity among all known metal oxides (20–22.5 kOe)^4,6 and could therefore be used as a magnetic recording material for high-density recording media. In addition, its magnetoelectric properties predestine that it could be useful in the production of multiple-state-memory elements^4,20. Finally, it exhibits ferromagnetic resonance in millimetre wave region, giving it potential applications in devices for suppressing electromagnetic interference and stabilizing electromagnetic transmittance^4,21,22,23.

Due to the diverse potential applications of all four known iron(III) oxide polymorphs, thermally induced phase transformations of the less thermodynamically stable polymorphs (β-, γ- and ε-Fe₂O₃) have been studied extensively. The results of these transformations depend on the intrinsic properties of the starting phase (polymorph structure, particle size, particle morphology, surface coating, particle aggregation, incorporation of particles within a matrix) and the nature of the applied treatment. In general, such transformations ultimately yield α-Fe₂O₃ as the final product and are frequently accompanied with the particle growth^5,24. The two rare iron(III) oxide polymorphs, β- and ε-Fe₂O₃, undergo direct thermal transitions to α-Fe₂O₃⁵. However, hollow β-Fe₂O₃ nanoparticles can be transformed into γ-Fe₂O₃^5,25, demonstrating that the morphology of the starting material can significantly affect the transformation process and permit the evolution of unexpected intermediate iron(III) polymorphs. In the case of γ-Fe₂O₃, both direct transformations into α-Fe₂O₃ and two-step transformations via ε-Fe₂O₃ have been observed, depending on the initial particle size and degree of interparticle interactions (i.e., aggregation, the extent to which the particles are spatially confined)^3,4,5.

Although thermally induced transformations of Fe₂O₃ polymorphs have been described at length in the literature, pressure-induced transitions have only been investigated for the most common polymorphs, α-Fe₂O₃ and γ-Fe₂O₃⁵. Several studies aiming to simulate the geophysical conditions in the Earth’s lower mantle have examined the behaviour of α-Fe₂O₃ under ultra-high pressures^{26,27,28,29,30,31,32,33}. For instance, Ito et al.²⁶ observed consecutive transitions of α-Fe₂O₃ to the Rh₂O₃-II-type Fe₂O₃ structure (Pbcn space group) and then to an orthorhombic structure (with a = 6.883 Å, b = 9.993 Å, c = 4.665 Å and V = 320.9 ų) as the applied pressure was increased from atmospheric levels to 58 GPa at 1400 K. Ono et al.²⁷ observed that the application of a pressure of 30 GPa with laser heating at 2000 K resulted in the formation of Fe₂O₃ with a perovskite-type structure (Pbnm space group). When the pressure was increased to 70 GPa and the temperature to 2500 K, other perovskite-like structures (so-called post-perovskites) with both orthorhombic (with a = 2.639 Å, b = 6.386 Å, c = 8.544 Å and V = 144.0 ų) and monoclinic (a = 5.282 Å, b = 6.385 Å, c = 4.471 Å, β = 107.22° and V = 144.0 ų) symmetry were identified. A further increase in pressure (to 96 GPa) and temperature (to 2800 K) resulted in the conversion of the perovskite structure to a CaIrO₃-type structure with an orthorhombic symmetry (Cmcm space group)²⁸. Recently, Bykova et al.³² reported the pressure-induced transformation of α-Fe₂O₃ single crystals into a cryolite double-perovskite-type phase with a monoclinic unit cell (P2₁/n space group) at about 54 GPa; the pressure-induced transition is accompanied by a large compression in the unit cell not previously observed for the α-Fe₂O₃/perovskite/post-perovskite pathway. In addition to these changes in the crystal structure and unit cell volume of α-Fe₂O₃, mild increases in the applied pressure have been observed to affect some of its other physical properties. Among other things, pressure treatment has been reported to increase its Morin transition temperature³⁴, both to increase and decrease its electrical conductivity^35,36,37, induce a high-spin to low-spin transition (i.e., a 5/2-to-1/2 spin crossover)²⁹ and cause the disappearance of a magnetically-ordered state³⁵. Most importantly, all previously reported high-pressure transformations of α -Fe₂O₃ (whether they occur at room temperature or under heating) were found to be reversible, i.e., the material recovered after the pressure was released had the original hexagonal crystal structure of α-Fe₂O₃ with almost unmodified physicochemical properties.

Conversely, high-pressure treatments of γ-Fe₂O₃ phase typically cause its irreversible transformation to α-Fe₂O₃, which is followed by the evolution of perovskite and post-perovskite structures identical to those formed during pressure treatment of α-Fe₂O₃^{38,39,40,41,42,43,44}. The pressure required to initiate the γ-Fe₂O₃-to-α-Fe₂O₃ phase transformation ranges from ~10 to ~37 GPa and appears to be highly dependent on the particle size of the transformed γ-Fe₂O₃ phase. For instance, Clark et al.³⁸ confirmed that the transition pressure increases as the size of the γ-Fe₂O₃ nanocrystals decreases: the pressures required to transform nanocrystals with dimensions of 7, 5 and 3 nm were 27 GPa, 34 GPa and 37 GPa, respectively. This trend was attributed to the higher surface energy of smaller nanocrystals.

There are no previous reports on the high-pressure transformations of the rare β-Fe₂O₃ and ε-Fe₂O₃ phases. Here, we describe the high-pressure transformations of rare β-Fe₂O₃ for the first time. β-Fe₂O₃ has a cubic crystal structure containing two non-equivalent octahedral cation sites, which have distinct symmetries and are referred to as b-sites and d-sites. They are filled with Fe³⁺ ions in a high-spin state (S = 5/2). On lowering the temperature, β-Fe₂O₃ passes from a paramagnetic to a magnetically ordered regime, adopting an antiferromagnetic state below ~110 K (the Néel temperature). The X-ray synchrotron data presented in this study show that when β-Fe₂O₃ is exposed to pressures above 30 GPa, a new iron(III) oxide polymorph designated zeta-Fe₂O₃ (ζ-Fe₂O₃) is formed. Remarkably, this new polymorph remains stable at room temperature, even after the pressure is released. It has a monoclinic crystal structure, belonging to the I2/a space group. Magnetization measurements indicate that it behaves as an antiferromagnet at temperatures below ~69 K. Its stability is explained due to the high surface energy it gains by being formed from smaller β-Fe₂O₃ nanoparticles and favourable changes in its chemical potential that occur during pressure treatment.

Results and Discussion

Before its pressure treatment, the purity and structural features of the synthesized β-Fe₂O₃ sample were checked using conventional X-ray powder diffraction (XRD) and ⁵⁷Fe Mössbauer spectroscopy. The room-temperature ⁵⁷Fe Mössbauer spectrum of the β-Fe₂O₃ sample is well deconvoluted into 3 spectral components – two dominant doublets whose isomer shift and quadrupole splitting values are characteristic of the b-sites and d-sites in the β-Fe₂O₃ crystal lattice (with an ideal spectral ratio of 1:3 in accordance with the complete occupation of individual crystallographically non-equivalent cation positions by Fe³⁺) and a minor sextet typical of α-Fe₂O₃ admixture (see Fig. 1a and Table 1). Based on the spectral areas of these components, the level of α-Fe₂O₃ admixture was 7 wt.%, a conclusion supported by the material’s XRD pattern (see Fig. 1b). The crystal structure of the β-Fe₂O₃ sample, derived by the Rietveld analysis of its XRD pattern, is shown in Fig. 1c. It has a cubic crystal structure within the space group, lattice parameters of a = b = c = 9.404 Å and a unit volume of V = 831.8 ų.

Table 1 Values of the Mössbauer hyperfine parameters, derived from the fitting of the recorded room-temperature ⁵⁷Fe Mössbauer spectrum of the initial b-Fe₂O₃ sample, where δ is the isomer shift, DE_Q is the quadrupole splitting, B_hf is the hyperfine magnetic field and RA is the relative spectral area of individual spectral components.

Figure 1

Characterization of the starting b-Fe₂O₃ sample.

(a) Room-temperature ⁵⁷Fe Mössbauer spectrum and (b) conventional XRD pattern of β-Fe₂O₃ sample before its treatment under high pressures. (c) Crystal structure of β-Fe₂O₃ (cubic, space group) projected along the b-axis. Red, blue and gray balls represent the octahedral Fe b-sites, octahedral Fe d-sites and oxygen sites, respectively. (d,e) TEM and HRTEM image showing fraction of smaller and larger β-Fe₂O₃ nanoparticles and (f) particle size distribution derived from TEM/HRTEM images, where bars correspond to experimentally observed nanoparticle sizes and red curve represents the best theoretical fit employing two lognormal distribution curves.

High-resolution transmission electron microscopy (HRTEM/TEM) images of the prepared β-Fe₂O₃ sample (see Fig. 1d,e) indicate that it contains nanoparticles of two distinct size classes (hereafter referred to as smaller and larger nanoparticle assemblies). It turns out that the two size fractions are well described (employing the χ²-test performed on a statistical level of confidence of 99%) in terms of lognormal distribution curves with average particle sizes of 15.6 and 52.3 nm and lognormal standard deviations of 0.34 and 0.41, respectively (see Fig. 1f). Assuming that the density of β-Fe₂O₃ is irrespective of particle size within the sample and nanoparticles are more or less spherical (see Fig. 1d,e), from the frequency vs. size distribution, the smaller and larger nanoparticle assemblies within the β-Fe₂O₃ sample account for 36.9(5) and 63.1(5) wt.%, respectively. It is believed that the formation of the two particle fractions is due to the solid-state reaction outlined in Eq. (1)⁴⁵. This thermally-induced process occurs at a temperature of 400 °C, which is well below the decomposition temperature of iron(III) sulphate and the transition temperature for the transformation of β-Fe₂O₃ into α-Fe₂O₃ (~500 °C)^2,5. However, the double sulphates formed by the reaction outlined in Eq. (1) may also be transformed into β-Fe₂O₃. As such, there are two parallel processes of β-Fe₂O₃ formation – the primary reaction and the subsequent decomposition of the double sulphates – leading to the formation of the two particle size fractions observed after complete removal of all the sulphate-based by-products via dissolution in water.

The effect of pressure treatment on the crystal structure of β-Fe₂O₃ was investigated using high-pressure synchrotron radiation XRD measurements. Representative high-pressure synchrotron XRD spectra are shown in Fig. 2a and the detailed Rietveld analyses of all the measured synchrotron radiation XRD patterns (including the values of the R_wp-factor) are depicted in Supplementary Figures S1–S7 in the Supplementary Material. At pressures of up to 10 GPa, the sample consist of β-Fe₂O₃ and α-Fe₂O₃ in approximately the same ratio as in the original sample (93/7 wt.%). This reflects the pressure stability of β-Fe₂O₃ up to 10 GPa. The crystal structure of β-Fe₂O₃ was determined and that of the increasing α-Fe₂O₃ phase was refined sequentially between 10 and 30 GPa. Some of the β-Fe₂O₃ nanoparticles undergo polymorphous transformation to α-Fe₂O₃ but no other iron(III) polymorphs are observed (see Fig. 2a,c). At 29.6 GPa, the fractions of β-Fe₂O₃ and α-Fe₂O₃ were 35.1(2)% and 64.9(2)%, respectively (see Fig. 2c). Given the relative volume (mass) ratio of the two β-Fe₂O₃ particle size fractions in the starting material (as determined from TEM/HRTEM analysis, see above), this implies that the smaller nanoparticle assembly remains untransformed but the larger nanoparticle assembly readily converts into α-Fe₂O₃. One might expect this trend for the conversion of β-Fe₂O₃ into α-Fe₂O₃ to continue as the pressure increases further. However, there was a dramatic shift in the mechanism of the pressure-induced transformation when the applied pressure was raised above 30 GPa, with both the α-Fe₂O₃ and β-Fe₂O₃ phases undergoing new structural transformations. Specifically, α-Fe₂O₃ was converted into Rh₂O₃-II-type Fe₂O₃ (RO-Fe₂O₃, orthorhombic, Pbcn space group) and post-perovskite Fe₂O₃ (PPV-Fe₂O₃, orthorhombic, Cmcm space group) structures (see Fig. 2a,c). Both these phases have previously been observed during high-pressure treatment of α-Fe₂O₃. The simultaneous formation of perovskite and post-perovskite structures can be understood by considering the particle size (and, hence, volume) distribution within the larger nanoparticle assembly. The applied pressure forces the crystal structure of α-Fe₂O₃ to change but the magnitude of the change is highly dependent on the level of strain inside the nanoparticles, which tends to resist structural alteration. This strain in turn varies considerably with particle size. The transformation process that will occur is that associated with the lowest overall Gibbs free energy, which is influenced by the strain and associated stress. In addition to these (post)-perovskite structures, some α-Fe₂O₃ nanoparticles remain untransformed because their size (and, hence, strain) is such that they can resist the effect of the applied pressure.

Figure 2

Synchrotron radiation XRD data for the treated b-Fe₂O₃ phase.

Synchrotron radiation XRD patterns with Rietveld analysis acquired (a) at various elevated pressures and (b) at 0.1 MPa (atmospheric pressure) after pressure release. Red dots, black lines and gray dots indicate the observed patterns, fitted patterns and the differences between them, respectively. Fitted patterns for each phase are shown with green lines (β-Fe₂O₃), brown lines (α-Fe₂O₃), navy blue lines (RO-Fe₂O₃), pink lines (PPV-Fe₂O₃) and light blue lines (ζ-Fe₂O₃). The tick marks indicate the calculated positions of the Bragg reflections for each phase. (d) Cell volume vs. pressure plots for β-Fe₂O₃ (green), ζ-Fe₂O₃ (light blue), α-Fe₂O₃ (brown), RO-Fe₂O₃ (navy blue) and PPV-Fe₂O₃ (pink). The open triangles show the cell volume of ζ-Fe₂O₃ (light blue) and α-Fe₂O₃ (brown) after pressure release. (e) SAED pattern of the sample after pressure release with identification of the most intense diffraction plane belonging to α-Fe₂O₃ and ζ-Fe₂O₃ (no diffraction corresponds to β-Fe₂O₃).

Surprisingly, β-Fe₂O₃ was found to transform into a completely new crystal structure following the Rietveld refinement of the synchrotron radiation XRD patterns recorded at pressures above 30 GPa (see Fig. 2a,c). The analyses were carried out adopting the following scenario. At high pressures (42.9–64.4 GPa), α-Fe₂O₃ transforms into RO-Fe₂O₃ and PPV-Fe₂O₃ with consistent transition pressures compared with the previous reports^{26,27,28,29,30,31,32,33}. On the other hand, β-Fe₂O₃ transits to a different new phase. By indexing these new peaks, we searched for its space group and found a suitable candidate of the crystal structure. The new peak pattern was found to belong to a single phase with a monoclinic crystal structure in the I2/a space group with the lattice parameters similar to β-Fe₂O₃. We know that the structure of this new phase is caused by lowering the symmetry from cubic β-Fe₂O₃ to a monoclinic structure. We designated this new phase as ζ-Fe₂O₃. ζ-Fe₂O₃ has thus a monoclinic crystal structure with a space group of I2/a and lattice constants of a = 9.17 Å, b = 9.30 Å, c = 8.50 Å, angle β = 97.6° and a unit volume of V = 718.4 ų at 42.9 GPa. While the new iron(III) oxide phase has several structural features that resemble those of its precursor (cubic β-Fe₂O₃), it also exhibits some unusual pressure-induced changes in its crystal lattice. In particular, the octahedral Fe b-site splits into two non-equivalent Fe sites (Fe1 and Fe4 with a 1:1 ratio), the octahedral Fe d-site splits into four non-equivalent Fe sites (Fe2, Fe3, Fe5 and Fe6 with a 1:1:2:2 ratio) and the β angle between the a- and c-axis increases to 98° compared to 90° for β-Fe₂O₃. At pressures of 42.9–64.4 GPa, the relative abundances of the four phases, i.e., α-Fe₂O₃, RO-Fe₂O₃, PPV-Fe₂O₃ and ζ-Fe₂O₃ were almost constant (see Fig. 2c). This indicates that the ζ-Fe₂O₃ polymorph is much more stable at high pressures than γ-Fe₂O₃, which is transformed into α-Fe₂O₃ (and then into perovskite or post-perovskite phases) once the pressure exceeds ~37 GPa³⁸. Moreover, the pressure dependence of each phase was found to be monotonous and the cell volume (see Fig. 2d) and cell parameters (see Supplementary Figure S8) gradually changed as the pressure increased. When examining the XRD data, it should be stressed that the XRD patterns of β-Fe₂O₃ and ζ-Fe₂O₃ are totally distinct with different number of peaks. This implies that the symmetry of the crystal structure became lower (cubic → monoclinic) and the XRD pattern of ζ-Fe₂O₃ cannot be reproduced by modifying the pattern of β-Fe₂O₃ considering the effects of strains and defects (peak shift, peak broadening, changing the peak intensities, etc.).

After releasing the pressure, both RO-Fe₂O₃ and PPV-Fe₂O₃ spontaneously reverted to the more thermodynamically stable α-Fe₂O₃ polymorph (see Fig. 2b). However, strikingly, the ζ-Fe₂O₃ phase retained its crystal structure after the pressure release, as shown in Fig. 2b,c. The presence of α-Fe₂O₃ and ζ-Fe₂O₃ phase in the sample after the pressure release was further evidenced by analyzing selective area electron diffraction (SAED) pattern (see Fig. 2e) where planes belonging to α-Fe₂O₃ and ζ-Fe₂O₃ phase were identified (other, not assigned planes most probably come from the matrix to which the sample was pressed). Moreover, the sample after pressure release features nanoparticles with an average particle size of 66.2 nm and log-normal particle size distribution from ~8 to ~150 nm (see Supplementary Figure S9), broadening significantly the diffraction lines in the XRD patterns. Then, the lattice constants of ζ-Fe₂O₃ at room temperature and atmospheric pressure were found to be a = 9.863 Å, b = 10.00 Å, c = 8.949 Å, β = 101.10° and V = 850.4 ų and its crystal structure falls within the I2/a space group. The crystal structure of stable ζ-Fe₂O₃ is shown in Fig. 3 and Supplementary Figure S10; the atomic coordinates of the iron and oxygen sites in the ζ-Fe₂O₃ crystal structure are listed, together with lattice parameters, cell volume and R_wp-factor, in Table 2.

Table 2 Crystal structure and lattice parameters of z-Fe₂O₃ phase at atmospheric pressure and room temperature with atomic coordination of iron and oxygen atoms.

Figure 3

Crystal structure of z-Fe₂O₃ (monoclinic, I2/a space group) after pressure release, projected along the b-axis.

Red, blue and gray balls represent Fe sites split from d-sites in β-Fe₂O₃ (designated Fe1, Fe4), Fe sites split from b-sites in β-Fe₂O₃ (designated Fe2, Fe3, Fe5, Fe6) and oxygen sites, respectively.

No iron(III) oxide phase with a monoclinic crystal structure that is stable at atmospheric pressure and room temperature has ever been identified before. Thus, ζ-Fe₂O₃ can be regarded as a new member of the iron(III) oxide polymorphic family. It is known that the stability of nanoscale Fe₂O₃ polymorphs is governed by two factors: the Gibbs free energy of the different i-Fe₂O₃ phases (i = α, β, γ, ε) and the energy barrier associated with the phase transformation²⁵. These two parameters in turn depend on many factors such as different phases’ kinetics of formation and (nano)structural properties of the phases’ particles such as their surface-to-volume ratios. The Gibbs free energy involves the chemical potential and the surface energy. It is generally accepted that surface energy and surface stress/strain are the key properties of nanoparticles that determine the formation and stability of crystalline phases. Because both parameters are strongly related to the nanoparticles’ dimensions, the extent to which a given applied pressure can modify the particles’ crystal structure depends on their size. Therefore, at high pressures, smaller β-Fe₂O₃ nanoparticles tend to transform into ζ-Fe₂O₃ while larger β-Fe₂O₃ nanoparticles primarily transform into α-Fe₂O₃ and then to perovskite and post-perovskite Fe₂O₃ phases (see Fig. 4). We hypothesize that the pressure treatment also affects the chemical potential of ζ-Fe₂O₃ and that this change (together with changes in the particles’ surface energy) causes the Gibbs free energy of the ζ-Fe₂O₃ phase to become lower than that of α-Fe₂O₃ and β-Fe₂O₃ over a wide range of pressures and temperatures. Consequently, it remains stable when the pressure is released. Additionally, pressure effect on nanoparticles may be considered as one of the reasons for the larger volume of ζ-Fe₂O₃ compared to β-Fe₂O₃ and for ζ-Fe₂O₃ to be maintained after pressure release. For example, in the case of CeO₂ nanoparticles, volume expansion by pressure application has been reported due to the difference in the pressure-induced stress between the surface and the core of the nanoparticles⁴⁶. Such nanosize effect may be contributing in the present system as well since β-Fe₂O₃ is obtained as nanoparticles. Furthermore, the remaining stress in the nanoparticles after pressure release may also explain the reason for ζ-Fe₂O₃ to remain under atmospheric pressure.

Figure 4

The proposed mechanism of b-Fe₂O₃ transformation under pressure.

The magnetic properties of the sample after the pressure release were investigated by measuring the temperature dependence of its mass susceptibility, χ (see Fig. 5a). Its χ profile contains two pronounced maxima, one at ~69 K (designated T_N) and the other at ~269 K (designated T_M). On moving away from these temperatures, χ decreases, indicating a transition to an antiferromagnetic state. The profile of the maximum at ~269 K resembles that of the Morin transition of α-Fe₂O₃, i.e. the transition from a weakly ferromagnetic regime to an antiferromagnetic state accompanied by a 90° spin reorientation (from the α-Fe₂O₃ basal plane to the c-axis direction). The second maximum is associated with a sharp peak and indicates the Néel temperature of the ζ-Fe₂O₃ phase. The transition at ~69 K shows features characteristic of a second-order thermodynamic transition, thus T_N can be regarded as the thermodynamic temperature of the passage from a magnetically disordered (paramagnetic) state to a magnetically ordered (antiferromagnetic) regime. The origin of the fall in the Néel temperature ζ-Fe₂O₃ with respect to β-Fe₂O₃ can be explained in terms of changes in the lattice volume. The lattice volume of ζ-Fe₂O₃ is larger than that of β-Fe₂O₃, resulting in the decrease in strength of superexchange interactions. Following the equation reported by Bloch⁴⁷, ∂ln J = γ × ∂ln V (where J is the superexchange parameter (exchange integral), V is the lattice volume and γ is the constant value given as –10/3), it turns out that the J value decreased as 93% by the phase transformation from β-Fe₂O₃ to ζ-Fe₂O₃. As T_N depends on J, the decrease in J is then macroscopically manifested in the decrease in T_N as experimentally evidenced from the temperature behavior of χ. Here, it should be stressed that the decrease in T_N does not result from strains or defects because the temperature differential of χ (i.e., ∂χ/∂T) at T_N (69 K) is sharper than that at 269 K belonging to T_M of α-Fe₂O₃ (see Fig. 5b). If the decrease in T_N in β-Fe₂O₃ was promoted by the local disorder, the peak at T_N should be broadened. Therefore, T_N at 69 K is considered as the magnetic transition temperature of the ζ-Fe₂O₃ phase.

Figure 5

Magnetic property.

Thermal evolution of the (a) magnetic susceptibility (χ) and (b) ∂χ/∂T of the sample after pressure release, monitored under an external magnetic field of 20 kOe (T_N marks the Néel temperature of ζ-Fe₂O₃ and T_M represents the Morin transition temperature of α-Fe₂O₃). Light blue and brown shades indicate the signals from ζ-Fe₂O₃ and α-Fe₂O₃, respectively.

Conclusions

The pressure-induced transformation of the rare β-Fe₂O₃ phase has been studied for the first time, leading to the identification of a new iron(III) oxide polymorph, ζ-Fe₂O₃. The transformation of β-Fe₂O₃ into ζ-Fe₂O₃ occurs above 30 GPa and the new phase withstands pressures of up to ~70 GPa, which is well above the thresholds for the pressure-induced transformations of α-Fe₂O₃ or γ-Fe₂O₃. More strikingly, ζ-Fe₂O₃ remains stable after pressure release and at room temperature. This remarkable observation is explained in terms of its Gibbs free energy (and surface energy), which is partly due to structural properties inherited from its precursor material (small β-Fe₂O₃ nanoparticles) and partly due to stabilizing structural changes that occur during high pressure treatment. Its stability is thus strongly linked to the nanoscale dimensions of its particles. It has a monoclinic crystal structure belonging to the I2/a space group (a = 9.683 Å, b = 10.00 Å, c = 8.949 Å, β = 101.10° and V = 850.4 ų). The ζ-Fe₂O₃ phase behaves in an antiferromagnetic manner with a Néel transition temperature of ~69 K. It may also have other interesting electronic, optical and transport properties that would lend themselves to practical applications. Thus, in future, two challenges are viewed to be of significance importance stimulating further research in the iron(III) oxide realm: (i) to develop new methods for preparing ζ-Fe₂O₃ from ultrafine β-Fe₂O₃ nanoparticles, possibly by exploiting spatial restrictions, controlling the level of interparticle interactions (aggregation) during transformation and using thermal rather than pressure treatment; and (ii) to study the pressure-induced transformations of rare ε-Fe₂O₃.

Methods

Synthesis of b-Fe₂O₃ nanoparticles

β-Fe₂O₃ nanoparticles were synthesized by the thermally-induced solid-state reaction of NaCl with Fe₂(SO₄)₃ in air followed by post-processing separation based on dissolution of all by-products in water as described previously⁴⁵.

In-situ high-pressure X-ray diffraction experiments with synchrotron radiation

High-pressure X-ray powder diffraction experiments with synchrotron radiation were performed using a diamond anvil cell high-pressure apparatus⁴⁸. A powdered β-Fe₂O₃ sample was loaded into a 50–100 μm diameter hole that was drilled into a rhenium gasket. Several ruby crystals were also put into the sample chamber. No pressure transmitting medium was used in this study. The applied pressure was determined by monitoring the fluorescence line of ruby⁴⁹ and ranged from 0.1 MPa to 64.4 GPa. In all cases, the desired pressure was established by gradually increasing the applied pressure. At selected pressures, the sample was probed by angle-dispersive X-ray diffraction using the NE1A synchrotron beam line at the Photon Factory in Japan. A monochromatic incident X-ray beam with a wavelength of λ ≈ 0.41 Å was used. The X-ray beams were collimated to a diameter of 30 μm and the angle-dispersive X-ray diffraction patterns were obtained on an imaging plate (Rigaku) with 3000 × 3000 pixels. The distance between the sample and the detector was ~320 mm. The observed intensities on the imaging plates were integrated as a function of 2θ in order to obtain conventional one-dimensional diffraction profiles; details of the experimental procedure are presented elsewhere⁵⁰.

For indexing the peaks and searching the space group, the JADE software from Materials Data., Inc., (MDI) was employed. Rietveld analyses were performed using the PDXL Integrated X-ray powder diffraction software package from Rigaku. When analyzing the synchrotron radiation XRD patterns, the phase fractions, lattice parameters, peak width and decay parameters were refined.

Conventional experimental techniques – X-ray powder diffraction, ⁵⁷Fe Mössbauer spectroscopy, magnetization measurements and transmission electron microscopy

XRD analysis of the initial β-Fe₂O₃ sample was recorded on a PANalytical X´Pert PRO diffractometer in the Bragg-Brentano geometry, equipped with an iron-filtered CoK_α radiation source, an X´Celerator detector, a programmable divergence and diffracted beam anti-scatter slits. Generally, 200 μL of a sample suspension was dropped onto a zero–background single–crystal Si slide, allowed to dry under vacuum at room temperature and scanned in continuous mode (resolution of 0.017° in 2θ, scan speed of 0.008° in 2θ per second, 2θ range from 20° to 105°) under ambient conditions. The commercially available standards SRM640 (Si) and SRM660 (LaB6) supplied by the National Institute of Standards and Technology (NIST) were used to evaluate line positions and instrumental line broadening, respectively. The acquired pattern was processed using the X´Pert HighScore Plus software package (PANalytical, The Netherlands) in combination with the PDF-4+ and ICSD databases.

The room-temperature transmission ⁵⁷Fe Mössbauer spectrum of the initial β-Fe₂O₃ sample was recorded with a Mössbauer spectrometer operating in constant acceleration mode and equipped with a 50 mCi ⁵⁷Co(Rh) source of γ-rays. The collected Mössbauer spectrum was fitted using Lorentzian line shapes with the MossWinn software package based on the least-square method. The isomer shift values were referenced to a metallic α-Fe sample at room temperature.

TEM images and SAED pattern were obtained using a JEOL JEM–2010 electron microscope operating at 160 kV with a point–to–point resolution of 1.9 Å. For each measurement, a drop of a very dilute dispersion of the sample was placed on a copper grid with a holey carbon film and allowed to dry under vacuum at room temperature. HRTEM images were obtained using a TITAN 60–300 high-resolution transmission electron microscope with an X-FEG type emission gun, operating at 80 kV. For HRTEM analyses, the powder β-Fe₂O₃ sample was dispersed in ethanol and ultrasonicated for 5 minutes. One drop of the resulting suspension was then placed on a copper grid covered with a holey carbon film and allowed to dry at room temperature.

A superconducting quantum interference device (SQUID) magnetometer (MPMS XL-7 type, Quantum Design, U.S.A.) was used to measure the magnetization of the β-Fe₂O₃ sample after pressure release. The temperature evolution of the sample magnetization was recorded under an external magnetic field of 20 kOe in the sweep mode at temperatures ranging from 5 to 300 K. The gathered data were corrected to account for the paramagnetic and diamagnetic contributions from the material the sample was pressed into.