1. Introduction

Iron oxide (Fe₂O₃) exists in a variety of structural polymorphs, each of which is utilized in a wide range of applications due to its distinct physical and chemical properties[1–6]. Representative polymorphs of Fe₂O₃, include the alpha (α), beta (β), gamma (γ), and epsilon (ε) phases, with their distinctive properties summarized in Table 1 [4, 5, 7–10]. α-Fe₂O₃ (hematite), the most thermodynamically stable phase, has a corundum structure (space group = R3¯c) and is widely utilized in pigments and photocatalysts. β-Fe₂O₃, a metastable phase with a bixbyite structure (space group = Ia3¯), has potential applications in advanced energy materials [11, 12] and environmental technologies[13, 14]. γ-Fe₂O₃ (maghemite), which adopts a spinel structure (space group = P4₁2₁2₁), is primarily utilized in the form of nanoparticles as a soft magnetic material for magnetic recording media and biomedical applications. Among the polymorphs, epsilon iron oxide (ε-Fe₂O₃) is distinguished by its orthorhombic structure (space group = Pna2₁) and its entirely unique electromagnetic properties, which recently garnered significant attention as a potential electromagnetic functional material (Fig. 1) [3, 5].

ε-Fe₂O₃ exhibits unique magnetic and electronic properties including extremely high magnetic anisotropy and coercivity (H_c) comparable to or exceeding rare earth-based magnets [6, 15]. These properties enable the utilization of ε-Fe₂O₃ in advanced magnetic applications such as millimeter-wave absorbers, high-density magnetic recording media, and spintronics [3, 16, 17]. Additionally, ε-Fe₂O₃ also exhibits semiconductor properties with a bandgap that allows it to absorb visible light, suggesting its potential as a photocatalyst for hydrogen production [9]. These characteristics suggest that ε-Fe₂O₃ is a versatile material capable of performing various functions beyond simple magnetic materials. However, to maximize the electromagnetic functionality of the iron oxides, it is essential to it is essential to synthesize high purity ε phase. As ε-Fe₂O₃ is an intermediate phase between α-Fe₂O₃ and γ-Fe₂O₃, it is only appears in the form of nanoparticles with diameters ranging from 8 to 50 nm, which requires more demanding synthesis conditions than other polymorphs [6]. Recent advances in wet/dry powder preparation and advanced deposition techniques have enabled ε-Fe₂O₃ materials to be stably prepared not only as nanoparticles but also as nanofilms, greatly expanding their potential for electromagnetic applications [18, 19].

In this paper, we will discuss the electromagnetic properties of ε-Fe₂O₃ among the many polymorphs of Fe₂O₃, and review the latest synthesis methods and applications to fully utilize its properties. Through this, we will revisit the scientific and technological significance of ε-Fe₂O₃ and investigate its potential as a next-generation electromagnetic functional material.

2. Properties and Synthesis of ε-Fe₂O₃

2.1. Structure and Properties of ε-Fe₂O₃

Among the polymorphs of iron oxide, ε-Fe₂O₃ exhibits distinct structural and magnetic properties that set it apart as a promising material for advanced applications. Notably, ε-Fe₂O₃ is characterized by its unique orthorhombic crystal structure, where Fe ions have four sites: three octahedral sites (Fe_AO₆, Fe_BO₆, Fe_CO₆) and one tetrahedral site (Fe_DO₄) located at the center of the crystal, with a spin of S=5/2[5].

Molecular-field theory calculations suggest that the magnetism of ε-Fe₂O₃ at room temperature arises from super-exchange interactions between Fe ions mediated by O. The spins of Fe_B and Fe_C are aligned upward along the a-axis, while Fe_A and Fe_D exhibit Neel P-type ferrimagnetism, with their spins aligned downward along the a-axis [3, 20]. ε-Fe₂O₃ exhibits a coercivity (H_c) of over 2 T at room temperature[6], which is fourfold greater than of conventional M-type ferrites and nearly twice that of NdFeB. This ultra-high coercivity can be explained by two factors. First, ε-Fe₂O₃ is only formed in a specific range of particle size conditions (8 nm to 50 nm), which corresponds to the single magnetic domain particle size that exhibits the maximum coercivity [5]. The actual iron oxide crystal phase undergoes a phase transition in the sequence of γ → ε → β → α phase depending on the size of the powder. This transition is driven by the relationship of the free energies of the nanopowders as shown in (Eq. 1).

(1)

Gi=GBi+6VmdGsi

Where G(i) is the free energy of the nanopowder, G_B(i) is the free energy of the bulk state, V_m is the molar volume, d is the diameter of the nanopowder, G_S(i) is the surface free energy, and i represents the phase type. Since the magnitudes of bulk and surface free energies vary across different crystalline phases (G_B(γ)>G_B(ε)>G_B(β)>G_B(α), G_S(γ)<G_S(ε)<G_S(β)<G_S(α), the relationship between the free energy G(i) of the nanopowder and the diameter (d) of the powder is given by (Fig. 2) [3]. Specifically, within a certain particle size range, the free energy G(ε) of the ε crystalline phase is the lowest, indicating that the ε-phase forms with high purity in the single magnetic domain particle size range (8 nm to 50 nm). Furthermore, the high coercivity of ε-Fe₂O₃ is attributed to its magnetic anisotropy fields (H_a). H_a is proportional to the crystal magnetic anisotropy constant (K₁) and inversely proportional to the saturation magnetization (M_s) value as expressed in (Eq. 2).

(2)

Ha=2K14πMs

The crystal magnetic anisotropy constant is a physical parameter that quantifies the anisotropy of energy in the magnetization direction, which depends on the crystal structure of the material. A higher value indicates greater stability of the magnetization direction. The crystal magnetic anisotropy constant (K₁) of epsilon iron oxide is calculated to be 2×10⁶ erg/cm³, which is significantly larger than the K₁ value of γ-Fe₂O₃ (~ 10⁴ erg/cm³) and the K₁ value of α-Fe₂O₃ (~ 10⁵ erg/cm³) [5]. This large K₁ of ε-Fe₂O₃ has been reported to be due to the strong orbital hybridization of Fe and O [20]. Generally, Fe³⁺ has a d⁵ electron configuration, resulting in zero orbital angular momentum (L). However, in the case of ε-Fe₂O₃, the orbital hybridization of Fe and O alters the Fe ion’s electron configuration to d^5+q, instead of the standard d⁵ electron configuration. This increases the orbital angular momentum of Fe, leading to stronger spin-orbit coupling and consequently, a high coercive force [21].

Similar to other iron oxide polymorphs ε-Fe₂O₃ exhibits distinct band gap characteristics and semiconductor properties. The valence band in ε-Fe₂O₃ primarily consists of O2p orbitals, while the conduction band is dominated by Fe3d orbitals. Density of states (DOS) analysis from previous studies[21] shows that in ε-Fe₂O₃, the occupied Fe3d band located between -8.3 and -6.6 eV, the O2p band spans from -6.3 to -0.8 eV, and the unoccupied Fe3d band extends from +0.7 to +3.2 eV relative to the Fermi level (Fig. 3(a)) [21]. The resulting band gap is estimated to be 1.6-1.9 eV for ε-Fe₂O₃, compared to the 2.0 to 2.2 eV band gap of α-Fe₂O₃, as shown in Table. 1. This band structure characterizes ε-Fe₂O₃ as a charge-transfer insulator, where electronic transitions involve electron transfer from O2p to Fe3d states [21, 22]. Such charge-transfer transitions enable efficient electron excitation under visible light, making ε-Fe₂O₃ suitable for photocatalytic applications. Density functional theory (DFT) calculations further reveal that the heterostructure of ε-Fe₂O₃ and α-Fe₂O₃ forms a type-III broken band-gap alignment, with the band edges of α-Fe₂O₃ positioned higher in energy than those of ε-Fe₂O₃ [22]. This alignment facilitates spontaneous spatial separation of photogenerated charge carriers, with electrons transferring from α-Fe₂O₃ to ε-Fe₂O₃ and holes moving in the opposite direction (Fig. 3(b)) [22]. Such charge separation reduces electron-hole recombination, enhancing photocatalytic efficiency compared to using a single material. These electronic and magnetic properties highlight the potential of ε-Fe₂O₃ as a multifunctional material. Its tunable electromagnetic properties enable diverse applications, as explored in subsequent sections.

2.2. Fabrication Methods for ε-Fe₂O₃

The unique crystal structure and excellent electromagnetic properties of ε-Fe₂O₃ show that it has significant potential as an electromagnetic functional material. However, unlike α-Fe₂O₃, which is abundant in nature, and γ-Fe₂O₃, which is easily synthesized, ε-Fe₂O₃ is strongly influenced by the surface free energy, and the ε crystalline phase is only forms under conditions that maintain a particle morphology with diameters in tens of nanometers. In 2004, high purity epsilon iron oxide was artificially synthesized by the Ohkoshi group using the reverse micelle sol-gel method [6]. In this process, surfactants such as cetyltrimethylammonium bromide (CTAB) and cetyltrimethylammonium chloride (CTAC) were used above the critical micelle concentration to create reverse micelles in oil. Then, a water-soluble iron salt solution was added to trap Fe ions in the micelles and disperse them in a lipophilic solvent, followed by the addition of silica precursors to initiate a sol-gel reaction [17, 23, 24]. This process involved emulsifying Fe ions with surfactants, trapping them in a silica matrix, which restricted the growth of iron oxide particles during subsequent heat treatment, leading to the formation of nanometer-sized ε crystalline phases. However, the use of surfactants and oils to create reverse micelle, combined with 12-hour sol-gel reaction is inefficient in terms of yield and cost. Accordingly, efforts have been made to reduce the overall reaction time of the silica sol-gel reaction to 2 hours, though a multi-step batch process [15]. Additionally, studies have been conducted on synthesizing ε-Fe₂O₃ using cost-effective raw materials and scalable production methods. The clay mineral nontronite, an iron-rich silicate, is inexpensive, readily available, and requires no additional pretreatment, facilitating its use in low-cost synthesis [25]. Similarly, the ball milling process, a scalable mechanochemical technique for producing fine powders, has been utilized to synthesize ε-Fe₂O₃ [26]. While both methods effectively produce the ε-Fe₂O₃, they may have limitations related to secondary phase formation and elemental substitution for Fe.

Recently, a study reported the continuous preparation of ε-Fe₂O₃ nanoparticles using a spray-drying method although this method only produced micrometer-sized powders (Fig. 4) [27]. In this approach, a water-soluble iron salt and silica precursor are homogeneously mixed to form a clear solution, which is continuously injected through a nozzle, and atomized in a high-temperature reactor. As the solution dries instantaneously, the silica forms a micrometer-sized xerogels, within which the Fe precursor is trapped, resulting in a composite powder. During subsequent heat treatment, when Fe ions inside the silica xerogel combine with oxygen to transition to the iron oxide crystalline phase. The silica matrix restricts particle growth, allowing the final particle size to be reduced 50 nm or less, leading to high purity (92.7%) formation of the ε crystalline phase. The aerosol method has been a widely used continuous manufacturing process that has been widely used in the industrial preparation of fine powders for pharmaceuticals, food, and chemicals, can produce powders in a single step without additional washing. Moreover, the method is versatile in combining different compounds into particles and allows the synthesis of composites containing dispersed Fe and Si elements without the use of surfactants. Therefore, it is expected that a commercialized technology capable of producing pure ε-phase at high yields will soon be realized. These advancements in cost-effective and scalable synthesis methods for ε-Fe₂O₃ hold significant potential for applications requiring ferromagnetic nanoparticles with high magnetic anisotropy. In particular, these nanoparticles are anticipated to find extensive use in fields such as magnetic recording media and electromagnetic wave absorbers, as further detailed in the following sections.

The ε-Fe₂O₃ crystalline phase, which occurs on the scale of a few to several tens of nanometers, has recently actively researched for the fabrication of thin films and their application in spintronic devices [28]. Spintronics devices, such as spin valves, magnetic tunnel junctions, and spin current devices, are integral to magnetoresistive random access memory (MRAM), a spintronic memory based on these devices, has been utilized as a non-volatile memory [29–32]. The operation principle of MRAMs is based on the tunnel magnetoresistance (TMR) effect, where the spin of electrons tunnels through an insulating layer located between two ferromagnetic layers, causing a change in electrical resistance depending on magnetization direction of the two layers. The change in resistance is used to store data as 0 and 1, and the magnetization direction is retained even when the power is turned off, ensuring non-volatility. One of the ferromagnetic layers has a fixed magnetization direction, requiring an anisotropic material, while the other is composed of a soft magnetic material to allow its magnetization direction to change in response to an external magnetic field or current[30]. In addition, to improve data stability and the efficiency of reading and writing, thin film fabricated in the form of single crystal rather than polycrystals, where magnetization directions may be distributed in multiple directions are more suitable for application in magnetic recording device.

Although ε-Fe₂O₃ exhibits high magnetic anisotropy, its crystalline phase, which only appears in nanoparticle form, has not been suitable for application in substrate-based devices. In previous studies, researchers attempted to immobilize a mixture of ε-Fe₂O₃ nanorods composites onto wafers using chemical vapor deposition (CVD) [33], but a significant reduction in magnetic properties was observed compared to the nanoparticle form. This reduction was speculated to result from decreased ordering of the crystal structure [18]. Therefore, it is crucial to study the preparation of ε-Fe₂O₃ in thin film form while preserving its unique magnetic properties, and various preparation methods have been tried. Gich group used pulsed laser deposition (PLD) to form single crystal thin films, stabilizing the ε crystalline phase through epitaxial modification on SrTiO₃ single crystal substrates, or by using a GaFeO₃ buffer layer with the same structure as ε-Fe₂O₃ to promote the growth of ε-Fe₂O₃ thin films. These thin film exhibited a coercivity of about 8 kOe [34]. Also, in 2017, Corbellini group successfully achieved the growth of (001)-oriented ε-Fe₂O₃ epitaxial thin films on single crystal zirconia (YSZ) (100) substrates using PLD (Fig. 5) [18]. This study is the result of growing ε-Fe₂O₃ thin films directly on the YSZ substrate, which are commonly used to grow oxides on silicon wafers in the (100) orientation. This suggests the potential for directly growth of ε-Fe₂O₃ thin films on silicon wafers, an important milestone for expanding applications to large-area devices. The Curie temperature of the prepared thin film was 460 K, showed a similar level to the Curie temperature observed in the nanoparticle form (approximately 490 K). However, the PLD has limitations, including the needs for high deposition temperature (700-900 K) and lattice-matched substrate. For the practical application of ε-Fe₂O₃ thin films, it is essential to develop methods that can form thin films at lower temperatures or to explore new fabrication technique compatible with a wider range of substrates.

Recently, a method for fabricating high-purity ε-Fe₂O₃ thin films without a lattice-matched substrate at a low temperature of 280°C using the atomic layer deposition (ALD) has been devised [19]. ALD is a chemical vapor deposition method that uses chemical precursors under relatively low vacuum (10⁻² to 10 mbar) and low temperature (< 400 °C), enabling atomically precise thickness control and fine composition control. The iron oxide thin films prepared by ALD are ε crystalline phase, with only 2.5% α crystalline phase impurities. As shown in the M-H curve (Fig. 6) [35], there was no kink resulting from soft magnetic phase (typically γ-Fe₂O₃) admixture, confirming the formation of high-purity ε-Fe₂O₃ thin films. Furthermore, ⁵⁷Fe Mössbauer measurements identified the four distinct Fe sites inherent to the ε crystalline phase. While the low-temperature transition observed in nanoparticles between 110 K and 150 K was shifted to the higher temperature range in the thin film, the transition was still present. This shift in transition temperature, characteristic of the thin film, was attributed to grain boundaries and substrate-induced strain. The use of ALD for ε-Fe₂O₃ thin film holds significant potential for broader applications, as it does not require a lattice-matched substrate. The development represents an advancement in expanding the application range of ε-Fe₂O₃. Furthermore, it is expected to play a critical role in advanced technological applications, such as photocatalytic devices that require its semiconducting properties and efficient charge carrier transport, as discussed in subsequent sections.

3. Applications of ε-Fe₂O₃

3.1 Millimeter-Wave Absorption Applications

The high magnetic anisotropy of ε-Fe₂O₃ has attracted attention for its potential application as a millimeter wave (30-300 GHz) absorber and is currently under active investigation. Electromagnetic wave absorbers are classified into conduction loss, dielectric loss, and magnetic loss materials, which attenuate the incident electromagnetic waves. Among these, magnetic loss materials offer the advantage of more effective electromagnetic wave absorption in high frequency bands, due to magnetic loss caused by the imaginary component of permeability, allowing them to selectively absorb electromagnetic signals in specific frequency bands. This phenomenon is attributed to the ferromagnetic resonance of magnetic materials, where ferromagnetic materials in a magnetic field selectively absorb specific frequencies. When a ferromagnet is exposed to electromagnetic waves, its spins precess around the magnetization axis, absorbing electromagnetic waves at a frequency that matches the precession. The natural resonance frequency (f_r) is proportional to the anisotropic field (H_a) as shown in the equation (3) below [35], where v is the gyromagnetic constant.

(3)

fr=v2π×Ha

Therefore, to adjust the frequency band to be absorbed, the magnetic anisotropy of the magnetic material must be adjusted, which can be achieved through control of the material composition. ε-Fe₂O₃ has been reported to exhibit ferromagnetic resonance absorption at approximately 180 GHz due to its high H_a [17]. The crystalline magnetic anisotropy of ε-Fe₂O₃ can also be controlled by substituting transition metal element at the Fe sites, and studies have been conducted to adjust the absorption band accordingly (Fig. 7(a)-(b)) [17, 36]. It has been reported that substitution Fe³⁺ (S = 5/2) with non-magnetic ion such as Ga³⁺ and Al³⁺ (S = 0) can modulate f_r over a wide range by substituting Fe_D or Fe_A sites. When substituted with Ga (ε-Ga_xFe_2-xO₃) and Al (ε-Al_xFe_2-xO₃), it has been reported that f_r can be adjusted from 35 GHz (x = 0.67) to 147 GHz (x = 0.10)[17] and from 182 GHz (x = 0.04) to 112 GHz (x = 0.4), respectively, depending on the content (x = 0.10 ~ 0.67) and (x = 0.04 ~ 0.4)[36]. Conversely, f_r was found to increase when the spin-orbit interaction was enhanced by substituting Rh³⁺, which has a larger orbital angular momentum than Fe³⁺ [37]. With Rh substitution, H_c increased up to 3.5 T [16], and for ε-Rh_0.14Fe_1.86O₃, a H_c of 2.7 T and f_r of 209 GHz were observed (Fig. 7(c)) [37].

In addition, complex substitution of various elements on ε-Fe₂O₃ is also being actively studied. A representative example is the co-substitution of Co²⁺(S=3/2), Ti⁴⁺ and Ga³⁺ (ε-Ga_xTi_yCo_yFe_2-x-yO₃). This combination is notable because the large magnetic anisotropy in the a-axis direction, caused by the orbital angular momentum of Fe³⁺ at the Fe_B site, is counteracted by the magnetic anisotropy of Co²⁺, which substitutes for Fe_D along the c-axis direction (Fig. 7(d)-(e)) [38]. As a result, the crystalline magnetic anisotropy is significantly reduced, lowering the coercivity to below 3 kOe [38]. In addition, to utilize this material as a more efficient millimeter-wave absorber, a study was conducted in which ceramic nanopowders with high conductivity (Ti₄O₇) were applied to achieve both dielectric loss and magnetic loss effect [16, 38]. This resulted in a high dielectric constant, as electron transfer on the Ti₄O₇ surface was blocked at the iron oxide nanoparticle interface, inhibiting the current flow. When prepared as a thin film (216 ㎛), it exhibited 99.8% absorption capacity (RL = -27.2 dB) at 80 GHz with a broadband width of 16.2 GHz (Fig. 8) [39]. ε-Fe₂O₃ can serve as an effective absorber in the millimeter-wave band due to a large magnetic anisotropy, and the absorption band can be tuned by controlling the substitutional composition. This demonstrates its potential as a multifunctional absorber capable of responding to wide band of electromagnetic waves, making it suitable for 6G mobile communication applications.

3.2 High-Density Magnetic Recording Media Applications

In the era of big data, the need for high-density magnetic recording media capable of reliably storing and managing large amounts of data over long periods become increasingly important [40–43]. Currently, magnetic recording tapes are mainly composed of magnetic materials such as spindle Co-Fe alloy nanoparticles or barium ferrite, which play an crucial role in long-term data storage across various fields such as insurance, finance, broadcasting, and web services [44–47]. To achieve high-density recording, it is essential to reduce the size of the magnetic filler particles used in these tapes. However, the reduction in size lead to several side effects. In the case of metallic alloys, a decrease in particle size increases the surface area, which raises the potential for oxidation and can result in pyrophoric character. On the other hand, iron oxides such as barium ferrite and magnetite do not exhibit pyrophoric character, but as their particle size decreases, they may lose magnetic ordering, potentially leading to a transition to superparamagnetism. As a solution to this issue, ε-Fe₂O₃ particles have been found to have a superparamagnetic limit at a particle size of 7.5 nm, allowing them to maintain a smaller size compared to barium ferrite (20-25 nm) [48]. This indicates that ε-Fe₂O₃ has significant potential as an ultra-high-density magnetic recording media.

Furthermore, magnetic particles in the form of multiple magnetic domains, rather than single magnetic domain are likely to have their spins only partially aligned due to their magnetic anisotropy. While this partial alignment does not affect data playback but it raises concerns about potential noise generation. Therefore, ε-Fe₂O₃ with its low superparamagnetic limit, well-maintained magnetic arrangement can be synthesized with single magnetic domain sphere sizes ranging from 8 to 30 nm, making it a suitable candidate for ultra-high-density magnetic recording media. However, ε-Fe₂O₃ exhibits a high coercivity of more than 20 kOe at room temperature, which can decrease the efficiency of information storage if used in its original form. To address this issue, research has been conducted to adjust the coercivity by substituting Fe with transition metal elements. Since magnetic fillers suitable for magnetic recording media typically require a coercivity of around 3 kOe, the Ohkoshi group has worked on modifying the coercivity by substituting element such as Ga, Ti and Co [16, 38, 49, 50].

In particular, the Ga_0.31Ti_0.05Co_0.05Fe_1.59O₃ case exhibits a magnetization value of 23.4 emu/g at 7 T, which represents a 44% increase from the original ε-Fe₂O₃ value of 16.2 emu/g. This increase is attributed to the substitution of Fe^3⁺ ions (S = 5/2) with Ga^3⁺ (S = 0), Ti^4⁺ (S = 0), and Co^2⁺ (S = 3/2) at rates of 48%, 10%, and 10%, respectively. The total magnetization of ε-Fe₂O₃ is determined by the sum of the positive sublattice magnetization (M_B, M_C > 0) at the Fe_B and Fe_C sites and the negative sublattice magnetization (M_A, M_D < 0) at the Fe_A and Fe_D sites. Ga substitution reduces the negative sublattice magnetization at Fe_D, thereby increasing the overall magnetization value. Furthermore, the coercivity (H_c) was adjusted to below 2.69 kOe through element substitution. This reduction in H_c is attributed to the compensating effect of the single-ion magnetic anisotropy of Fe^3⁺ and Co^2⁺. First-principles calculations indicate the ε-Fe₂O₃ crystal phase originally exhibits a strong magnetic anisotropy along the a-axis, which arises from the orbital angular momentum resulting from the hybridization between Fe^3⁺ and O^2⁻ions at the Fe_B site. However, the substitution of Co^2⁺ ions at Fe_D sites introduces magnetic anisotropy along the c-axis, which appears to offset the magnetic anisotropy along the a-axis, thereby reducing the coercivity to below 3 kOe [38]. The resulting ε-Fe₂O₃ nanoparticles were dispersed in a urethane resin and vinyl chloride polymer and then coated on a non-magnetic polyethylene (PE) film while an external magnetic field was applied (Fig. 9) [38]. During this process, the ε-Fe₂O₃ nanoparticles were oriented vertically, forming a coated magnetic tape. Compared to commercial CoFe alloys from Dowa Electric Materials, the magnetic recording tape produced exhibited a sharper power spectrum and a higher signal-to-noise ratio (S/N) than existing commercial products. This indicates the high potential of ε-Fe₂O₃ nanoparticles as a magnetic recording media. This study demonstrates that magnetic recording media utilizing the properties of ε-Fe₂O₃ can outperform current commercial materials, highlighting its potential application as a next-generation high-density magnetic recording media.

3.3 Visible light catalytic Applications

Solar hydrogen production is an environmentally friendly method to produce hydrogen by utilizing clean, renewable energy sources. It has gained significant attention as a key technology for building a sustainable hydrogen economy. The main approaches include photoelectrochemical (PEC) water splitting, photocatalytic water splitting, and photo-reforming using oxygenated organic compounds (OOCs), all of which aim to minimize greenhouse gas emissions and maximize energy efficiency. Recently, research has focused on the potential of ε-Fe₂O₃, which can absorb visible light.

Currently, PEC water splitting has attracted considerable attention from researchers, and it has been found that ferroelectric materials can play an important role as PEC catalysts. This is due to the internal electric field induced by their spontaneous electric polarization, which can enhance the separation of electron-hole pairs generated by sunlight[51-53]. Conventional ferroelectrics such as BaTiO₃ and Pb(Zr_x,Ti_1-x)O₃ have bandgaps greater than 3 eV, making them unsuitable for use in the visible light range. In contrast, BiFeO₃ possesses a relatively narrow bandgap (≈2.2–2.8 eV) and exhibits the largest spontaneous polarization (100 μC cm^-2) [54]. However, the solar energy conversion efficiency of BiFeO₃ remains limited due to the rapid recombination rate of photoexcited charge carriers[50]. On the other hand, Fe₂O₃ is a well-known photocatalyst for solar hydrogen production due to its visible light absorption (bandgap ≈ 2.1 eV), non-toxicity, and low cost [55-58]. However, its primary drawbacks low electrical conductivity, hole diffusion length of 2-4 nm, and low absorption coefficient have resulted in solar-to-hydrogen conversion values that are lower than theoretically predicted, with photocurrent reaching up to 12 mA cm^-2 [59, 60]. To overcome the limitations of these single photocatalysts and enhance solar energy conversion efficiency, a study on nanocomposite photocatalysts based on heterojunction of ε-Fe₂O₃ and BiFeO₃ was published[61]. In this study, a photoelectrode for solar water splitting was fabricated based on a heterostructure in which BiFeO₃ nanopillars were vertically embedded into an ε-Fe₂O₃ matrix thin film using the PLD (Fig. 10) [61]. By optimizing the photodegradation performance through the adjustment of the ratio between the two materials, the BiFeO₃-ε-Fe₂O₃ photoelectrode containing 9% ε-Fe₂O₃ exhibited more than double the photocurrent (0.19 mA cm^-2) compared to the single-material sample. Additionally, it maintained high stability under continuous light irradiation for 25 hours. This result can be explained by the charge separation effect caused by band-bending between the two materials. When the two materials are contacted, free charge carriers are redistributed across the interface until equilibrium is achieved, leading to the realignment of the Fermi (Fig. 10(g)-(h)) [61]. Since the work function of BiFeO₃ (approximately 4.95 eV) is smaller than that of ε-Fe₂O₃ (approximately 5.72 eV) [61-63], electrons migrate from BiFeO₃ to ε-Fe₂O₃. This migration forms a charge region at the heterointerface and induces band-bending, which results in hole accumulation (electron deficiency) at the BiFeO₃ interface. Due to this band-bending, photoexcited electrons can easily migrate from BiFeO₃ to ε-Fe₂O₃, while holes can migrate from ε-Fe₂O₃ to BiFeO₃, significantly inhibiting electron-hole recombination. In other words, ε-Fe₂O₃, with a bandgap corresponding to visible light, can be utilized as a photocatalytic material. Furthermore, it is a promising material for enhancing photocurrent by inducing charge separation when combined with existing materials.

ε-Fe₂O₃ has also been actively studied as a photocatalyst for the photoreforming of oxygen-containing organic compounds such as ethanol, glycerol, and glucose [9]. This approach to hydrogen production via biomass photoreforming offers significant environmental benefits by simultaneously removing pollutants and processing waste from the biomass industry [64-66]. Although α-Fe₂O₃ has traditionally been used as visible light catalyst due to its low fabrication cost and ease of processing into various nanostructures, its performance has been constrained by factors such as a low absorption coefficient, high recombination losses of charge carriers, and a short diffusion length of charge carriers [2, 67]. To retain the advantages of α-Fe₂O₃ while overcoming its disadvantages, ε-Fe₂O₃ and β-Fe₂O₃ polymorphs of Fe₂O₃ were prepared by chemical vapor deposition (CVD) and their performance was compared in this study. The catalytic activity of the Fe₂O₃ polymorphs increased in the order of α-Fe₂O₃ < ε-Fe₂O₃ < β-Fe₂O₃ with the average hydrogen production rate of ε-Fe₂O₃ reaching 125 mmol h^-1 m^-2, which was significantly higher than that of α-Fe₂O₃ [9]. Additionally, ε-Fe₂O₃ demonstrated better photocorrosion resistance than α-Fe₂O₃, indicating its ability to maintain photocatalytic activity over an extended period. Stable hydrogen generation was observed for at least 20 hours of continuous irradiation, demonstrating the robustness and reliability of ε-Fe₂O₃ as a photocatalyst. This study shows that ε-Fe₂O₃ is a promising alternative material that can overcome the limitations of α-Fe₂O₃ and make an important contribution to the development of high-performance photocatalysts.

4. Summary

Epsilon iron oxide (ε-Fe₂O₃) has attracted attention for various advanced applications, including millimeter-wave absorbers, high-density magnetic storage media, and photocatalysis, due to its unique magnetic and electronic properties. Specifically, its high magnetic anisotropy and coercivity set ε-Fe₂O₃ apart from conventional magnetic materials, enhancing its potential as an electromagnetic wave absorber in the millimeter-wave band and as a high-density magnetic storage media. In addition, its semiconducting property of visible light absorption suggests potential applications as a catalyst for solar hydrogen production. To maximize the electromagnetic functionality of ε-Fe₂O₃, it is crucial to synthesize it in high purity and stable form. Various synthesis methods have been developed, successfully yielding nanoparticles and thin films. Advances in the continuous manufacturing process of ε phase nanoparticles, which are particularly challenging to prepare, are expected to further promote the practical applications of ε-Fe₂O₃ across various applications.

Future research will need to focus on optimizing the properties of ε-Fe₂O₃ for each application and developing structural design and fabrication techniques to enhance performance. For example, magnetic property optimization and composite structuring to enhance absorption in high-frequency band, control of thin film structure for spintronics device applications, and design of heterostructure to improve efficiency as photocatalysts will become a key area of research. With continued research, ε-Fe₂O₃ is expected to become a core material in various industries including 6G wireless communication, spintronic devices, high-density magnetic storage, and hydrogen production, serving as a next-generation electromagnetic functional material.

Article information

Funding

This research was supported by the Basic Research Program (PNK9960) of Korea Institute of Materials Science.

Conflict of Interest Declaration

The authors declare no relevant conflicts of interest.

Author Information and Contribution

Jihyeong Jeong : Writing a manuscript and researching previous studies, Hwanhee Kim and Jung-Goo Lee: data organization and researching previous studies, Youn-Kyoung Baek: Supervision, reviewing and editing.

Acknowledgement

None.

Fig. 1.

(a) Graphical representations of the crystal structures of α-Fe₂O₃, ε-Fe₂O₃, and γ-Fe₂O₃. Adapted with permission from [5] Copyright 2009 American Chemical Society. (b) Crystallographic structure of the ε-Fe₂O₃ phase represented by the cation polyhedral. Adapted with permission from [3] Copyright 2010 American Chemical Society.

Fig. 2.

Stability of individual polymorphs of Fe₂O₃ based on the calculated dependence of the free energy per volume (i.e., G/V) on the size (d) of the iron(III) oxide nanoparticles of a particular polymorph. Reproduced with permission from [3] Copyright 2010 American Chemical Society.

Fig. 3.

(a) Density of states (DOS) of ε-Fe₂O₃, with the total DOS (black), iron DOS (red), and oxygen DOS (blue). Adapted with permission from [21] Copyright 2012 American Chemical Society, (b) Schematic diagram of band bending and the charge flow at the interface of ε-Fe₂O₃ and α-Fe₂O₃ heterostructure after connection. Adapted with permission from [22] Copyright 2020 Royal Society of Chemistry.

Fig. 4.

(a) Schematic illustration of the preparation of ε-Fe₂O₃ nanoparticles via spray drying. (b) and (c) Scanning electron microscopy (SEM) images of as-spray-dried precursor particles with a precursor molar ratio (Fe/Si) of 0.4:1. (d) and (e) Field-emission transmission electron microscopy (FE-TEM) images of ε-Fe₂O₃ nanoparticles embedded in SiO₂ particles after annealing at 1180°C for 4 h. (f) and (g) TEM images of the corresponding ε-Fe₂O₃ NPs after SiO₂ removal. Reproduced with permission from [27] Copyright 2022 Royal Society of Chemistry.

Fig. 5.

(a) STEM image of an approximately 100-nm-thick film of ε-Fe₂O₃ on YSZ (100) highlighting the formation of pillar-like twins, (b) details of the interface between the substrate and the film, evidencing the formation of “bubbles” of a foreign phase (most likely Fe₃O₄) at the interface. Reproduced with permission from [18] Copyright 2017 Nature.

Fig. 6.

(a) Magnetization hysteresis loop at 300 K of ε-Fe₂O₃/AlFeO₃//Nb:STO(111). (b) Ferroelectric hysteresis loop measured at 300 K and 10 Hz with 100 ms of delay time. Reproduced with permission from [34] Copyright 2014 Wiley-VCH.

Fig. 7.

Absorption spectra of (a) ε-Ga_xFe_2-xO₃. Adapted with permission from [17]. Copyright 2007 Wiley-VCH. (b) ε-Al_xFe_2-xO₃. Adapted with permission from [36]. Copyright 2008 American Chemical Society. (c) ε-Rh_xFe_2-xO₃. Adapted with permission from [37]. Copyright 2012 Nature. Magnetic structure of (d) ε-Fe₂O₃ and (e, left) ε -Ga_0.31Ti_0.05Co_0.05Fe_1.59O₃. Red and blue arrows denote the sublattice magnetizations of Fe³⁺ and Co²⁺, respectively. Black arrows show the total magnetization. (e, Right) The direction of the single-ion anisotropies (K, red and blue arrows) for Fe³⁺ at the B site along a-axis and Co²⁺ at the D site along c-axis. Adapted with permission from [38] Copyright 2016 Wiley-VCH.

Fig. 8.

(a) Schematic illustration of phase matching for absorption, (b) photograph, and (c) observed RL spectrum of the flexible millimeter-wave–absorbing ultrathin film composed of the Ti₄O₇@ε-Ga_0.21Ti_0.05Co_0.05Fe_1.69O₃ composite and acrylic acid ester polymer on a copper foil. Adapted with permission from [39] Copyright 2021 Wiley-VCH.

Fig. 9.

(a) Photograph of the manufactured magnetic recording tape composed of ε-Ga_0.31Ti_0.05Co_0.05Fe_1.59O₃. (b) Schematic illustration of the LTO-3 AMR head. (c) Cross-section SEM image of the manufactured magnetic tape. (d) Power spectrum of the signal of the ε-Ga_0.31Ti_0.05Co_0.05Fe_1.59O₃ tape (red line) and cobalt–iron alloy tape (gray line) measured with the LTO-3 AMR head. Reproduced with permission from [38] Copyright 2016 Wiley-VCH.

Fig. 10.

(a) Cross-sectional TEM image of BFO-FO heterostructures on STO substrate. (b) High-resolution TEM image showing the interfaces between films and STO substrates. The corresponding FFT patterns of (c) BFO and (d) ε-FO. (e) Top view. (f) Schematic of the self-assembled BFO-FO vertical heterostructures with BFO pillars embedded in the ε-FO matrix. (g), (h) Schematic diagrams illustrating the energy band alignment and the expected charge flow at the BFO-FO heterojunction under light excitation. Adapted with permission from [61] Copyright 2016 Wiley-VCH.

References

• 1. S. Piraman, S. Sundar, R. Mariappan, Y. Y. Kim and K. Min: Sens. Actuators, B Chem., 234 (2016) 386.Article
• 2. K. Sivula, F. Le Formal and M. Grätzel: ChemSusChem, 4 (2011) 432.Article
• 3. J. Tuček, R. Zbořil, A. Namai and S. I. Ohkoshi: Chem. Mater., 22 (2010) 6483.Article
• 4. L. MacHala, J. Tuček and R. Zbořil: Chem. Mater., 23 (2011) 3255.Article
• 5. S. Sakurai, A. Namai, K. Hashimoto and S. I. Ohkoshi: J. Am. Chem. Soc., 131 (2009) 18299.Article
• 6. J. Jin, S. I. Ohkoshi and K. Hashimoto: Adv. Mater., 16 (2004) 48.Article
• 7. A. S. Teja and P. Y. Koh: Prog. Cryst. Growth Charact. Mater., 55 (2009) 22.Article
• 8. I. Ahamed, R. Pathak, R. Skomski and A. Kashyap: AIP Adv., 8 (2018) 148.
• 9. G. Carraro, C. MacCato, A. Gasparotto, T. Montini, S. Turner, O. I. Lebedev, V. Gombac, G. Adami, G. Van Tendeloo, D. Barreca and P. Fornasiero: Adv. Funct. Mater., 24 (2014) 372.Article
• 10. O. Malina, J. Tuček, P. Jakubec, J. Kašlík, I. Medřík, H. Tokoro, M. Yoshikiyo, A. Namai, S. I. Ohkoshi and R. Zbořil: RSC Adv., 5 (2015) 49719.Article
• 11. C. Liu, N. Zhang, Y. Li, R. Fan, W. Wang, J. Feng, C. Liu, J. Wang, W. Hao, Z. Li and Z. Zou: Nat. Commun., 14 (2023) 1.
• 12. G. Carraro, D. Barreca, M. Cruz-Yusta, A. Gasparotto, C. MacCato, J. Morales, C. Sada and L. Sánchez: ChemPhysChem, 13 (2012) 3798.Article
• 13. M. M. Rahman, A. Jamal, S. B. Khan and M. Faisal: Superlattices Microstruct., 50 (2011) 369.Article
• 14. M. Li, J. Wu and G. Shen: Catal. Sci. Technol., 12 (2022) 2659.Article
• 15. E. Gorbachev, M. Soshnikov, M. Wu, L. Alyabyeva, D. Myakishev, E. Kozlyakova, V. Lebedev, E. Anokhin, B. Gorshunov, O. Brylev, P. Kazin and L. Trusov: J. Mater. Chem. C, 9 (2021) 6173.Article
• 16. H. Tokoro, A. Namai and S. I. Ohkoshi: Dalt. Trans., 50 (2021) 452.ArticlePDF
• 17. S. Ohkoshi, S. Kuroki, S. Sakurai, K. Matsumoto, K. Sato and S. Sasaki: Angew. Chem., 119 (2007) 8544.Article
• 18. L. Corbellini, C. Lacroix, C. Harnagea, A. Korinek, G. A. Botton, D. Ménard and A. Pignolet: Sci. Rep., 7 (2017) 3712.Article
• 19. T. Jussila, A. Philip, J. Lindén and M. Karppinen: Adv. Eng. Mater., 25 (2023) 2201262.Article
• 20. S. I. Ohkoshi, A. Namai and S. Sakurai: J. Phys. Chem. C, 113 (2009) 11235.Article
• 21. M. Yoshikiyo, K. Yamada, A. Namai and S. I. Ohkoshi: J. Phys. Chem. C, 116 (2012) 8688.Article
• 22. I. Ahamed, N. Seriani, R. Gebauer and A. Kashyap: RSC Adv., 10 (2020) 27474.Article
• 23. S. Sakurai, K. Tomita, K. Hashimoto, H. Yashiro and S. I. Ohkoshi: J. Phys. Chem. C, 112 (2008) 20212.Article
• 24. S. Sakurai, J. I. Shimoyama, K. Hashimoto and S. I. Ohkoshi: Chem. Phys. Lett., 458 (2008) 333.Article
• 25. K. Kelm and W. Mader: Z. Anorg. Allg. Chem., 631 (2005) 2383.Article
• 26. Y. Zhao and G. Wen: J. Magn. Magn. Mater., 512 (2020) 167039.Article
• 27. G. R. Jo, M. B. Yun, Y. Hun Son, B. Park, J. G. Lee, Y. G. Kim, Y. G. Son and Y. K. Baek: Chem. Commun., 58 (2022) 11442.Article
• 28. S. M. Suturin, A. M. Korovin, S. V. Gastev, M. P. Volkov, A. A. Sitnikova, D. A. Kirilenko, M. Tabuchi and N. S. Sokolov: Phys. Rev. Mater., 2 (2018) 1.Article
• 29. S. Bhatti, R. Sbiaa, A. Hirohata, H. Ohno, S. Fukami and S. N. Piramanayagam: Mater. Today, 20 (2017) 530.Article
• 30. S. A. Wolf, D. D. Awschalom, R. A. Buhrman, J. M. Daughton, S. Von Molnár, M. L. Roukes, A. Y. Chtchelkanova and D. M. Treger: Science, 294 (2001) 1488.Article
• 31. S. A. Wolf, J. Lu, M. R. Stan, E. Chen and D. M. Treger: IEEE, 98 (2010) 2155.Article
• 32. A. Fert: Rev. Mod. Phys., 80 (2008) 1517.Article
• 33. D. Barreca, G. Carraro, D. Peeters, A. Gasparotto, C. Maccato, W. M. M. Kessels, V. Longo, F. Rossi, E. Bontempi, C. Sada and A. Devi: Chem. Vap. Depos., 20 (2014) 313.Article
• 34. M. Gich, I. Fina, A. Morelli, F. Sánchez, M. Alexe, J. Gàzquez, J. Fontcuberta and A. Roig: Adv. Mater., 26 (2014) 4645.Article
• 35. C. Kittel: Phys. Rev., 71 (1947) 270.Article
• 36. A. Namai, S. Sakurai, M. Nakajima, T. Suemoto, K. Matsumoto, M. Goto, S. Sasaki and S. I. Ohkoshi: J. Am. Chem. Soc., 131 (2008) 1170.Article
• 37. A. Namai, M. Yoshikiyo, K. Yamada, S. Sakurai, T. Goto, T. Yoshida, T. Miyazaki, M. Nakajima, T. Suemoto, H. Tokoro and S. I. Ohkoshi: Nat. Commun., 3 (2012) 1035.Article
• 38. S. ichi Ohkoshi, A. Namai, M. Yoshikiyo, K. Imoto, K. Tamazaki, K. Matsuno, O. Inoue, T. Ide, K. Masada, M. Goto, T. Goto, T. Yoshida and T. Miyazaki: Angew. Chem. Int. Ed., 55 (2016) 11403.ArticlePDF
• 39. R. Kinugawa, K. Imoto, Y. Futakawa, S. Shimizu, R. Fujiwara, M. Yoshikiyo, A. Namai and S. I. Ohkoshi: Adv. Eng. Mater., 23 (2021) 2001473.Article
• 40. G. Cherubini, C. C. Chung, W. C. Messner and S. O. R. Moheimani: IEEE Trans. Control Syst. Technol., 20 (2012) 296.Article
• 41. H. Nishio and H. Yamamoto: IEEE Trans. Magn., 46 (2010) 3747.Article
• 42. S. I. Ohkoshi, K. Imoto, A. Namai, M. Yoshikiyo, S. Miyashita, H. Qiu, S. Kimoto, K. Kato and M. Nakajima: J. Am. Chem. Soc., 141 (2019) 1775.Article
• 43. S. ichi Ohkoshi, M. Yoshikiyo, K. Imoto, K. Nakagawa, A. Namai, H. Tokoro, Y. Yahagi, K. Takeuchi, F. Jia, S. Miyashita, M. Nakajima, H. Qiu, K. Kato, T. Yamaoka, M. Shirata, K. Naoi, K. Yagishita and H. Doshita: Adv. Mater., 32 (2020) 139821.
• 44. I. Tudosa, C. Stamm, A. B. Kashuba, F. King, H. C. Siegmann, J. Stöhr, G. Ju, B. Lu and D. Weeler: Nature, 428 (2004) 831.ArticlePDF
• 45. D. Alloyeau, C. Ricolleau, C. Mottet, T. Oikawa, C. Langlois, Y. Le Bouar, N. Braidy and A. Loiseau: Nat. Mater., 8 (2009) 940.ArticlePDF
• 46. M.L. Plumer, J. van Ek and D. Weller, The Physics of Ultra-High-Density Magnetic Recording, Springer Science and Business Media, 41(2001).
• 47. S. U. T Hayashi, S. Hirono and M. Tomita: Nature, 381 (1996) 772.ArticlePDF
• 48. S. I. Ohkoshi, A. Namai, K. Imoto, M. Yoshikiyo, W. Tarora, K. Nakagawa, M. Komine, Y. Miyamoto, T. Nasu, S. Oka and H. Tokoro: Sci. Rep., 5 (2015) 14414.Article
• 49. M. Yoshikiyo, A. Namai, K. Imoto, H. Tokoro and S. I. Ohkoshi: Eur. J. Inorg. Chem., 2018 (2018) 847.ArticlePDF
• 50. M. Yoshikiyo, Y. Futakawa, R. Shimoharai, Y. Ikeda, J. MacDougall, A. Namai and S. Ohkoshi: Chem. Phys. Lett., 803 (2022) 139821.Article
• 51. W. Yang, Y. Yu, M. B. Starr, X. Yin, Z. Li, A. Kvit, S. Wang, P. Zhao and X. Wang: Nano Lett., 15 (2015) 7574.Article
• 52. D. Cao, Z. Wang, L. Wen, Y. Mi and Y. Lei: Angew. Chem., 126 (2014) 11207.Article
• 53. W. Ji, K. Yao, Y. F. Lim, Y. C. Liang and A. Suwardi: Appl. Phys. Lett., 103 (2013) 103.Article
• 54. C. B. E., R. R. T. Zhao, A. Scholl, F. Zavaliche, K. Lee, M. Barry, A. Doran, M. P. Cruz, Y. H. Chu, C. Ederer, N. A. Spaldin, R. R. Das, D. M. Kim and S. H. Baek: Nat. Mater., 5 (2006) 823.
• 55. Y. Ling, G. Wang, D. A. Wheeler, J. Z. Zhang and Y. Li: Nano Lett., 11 (2011) 2119.Article
• 56. G. Wang, Y. Ling, D. A. Wheeler, K. E. N. George, K. Horsley, C. Heske, J. Z. Zhang and Y. Li: Nano Lett., 11 (2011) 3503.Article
• 57. D. A. Wheeler, G. Wang, Y. Ling, Y. Li and J. Z. Zhang: Energy Environ. Sci., 5 (2012) 6682.Article
• 58. A. R. H. Dotan, O. Kfir, E. Sharlin, O. Blank, M. Gross, I. Dumchin and G. Ankonina: Nat. Mater., 12 (2013) 158.ArticlePDF
• 59. M. S. Prévot and K. Sivula: J. Phys. Chem. C, 117 (2013) 17879.Article
• 60. K. Sivula, R. Zboril, F. Le Formal, R. Robert, A. Weidenkaff, J. Tucek, J. Frydrych and M. Grätzel: J. Am. Chem. Soc., 132 (2010) 7436.Article
• 61. L. T. Quynh, C. N. Van, Y. Bitla, J. W. Chen, T. H. Do, W. Y. Tzeng, S. C. Liao, K. A. Tsai, Y. C. Chen, C. L. Wu, C. H. Lai, C. W. Luo, Y. J. Hsu and Y. H. Chu: Adv. Energy Mater., 6 (2016) 1600686.Article
• 62. S.J. Clark and J. Robertson: Appl. Phys. Lett., 90 (2007) 132903.Article
• 63. Y. Xu and M. A. A. Schoonen: Am. Mineral., 85 (2000) 543.Article
• 64. M. Cargnello, A. Gasparotto, V. Gombac, T. Montini, D. Barreca and P. Fornasiero: Eur. J. Inorg. Chem., 28 (2011) 4309.Article
• 65. D. I. Kondarides, V. M. Daskalaki, A. Patsoura and X. E. Verykios: Catal. Letters, 122 (2008) 26.ArticlePDF
• 66. A. Patsoura, D. I. Kondarides and X. E. Verykios: Catal. Today, 124 (2007) 94.Article
• 67. M. J. Katz, S. C. Riha, N. C. Jeong, A. B. F. Martinson, O. K. Farha and J. T. Hupp: Coord. Chem. Rev, 256 (2012) 2521.Article

Figure & Data

References

Citations

Citations to this article as recorded by