Chemical Vapor Deposition of Ferrimagnetic Fe₃O₄ Thin Films

Abstract

Ultrathin magnetic materials with room-temperature ferromagnetism/ferrimagnetism hold great potential in spintronic applications. In this work, we report the successful controllable growth of Fe₃O₄ thin films using a facile chemical vapor deposition method. Room-temperature ferrimagnetism was maintained in the as-grown Fe₃O₄ thin films down to 4 nm. Raman spectroscopy, X-ray diffraction and X-ray photoelectron spectroscopy analysis were conducted to reveal the structure and quality of the Fe₃O₄ film. Magnetization measurement showed ferrimagnetic hysteresis loops in all Fe₃O₄ thin films. A saturation magnetization of 752 emu/cm³ was observed for the 4 nm Fe₃O₄ film, which was higher than that of bulk Fe₃O₄ materials (480 emu/cm³). Additionally, the Verwey transition at ~120 K was visible for the Fe₃O₄ thin films. This work provides an alternative method of synthesizing ferrimagnetic ultrathin films for electronic, spintronic, and memory device applications.

1. Introduction

Room-temperature ferromagnetic/ferrimagnetic thin films have greatly promoted the development of electronic devices and spintronic devices [1,2,3]. The recent emergence of two-dimensional magnetic materials provides possibilities for new physics and new devices [4,5]. Several atomically thin 2D magnetic materials, e.g., CrI₃ [6,7], Cr₂Ge₂Te₆ [8], Fe₃GeTe₂ [9,10] and Cr₂S₃ [11], have been discovered. However, the Curie temperature of most 2D magnetic materials is still far below room temperature. Another alternative approach to obtain room-temperature ferromagnetic/ferrimagnetic films is to grow non-layered ultrathin materials with maintained magnetic properties.

Magnetite (Fe₃O₄) has been widely studied due to its half metallicity, high Curie temperature (T_C = 860 K) and metal–insulator transitions (Verwey transition, T_V ≈ 120 K) [12,13,14]. These properties make Fe₃O₄ a superior candidate for spintronics devices [15,16,17,18,19]. Above T_V, Fe₃O₄ possesses the cubic-inverse-spinel structure with lattice constants a = 8.3967 Å and belongs to the space group Fd-3m [20]. In the crystal structure of Fe₃O₄, oxygen ions are closely packed in a cubic arrangement, and Fe ions fill in the interstices. There are two types of interstices: a tetrahedral interstice (the Fe ion is surrounded by four oxygen ions) and an octahedral interstice (the Fe ion is surrounded by six oxygen ions). In a unit cell of Fe₃O₄, there are 64 tetrahedral interstices and 32 octahedral interstices. One eighth of the tetrahedral interstices are occupied by half of the Fe³⁺ ions. Half of the octahedral interstices are occupied by the other half of the Fe³⁺ ions and all of the Fe²⁺ ions.

The magnetic property of Fe₃O₄ thin films varies with the thickness. Fe₃O₄ thin films (with thicknesses less than 20 nm) epitaxially deposited on MgO substrate are still ferrimagnetic with high magnetic moments. The saturation magnetization (Ms) for a 5 nm Fe₃O₄ film is found to be 922 emu/cm³, which is much greater than that of bulk Fe₃O₄ [21]. Epitaxial Fe₃O₄ films on SrTiO₃ substrate showed an increase in Ms for thickness below 15 nm. The Ms of 3 nm Fe₃O₄ film reached 1017 emu/cm³. As for films with thicknesses above 15 nm, Ms is close to the bulk Fe₃O₄ value [22]. According to previous reports, film thickness has a great influence on the magnetic properties of Fe₃O₄.

Fe₃O₄ thin films are usually synthesized with pulsed laser deposition (PLD) [23,24], molecular beam epitaxy (MBE) [25], magnetron sputtering [26], sol-gel and spin-coater [27] as well as chemical vapor deposition (CVD). PLD, MBE and magnetron sputtering can produce high-quality films but with high costs. Sol-gel and spin-coater techniques are low-cost and convenient methods, but the film quality is limited. CVD is an intermediate method with low costs and good film quality. There are several works on CVD growth of Fe₃O₄ films. In Mantovan’s work [28], the magnetoresistance of polycrystalline Fe₃O₄ thin films by using the cyclohexadiene iron tricarbonyl liquid precursor was mainly studied. In another report [29], the Ms of Fe₃O₄ film prepared under ambient conditions by using a mist CVD method was 500 emu/cm³. A space-confined CVD method developed by Xiong’s group [30] obtained ultrathin Fe₃O₄ nanosheets, with the size of about 2–4 µm, whose ultrabroadband photoresponse was demonstrated. Here, we choose the CVD method using Fe, PtCl₂ and KMnO₄ as precursors to prepare Fe₃O₄ thin films and thin triangles and investigate their electrical and magnetic properties.

Here, we report CVD growth of Fe₃O₄ thin films with thicknesses down to a few nanometers on (0001) α-Al₂O₃ substrate. Raman spectroscopic characterization and XRD measurement showed that Fe₃O₄ films were compressive strained due to the lattice mismatch between Fe₃O₄ and Al₂O₃. A strain of −6.50‰ was observed for the 4 nm Fe₃O₄ thin film. Magnetic measurement revealed robust ferrimagnetism in the 4 nm Fe₃O₄ thin film. The Ms of thinner Fe₃O₄ film (4 nm thickness) was significantly larger than that of thicker film.

2. Materials and Methods

2.1. Materials

Fe (99.99%) foil with 0.03 mm thickness was purchased from Taizhou Senuo Material Technology Co., Ltd., Beijing, China. KMnO₄ (CAS:7722-64-7, 99.8%) was purchased from Sinopharm Chemical Reagent Beijing Co., Ltd., Beijing, China. PtCl₄ (CAS:13454-96-1, Pt 58.0%) was purchased from Beijing Yanuo Xincheng Technology Co., Ltd., Beijing, China. All the materials were analytical grade and were used without further purification.

2.2. CVD Growth of Fe₃O₄

An alumina boat containing KMnO₄ (~0.40 g) was put outside of the furnace and heated to 180 ℃. Another alumina boat containing PtCl₄ powder (~0.02 g) was placed in the upstream of the furnace, and sapphire substrates were placed in the center of the furnace. Iron foil was placed in the downstream of the furnace. The temperature of the furnace was raised to 400 °C in 20 min and kept at 400 °C for 30 min. Then, the furnace was naturally cooled down to room temperature. The carrier gas was 30 sccm Ar. The growth was carried out at atmospheric pressure.

2.3. Characterization

The crystal structure and composition of the as-grown films were checked using X-ray photoelectron spectroscopy (ESCALAB250Xi, Thermo Fisher Scientific Co., Ltd., Loughborough, England) and X-ray diffraction (XRD). XRD patterns were collected using an X-ray diffractometer (D/MAX-TTRIII, Rigaku Co., Tokyo, Japan) with a scan rate of 6°/min. Atomic force microscopic (AFM) images were captured on a multimode scanning probe microscope (Bruker Multimode-8, Bruker Co., Billerica, MA, USA). Energy dispersive spectroscopy (EDS) elemental mapping were taken on a Hitachi S4800 equipped with an energy dispersive X-ray spectroscopy detector.

Raman spectroscopic measurements were performed on a home-made vacuum, variable temperature, low-wavenumber Raman system with 532 nm excitation. An NA = 0.82 low temperature objective (LT-APO/VIS/0.82, attocube systems AG, Haar, Germany) was used for laser focusing and signal collection.

Resistance measurements were performed in a physical property measurement system (PPMS-9, Quantum Design Inc., San Diego, CA, USA) under He-purged conditions. The copper transmission electron microscopy grid was used as a shadow mask. For electrical measurement, 10 nm Ti and 60 nm Au was electron-beam evaporated on Fe₃O₄ film. The magnetic properties were measured using the vibrational sample magnetometer at different temperatures.

3. Results

Fe₃O₄ thin film was prepared via the oxidation of iron species in a CVD system, in which Cl₂ decomposed from PtCl₄ was used to produce volatile iron species from the iron foil (Figure 1a). The distance between the substrate and the iron foil was changed to tune the thickness of Fe₃O₄ films. Figure 1b,c show optical microscopic images and AFM images of Fe₃O₄ films grown on sapphire. The EDS elemental mapping of Fe, O and Al (Figure S1 in the Supplementary Materials) illustrate that the Fe and O atoms were homogeneously distributed on the substrate. The synthesis of ultrathin magnetic films via CVD is simple and convenient, and the synthesized materials can be transferred and stacked to prepare heterojunctions and functional devices.

Figure 1d presents a schematic diagram of (111) faced Fe₃O₄ film on (0001) plane of Al₂O₃. The distance between adjacent oxygen atoms is 2.96 Å in Fe₃O₄, while it is 2.89 Å in Al₂O₃, yielding a compressive lattice mismatch of about 2.3% for Fe₃O₄. Sapphire is composed of identical layers of face-sharing AlO₆ octahedron. For the Fe₃O₄ film grown on sapphire, the diffraction peaks at 18.518°, 37.308°, 57.208° and 79.213° are clearly observed (Figure 1e) and assigned to (111), (222), (333) and (444), respectively. Other peaks were from the α-Al₂O₃ substrate. The above results suggest that the Fe₃O₄ films are epitaxially grown on (0001) α-Al₂O₃ substrates, and the preferred orientation of the obtained epitaxial Fe₃O₄ films is the [111] crystal orientation.

X-ray photoelectron spectroscopy (XPS) is used to determine the valence states of Fe and the proportion of different valence states in Fe₃O₄ films. Figure 1f shows the Fe2p core level XPS of Fe₃O₄ thin film. The Fe2p splits into Fe2p_3/2 and Fe2p_1/2 due to spin–orbit coupling [31]. The binding energies positioned at 711 eV and 724 eV could be denoted as Fe2p_3/2 and Fe2p_1/2 of Fe²⁺ and Fe³⁺, respectively, which is consistent with a literature report [32]. The binding energy of 2p_3/2 and 2p_1/2 of Fe²⁺ are about 710.06 eV and 723.35 eV, while those of Fe³⁺ are about 711.47 eV and 724.97 eV, respectively. The ratio of Fe²⁺/Fe³⁺ calculated from the peak area is 0.468 (1:2.1), which is lower than 1/2 of stoichiometric Fe₃O₄, which indicates that the film is fully oxidized. Only the binding energies at 711 eV and 724 eV, corresponding to the Fe2p_3/2 and Fe 2p_1/2 of Fe₃O₄, are observed, with no satellite peaks of FeO (at 715.5 eV) or γ-Fe₂O₃ (at 719.1 eV), which finally confirms that all samples are Fe₃O₄ rather than γ-Fe₂O₃.

The strain of the films can be measured using XRD. Figure 2a,b show XRD patterns of the Fe₃O₄ thin films with various thicknesses grown on (0001) α-Al₂O₃ substrates. The positions of diffraction peak and the calculated lattice parameters are shown in Table S1 in the Supplementary Materials. The lattice parameter decreases from 8.3796 Å to 8.3414 Å when the thickness decreases from 16 nm to 4 nm (Figure 2c). The diffraction peak of the film with 4 nm is narrower, indicating a narrower distribution of the lattice parameters. All the lattice parameters of Fe₃O₄ thin films are smaller than that of the bulk Fe₃O₄ (8.3967 Å) [20], which confirms a compressive strain in the thin films [33]. The compressive strain of the 4 nm film and 16 nm film is 6.50‰ and 1.95‰, respectively (Figure 2c).

In addition, the Raman spectra can reflect the change in the strain in the sample [34]. Figure 2d,e display the Raman spectra of the Fe₃O₄ grown on sapphire. The Raman peaks located at 665 cm⁻¹, 670 cm⁻¹ and 678 cm⁻¹ are observed, corresponding to the A₁g of bulk Fe₃O₄, 16 nm and 4 nm Fe₃O₄ film, respectively. The relationship between Raman peak position and strain are shown in Figure 2f. The Raman peak shifts to a higher frequency at a stronger compressive strain.

In addition to the Fe₃O₄ films, triangular Fe₃O₄ nanosheets down to 5 nm thickness were grown on sapphire via CVD. Figure 3a displays the Raman spectra of the triangle Fe₃O₄ nanosheets with different thicknesses grown on sapphire at room temperature. The optical microscopic images of triangle Fe₃O₄ nanosheet grown on sapphire are shown in Figure 3b. A strong peak at about 665 cm⁻¹ is observed in triangle Fe₃O₄ nanosheets, which is assigned to the A_1g mode [30]. Other Raman peaks located at 194, 300, and 536 cm⁻¹ correspond to the T_2g(1), E_g, and T_2g(2) modes, respectively. The A_1g mode is the symmetric stretch of oxygen atoms along Fe-O bonds; the T_2g(2) mode is the asymmetric stretch of Fe and O; the E_g mode is the symmetric bend of oxygen with respect to Fe; T_2g(1) is the translatory movement of the whole FeO₄. The thickness of triangle Fe₃O₄ nanosheets with different thickness is presented in Figure 3c–f. No significant Raman peak shift is observed as the thickness changes. The absence of strain in the Fe₃O₄ nanosheets is very possibly due to a different interface structure.

To determine the magnetic properties of Fe₃O₄ ultrathin films, we measured the temperature-dependent magnetization curves of Fe₃O₄ films. The microscopic origin of the change in the magnetic properties of Fe₃O₄ can been studied in detail via XAS and XMCD advanced characterization techniques [27,35]. Currently, we study the macroscopic magnetic properties of Fe₃O₄ films using the conventional vibrational sample magnetometer technique due to experimental conditions. The magnetic hysteresis (M-H) loops are shown in Figure 4. At 300 K, the Ms of 16 nm Fe₃O₄ film is about 445 emu/cm³, which is close to the value obtained in bulk Fe₃O₄ (480 emu/cm³) [36]. Surprisingly, the Ms of 4 nm Fe₃O₄ film is dramatically increased to 752 emu/cm³. Compared with Fe₃O₄ films synthesized using different methods (

Table 1. Saturation magnetization and thickness of Fe₃O₄ films synthesized by different methods.

Method	Substrate	Thickness (nm)	Ms (emu/cm3)	Reference
MBE	MgO	5	922	[21]
MBE	SrTiO3	3	1017	[22]
PLD	MgO	200/150	119.7/109.2	[24]
MS	SiO2	15	301	[26]
SGSC	Si	650	1.9 µB/u.c.	[27]
CVD	Si/SiO2	100/24.8/13.4	Not Measured	[28]
CVD	SrTiO3	200–30	500	[29]
CVD	mica	1.95/4.2/15	Not Measured	[30]
CVD	Al2O3	16/4	445/752	In this work

Notes: MBE: molecular beam epitaxy; PLD: pulsed laser deposition; MS: magnetron sputtering; SGSC: sol-gel and spin-coater; CVD: chemical vapor deposition.

), ultra films with high Ms were obtained in our work. As shown in Table 1, Ms for a 5 nm Fe₃O₄ film grown on MgO substrate by MBE is 922 emu/cm³ [21]; epitaxial 3 nm Fe₃O₄ film on SrTiO₃ substrate shows a Ms of 1017 emu/cm³ [22]. The Ms of Fe₃O₄ film on SrTiO₃ via CVD is about 500 emu/cm³ [29]. The Ms obtained in the above experiments are all larger than the Ms of the bulk Fe₃O₄. In our work, 4 nm Fe₃O₄ film was obtained, and the Ms is 752 emu/cm³. At 2 K, the Ms of Fe₃O₄ films exhibit a similar thickness dependence. Representative in-plane and out-of-plane M-H loops of the 4 nm film are shown in Figure 4c,d. The magnetic coercivity of the film increased with the cooling down process, from 238 Oe (or 912 Oe) at room temperature to 772 Oe (or 2702 Oe) at 2 K in the in-plane (or out-of-plane) magnetic field. This is normal since the reverse of magnetic moment requires more energy at low temperature due to the suppression of thermal disturbance. The in-plane M-H loops of the 4 nm Fe₃O₄ film show better rectangularity than that of out-of-plane M-H loops at 300 K and 2 K. At 300 K, in-plane and out-of-plane M-H curves of 4 nm Fe₃O₄ are saturated at 0.1359 T and 0.2679 T, respectively. This indicates that the in-plane M-H curve is easy to saturate under a small applied magnetic field, while the out-of-plane M-H curve is saturated at a larger magnetic field. This suggests that the easy axis of Fe₃O₄ film lies in the film plane with a small magnetic anisotropy.

As shown in Figure 5a, the resistance of a 4 nm Fe₃O₄ film first increases gradually as the temperature decreases and then rapidly rises at T_V. The conductivity of an Fe₃O₄ film is mainly contributed to the electron hopping between Fe²⁺ and Fe³⁺ at octahedral positions. Above T_V, the electrons are easily exchanged between the Fe²⁺ and Fe³⁺ on the octahedral sites, which accounts for the relatively high electrical conductivity of cubic Fe₃O₄ at high temperatures. Low temperatures can freeze the electron hopping and thus result in much higher resistance.

Figure 5b shows the angular−dependent resistance for the 4 nm Fe₃O₄ film with a fixed magnetic field of 3 T. The △R/R_max (%) is defined as △R/R_max (%) = (R_θ − R_max)/R_max × 100), where R_max is the maximum of resistivity for each scan. θ is the angle between the applied magnetic field and film normal direction. θ = 0° corresponds to the out-of-plane field, and θ = 90° is the in-plane magnetic field. The curves maintain two-fold symmetry at different temperatures near T_V (140 K, 120 K, 100 K), following the typical sin²θ dependence with valleys at 0°/180° and peaks at 90°/270°, respectively. The details of the origin resistance of 4 nm Fe₃O₄ film are seen in Figure S2. Sin²θ symmetry indicates that the thin films have a uniaxial magnetic anisotropy.

In addition, we measured magnetoresistance (MR) at various applied fields. The MR is defined as MR = (R_H − R₀)/R₀ × 100%. A representative MR of the 4 nm Fe₃O₄ film as a function of magnetic field measured at different temperatures (300 K, 180 K and 120 K) are shown in Figure S3. As shown in Figure S3a, in-plane MR at 120 K shows abnormal behavior compared with 180 K and 300 K, which is due to the occurrence of phase transfer at about 120 K. However, the abnormal behavior at 120 K is absent when the applied magnetic field is perpendicular to the film (Figure S3b). The possible reason is that the out-of-plane direction is the hard magnetization direction.

The effect of temperature and magnetic field on the anisotropic magnetoresistance (AMR, AMR (%) = (R_θ=90° − R_θ=0°)/R_average × 100), where R_θ=90° and R_θ=0° are the resistivities for magnetic field applied parallel and perpendicular to the current direction, respectively, and R_average = R_θ=90°/3 + 2 × R_θ=0°/3, should be further explored. The in-plane and out-of-plane resistance as a function of temperature at different fields were investigated (Figure S4). As shown in Figure 5c, the AMR is enhanced at higher magnetic fields and is suppressed at lower temperatures.

4. Conclusions

In summary, we report the growth of ultrathin Fe₃O₄ films and triangle nanosheets using a low-cost and highly efficient CVD method. The 4 nm Fe₃O₄ ultrathin film exhibits room-temperature ferrimagnetism and a higher magnetic moment. This work contributes greatly to the synthesis of ultrathin magnetic films from non-layered magnetic materials for next-generation electronic and spintronic devices.