Ultra-narrow ferromagnetic resonance in organic-based thin films grown via low temperature chemical vapor deposition

There is a strong interest in magnetic materials for planar microwave devices, such as circulators, isolators, phase shifters, and patch antennas, with much of the focus on soft ferrite materials.¹ In particular, yttrium iron garnet (Y₃Fe₅O₁₂, YIG) has enabled the development of high frequency electronic devices that take advantage of its electrically insulating properties, extremely narrow ferromagnetic resonance (FMR) linewidth, and consequent low microwave loss.² However, only films grown on specific lattice-matched substrates (such as gadolinium gallium garnet, GGG) yield optimal properties.³ These properties make YIG the material of choice for a range of microwave applications but also limit it to device structures that can accommodate these specific substrates. Specifically, devices must either be compatible with high temperature growth or be integrated with a separately grown YIG layer. Given these constraints, there is space for a narrow linewidth magnetic material that can be deposited on varied substrates, ideally post fabrication, to complement YIG.

Organic and organic-based conductors and semiconductors have been shown to tolerate a wide variety of substrates and are compatible with low temperature and ambient pressure deposition. Recently, these materials have made inroads in consumer electronics in the form of organic light-emitting diodes (OLEDs),⁴ organic photovoltaics (OPVs),⁵ and organic field effect transistors (FETs).^6,7 In this light, an organic material with magnetic resonance properties comparable to YIG could similarly enable future organic-based microwave devices. A promising candidate for these purposes is V[TCNE]_x (x ∼ 2, TCNE: tetracyanoethylene), an organic-based ferrimagnetic semiconductor with room temperature magnetic ordering and E_G ∼ 0.5 eV.⁸ V[TCNE]_x∼2 can be grown via a low temperature (60 °C) chemical vapor deposition (CVD) process, enabling conformal deposition on a variety of surfaces such as flexible and three-dimensional substrates.^8–10 However, previous studies of FMR in V[TCNE]_x∼2 have yielded complex resonance spectra that evolve as a function of the external field orientation and vary significantly from sample to sample. Here, we present DC magnetometry and FMR studies of optimized V[TCNE]_x∼2 thin films that show a Curie temperature above 600 K, sharp magnetization switching, and a single resonance with full width at half maximum (FWHM) linewidth of 1.4 G. These results demonstrate the potential of V[TCNE]_x∼2 to provide the materials basis for future hybrid and organic microwave devices.

One of the more surprising aspects of the magnetic ordering in V[TCNE]_x∼2 is that it is quite robust, even in the original powder synthesis,⁸ despite a significant degree of structural disorder. Specifically, extended x-ray absorption fine-structure (EXAFS) studies¹¹ show that the vanadium atoms are locally coordinated with six nitrogen atoms in a distorted octahedral environment with a vanadium-nitrogen spacing of 2.084(5) Å, Fig. 1(a). In contrast, x-ray diffraction and transmission electron microscopy (TEM)¹² show no signatures of long-range structural order. While unusual, the presence of strong long-range magnetic order without corresponding long-range structural order makes V[TCNE]_x∼2 particularly amenable to deposition on a variety of both crystalline and amorphous substrates. Figure 1(b) shows the electronic structure resulting from this local coordination. The highest occupied molecular orbital of the [TCNE]⁻ is the π* level, which is split by Coulomb repulsion energy U_c and singly occupied, leading to a net spin of 1/2 (Fig. 1(a), blue arrows). This π* state is populated via hybridization with the vanadium t_2g level, which falls in the gap between the π* and π* + U_c [TCNE]⁻ levels for a net charge transfer of two electrons, leaving the vanadium in the spin 3/2 V²⁺ state (Fig. 1(a), red arrows). The antiferromagnetic exchange interaction between the t_2g electrons on the V²⁺ and the π* electrons on the [TCNE]⁻ molecules results in a net ferrimagnetic ordering with the [TCNE]⁻ acting as a superexchange pathway for adjacent V²⁺ (Fig. 1(b) green arrow). The V[TCNE]_x∼2 films reported here are grown by a low temperature (60 °C) CVD process in a custom built reactor shown in Fig. 1(c). The films are formed through the reaction of precursors TCNE and V(CO)₆. Commercially available TCNE is purified through sublimation, and V(CO)₆ is synthesized from tetraethylammonium vanadium hexacarbonyl and phosphoric acid.¹³ 

Initial reports of V[TCNE]_x∼2 grown via CVD yielded the first thin film organic magnetic material with a T_C above room temperature (T_C ∼ 400 K (Ref. 9)). It was observed that the CVD grown films show sharper magnetization switching compared to previous solvent-based powders, which is attributed to the absence of the noncoordinating CH₂Cl₂ solvent required for powder synthesis.¹⁴ Previous studies have also shown that the sublimation rate and ratio of the V[TCNE]_x∼2 precursors affect the coercivity and sharpness of the switching for M(H) measurements.¹⁵ Here, we extend these studies by further exploring the impact of the relevant conditions including: the temperature of the precursors, the ratio of the precursors, and the impact of a clean solvent-free substrate. The heating coil shown in Fig. 1(c) creates a temperature gradient determined by the density of the coil windings. For ideal films, (1) the TCNE precursor temperature is kept at 50 °C, while the V(CO)₆ is kept at 10 °C; (2) the ratio of TCNE to V(CO)₆ is set at 10:1 by weight; and (3) the film is deposited on a clean, solvent-free substrate. Depositions that deviated from these conditions generated films with magnetic properties in line with previous reports. Over 70 depositions, we consistently observe that these growth conditions generate more magnetically homogenous samples, exhibiting behaviors such as sharper switching in hysteresis loops and higher extrapolated T_C (Fig. 2), with extremely sharp ferromagnetic resonance features (Fig. 3).

Due to the air sensitive nature of the films, the CVD process is performed inside a series of argon glove boxes (O₂, H₂O < 0.5 ppm). Films of thicknesses between 250 and 350 nm are deposited onto glass slides for magnetic characterization and sapphire substrates for FMR measurements. Glass is used for SQUID measurements to make direct comparisons with earlier work,^9,11,14,15 while sapphire is used for FMR because of its lower microwave absorption. All samples are sealed in evacuated ESR grade quartz tubes to prevent air exposure during transport; for FMR measurements, samples are mounted on a custom sample mount to maintain their orientation. The results of DC magnetometry measurements as a function of both temperature and applied magnetic field are shown in Figure 2(a). The net magnetization is consistent with the theoretical prediction of 1 Bohr magneton per formula unit (Fig. 1(b) (Ref. 9)) and shows only modest suppression from 5 K to room temperature. The slight peak in the magnetization observed at ∼150 K is consistent with previous reports of correlated spin glass behavior at low temperatures.¹⁶ Higher temperature measurements are precluded due to film decomposition, preventing a direct measurement of the Curie temperature, however, an estimate of T_C can be obtained from a fit above 150 K to the Bloch Law (Fig. 2(a), solid line; note that this approach yields a lower estimate of T_C than a simple extrapolation of the data),

where the saturation magnetization is given by M_s(T) and β is a fitting parameter. From this fit, we find β = (6.7 ± 0.1)× 10⁻⁵ K^−3/2, which corresponds to a T_C of 606 K, comparable to 550 K for thin-film YIG.¹⁷ 

The inset of Fig. 2(a) shows magnetization versus field for the same sample; data were taken to fields of ±1 T, but since the magnetization saturates by 100 Oe only low field data are shown. In contrast, previous films have shown much broader switching and a high field paramagnetic background, consistent with a magnetically inhomogeneous sample or complex domain structure.^9,14 Figure 2(b) shows reported T_C of V[TCNE]_x∼2 at various stages of development over the past 25 years, from powder-based (open circle⁸) through advancements in thin film deposition (open squares^9,16,18) to this work (solid star). While T_C is above 300 K in all cases, the T_C of 606 K we observe here is more than 100 K above previous reports.

The magnetically homogenous nature of the films is also well captured by FMR measurements. Figure 3(a) shows the FMR spectrum of a thin film of V[TCNE]_x∼2 (black) as well as the precursors used in the V[TCNE]_x∼2 CVD process, V(CO)₆ (red), and TCNE powder (blue). Measurements are made using a Bruker ESR spectrometer at a temperature of 300 K with 200 μW of applied microwave power at 9.63 GHz and H_ext applied in-plane. The applied microwave frequency is tuned between 9 and 10 GHz for optimal microwave cavity performance before starting the measurement. The inset of Fig. 3(a) shows the precursor spectra at a 1000× expanded scale. The peak in the V[TCNE]_x∼2 spectrum has an exceptionally narrow FWHM linewidth of 1.4 G, and there is no feature in the two precursor spectra that resembles the sharpness and position of the resonance peak seen in the V[TCNE]_x∼2 spectrum. Previous reports of FMR in V[TCNE]_x∼2 thin films found spectra with similarly sharp features but with multiple resonances that evolve with the field orientation with respect to the plane of the film.¹⁸ 

In contrast, Fig. 3(b) shows the FMR spectrum of an optimized film at various angles of the applied microwave and DC fields, rotating from in-plane (90°) to out-of-plane (0°) at 300 K with an applied microwave frequency of 9.39 GHz. The in-plane resonance field is shifted from the spectrum shown in Fig. 3(a), which can be attributed to slightly different cavity conditions and sample-to-sample variation. The only variation with angle is the expected shift in resonance field predicted by the Kittel equation, assuming that there is little or no crystalline anisotropy and shape anisotropy is dominant.¹⁹ Figure 3(c) shows the extracted resonance field as a function of angle at 300 K. The expected angular-dependent resonance condition for a thin film ferromagnet rotated from in-plane to out-of-plane when H ≫ 4πM_s (as is the case here) is

where ω_res is the resonance frequency in MHz, γ is the gyromagnetic ratio 2π × 2.8 MHz/Oe, M_s is the saturation magnetization in emu/cm³, H is the applied external field in Oe, and θ_H is the angle of the applied external field with respect to the normal of the film plane.^19,20 The black line in Fig. 3(c) shows the expected angular dependence for 4πM_S = 95 G (7.56 emu/cm³), which fits the angular data well and is consistent with SQUID data (Fig. 2) within our ability to estimate the volume of the sample. This validates the assumption that the magnetization angle exactly follows the external field angle since the shape anisotropy field 4πM_s is much smaller than the applied field. This angular variation is weak compared to other inorganic ferromagnets as a result of the lack of crystalline anisotropy and the low M_s (for comparison, thin film YIG has a saturation magnetization of 1750 G at room temperature³). Finally, the observed linewidth of the V[TCNE]_x∼2 film is comparable to that of YIG films of similar thickness (1.1 Oe)^2,3 and substantially better than all other known thin film ferromagnets.

In summary, this work demonstrates the capability to grow magnetically homogenous films of V[TCNE]_x∼2 that exhibit extremely narrow linewidths on the order of 1 G, sharply switching magnetization, and Curie temperature of 606 K, far above room temperature. Through control of deposition conditions, we are able to consistently grow films with magnetic resonance properties that rival those of YIG despite the lack of crystalline structure in V[TCNE]_x∼2, while also offering the benefits of organic materials, including low temperature conformal CVD deposition on a wide variety of substrates. Furthermore, V[TCNE]_x∼2 belongs to a class of materials that has demonstrated a high degree of flexibility, with multiple deposition methods^8,9,21,22 and the ability to modify the organic ligand.²³ These properties make V[TCNE]_x∼2 a valuable material complement to YIG with the potential for hybrid and all-organic high frequency electronic and spintronic applications. While the realization of this potential will require further work on challenges such as air sensitivity and low thermal tolerance, these constraints have been addressed for other commercially viable organic materials. For example, organic light-emitting diodes, photovoltaics, and thin film transistors have been encapsulated^24–26 and incorporated with supporting inorganic circuits.^4–7 

This work was supported by NSF Grant No. DMR-1207243, the NSF MRSEC program (DMR-0820414), DOE Grant No. DE-FG02-03ER46054, and the OSU-Institute for Materials Research. The authors acknowledge the NanoSystems Laboratory at Ohio State University, valuable discussions with M. Flatté and H. Tang, and technical assistance from C. Y. Kao.