Competitive coexistence of ferromagnetism and metal–insulator transition of VO₂ nanoparticles

Vanadium dioxide (VO₂) exhibits a metal-insulator transition (MIT) at $\sim$ 340 K [1, 2, 3, 4, 5, 6, 7]. The MIT is accompanied by a structural phase transition (SPT) from a high-temperature tetragonal metallic rutile phase to a low-temperature insulating monoclinic one, and the value of the electric resistivity changes by more than several orders of magnitude [1, 8, 9, 10, 11]. VO₂ is expected to be applied to novel electronic devices using MIT [12, 13, 14, 15, 16, 17, 18, 19, 20]. Recently, the MIT of low-dimensional VO₂, such as thin films [21, 22, 23, 24], nanobeams[25, 26, 27], and nanoparticles [28, 29], has been studied. More recently, ferromagnetism has also been reported in VO₂ nanoparticles[30]. Although ferromagnetism in nanoparticles may occur because of many defects on the surface of VO₂ nanoparticles[30], its detailed mechanism has not yet been completely elucidated. Thus, both the magnetic and electric properties of VO₂ nanoparticles must be analyzed. The magnetic properties of nanoparticles can easily be determined using a superconducting quantum interference device (SQUID). Meanwhile, one simple way to observe the MIT of VO₂ nanoparticles is to measure the variations of electric resistance values. However, measuring electric resistance by attaching electrodes to VO₂ nanoparticles is difficult. Therefore, there have been no experimental results that measure ferromagnetism by the magnetic measurement and the MIT by the electric measurement.

In this study, we prepared VO₂ nanoparticles by milling VO₂ powder, and the magnetic field dependence of the magnetization of VO₂ nanoparticles was observed using SQUID. Furthermore, VO₂ samples were fabricated by bridging between submicron-sized nanogap electrodes with the VO₂ nanoparticles. Then, the temperature dependence of the resistance of the VO₂ samples was measured. Finally, the relation between ferromagnetism and MIT of the VO₂ nanoparticles was discussed.

VO₂ nanoparticles were prepared by milling a commercial powder product (Kojundo Chemical Lab, 99.9%)[31] using ZrO₂ balls and a ZrO₂ vessel at a rotation speed of 400 rpm for 1 h. Here, we confirmed that ZrO₂ shows paramagnetism and not ferromagnetism by measuring the magnetic field dependence of the magnetization of ZrO₂ balls[30]. Figure 1 (a) shows a scanning electron microscope (SEM) image of the VO₂ powder. As shown in Fig. 1 (a), the VO₂ powder was composed of micrometer-sized particles with smooth surfaces. Meanwhile, the SEM image of the milled VO₂ nanoparticles is shown in Fig. 1 (b). The VO₂ powder was crushed to nanometer-sized particles with pebble-grained surfaces. The average diameter of the VO₂ nanoparticles was estimated as 42 nm using the Scherrer equation[30].

The VO₂ samples were fabricated using the milled VO₂ nanoparticles. First, a thermal oxide with a thickness of 50 nm was grown on an n-type Si substrate. Then, nanogap electrodes with Ti and Au were created on substrates using 100 kV electron beam lithography (ELIONIX, ELS-G100-SP) and a metal evaporator. Figure 2 (a) shows the SEM image of typical nanogap electrodes. The distances between the two electrodes were $60-100$ nm, and the thicknesses of Ti and Au were 5 and 20 nm, respectively. Then, the prepared VO₂ nanoparticles were dissolved in ethanol by stirring with ultrasonic waves, and a VO₂ solution was created. Next, the VO₂ solution was sprayed onto the SiO₂, where the nanogap electrodes were located. Finally, ethanol was evaporated at room temperature, and VO₂ samples were fabricated. The schematic of a VO₂ sample is shown in Fig. 2 (b). The average diameter of the VO₂ nanoparticles was 42 nm. Therefore, several nanoparticles must be located between the nanogap electrodes to form a current path. We confirmed current flows in two samples (samples A and B).

We measured the magnetization ($M$) of VO₂ powder and nanoparticles as a function of the magnetic field ($H$). Figure 3 (a) shows the $H$ dependence of $M$ of the VO₂ powder and milled VO₂ nanoparticles at $T=$ 300 K. As shown in Fig. 3 (a), no magnetization was observed for the VO₂ powder. Meanwhile, the variation of $M$ with hysteresis was observed for the VO₂ nanoparticles, indicating ferromagnetism for the insulating phase. The $H$ dependence of $M$ of the VO₂ powder and nanoparticles at $T=$ 380 K is shown in Fig. 3 (b). In this figure, we observed accreted hysteresis loops for the VO₂ nanoparticles but not for the VO₂ powder. This also indicated ferromagnetism for the metallic phase in the VO₂ nanoparticles. Therefore, the VO₂ nanoparticles exhibited ferromagnetism independent of the crystal structure (i.e., insulating and metallic phases). This ferromagnetism may have been due to many defects on pebble-grained surface[30].

To investigate whether defects are formed by milling the VO₂ powder, we analyzed VO₂ starting powder and nanoparticles using X-ray photoelectron spectroscopy (XPS). Figure 4 shows the XPS spectrum and the peak fitting of the VO₂ starting powder and nanoparticles. The dots are the measurement results. The background was obtained by the Shirley method, and the red curves are the fits to the XPS spectrum. The blue, purple, green, and dark blue curves show the contributions of the V 2p_3/2, V 2p_1/2, O 1s and OH peaks, respectively. The light blue and yellow curves show the satellite peaks of V 2p_3/2 and V 2p_1/2, which appear due to the strong hybridization between the V 2p_3/2 or V 2p_1/2 and O 1s. In this figure, the V 2p_3/2 peak of VO₂ nanoparticles redshifts by 0.3 eV with respect to the starting powder. This redshift may be due to the presence of oxygen defects. However, the VO₂ starting powder is milled in air, so the VO₂ nanoparticles are pre-exposed to atmospheric oxygen. Therefore, it is difficult to unambiguously claim that oxygen defects present on the surface of VO₂ nanoparticles.

Next, we measured the resistance ($R$) of the two samples as a function of $T$. Here, the source-drain voltage was fixed at 100 mV. The $T$ dependence of $R$ of sample A is shown in Fig. 5 (a). Here, the two arrows indicate the direction of the increase and decrease in $R$. As $T$ increased from $\sim$295 K up to $\sim$340 K, $R$ gradually decreased from $\sim$27 M$\Omega$. Then, when $T$ exceeded $\sim$340 K, $R$ rapidly decreased. As $T$ further increased up to $\sim$360 K, the value of $R$ slowly decreased again. Finally, the value of $R$ was $\sim$500k$\Omega$ at $T=$ 370 K, which was approximately 60 times lower than that at $T=$ 295 K. Next, when $T$ decreased from $\sim$370 K, $R$ gradually increased until $T$ reached approximately 340 K. When $T$ decreased from $\sim$340 K to $\sim$320 K, $R$ rapidly increased and then increased slowly again until 295 K. The value of $R$ returned to the initial one and the variation of $R$ showed hysteresis. Therefore, the MIT occurs in sample A. Similarly, the $T$ dependence of $R$ of sample B is shown in Fig. 5 (b). As shown in this figure, the $T$ dependence of $R$ of sample B was almost the same as that of sample A. However, the temperature range where the MIT occurred was wider than that of sample A. Moreover, the minimum value of $R$ was $\sim$1.1 M$\Omega$, which $\sim$30 times lower than that at $T=$ 295 K. For the MIT of bulk VO₂, the value of its resistance changes by more than five orders of magnitude in a very narrow temperature range of several K at approximately $T=$340 K [1, 8, 9, 10, 11]. Therefore, the $R$ of the VO₂ nanoparticles changed more slowly than that of bulk VO₂ as $T$ changed.

To discuss the electric proprieties of the VO₂ samples in detail, the SEM images of samples A and B are shown in Figs. 6 (a) and (b), respectively. In these figures, some VO₂ nanoparticles were not located just between a spacing of $\sim$100 nm of two nanogap electrodes, which corresponded to the narrowest width between them. Instead, many VO₂ nanoparticles were accidentally located between approximately 2 ${\rm\mu}$m of the two nanogap electrodes, which corresponded to the second narrowest width between them as shown in Fig. 2 (a) [33]. Because the inner part of the VO₂ nanoparticles at high temperatures is metallic, the values of the contact resistances between the VO₂ nanoparticles are higher than that of the inner part of the VO₂ nanoparticles. Here, the high contact resistances may be caused by electrons trapped in many defects on the pebble-grained surfaces. Therefore, the values of $R$ of the two samples were determined by the contact resistance between the VO₂ nanoparticles, not the resistances of the inner part of the VO₂ nanoparticles. As described above, many VO₂ nanoparticles were located between the two electrodes. Thus, the values of $R$ of the two samples were much larger than those of typical bulk VO₂. Besides, as in Fig. 1 (b), the diameters of most of VO₂ nanoparticles were lower than approximately 100 nm. However, some of them appeared to be aggregated to micrometer size, and micrometer-sized VO₂ particles were located between the two nanogap electrodes. Therefore, the two samples had different hysteresis shapes and different minimum resistance values maybe because the VO₂ particles had not only nanometer sizes but also dispersed sizes due to aggregation.

We made VO₂ nanoparticles by milling VO₂ powder and fabricated VO₂ samples using these nanoparticles. Ferromagnetism was observed by measuring the magnetic field dependence of the magnetization of VO₂ nanoparticles. Many defects on the pebble-grained surface of the VO₂ nanoparticles may have resulted in the exhibition of ferromagnetism. However, although MIT was also obtained in the temperature dependence of the resistance of the VO₂ samples, it was exhibited over a wider temperature range than that of bulk VO₂. This wider hysteresis of MIT may be because the VO₂ particles had not only nanometer sizes but also dispersed sizes due to aggregation. We must thus continue researching the reasons why ferromagnetism and MIT are exhibited in VO₂ nanoparticles. In conclusion, this research is the first demonstration of competitive coexistence of ferromagnetism and MIT in VO₂ nanoparticles. Besides, the utilization of the ferromagnetism and phase transition of VO₂nanoparticles, is expected to create novel spintronic devices, such as magneto-resistive sensor devices whose resistance can be controlled by temperature.