Physical properties of V$_{1-x}$Ti$_{x}$O$_{2}$ (0 $<$ x $<$ 0.187) single crystals

Free standing, low strain, single crystals of pure and titanium doped VO$_{2}$ were grown out of an excess of V$_{2}$O$_{5}$ using high temperature solution growth techniques. At $T_{MI} \sim$ 340 K, pure VO$_{2}$ exhibits a clear first-order phase transition from a high-temperature paramagnetic tetragonal phase (R) to a low-temperature non-magnetic monoclinic phase (M1). With Ti doping, another monoclinic phase (M2) emerges between the R and M1 phases. The phase transition temperature between R and M2 increases with increasing Ti doping while the transition temperature between M2 and M1 decreases.


1
The metal-insulator (MI) transition in VO 2 at around 340 K was first reported by Morin in late 1950s [1]. Ever since, great effort has been made to understand the mechanism behind this MI transition as well as to explore its potential application in electronic devices [2,3]. Samples in various forms have been synthesized: bulk (polycrystalline and single-crystalline)[1, [4][5][6][7], thin films and nano-structures [3]. At high-temperatures, VO 2 is in a paramagnetic state with a tetragonal (P4 2 /mnm) rutile structure (R). Below the MI transition the V 4+ ions dimerize into non-magnetic pairs and cant/twist into a monoclinic (P2 1 /c) structure (M1) [2]. Fig. 1 shows schematics of V-V pairing of VO 2 in different phases.
Another intermediate monoclinic phase (M2) with only half of the V 4+ dimerized and the other half canting was first reported in Cr doped VO 2 [8,9]. Later, the M2 phase was also found to be stable under certain conditions, for example, by applying very small uniaxial stresses to pure VO 2 [10] or other transition metal substitutions involving lower oxidation states [11]. The uniaxial stress measurements were exceptionally significant for two very different reasons. On the fundamental side they demonstrated that in pure VO 2 , at 340 K, there is a near degeneracy of the R, M1 and M2 phases. This observation has been used to argue that argue that VO 2 is as clear example of a Mott-Hubbard insulator and can also be used to argue that VO 2 is an example of a boot-strapped spin-Peierls transition. On the applied/operational side, the profound strain sensitivity of VO 2 requires strain free samples for measurements of intrinsic properties and offers the possibility of using strain, e.g. in thin films via epitaxial mismatch, to tune/modify the system. VO 2 doping with Ti has been demonstrated to be one of the ways to stabilize the M2 phase in between the R and M1 phases at remarkably low Ti doping levels. However, so far, samples have been primarily studied in thin film and polycrystalline form [12][13][14]. In this paper, we present the details of how to grow pure and Ti-doped single crystals of VO 2 in as low strain of a state as possible. Given the profound sensitivity of VO 2 to strain, the availability of such samples is vital for providing intrinsic, bulk comparisons to the growing number of thin film studies of pure and doped VO 2 . In addition, we demonstrate the effect of Ti-valence on doping level when using solution growth out of V 2 O 5 .
Single crystals of V 1−x Ti x O 2 were grown using a high-temperature solution growth technique [15,16]. Typical starting materials for a pure VO 2 growth were roughly 1 gram of VO 2 lump, which was obtained by reducing V 2 O 5 in a N 2 atmosphere, and 8.1 grams of V 2 O 5 powder. The sealed silica tube that holds the mixture of materials was heated   As can be seen there is a roughly linear dependence of x-WDS versus x-nominal, but the slope is close to eight. The black squares plot x-WDS versus x-nominal value determined by comparing the Ti level to the V 4+ level in the melt (i.e. comparing the Ti 4+ level from the TiO 2 to the V 4+ level from the VO 2 ). As can be seen in this case the data fall very close to a line with a slope of unity. This result makes sense considering that Ti cannot have a higher oxidation level and is essentially trapped in the Ti 4+ state by stoichiometry and the V 2 O 5 melt. x W DS values are used throughout this paper to identify the samples.  5 shows the room-temperature powder X-ray diffraction data of the pure VO 2 over a 2θ range of 20-100 • . All peaks can be fitted to the M1 monoclinic structure of VO 2 . Upon doping with Ti, the M2 phase [13,14] appears in between the R and M1 phases. The phase boundaries between R-M2 and M1-M2 split with increasing amounts of Ti substitution, as is shown in Fig. 7 below. Above about 15% of Ti substitution, the M1-M2 phase boundary Temperature-dependent dc magnetic susceptibilities measured both on cooling and warming in 10 kOe are presented in Fig. 6. For most of the measurements, a transparent plastic capsule was used to hold a collection of crystalline rods (see Fig. 3) in order to acquire a large enough signal. Therefore, apart from pure VO 2 , the data shown in Fig. 6 also contains a small diamagnetic background signal from the sample holder. It should be pointed out that although Fig. 6 plots data between 300 K and 375 K for clarity, for x = 0.187, data were collected down to T = 200 K and no signature of a lower-temperature transition was found.
A sharp first-order transition is clearly observed in VO 2 at ∼ 340 K, which corresponds to the metal-insulator, structural R-M1, phase transition. In comparison, with Ti doping such as x = 0.059 and 0.082, the single, first-order transition splits into two, sharp, well defined, first-order transitions. In between these two, first-order transitions, the M2 phase is stabilized [13,14]. Taking the peak positions of the derivatives of the temperature-dependent magnetization as transition temperature values, the evolution of the transition temperatures can be plotted as a function of Ti concentration. In Fig. 7, the transition temperatures obtained in this study are plotted together with recent results from a study of polycrystalline samples [14]. Since the sample holder's signal is essentially temperature-independent over this temperature range, we can also look at the magnetic susceptibility change associated with each phase transition. Fig. 8 shows the size of magnetic susceptibility jump at each transition plotted as a function of Ti substitution level.
With increasing Ti substitution, the R-M2 transition temperature moves higher while the M1-M2 phase transition temperature moves lower (Fig. 7). In between, the M2 phase is The total loss of magnetization from the R phase to M1 decreases with increasing amount of Ti. This can be roughly understood as a consequence of replacing magnetic V 4+ with nonmagnetic Ti 4+ . Ti substitution results in a decrease in magnetization in the paramagnetic R phase by reducing the amount of V 4+ , and an increase of magnetization in the non-magnetic M1 phase by increasing the amount of un-paired V 4+ ions. The un-paired V 4+ also give rise to a clear Curie tail at low temperatures [14]. It worth pointing out, however, by looking at Black lines show the trend from the polycrystalline study [14]. Note: no signature of the M1-M2 transition was found down to 200 K for x = 0.187.
samples, unlike thin film samples, strain is not playing a significant role [3].
In conclusion, we've used a high-temperature solution technique to grow large, low strain, diffraction study will be needed to provide more details about the Ti doping effect on the structure and stability of VO 2 in these phases.