Shape-controlled synthesis and influence of W doping and oxygen nonstoichiometry on the phase transition of VO2

Monoclinic VO2(M) in nanostructure is a prototype material for interpreting correlation effects in solids with fully reversible phase transition and for the advanced applications to smart devices. Here, we report a facile one-step hydrothermal method for the controlled growth of single crystalline VO2(M/R) nanorods. Through tuning the hydrothermal temperature, duration of the hydrothermal time and W-doped level, single crystalline VO2(M/R) nanorods with controlled aspect ratio can be synthesized in large quantities, and the crucial parameter for the shape-controlled synthesis is the W-doped content. The dopant greatly promotes the preferential growth of (110) to form pure phase VO2(R) nanorods with high aspect ratio for the W-doped level = 2.0 at% sample. The shape-controlled process of VO2(M/R) nanorods upon W-doping are systematically studied. Moreover, the phase transition temperature (Tc) of VO2 depending on oxygen nonstoichiometry is investigated in detail.

for all the nanostructures. Gao and co-workers have also regulated the hysteresis width through the nano-size effect 13 , which provides a key that nanoscale VO 2 (M 1 /R) possesses the probability of tuning hysteresis width for obtaining a sharper, more reproducible phase transition. Up to now, more than 20 compounds of vanadium oxide (VO, V 2 O 3 , VO 2 , V 6 O 13 , V 8 O 15 , V 2 O 5 and so on 14 ) and 10 polymorphs of VO 2 (B, A, T, M 1 , M 2 , R phase and so on 15 ) had been reported. Only the VO 2 (M/R) (the M 1 phase is referred to as the M phase of VO 2 in this study) experiences a fully reversible MIT at the vicinity of room temperature (RT). Moreover, low temperature synthetic method has usually generated VO 2 (B) nanobelts and subsequently can be transformed to VO 2 (M/R) by the post-heating treatment, but the nanostructure has been nearly destroyed [16][17][18] . So it should be a challenge to synthesize pure phase VO 2 (M/R) with a shape controlled nanostructure.
The ongoing debate associated to the fundamental origin of the phase transition behavior in VO 2 involves electron-correlation-driven (Mott transition) 19,20 , structure-driven (Peierls transition) 21,22 , or the cooperation of both 23 . W doping is known as an effective route to regulate electron density in the conduction band for decreasing T c by approx. 20-26 °C/at% W for the bulk and by 50-80 °C/at% W in nanostructures [24][25][26][27] . Synthesis of VO 2 (M/R) by controlling both the shape of nanostructures and the amount of W dopant could be a good strategy to narrow the hysteresis width while reducing T c for obtaining an excellent phase transition property of VO 2 -based materials. Of note, systematically experimental investigation of nonstoichiometric effect in VO 2 has been insufficient. The phase transition behavior has been demonstrated to be also sensitive to vanadium or oxygen related vacancies, even a deviation in the oxygen stoichiometry by a few percent can cause the lattice structure change and result in several orders of magnitude difference in the resistivity transition or the phase transition temperature shift 28,29 . Therefore, studying on oxygen nonstoichiometry induced reduction of T c will contribute to the general understanding of the intrinsic MIT mechanism in VO 2 .
In this study, we successfully explored a one-pot hydrothermal method to prepare VO 2 (M/R) with desired morphology. It is inspiring to discover that the W dopant promotes the generation of pure phase VO 2 (M/R) nanorods with high aspect ratio. Moreover, the effect of oxygen nonstoichiometry on the structural phase transition and subsequently T c of VO 2 is discussed in detail.

Results
Shape-controlled synthesis and phase metamorphosis behavior upon W doping. Figure 1 shows the crystalline phase metamorphic behavior of W x V 1−x O 2 with x = 0, 0.5, 1.0 and 2.0 at% respectively where the temperature being kept at 280 °C but the different duration of the hydrothermal time being applied. For the undoped VO 2 , pure phase VO 2 (B) is obtained for the duration of the hydrothermal time for 6 h. By increasing the duration of the hydrothermal time from 12 to 72 h, the peak of {011} for VO 2 (M) (M {011} at around 27.8°) appears and becomes more significant. However, there always exists the secondary phase VO 2 (B) in the final product. Serial SEM images in Fig. 2 show the morphology transition behavior of the undoped VO 2 upon increasing the duration of the hydrothermal time. Products of the metastable VO 2 (B) are the tangled nanobelts in the morphology for the 6 h-sample. By increasing the duration of the hydrothermal time from 12 to 72 h, VO 2 (B) nanobelts always exist as partial morphology except the block or snowflake VO 2 (M). In conclusion, we could not synthesize pure phase VO 2 (M) without W doping.   data confirm the pure phase VO 2 (M/R) is exactly free from the existing of the other V-O compounds and other VO 2 phases, which was reported by our group in the recently study 30 . As an overall comparison, the schematic illustration of the morphology metamorphic behavior of VO 2 is summarized in Fig. 6.     Table 1. It increases significantly from 2.1 to 5.7 with increasing the W-doped level from 1.0 to 2.0 at%. The strong intensity of the {110} reflections points to the strongly preferential growth direction of the structures, as has also been noted previously for VO 2 nanowires prepared at high temperatures by vapor transport [31][32][33] . Simultaneously, the aspect ratio of the VO 2 (R) nanorods increases from nearly 5 to 10 with the increased dopant. Whereas, if the W-doped level increases from 4.0 to 10.0 at%, the intensity ratio {110}/{101} decreases from 2.5 to 1.1. Meanwhile, the aspect ratio of nanorods decreases with the increased W dopant as shown in Fig. 8. Finally, the bulk crystal of VO 2 (R) is grown for the 10.0 at% sample. The results indicate a certain doping level of W can promote the preferential growth of R {110} and the increased aspect ratio of VO 2 (R) nanorods, whereas the excess W would restrain.
The shape-controlled mechanism revealed by TEM. The length of W-doped 4.0 at% VO 2 nanorods (synthesized at 280 °C for 72 h) is about 2.5 μ m with 600 nm in diameter as shown in the low magnification TEM image in Fig. 9A. The single-crystalline nanorods is confirmed by the lattice images of HRTEM and the inset SAED pattern as shown in Fig. 9. The lattice constants observed in Fig. 9B are 0.3236 and 0.2430 nm respectively, which can be indexed to the spacing of R {110} and R {101}, and the angle between the two lattice images is 67.9° in arc and this corresponds to the angle between the designated crystal planes of R (110) and R (101). In addition, the (001) plane orientation is just perpendicular to the nanorod' growth direction R (110), and revealing the preferential growth direction of the VO 2 (R) nanorods is along [001]. The results demonstrate that the preferential growth of nanorod' growth direction R (110) is responsible for the increased aspect ratio of VO 2 (R) nanorods. It is generally known that the greater the d-spacing, the atom arrange more closely on the crystal plane. For the body-centered tetragonal VO 2 (R), (110) with the largest d-spacing contributes to the lowest surface energy for the preferential growth of VO 2 grains. According to the SAED pattern shown in the inset of Fig. 9A, the bright diffraction spots reveal the good crystallinity of the sample. Based on the Bragg equation, the diffraction spots can be ascribed to different crystal planes of VO 2 (R). The three Bravais lattice points shown in the SAED of the inset of Fig. 9A correspond to crystal planes of R (110), R (101) and R (211) respectively as indexed therein. This definitely demonstrates the nanorods belong to VO 2 (R). Moreover, no fringe  Influence of oxygen nonstoichiometry on the phase transition behavior. Figure  To study the unusual low T c for the HTh 1 synthesized undoped sample, we directly compare the DSC for this sample by the after annealing (HTh 1 + Annealing) with that for the hydrothermal undoped one treated at 160 °C for 72 h and after annealing (HTh 2 + Annealing) (annealing at 500 °C for 1 h in high-purity argon and this being also prepared by our group 34 ) as shown in Fig. 10B. For the sake of comparison, we list Table 2 to show T c and hysteresis width depending on the W-doped level and fabrication processes. When the undoped sample is synthesized by the (HTh 1 + Annealing) process, T c,h and T c,c is about 59.7 and 47.2 °C respectively. Therefore, the T c is about 53.5 °C, which is also lower than the (HTh 2 + Annealing) fabricated undoped one (T c being c.a 63.0 °C as shown in Table 2). Figure 10C shows the XRD patterns of the undoped samples synthesized by the designated two fabrication processes. The peaks of VO 2 (B) vanish and all of the peaks can be indexed to pure phase VO 2 (M) for the HTh 1 synthesized undoped sample by the after annealing, which is similar to the (HTh 2 + Annealing) fabricated one. The inset close-up shows that M (011) peak shifts to low angles when comparing the (HTh 1 + Annealing) synthesized undoped sample with those by the (HTh 2 + Annealing) synthesized one, which indicates the lattice spacing of M (011) increases. Both micron-sized snowflake and block-like morphologies are observed for the (HTh 1 + Annealing) synthesized undoped sample as shown in Fig. 10D. Whereas, nanostructure is grown by the (HTh 2 + Annealing) fabrication process. Thanks to the formation energies of oxygen vacancies in rutile oxides are very high, the high hydrothermal temperature (280 °C) and reductive hydrothermal atmosphere for the (HTh 1 + Annealing) method may contribute to the generation of oxygen vacancies to form nonstoichiometric VO 2-δ compared to the (HTh 2 + Annealing) process (160 °C), and this would promote the lattice structural transition 35,36 . Discussion. To determine the oxygen stoichiometry, the thermogravimetric analysis of the samples was conducted as shown in Fig. 11. According to the TG curves, it can be found there exists one stage for the complete oxidization of the samples in the range of 300-600 °C. The weight gain (Δ TG ) is about 10.4 %, 10.5 % and 9.6 % for the HTh 1 synthesized undoped VO x , (HTh 1 + annealing) undoped VO y and (HTh 2 + annealing) undoped VO z respectively. The reaction equations for the oxidization of the samples can be given as follows (1): Where M O and M VOx represent molar mass of oxygen and VO x respectively. When combining the above formulas (2) and experimental results, we can work out x = 1.96, y = 1.95 and z = 2.00 respectively. The fact demonstrates that oxygen deficiency is formed in the HTh 1 synthesized undoped VO 1.96 and (HTh 1 + annealing) undoped VO 1.95 , and the precisely stoichiometric VO 2.00 is formed in the (HTh 2 + annealing) undoped sample. Son and co-workers have synthesized monoclinic VO 2 micro-and nanocrystals by optimizing the hydrothermal conditions 37 . In their research, the phase transition temperature of stoichiometric VO 2.00 microrods is around 68 °C. Usually, the T c of MIT for VO 2 is affected by doping, nanoscaling, nonstoichiometry, strain and etc 12,30,38,39 . For the HTh 1 synthesized undoped micron-sized VO 1.96 , the reason for the unusual low T c may be due to the oxygen nonstoichiometry. The nano-size effect may be responsible for the relative lower T c (63 °C) of the (HTh 2 + Annealing) synthesized stoichiometric VO 2.00 nanostructure.  Table 2. DSC parameters of the HTh 1 synthesized sample with W-doped at 0.0 at% and of the undoped samples synthesized by the (HTh 1 + Annealing) method and the (HTh 2 + Annealing) process respectively. Figure 11. The thermogravimetric analysis of the samples.
Scientific RepoRts | 5:14087 | DOi: 10.1038/srep14087 Figure 12A,B shows the Raman spectra of the samples depending on dopant level and fabrication processes. The peaks in the Raman spectra are all identified as 144 (B 1g ), 191 (A g ), 223 (A g ), 260 (A g ), 308 (A g ), 338 (A g ), 388 (A g ), 437 (A g ), 442 (E g ), 499 (A g ), 617 (A 1g ), and 826 (B 2g ) cm −1 respectively, and these Raman-active modes are the clear evidence of the existing of VO 2 (M) belonging to space group C 2h 5 , which agrees with the identified Raman peaks by other researchers [40][41][42][43] . The intensity ratio between the peak of 191 and that of 223 cm −1 (191/223) of the HTh 1 synthesized undoped VO 1.96 is 1.6. When comparing the (HTh 2 + Annealing) synthesized undoped VO 2.00 with those by the (HTh 1 + Annealing) synthesized undoped VO 1.95 , the intensity ratio decreases from 2.3 to 1.3. H. T. Kim and co-workers have studied Raman spectra for the MIT of the undoped VO 2 in detail and deduced the conclusion that the Raman-active A g modes at 191 and 223 cm −1 were explained by the pairing and the tilting of V cations respectively 43 . Hence, the decreased relative intensity of 191 cm −1 peak suggests the depairing of V cations and the occurring of the localized structural phase transition (SPT, induced possibly by oxygen nonstoichiometry for the HTh 1 synthesized undoped VO 1.96 and (HTh 1 + annealing) undoped VO 1.95 ), and this might cause the transformation from the intrinsic structure of the matrix of VO 2 (M) to the localized rutile structure. In addition, the local rutile structure is the structure-guided domain, which will act as the initial nucleation site for the whole SPT 44 . This process might promote MIT for the origin of the lowering T c . However, this origin is still under the debate among the concerned experts as cited in the literature by Y. Xie et al. for an example, who pointed out that the atomic structure of isolated W dopant play a role in driving the nearby symmetric monoclinic VO 2 lattice towards rutile phase, resulting in the depression of T c 45,46 . Hence the exact mechanism for the observed unusual phenomena requires our further investigation.

Conclusions
In this study, pure phase VO 2 (M/R) with controlled morphology were successfully prepared via one-step hydrothermal method. The addition of a certain level of W (0.5-2.0 at%) is vital to synthesize the pure phase VO 2 (M/R) nanorods. The assured level of W doping can promote the preferential growth of {110} to form VO 2 (M/R) nanorods with high aspect ratio. It must be emphasized that the unusual low T c equals to 55.8 and 53.5 °C is observed for the nonstoichiometric VO 1.96 and VO 1.95 in the bulk respectively, and the T c is 63.0 °C for the precisely stoichiometric VO 2.00 nanostructure. The present study demonstrates an improvement of the phase transition behavior and reduces the hindrances for the advanced applications of VO 2 -based materials.  The preparation process. The detail of this part has been described in previous report 30 . Briefly, V 2 O 5 and oxalic acid (1: (1-3) in molar ratio) were directly added to 75 ml deionized water at RT. Then, a certain amount of W dopant was dispersed into the above solution with magnetic stirring. After mixing for 1 h, the resulting precursor was transferred into a 100 mL stainless steel autoclave with polyphenylene cup, then being sealed and maintained at 280 °C for 6-72 h. After the autoclave cooling to RT, a dark blue precipitate was obtained. The product was washed with deionized water and acetone for several times, then centrifuged at 8000 rpm for 8 min and dried in vacuum at 60 °C for 6 h.
In this study, (NH 4 ) 5 H 5 [H 2 (WO 4 ) 6 ]·H 2 O was used as the W dopant, and the reported W-doped content here is based on the quantity of W atoms added in the feed. The sample synthesized by the duration of the hydrothermal time for 6 or 72 h is simplified to the 6 or 72 h-sample.
Characterization techniques. The phase purity of the products was examined by an X-ray diffractometer (XRD, PANalytical X'pert Pro MPD) in the 2θ range of 5-80° with the step of 0.0083° using Cu-Kα radiation (λ = 1.54178 Å). The operating voltage and current were kept at 40 kV and 40 mA, respectively. The morphology and dimensions of the products were investigated using a field emission scanning electron microscope (FESEM, S-4800, Hitachi Japan) under the operating voltage of 2 kV. A JEOL-2100F instrument operated at 200 kV was used to acquire high-resolution transmission-electron-microscopy (HRTEM) images and selected area electron diffraction (SAED) patterns. Raman scattering spectra of the samples were recorded on a LabRAM HR800 micro-Raman spectrometer using a 532 nm wavelength YAG laser. The phase transition properties depending on the surrounding temperature of the as-prepared VO 2 were studied by differential scanning calorimetry (HDSC, PT500LT/1600) under the temperature range from 25 to 100 °C under the circulatory heating/cooling cycles. The thermogravimetric analysis (TG) of the samples was conducted on a Nicolet 6700-Q50 thermal analyzer under dry air flow in the range of 50-650 °C with a heating rate of 5 °C min −1 .