Facile synthesis and phase transition of V₂O₃ nanobelts

Abstract

The controlled synthesis of high quality one-dimensional V₂O₃ materials is critical for addressing the fundamental aspects of metal–insulator transition (MIT) in these materials. Here we developed a controllable approach for synthesizing pristine V₂O₃ nanobelts by reducing hydrothermally synthesized V₂O₅ nanobelts in the gas phase. The obtained V₂O₃ nanobelts showed high crystallinity and purity with length up to tens of micrometers. We investigated the changes in electrical conductance, magnetic susceptibility and Raman spectra that accompanied the phase transition of nanobelts with variable-temperature measurements. The MIT behaviour of V₂O₃ nanobelts was very similar to that of the bulk materials while the magnetic phase transition of V₂O₃ was suppressed in the nanobelts due to the finite size effect. This simple and reliable synthesis approach makes the V₂O₃ nanobelts easily accessible for exploring their fundamental properties and potential applications in novel electronic devices.

---

Transition metal oxides with metal–insulator transition (MIT) provided unique models for understanding correlation effects associated with the electron–electron interactions.¹ The dramatic changes in conductivity accompanied with the MIT also allow for their potential applications in novel electronic devices, such as switches,² memory devices³ and field-effect transistors (FETs).⁴ Vanadium sesquioxide (V₂O₃) is a typical MIT material exhibiting phase transition from the metallic and paramagnetic to insulating and antiferromagnetic state at a transition temperature of ∼150 K.⁵ A lot of efforts have been devoted to tuning the phase transition temperature (T_c) of V₂O₃ by strain,⁶ pressure⁷ and doping.⁸ But the effect of dimensionality on the MIT of V₂O₃ was still unknown. Quasi-one dimensional (1D) MIT materials are expected to show different transition behaviour to their bulk counterparts due to the large quantum fluctuations at low dimensions.¹ Therefore, compared with V₂O₃ bulk materials and thin films it is important to investigate the MIT of 1D V₂O₃ materials due to its feasible applications in future nanoelectronics. The first step towards the fundamental studies and applications of 1D V₂O₃ is to realize the controlled synthesis of these materials. V₂O₃ nanotubes⁹ and nanorods¹⁰ were synthesized by using VO₂ nanostructures as templates, but the obtained products were short and highly defective, which is undesirable for characterizing the pristine properties of 1D V₂O₃. Currently, the synthesis of high quality 1D V₂O₃ still remains a challenge. Here we report a controllable approach on the synthesis of pristine V₂O₃ nanobelts with length up to tens of micrometers at a large scale. The obtained V₂O₃ nanobelts were characterized by various microscopic and spectroscopic methods. The phase transition of these materials was also investigated in details.

The challenge of the synthesis of high quality 1D V₂O₃ arises from the difficulties in controlling the oxidation state of V and the morphology of the obtained V₂O₃. To address this challenge, we developed a two-step synthesis approach as shown in Fig. 1a. We first prepared highly crystalline V₂O₅ by a well-established hydrothermal method using ammonium metavanadate (NH₄VO₃) as precursor.¹¹ After that, the obtained V₂O₅ nanobelts were used as templates to synthesize V₂O₃ nanobelts by reducing V₂O₅ to V₂O₃ using the mixture of hydrogen and sulfur vapor at a temperature of 450 °C. The production of V₂O₃ nanobelts could be carried out on the surface of substrates for device fabrication and characterizations or in bulk for the large-scale production (see ESI† for experimental details).

Fig. 1 (a) Schematic for the synthesis of V₂O₃ nanobelts via a two-step approach. V₂O₅ nanobelts were obtained by hydrothermal synthesis and then were used as templates to grow V₂O₃ nanobelts by the reduction of V₂O₅ in the mixture of H₂ and S vapor. (b) and (c) TEM images of V₂O₅ and V₂O₃ nanobelts, respectively. Insets, SAED patterns of V₂O₅ and V₂O₃ nanobelts.

We utilized transmission electron microscopy (TEM) and selected area electron diffraction (SAED) to characterize the morphology and structure of the obtained nanobelts. The hydrothermally synthesized V₂O₅ nanobelts were single crystalline with width of ∼20–500 nm and length up to tens of micrometers (Fig. 1b). After the reduction, the morphology of the V₂O₅ templates was well-maintained and the SAED pattern taken from an individual nanobelt along the [4 2 −1] zone axis demonstrated that the obtained V₂O₃ nanobelts were still single crystalline (Fig. 1c). The successful conversion of V₂O₅ nanobelts to V₂O₃ nanobelts was attributed to the use of the mixture of H₂ and S vapor as reducing agents. Using H₂ (50 sccm H₂ with 200 sccm Ar) which is a widely used reductant in gas phase, we observed a lot of cracks on the obtained V₂O₃ nanobelts after reduction even at low reaction temperature (Fig. S1 in ESI†), which indicated that the reduction with H₂ was too harsh for maintaining the morphology of the nanobelts. As a mild reductant, S vapor can reduce metal oxides in a more controllable manner as confirmed in our previous work, where MoO₃ was reduced to MoO₂ successfully with S vapor.¹² Therefore, the pure S vapor was also used to reduce V₂O₅ nanobelts, but it was not successful. Only with the mixture of H₂ and S vapor at a suitable ratio, the reduction of V₂O₅ was found to be efficient and the morphology of nanobelts was well-maintained. The effects of reaction temperature and time on the formation of V₂O₃ nanobelts were also systematically investigated and the optimized temperature and time for the reduction was found to be at 450 °C for 1 h.

To exclude the coexistence of vanadium oxides with other oxidation states, such as VO₂ and V₆O₁₃ and vanadium sulfides (VS_x), we carried out X-ray photoelectron spectroscopy (XPS) and X-ray diffraction (XRD) measurements on the obtained products. As shown in the XPS spectra of V₂O₅ nanobelts before and after the reduction (Fig. 2a), the peak corresponding to V 2p_3/2 shifted from ∼517.2 eV to ∼515.8 eV, indicating a change of V⁵⁺ to V³⁺ after the reduction.¹³ The broader full width at half maximum (FWHM) for the V 2p_3/2 peak after reduction also reflected the decrease of oxidation degree.¹³ In addition, no peak was observed in the region of the binding energy of S 2p_1/2 and S 2p_3/2 (∼165–160 eV) after reduction, indicating that no vanadium sulfides coexisted with V₂O₃. XRD patterns of the V₂O₅ and V₂O₃ nanobelts were also collected to reveal the changes in crystal structures accompanied with the reduction. As shown in Fig. 2b (top), the diffraction peaks were well-indexed to V₂O₅ with an orthorhombic structure (JCPDS card no. 41-1426). The calculated lattice parameters were a = 11.516 Å, b = 3.566 Å and c = 4.373 Å. Both the smooth base line and sharp peaks indicated a high purity of the as synthesized V₂O₅ nanobelts. The XRD pattern after reduction (Fig. 2b, bottom) was consistent with a rhombohedral V₂O₃ with the lattice parameters of a = 4.954 Å, b = 4.954 Å, c = 14.008 Å (JCPDS card no. 34-0187). No other substances were detected by XRD measurements, further confirming that the V₂O₅ precursors were completely converted to V₂O₃ without by-products after reduction.

Fig. 2 (a) XPS measurements for the binding energies of V₂O₅ and V₂O₃ nanobelts. Note that no peak appears at the binding energies of S 2p_1/2 and S 2p_3/2 (∼165–160 eV), indicating no S was detected in the V₂O₃ nanobelts. (b) XRD pattern of V₂O₅ and V₂O₃ nanobelts.

The pure V₂O₃ nanobelts with lengths up to tens of micrometers greatly facilitated the investigation of the MIT temperature of V₂O₃ nanobelts by variable-temperature measurements of the conductance of individual nanobelts. We fabricated two-terminal devices with ∼200–500 nm channel lengths on individual V₂O₃ nanobelts on 300 nm SiO₂/Si substrates by electron beam lithography (EBL) and electron beam deposition of 5 nm Ti/50 nm Au (Fig. 3a, inset). The obtained devices were annealed at 400 °C in vacuum for 30 min to improve the contacts between nanobelts and electrodes. The electrical conductance of these devices was measured in a vacuum probe station at a pressure of ∼10⁻⁶ mbar to prevent the oxidation of the nanobelts at temperatures ranging from 300 to 80 K. The resistivity of the V₂O₃ nanobelts increased upon cooling with a total resistivity change of ∼10³ (Fig. 3a), indicating a metal–insulator transition of the nanobelt during the cooling. The phase transition temperature was estimated to be ∼150 K by taking the maximum in the derivative of the resistivity plot.¹⁴ The measured T_c of V₂O₃ nanobelts was similar to that of the bulk materials and thin films of V₂O₃,^5,14 indicating that the width of the nanobelt (∼100 nm) was still above the critical size of V₂O₃ to show obvious quantum fluctuations.

Fig. 3 (a) The dependence of resistivity on temperature of the V₂O₃ nanobelt shown in the inset during the cooling circle. Inset, optical image of a two-terminal device made on a single V₂O₃ nanobelt with a width of ∼100 nm. (b) The dependence of magnetic susceptibility on temperature of V₂O₃ commercially purchased powder and nanobelts with applied field of ∼1 T.

Accompanied with the metal–insulator transition, the commercially purchased V₂O₃ powder (Aldrich Sigma, product no. 463744-5G, 99.99%, see XRD pattern in Fig. S3†) also shifted from paramagnetic phase to antiferromagnetic phase as indicated by the sudden change in the magnetic susceptibility at ∼160 K measured using superconducting quantum interference device (SQUID) with applied field of 1 T (Fig. 3b, blue curve). This magnetic transition behaviour was obviously different from other vanadium oxide phases as discussed in the ESI.† Different from the powder materials, the magnetic susceptibility of V₂O₃ nanobelts kept on increasing after reaching a platform at ∼130 K during the cooling cycle (Fig. 3b, red curve), which indicated that the paramagnetic to antiferromagnetic transition at low temperature was suppressed in the nanobelts. The suppression of magnetic phase transition in V₂O₃ was also observed in hydrothermally synthesized V₂O₃ nanoparticles.¹⁵ The antiferromagnetism of the low temperature (monoclinic) phase of V₂O₃ was originated from the coupling of adjacent ferromagnetic (010) plane with opposite magnetic moments.¹⁶ The finite size of V₂O₃ nanomaterials could break the long range magnetic ordering in the low temperature phase and therefore, the magnetic transition introduced by the antiparallel alignment of magnetic moments in adjacent planes might not occur with the structural transition at the same time¹⁵ and result in the different magnetic susceptibility-temperature dependence in the nanomaterials from the bulk materials. Different from V₂O₃ single crystal, the crystallographic, electronic, and magnetic transitions in nanomaterials may not occur simultaneously due to finite size effect, stress and defect states in these nanomaterials.^15,17

To further explore the phase transition of the V₂O₃ nanobelts, we also utilized variable-temperature Raman spectroscopy (VT-Raman) to measure the T_c of individual V₂O₃ nanobelts. Compared to other methods such as variable-temperature XRD,⁹ optical¹⁸ and electrical measurements,¹⁴ VT-Raman is capable of detecting individual material with high spatial resolution and does not require complicated lithographic fabrications. So this method was widely used to trace the phase transition of low dimensional materials,¹⁹ including V₂O₃.^20,21 We first measured the Raman spectra of V₂O₃ nanobelts sitting on 300 nm SiO₂/Si substrates in vacuum at room temperature by mapping a few V₂O₃ nanobelts (Fig. 4a) using 514 nm laser excitation with prolonged exposure time with a laser power of ∼25 mW. But no Raman signal except for the peak at ∼520 nm which was originated from the Si substrates was found under this laser power (Fig. 4b). After that, we collected Raman spectra of V₂O₃ nanobelts in air at room temperature, and observed multiple Raman peaks which were exactly the same as the peaks of V₂O₅ nanobelts²² (Fig. S2 in the ESI†). The Raman spectra of V₂O₃ commercial powders were also collected in vacuum and in air (Fig. S4†). We concluded that the pristine V₂O₃ nanobelts were not Raman active with a laser power of ∼25 mW and the V₂O₃ nanobelts were partially oxidized to V₂O₅ under the laser irradiation during the Raman measurement in air and therefore, exhibited the characteristic Raman peaks of V₂O₅. However, different from at room temperature, no Raman peak was detected from the V₂O₃ nanobelts at 100 K in air, indicating that the V₂O₃ nanobelts in low temperature phase were not oxidized. So we inferred that the reactivity of V₂O₃ with O₂ under the laser irradiation might be different in the two phases. To confirm this, we collected the Raman spectra of the V₂O₃ nanobelts at varied temperature (296–100 K) in air. To avoid the oxidization introduced by previous measurements, the location of the laser on the nanobelts was changed with the change of temperature. We found that the V₂O₃ nanobelts showed the typical peaks of V₂O₅ at above 150 K and most peaks disappeared suddenly at below ∼150 K which was exactly the same as the T_c measured by electrical method (Fig. 4c). To exclude the possible phase transition of V₂O₅ occurred at ∼150 K, the same VT-Raman measurements on V₂O₅ nanobelts were done and the intensities of V₂O₅ nanobelts were not found to change with temperature (Fig. 4d). The disappearance of Raman peaks of V₂O₃ nanobelts at ∼150 K indicated that the insulating phase of V₂O₃ was more chemically inert than the metallic phase due to the obvious differences in the band structures of the two phases.²³ Therefore, VT-Raman offers an alternative tool to investigate the phase transition of the V₂O₃ materials by tracing the transition in reactivity accompanied with MIT.

Fig. 4 (a) Optical image of a few individual V₂O₃ nanobelts deposited on a SiO₂/Si substrate. (b) Raman mapping image acquired at the same region of (a) with 520 cm⁻¹ (Si peak) peak intensity using 514 nm laser excitation in vacuum. No Raman peak expect for the Si peak was detected as shown in the Raman spectrum in the inset, indicating pristine V₂O₃ nanobelts did not show Raman peak. (c) and (d) Raman spectra of V₂O₃ and V₂O₅ nanobelts at varied temperature in air, respectively.

In summary, we developed a controllable approach for the synthesis of high quality V₂O₃ nanobelts by using V₂O₅ nanobelts as template. The morphology of nanobelts was well-maintained and the oxidation state of V was well-controlled in the V₂O₃ nanobelts by using carefully chosen reducing agents as indicated by TEM, XPS and XRD measurements. The obtained V₂O₃ nanobelts exhibited MIT at ∼150 K estimated by variable temperature electrical measurements, similar to that of bulk V₂O₃. The magnetic phase transition in the nanobelts was found to be dramatically different from the bulk materials, which is like due to the finite size effect of nanomaterials. In addition to the changes in electrical and magnetic properties, we also observed that the chemical reactivity of V₂O₃ nanobelts changed obviously associated with phase transition as indicated by the VT-Raman measurements. This simple approach opens up a new avenue for controlled synthesis of 1D V₂O₃ materials, and will make this high quality material easily accessible for fundamental aspects and various applications.

Acknowledgements

X.C. acknowledges the financial support from NSFC (no. 21373127). L.J. acknowledges the NSFC (no. 21322303, 51372134) and National Program for Thousand Young Talents of China.

Notes and references

1. M. Imada, A. Fujimori and Y. Tokura, Rev. Mod. Phys., 1998, 70, 1039–1263.
2. G. Stefanovich, A. Pergament and D. Stefanovich, J. Phys.: Condens. Matter, 2000, 12, 8837–8845.
3. M. J. Lee, Y. Park, D. S. Suh, E. H. Lee, S. Seo, D. C. Kim, R. Jung, B. S. Kang, S. E. Ahn, C. B. Lee, D. H. Seo, Y. K. Cha, I. K. Yoo, J. S. Kim and B. H. Park, Adv. Mater., 2007, 19, 3919–3923.
4. S. Hormoz and S. Ramanathan, Solid-State Electron., 2010, 54, 654–659.
5. J. Feinleib and W. Paul, Phys. Rev., 1967, 155, 841–850.
6. H. Schuler, S. Klimm, G. Weissmann, C. Renner and S. Horn, Thin Solid Films, 1997, 299, 119–124.
7. D. Grieger, C. Piefke, O. E. Peil and F. Lechermann, Phys. Rev. B: Condens. Matter Mater. Phys., 2012, 86, 155121.
8. Y. Guo, S. J. Clark and J. Robertson, J. Chem. Phys., 2014, 140, 054702.
9. S. A. Corr, M. Grossman, J. D. Furman, B. C. Melot, A. K. Cheetham, K. R. Heier and R. Seshadri, Chem. Mater., 2008, 20, 6396–6404.
10. A. C. Santulli, W. Xu, J. B. Parise, L. Wu, M. C. Aronson, F. Zhang, C. Y. Nam, C. T. Black, A. L. Tiano and S. S. Wong, Phys. Chem. Chem. Phys., 2009, 11, 3718–3726.
11. X. K. Hu, D. K. Ma, J. B. Liang, Q. Xie, Y. C. Zhu and Y. T. Qian, J. Phys. Chem. C, 2007, 111, 5882–5885.
12. X. Wang, H. Feng, Y. Wu and L. Jiao, J. Am. Chem. Soc., 2013, 135, 5304–5307.
13. J. Mendialdua, R. Casanova and Y. Barbaux, J. Electron Spectrosc. Relat. Phenom., 1995, 71, 249–261.
14. B. S. Allimi, S. P. Alpay, C. K. Xie, B. O. Wells, J. I. Budnick and D. M. Pease, Appl. Phys. Lett., 2008, 92, 202105.
15. V. A. Blagojevic, J. P. Carlo, L. E. Brus, M. L. Steigerwald, Y. J. Uemura, S. J. L. Billinge, W. Zhou, P. W. Stephens, A. A. Aczel and G. M. Luke, Phys. Rev. B: Condens. Matter Mater. Phys., 2010, 82, 094453.
16. R. Moon, Phys. Rev. Lett., 1970, 25, 527–529.
17. X. B. Chen, J. H. Shin, H. T. Kim and Y. S. Lim, J. Raman Spectrosc., 2012, 43, 2025–2028.
18. M. K. Stewart, D. Brownstead, S. Wang, K. G. West, J. G. Ramirez, M. M. Qazilbash, N. B. Perkins, I. K. Schuller and D. N. Basov, Phys. Rev. B: Condens. Matter Mater. Phys., 2012, 85, 205113.
19. A. Ridah, M. D. Fontana and P. Bourson, Phys. Rev. B: Condens. Matter Mater. Phys., 1997, 56, 5967–5973.
20. A. Okamoto, Y. Fujita and C. Tatsuyama, J. Phys. Soc. Jpn., 1983, 52, 312–317.
21. N. Kuroda and H. Y. Fan, Phys. Rev. B: Solid State, 1977, 16, 5003–5008.
22. C. V. Ramana, R. J. Smith, O. M. Hussain and C. M. Julien, J. Vac. Sci. Technol., A, 2004, 22, 2453–2458.
23. Y. Guo, S. J. Clark and J. Robertson, J. Chem. Phys., 2014, 140, 054702.

---

Footnote

† Electronic supplementary information (ESI) available: Experimental details and supporting Fig. S1–4. See DOI: 10.1039/c4ra13707h

---