Huafang Zhanga,
Quanjun Li*a,
Benyuan Chengb,
Zhou Guana,
Ran Liua,
Bo Liua,
Zhenxian Liuc,
Xiaodong Lid,
Tian Cuia and
Bingbing Liu*a
aState Key Laboratory of Superhard Materials, Jilin University, Changchun 130012, China. E-mail: liubb@jlu.edu.cn; liquanjun@jlu.edu.cn; Fax: +86-0431-85168256; Fax: +86-18043176111; Tel: +86-0431-85168256 Tel: +86-18043176111
bChina Academy of Engineering Physics, Mianyang, Sichuan 621900, China
cU2A Beam Line, Carnegie Institution of Washington, Upton, New York 11973, USA
dBeijing Synchrotron Radiation Facility, Institute of High Energy Physics, Chinese Academy of Sciences, 100049, Beijing, China
First published on 24th October 2016
In this study, the high-pressure behavior of monoclinic vanadium dioxide (VO2 (M1)) was revisited using infrared reflectivity (IR) spectroscopy, Raman spectroscopy and in situ synchrotron X-ray diffraction (XRD) up to 64.7 GPa. Upon compression, VO2 (M1) follows the expected the structure transition sequence, M1 → M′1 → X, and we found that the structural transition from M′1 to X phase is completed at about 59 GPa. Moreover, our IR data demonstrated that the M′1 phase is a semiconductor within the pressure region of 11.4–43.2 GPa and became metallic with further compression, and that the X phase is metallic. Further analysis suggests that the pressure-induced metallization (PIM) of the M′1 phase is associated with electron–electron correlations, while the PIM from the M′1 to the X phases is relate to structural phase transitions. These results provide further insight into the PIM of VO2 (M1).
Pressure, another important tuning parameter, has great influence on the properties and structures of substances.19–21 Previous studies have observed evidences of PIM process on VO2 (M1) under pressure.22–25 Below ∼10 GPa, M1 phase clearly shows semiconductor behavior.26–28 Under higher pressure, Arcangeletti at al. carried out high-pressure MIR up to 14 GPa and Raman measurements up to 19 GPa and found that MIR reflectivity R(ω) starts to increase and transmittance T(ω) abruptly decreases to very small values above 10 GPa, accompanied with subtle modification of V ion arrangements, the onset of PIM was suggested.23,24 Later, Mitrano et al. proposed that the structure above 10 GPa (named M′1 phase) has the same crystal symmetry with monoclinic M1 phase by X-ray diffraction up to 20 GPa, and inferred a pressure-induced charge delocalization above 12 GPa by analyzing the axial bulk modulus.22 Recently, Bai et al. have further investigated the PIM and pressure-induced phase transitions of VO2 based on electrical resistivity, XRD and Raman spectroscopy.29 The resistivity drops with increasing pressure and drops faster above 14 GPa, then, becomes nonlinearly dropping upon compressing above 23.3 GPa and levels off at the value of 4 to 6 × 10−6 Ωm above 37.7 GPa, suggesting sample being metallic above this pressure. From their XRD and Raman measurements, the phase transitions from M1 to M′1, and M′1 to a new monoclinic X occur at about 13.0 and 34 GPa, respectively, and latter transition is not completed up to the highest pressure of 55 GPa. They pointed out that the changes in resistivity at 14 and 23.3 GPa are related to the transition from M1 to M′1, and M′1 to X phase, respectively, M′1 phase is semiconductor under pressure, and X phase is responsible for the metallization under pressure. However, it is clear from their XRD study that the metallization observed at 37.7 GPa occurs in the pressure region where M′1 and X phase coexist, which naturally hinders the recognition of the origin of PIM. Lately, Zhang et al. showed that resistivity in M′1 phase decreases with increasing temperature under pressure by temperature-dependent in situ resistivity measurements up to 18.1 GPa, which illustrated that the M′1 phase is a semiconductor within low pressure region of 10.4–18.1 GPa, unfortunately, corresponding measurements on M′1 phase in high pressure region and on pure X phase is not reported.28 Therefore, investigating the metallicity of M′1 and X phase under high pressure is still an open subject.
IR spectroscopy is a powerful tool for clarifying a phase is metallic or semimetallic, with showing the corresponding IR reflectivity and its pressure insensitivity.30,31 Early studies show that different from electrical resistivity methods, which could bring controversies from grain boundaries and measuring conditions, the IR spectroscopy decisively dependent on the configuration of atom in the structure, giving more detailed information of measured samples.32,33 Recently, the IR spectroscopy has been widely used in characterizing the metallization process under extreme pressure, such as MnO, SiH4, V3O5, and VO2 (A),30,34–37 for VO2 (M1), nevertheless, only the IR measurements within the 0–14 GPa range have been reported,23 there is an complete lack of corresponding IR data under higher pressure. Thus, here, we revisit the PIM of VO2 (M1) up to 61.2 GPa using IR spectroscopy, accompanied with XRD and Raman spectroscopy characterizing corresponding structure transitions. In our IR study, pure sample was used for measuring, which allows us to detect more details of Rsd(ω) spectra, moreover, the deviation of IR measurements was also considered, which ensure the reliability of our data. We find that the semiconductor M′1 phase transforms into a metallic M′1 phase at about 42.3 GPa, and demonstrate that the X phase is metallic. These results provide a further insight into the PIM of VO2 (M1) under pressure.
Because the direct contact between the sample and diamond culet, the reflectance from the sample–diamond interface (Rsd(ω)) can be calculated based on:38
(1) |
(2) |
In practice, was measured and calculated at ambient pressure before samples loading and the ration of was fixed at 0.18. With the assumption of the pressure independent ratio of , the reflectance at each pressure can be determined accurately based on the ratio of measured at each corresponding pressure, multiply obtained at ambient pressure. The schematic geometry for the Rsd(ω) measurements in a diamond anvil cell is shown in Fig. 1a.
As a good estimation, the percentage of the deviation of the intensity reflected from the air–diamond interface was calculated as:
(3) |
Fig. 2 The reflectivity Rsd(ω) at selected pressure collected upon compression. Inset: pressure dependence of reflection at the point of 1300 cm−1. |
To further verify the PIM for monoclinic VO2, we presented the transmittance T(ω) spectra as a function of pressure in Fig. 3a. As illustrated above, the data below 1000 cm−1, between 1700 cm−1 and 2700 cm−1 were cut out from T(ω) spectra due to the absorption of diamond and the limit of the measuring range of our MIR detector, respectively. The bump at about 3500 cm−1 is result of the bad compensation of the diamond absorption.24 With increasing pressing, T(ω) gradually decreases up to 11.4 GPa. Upon further compressing, T(ω) starts to decrease more obviously. Compare with the data reported in early study (0–14), T(ω) show similar pressure dependence.23 When applied pressure increased to 20.9 GPa, T(ω) decreases to small values. And there almost no transmitted light in the whole spectrum range above 35.2 GPa (the enlarged spectra in the region of 800–1700 cm−1 is shown in the inset of Fig. 3a). These results indicate that the bandgap decreased to a value less than the minimum measured range 1000 cm−1 (0.12 eV). To follow the variation of bandgap under pressure, we also undertook estimation of the bandgap based on T(ω) spectra. Because of the presence of the bump at about 3500 cm−1 and the interference fringes in spectrum, it would yield relatively great errors in bandgap calculations using the methods employed in early study.6,7 Thus, we use the frequency (ωc), above which T(ω) = 0, to estimate the bandgap. The pressure dependence of ωc is shown in Fig. 3b. The ωc (P = 0) value is about 5270 cm−1 (0.65 eV), obtained by a rough linear extrapolation of the ωc values below 10 GPa in Fig. 3b, which is 0.05 eV larger than the bandgap value of VO2 at ambient pressure reported in early study.36 Therefore, the values of ωc (in eV) subtract 0.05 eV giving the corresponding approximate bandgap values at different pressure (Fig. 3c). Below 10.2 GPa, the band gap is weakly pressure dependent. On further increasing pressure, clear changes on slope at about 10.2 GPa and 25 GPa are observed, which could be related to structure transition from the M1 to M′1, and from M′1 to X phase, respectively. The band gap decrease to approximately 0.13 eV at about 29.6 GPa. Unfortunately we cannot calculate the smaller band gap values under higher pressure due to the limit of the measuring range of our MIR detector. We note that a rough linear extrapolation of the data gives a closed band gap at about 40–45 GPa, suggesting samples being metallic in this pressure range, which agree well with the results observed in our Rsd(ω) study.
To have a better understand about this PIM process, we carried out high-pressure XRD measurements in characterizing the crystal structure transitions upon compression (Fig. 4). With increasing pressure, the expected the structure transition from M1 to M′1, and M′1 to the X phase were observed at about 13.1 GPa and 26.0 GPa, respectively. We note that the peak of the M′1 phase gradually decreases with increasing pressure above 26 GPa, whereas the intensity of the main M′1 phase diffraction line is almost the same as that of X phase at about 47.3 GPa, demonstrating that the M′1 phase is still the dominant phase at the critical pressure P = 43.2 GPa, where samples being metallic. These results indicate that the semiconductor M′1 transforms into a metallic M′1 phase. Goodenough reported a critical V–V separation value (2.93 ± 0.04 Å) for VO2, below which the V–V interaction is strong enough to delocalize the 3d electrons.12 At ambient pressure, a short (2.65 Å) and a long (3.12 Å) V–V separation alternating along a axis of M1 phase, leading to the insulating behavior. The pressure dependence of lattice parameters a, b, c and unit-cell volume are shown in Fig. 5. The a axis continuous decrease with pressure, whereas the b and c axes show noticeable changes in slope at about 13.1 GPa, indicating the anisotropy of M′1 phase. The unit-cell volume exhibits a regular continuous decrease under pressure, which can lead to a gradual decrease in atoms distance, thus, it is likely that the V–V separation become less than the critical value at about 43.2 GPa, resulting in the delocalization of the 3d electrons. These results suggest that the PIM in M′1 phase maybe mainly driven by electron–electron correlation, which agrees with the results reported by Baldini et al.25 and the results observed in our VO2 nanoparticles.41 Further, the structure transition from the M′1 to the X phase is completed above 59.0 GPa, where VO2 samples are metallic, revealing that the X phase is metallic.
Fig. 4 (a) XRD diffraction pattern of VO2 at selected pressures upon compression. Stars indicate diffraction peaks from Ar. |
Fig. 5 Pressure dependence of lattice parameters a, b, c and unit-cell volume [panels (a), (b), (c) and (d)] for VO2 samples. |
To gain further insight into the structure transition and metallization of VO2 (M1) under pressure, we carried out high pressure Raman measurements up to 64.7 GPa shown in Fig. 6a. The spectrum at 3.7 GPa is in full agreement with previous Raman data on VO2 in M1 phase.23 The Raman mode frequencies are plotted as a function of pressure in Fig. 6b. Upon compression to 12.9 GPa, a slight change in slope was observed and two weak Raman peaks disappear, suggesting a structural transition from M1 phase to M′1 phase occurs. These results compare well with our XRD study and early reports on VO2 samples.23,29 It is reported that the Raman modes A and B are associated with the paring and tilting motions of V-ions motion in the dimerized chins.23,24,29 When pressure increased above 27.0 GPa, the Raman modes A and B gradually broaden and weaken, moreover, the Raman modes B displays a sharp slope change (Fig. 6b), suggesting the onset of the structure transition from M′1 to the X phase. The most important evidence is that the Raman modes A and B are still well distinguished up to about 50.5 GPa, demonstrating that the V–V dimers exist up to this pressure. Recently, M. Baldini et al. found that the V–V dimers is not suppressed in M′1 phase by means of high pressure atomic pair distribution function measurements up to 22 GPa, and pointed that the suppression of Peierls distortion is not correlated with the change of the electronic properties within the pressure range under investigation.25 In our Raman study, it is clearly shown that the V–V dimers exist up to 50.5 GPa, above the pressure (43 GPa) where the PIM transition occurs, thus, demonstrating that the PIM in M′1 phase is not related to the suppression of Peierls distortion.
Fig. 6 (a) Raman spectra of VO2 at selected pressures upon compression. (b and c) Pressure dependence of the frequency of the observed Raman modes upon compression. |
Moreover, when the pressure increased above 58.3 GPa, all Raman modes, including the A and B, disappear, suggesting the suppression of Peierls distortion. It is shown in our XRD study that all the structure transition from M′1 phase into the high pressure X phase is completed at about 59 GPa, thus, we can guess that the Peierls distortion is suppressed in X phase and the metallization in X phase is mainly related to structure transition.
This journal is © The Royal Society of Chemistry 2016 |