Identification of surface oxygen vacancy-related phonon–plasmon coupling in TiO2 single crystal
Guo Jun-Hong1, Li Ting-Hui2, Hu Fang-Ren1, †, , Liu Li-Zhe3, ‡,
School of Optoelectronic Engineering and Grüenberg Research Center, Nanjing University of Posts and Telecommunicates, Nanjing 210023, China
College of Electronic Engineering, Guangxi Normal University, Guilin 541004, China
Nanjing National Laboratory of Microstructures and Department of Physics, Nanjing University, Nanjing 210093, China

 

† Corresponding author. E-mail: hufr@njupt.edu.cn

‡ Corresponding author. E-mail: lzliu@nju.edu.cn

Project supported by the National Natural Science Foundation of China (Grant Nos. 61574080, 11404162, 61505085, and 61264008) and the Natural Science Foundation of Jiangsu Province, China (Grant No. BK20130549).

Abstract
Abstract

Oxygen vacancies (OVs) play a critical role in the physical properties and applications of titanium dioxide nanostructures, which are widely used in electrochemistry and photo catalysis nowadays. In this work, OVs were artificially introduced in the surface of a pure TiO2 single crystal by pulsed laser irradiation. Raman spectra showed that the intensity of Eg mode was enhanced. Theoretical calculations disclose that this was caused by the strong coupling effect between the phonon vibration and plasmon induced by the OVs-related surface deformation, and good agreement was achieved between the experiments and theory.

1. Introduction

Metal oxides have played important roles in scientific research and industrial production, because of their potential use in optical devices, catalysts, batteries, gas-sensors, and biosensors.[15] Among the different metal oxides, titanium dioxide (TiO2) has attracted much attention due to its interesting applications such as water splitting, photo catalysis and gas sensing.[610] In various technological applications, nanosized materials have yielded excellent performance because of their small dimensions, i.e., the large surface-to-volume ratio. Nanoscale titanium dioxide such as nanotubes (NTs), nanocrystals (NCs), nanodisks, and nanowires have been fabricated and their optical and electronic characteristics and potential applications have been extensively studied.[1114] Among these, many novel physical characteristics are simply ascribed to its oxygen vacancies, but cannot be confirmed with evidence.[1518] Because the nanostructure is very complex, its intrinsic physical mechanism about OVs is still not clarified. Therefore, how to directly characterize and analyze the existing states of OV has become a crucial problem.

In order to clearly investigate the phonon characteristics of metal oxide nanomaterials with different OV types, we introduced OVs in the surface of a pure TiO2 single crystal artificially by pulsed laser irradiation. Spectral analysis and theoretical calculations showed that strong coupling effect existed between the phonon relating to surface OVs and the plasmon induced by the surface deformation potential. This effect caused the intensity of Raman active mode Eg to increase while A1g mode decreased. This work improves our understanding of the OVs properties in TiO2 and is beneficial to further applications.

2. Experiment

A 248-mm laser beam with a 1-Hz repetition rate was selected to irradiate the TiO2 single crystal (named sample A) via a 90° prism. The laser beam was a Gauss distribution of linear polarized light with the power of 451 mJ/pulse and the irradiation time was 5 s. A 20-times microscope objective was added to converge the laser to TiO2 single crystal and also to enhance the laser intensity. After irradiation, the sample A was annealed in O2 at 1000 °C for 5 h to prepare sample C. The Raman spectra of irradiated crystal (sample A), non-irradiated crystal (sample B), and the annealed crystal (sample C) were acquired on a T64000 triple Raman system at backscattering geometries using the 514.5-nm line of an Ar–ion laser as the excitation source. The diameter and power of the beam spot were 3 mm and 4.6 mW, respectively. The resolution of the spectrometer is 0.5 cm−1. X-ray photoelectron spectroscopy (XPS) was performed on the PHI 5000 Versa Probe.

3. Results and discussion

Figure 1(a) provides the corrected XPS spectra of Ti 2p from samples A and B. It was clear to see the 2p3/2 peak at 458.9 eV and 2p1/2 peak at 464.6 eV, and the Ti 2p peaks in all samples had only one component. Figure 1(b) is a larger image of the binding energies of Ti 2p3/2 in Fig. 1(a). The binding energy of sample A had a deviation of 0.2 eV compared to sample B, which resulted in the existence of OVs that changed the charge distribution around titanium atoms.[19] Based on the method introduced to calculate the OVs in SnO2,[20,21] and according to the relationship between the binding energies of Ti 2p3/2 level versus the valences of Ti from pure TiO2−x, the OV in sample A was 0.77%.

Fig. 1. (a) XPS spectra from samples A and B, and (b) larger image of the Ti 2p3/2 spectra, (c) PL spectra from samples A and B, (d) EPR spectra of sample A. The unit 1 Gs = 10−4 T.

To clarify the existence of OVs, PL spectra were acquired from the two samples and shown in Fig. 1(c). The PL features composed of a double-peak structure could be observed and the energy separation between two subpeaks increases after irradiation. Since a large number of OVs will exist after irradiation, both the adsorbed oxygen atoms and OVs on the surface of crystals may be responsible for the double-peak PL. The difference in the applied stress caused by OVs on the inner and outer surfaces of crystals increases and it causes the increased splitting and shifting of the PL peak. To further investigate OVs existed on the surface of crystals, the EPR was shown in Fig. 1(d). The Landeg value was obtained to be 2.002 which has good consistency with previous reports.[22] The above experiments further confirmed the existence of OVs on the surface of TiO2 crystals. However, the type of OVs in sample A was still not clear. In order to investigate the type of OVs produced by irradiation, Raman spectra were acquired from samples A and B and the corresponding results were shown in Fig. 2(a). The peaks at 235, 449, 611 cm−1 correspond to the well-known multi-phonon process, Eg and A1g optical modes of the TiO2 crystals.[23] However, the intensity of Eg mode had enhanced after irradiation. In addition, the Raman spectrum of annealed sample A was also shown in Fig. 2(a). Comparing the Raman spectra of sample, A and the annealing sample A, the intensity of Eg mode decreased obviously after annealing. To further investigate the intensity changes of vibration modes, we examined the temperature-dependent Raman spectra of the irradiated sample (sample A) and the results were displayed in Fig. 2(b). With decreasing measurement temperature from 300 K to 77 K, the location and peak shape of the Raman mode were relatively unchanged, but the intensity underwent an obvious increase. However, it is generally assumed that the effects of thermal expansion and a harmonic coupling to other phonons during the temperature increasing can cause the intensity reduction and simultaneous red-shift of the Raman spectra,[24,25] which is obviously contrary to our results. So this anomalous behavior needs to be clarified.

Fig. 2. (a) Raman spectra of samples A and B at room temperature and Raman spectra of annealed sample A, (b) temperature-dependent Raman spectra of sample A.

Density functional theory (DFT) calculations were conducted to theoretically confirm the anomalous behavior of Raman spectra. The optimized primitive TiO2 with a = 4.59 Å, b = 2.96 Å lattice constant was shown in Figs. 3(a) and 3(b). The optimized structure lead to two typical Raman modes Eg (449 cm−1) and A1g (611 cm−1), as shown in Figs. 3(c) and 3(d), respectively. We can infer that those Raman modes with different vibration direction can exhibit different Raman features, which also strongly depended on its surficial structure deformation.

Fig. 3. (a) Optimized structure of TiO2 cell used in the DFT calculations, (b) optimized structure of TiO2 along the direction of (001), (c) vibration of Eg Raman mode, and (d) vibration of A1g Raman mode.

Our calculations were performed under the framework of DFT as implemented in the CASTEP package. Electron–ion interactions were described by the projector augmented plane wave method, and the wave function was expanded in a plane wave basis set with an energy cutoff of 500 eV. The k points in the Brillouin zone were sampled on a 5 × 5 × 8 mesh. The norm-conserving pseudo-potential method was chosen together with the gradient correction and the Perdew–Burke–Ernzerhof potential function.[26] Finally, the optimized geometrical structure was employed for construction and the diagonalized Hessian matrix:

where were the matrix force constants related to bond length and bond angle. The vibrational frequencies were obtained as the square roots of the phonon wave-vector q = 0. The ratio of atom mass and bond length of Ti and O atoms were respectively. The optimized cell eigenfrequency can be simply written as:[27]

where mTi(O) was Ti(O) atom mass and kTi(O) is the corresponding parameters determined by Ti–O bond length and force constants. From the vibration of Eg and A1g modes in Figs. 3(c)3(d), we can see that its Raman mode behavior strongly depended on their symmetrical structure change. The TiO2 structures were characterized by two different types along (001) direction, as shown in Fig. 3(b).

In Fig. 4(a), we showed the calculated Raman spectra of TiO2 at different temperature. According to the calculated Raman spectra, the intensity of the Eg and A1g modes both increased when temperature decreased. But the intensity changes of both modes were basically the same, which was quite different from the experiment results shown in Fig. 2(b).

Fig. 4. (a) The Raman spectra of TiO2 surface calculated by DFT, (b) the Raman spectra of TiO2 surface calculated by phonon–plasmon coupling mechanism.

By considering the deformation potential in the surface layer caused by the phonon–plasmon coupling mechanism, the Raman spectrum could be calculated by[28]

where the Faust–Henry coefficient CFH had been modified into

which depends on the Eg and A1g mode frequencies ωE and ωA and the high-frequency dielectric constant εm of the surface layer , and the total dielectric function ε(ω) was considered to be due to contributions from phonons and plasmons:

Here ωp was the plasma frequency, χph was the free-carrier susceptibility, χfc was the ionic susceptibility, and Γ and γ were the phonon and plasmon damping constants, respectively. In the calculation, ωE = 449 cm−1, ωA = 611 cm−1, ε = 7.95, ωp = 17.5 cm−1, and γ = 60 cm−1.[29] Γ = 28 cm−1 and C = 0.93 were adjustable parameters related to the deformation potential induced by the lattice imperfection near the surface and carrier changes. After considering the coupled phonon–plasmon effect, the measured Raman spectra of irradiated TiO2 can be fitted well, as shown in Fig. 4(b). Those experimental and theoretical results implied that the superficial plasmon induced by OVs played an important role in its Raman behaviors.

4. Conclusions

In conclusion, we have revealed that when OVs are introduced in the surface of TiO2, the phonon vibration of OVs and the plasmon induced by the surface deformation potential must be considered together. This leads to the large enhancement of the Eg mode intensity.

Reference
1Reddy B MKhan A 2005 Catal. Rev. Sci. Eng. 47 257
2Idota YKubota TMatsufuji AMaekawa YMiyasaka T 1997 Science 276 1395
3Brovelli SChiodini ALauria AMeinardi FPaleari A 2006 Phys. Rev. 73 073406
4Kucheyev S OBaumann T FSterne P AWang Y MBuuren THamza A VTerminello L JWilley T M 2005 Phys. Rev. 72 035404
5Fröschl THörmann UKubiak PKučerová GPfanzelt MWeiss C KBehm R JHüsing NKaiser ULandfesterd KWohlfahrt-Mehrens M 2012 Chem. Soc. Rev. 41 5313
6Sumita MHu C PTateyama Y 2010 J. Phys. Chem. 114 18529
7Kato HAsakura KKudo A 2003 J. Am. Chem. Soc. 125 3082
8Fujishima ARao T NTryk D A2000J. Photochem. Photobiol. C11
9Zeng WLiu TWang Z CTsukimoto S SSaito MIkuhara Y 2010 Mater. Trans. 51 171
10Linsebigler A LLu G QYates J T 1995 Chem. Rev. 95 735
11Jeon K SOh S DSun Y DYoshikawa H YMasuhara HYoon M 2009 Phys. Chem. Chem. Phys. 11 534
12Bavykin D VGordeev S NMoskalenko A VLapkin A AWalsh F C2008J. Phys. Chem. B1098565
13Wu J MShih H CWu W T 2006 Nanotechnology 17 105
14Abazović N DCǒmor M IDramicánin M DJovanović D JAhrenkiel S PNedeljković J M 2006 J. Phys. Chem. 110 25366
15Wang D ZWen S LChen JZhang S YLi FQ 1994 Phys. Rev. 49 14282
16Maki-Jaskari M ARantala T T 2002 Phys. Rev. 65 245428
17Verma RMantri BRamphalSrivastava A K2015Adv. Mater. Lett.6324
18Zhao Z HTian JSang YCabot ALiu H 2015 Adv. Mater. 27 2557
19Liu L ZWu X LLi T HShen J C 2015 Appl. Surf. Sci. 347 265
20Wang D NMiller A CNotis M R 1996 Surf. Interface Anal. 24 127
21Liang C HShimizu YSasaki TKoshizaki N 2003 J. Phys. Chem. 107 9220
22Li Y ZHwang D SLee N HKim S J 2005 Chem. Phys. Lett. 404 25
23Arsov L DKormann CPlieth W 1991 J. Raman Spectra 22 573
24Samanta KBhattacharya PKatiyar R S 2007 Phys. Rev. 75 035208
25Wu X LXiong S JSun L TShen J CChu P K 2009 Small 5 2823
26Liu L ZLi T HWu X LShen J CChu P K 2012 J. Raman Spectra 43 1423
27Liu L ZWu X LShen J CLi T HGao FChu P K 2010 Chem. Commun. 46 5539
28Katayama SMurase K 1977 J. Phys. Soc. Jpn. 42 886
29Mitev P DHermansson KMontanari BRefson K 2010 Phys. Rev. 81 134303