Magnesium Doping for the Promotion of Rutile Phase Formation in the Pulsed Laser Deposition of TiO2 Thin Films

Akihiro Ishii1,*1, Itaru Oikawa1, Masaaki Imura2, Toshimasa Kanai2 and Hitoshi Takamura1,*2

1Department of Materials Science, Graduate School of Engineering, Tohoku University, Sendai 980–8579, Japan2Thin Films Division, Nippon Electric Glass Co., Ltd., Nagahama 529–0292, Japan

The preparation of a transparent and smooth rutile-type TiO2 thin �lm without the use of the crystallographical effect of the substrate is a challenge for the advanced utilization of TiO2 in the �elds of optics and solid state ionics. Because acceptor doping leads to the formation of oxygen vacancies, this method has promise as a new approach to promote the formation of rutile-type TiO2. Mg2+-doped TiO2 thin �lms were prepared by pulsed laser deposition, and the effects of Mg2+ doping on the phases present, the microstructure, the optical properties, and the surface roughness of the �lms were investigated. Particular attention was paid to the Mg2+ distribution in the prepared �lms. The formation of the rutile phase was promoted by 2.7 mol% and 5.5 mol%Mg2+ doping. The negligible segregation of Mg2+ and absence of change in the ex-tinction coef�cient by Mg2+ doping indicate that Mg2+ worked as the acceptor and induced oxygen vacancies for charge compensation, which promoted the formation of the rutile phase. Given that Mg2+ is a doubly charged acceptor, Mg2+ doping is a more effective method for promot-ing the formation of the rutile phase than trivalence doping. Besides the excellent optical properties (n ≈  3.03 and k <  0.02 at λ =  400 nm) of the 2.7%Mg2+-doped rutile-type TiO2 thin �lm deposited at 350°C, the �lms were smooth, with a roughness index of only approximately 0.8 nm. This method of preparing smooth rutile-type TiO2 thin �lms has potential for the further development of TiO2-based resistive memory devices.　[doi:10.2320/matertrans.MB201704]

(Received August 2, 2017; Accepted November 10, 2017; Published December 8, 2017)

Keywords:　 TiO2 thin �lm, phase control, rutile, pulsed laser deposition

1.　 Introduction

TiO2 has long been studied since it can be used in a wide variety of applications: in white pigments1), high-refrac-tive-index coatings2) and resistive switches3,4), in the anodes of Li-ion batteries5), in the electrolytes in the proton ex-change membrane fuel cell6) and in photocatalytic pro-cesses7). The functionality of TiO2 improves when its phases are controlled. A typical example is the higher photocata-lytic activity of metastable anatase-type TiO2

8,9). The anode performance of this particular form of TiO2 in Li-ion batter-ies has been shown to be superior to that of a thermodynam-ically stable rutile phase5,10). While the determination of the optimum phase should be done for all applications of TiO2, particularly for resistive memory elements or surface proton conductors, little attention has been paid to rutile-type TiO2. This is presumably due to a combination of the necessity of thin TiO2 �lms for reducing electrical resistance and the rar-ity of rutile-type TiO2 crystals in TiO2 thin �lms11–14). In or-der to investigate the optimum phase in applications to be-come standard practice, and for the rutile phase to be considered, an evolution in the preparation technique of ru-tile-type TiO2 thin �lms is essential.

In contrast to the �eld of the solid state ionics, rutile-type TiO2 has been the focus of signi�cant attention in the �eld of optics due to its transparency and a higher refractive in-dex (n) than that of the anatase phase (nRutile  ≈  2.75, nAnatase ≈  2.54 at λ =  550 nm15)). This higher-n of transparent rutile-type TiO2 is expected to improve the controllability of the re�ection spectrum for optical coatings16). In the optics �eld, therefore, there is a clear demand for better methods for the preparation of rutile-type TiO2 thin �lms. As men-

tioned above, TiO2 thin �lms are generally employ ana-tase-type TiO2. This is largely because the rutile phase has higher surface energy than the anatase phase17). Three differ-ent preparation techniques have been reported for preparing rutile-type TiO2 thin �lms: 1) deposition at high temperature around 700°C11,18), 2) deposition on single crystal sub-strates19,20), and 3) the introduction of oxygen vacancies, which promotes the formation of the rutile phase21,22) based on the following equation23–25):

OXO → V••O + 2e− +

12

O2 (1)

However, each of these three techniques is associated with serious �aws. The high temperature required for the �rst technique gives a rough surface, which results in signi�cant optical loss11,16), the cost of single crystal substrates is im-practically high, and the introduction of the oxygen vacan-cies through eq. (1) results in a decrease in the transparency of the TiO2 thin �lm due to the excess of conduction elec-trons11,26). Preparation of transparent rutile-type TiO2 thin �lms at low temperature without the use of the crystallo-graphical information of the substrate has proved to be challenging.

In a new approach to the low-temperature preparation of transparent rutile-type TiO2 thin �lms, acceptor doping in the TiO2 thin �lms has been investigated by our group. This approach was adopted because the acceptor Mx+ (1 ≤  x ≤  3) can lead to the formation of oxygen vacancies without the introduction of excess electrons according to the following27):

M2O3 → 2MTi + V••O + 3OO (when x = 3 in Mx+) (2)

It has been reported that doping with 10 mol% Al3+, a sin-gle-phase rutile-type TiO2 thin �lm with a high n ≈  3.05 (at λ =  400 nm) and a low extinction coef�cient (k, ≈0.01) was

*1 Graduate Student, Tohoku University*2 Corresponding author, E-mail: takamura@material.tohoku.ac.jp

Materials Transactions, Vol. 59, No. 1 (2018) pp. 33 to 38 Special Issue on Recent Advances in Solid State Ionics and Its Applications ©2017 The Japan Institute of Metals and Materials

successfully prepared on glass by pulsed laser deposition at 350°C. This single-phase rutile-type TiO2 thin �lm showed negligible surface roughness (≈0.8 nm) and optical loss was insigni�cant16). This suggests that acceptor doping in the TiO2 thin �lms makes it possible to produce optical-coat-ing-ready high-n rutile-type TiO2 thin �lm at low tempera-ture without the use of the crystallographical effect of the substrate. In the case of rutile-type TiO2 thin �lm prepared by the 10 mol%Al3+ doping, this trivalent substitution would have resulted in a decline in the n value due to the decrease in the Ti4+ content (nAl2O3 nTiO2)

16). By decreasing the concentration of the doping element, it is expected that the decline in the n value will be minimized.

In this investigation, the focus is on Mg2+ doping as an al-ternative to Al3+ doping in the production of rutile-type TiO2 thin �lms. Because MgTi is a doubly charged acceptor, the appropriate dopant concentration of Mg2+ for the preparation of the rutile-type TiO2 thin �lm can likely be reduced by half that required for Al3+ doping. Even though the TiO2–MgO phase diagram indicates the solubility of Mg2+ into TiO2 is very limited28), solubility can be maximized using the pulsed laser deposition (PLD) method for the preparation of TiO2 thin �lms containing Mg2+ as the acceptor. The PLD method involves a rapid cooling process, and therefore the thermal equilibration would be insuf�cient for the segregation of Mg2+ to take place. This study of the TiO2 thin �lm prepared by PLD investigates the effects of Mg2+ doping on the phases, the microstructure, and in particular of the distribu-tion of Mg2+. The optical properties and the surface rough-ness were also investigated. The phase control technique for preparing TiO2 thin �lms revealed by this study opens a new direction in the development of rutile-type TiO2-based appli-cations in the �eld of solid state ionics.

2.　 Experimental Procedure

Undoped, 2.7 and 5.5 mol%Mg2+-doped TiO2 targets were prepared by a solid-state reaction of rutile-type TiO2 powder (Kojundo Chemical Laboratory Co., Ltd: purity 99.9%) and MgO powder (Kojundo Chemical Laboratory Co., Ltd: purity 99.9%). The sintering condition for the un-doped-TiO2 target was at 1300°C for 10 h in air, and those for the Mg2+-doped TiO2 targets were at 1500°C for 36 h in air. The PLD process was conducted using a KrF excimer la-ser (COMPexPro205, COHERENT, Inc: λ =  248 nm) and a highly vacuumed deposition chamber (≈10−5 Pa) (PLAD-242, AOV Co., Ltd). The laser was emitted at a frequency of 5 Hz at a �uence of 5.8 J·cm−2 for 30 min. The substrate was an alkali-free glass (OA-10G, Nippon Electric Glass Co., Ltd: 15 mm ×  15 mm ×  0.7 mm) cleaned by using a sonica-tor with acetone. The deposition temperature was controlled by heating the substrates with an infrared lamp heater, and a Si plate was mounted on the backside of the substrates to transmit the heat. The distance between the substrate and the targets was kept at 50 mm in all cases. The deposition atmo-sphere was controlled using oxygen gas (purity 6N) at 0.5 Pa, which has been shown to be suitable for the prepara-tion of the transparent rutile-type TiO2 thin �lms16). The thickness of the TiO2-based thin �lms was controlled at ap-proximately 100 to 150 nm under these conditions.

The phases present in the TiO2-based targets was clari�ed by θ–2θ X-ray diffraction method (XRD, D8 Advance, Bruker AXS) and that for the thin �lms was clari�ed by grazing-incidence (α  =  2°) XRD. To examine the phases present of the TiO2-based thin �lms in detail, in-situ high-temperature XRD (HT-XRD) was also carried out in air. The X-ray source was Cu-Kα radiation (λ =  1.5418 Å). Micro-Raman spectroscopy (HR-800, Horiba Jobin Yvon S.A.S.) was also conducted to support the phase identi�ca-tion. The Raman spectra were generated using a He-Ne laser (λ  =  632.8 nm). The microstructure was observed by �eld emission scanning electron microscope (FE-SEM, JSM-7800F, JEOL Ltd) and transmission electron microscope (TEM)–aberration-corrected annular dark-�eld scanning TEM (ADF-STEM) combined apparatus (JEM-ARM200F, JEOL Ltd) with an energy dispersive spectrometer (EDS). The surface morphology was observed by atomic force mi-croscope (AFM, JSPM- 5200, JEOL Ltd).

The optical properties were clari�ed by spectroscopic el-lipsometry (M-2000, J. A. Woollam Co., Ltd). This measure-ment included obtaining an ellipsometric parameter (Φ, Δ) with incident angles from 50° to 70° and obtaining transmit-tance spectrum for a wavelength range of 250–1000 nm. Using these two data and by taking the back-side re�ection of the substrate into account, the wavelength dispersion of n and k values was determined. In this study, the values of n and k at λ =  400 nm are discussed as typical values to ensure the transparency of the �lm.

3.　 Results and Discussions

To clarify the effect of Mg2+ doping on the phases present in the TiO2 thin �lms, the Mg2+ concentration dependence on the XRD patterns was investigated. Figure 1 shows the XRD patterns of the undoped, 2.7 mol% and 5.5 mol% Mg2+-doped TiO2 targets (Fig. 1(a)) and thin �lms deposited at 350°C (Fig. 1(b)). In Fig. 1(a), XRD peaks attributed to the rutile-type TiO2 were observed in the undoped TiO2 tar-get. In addition, MgTi2O5 was observed in the Mg2+-doped TiO2 target. These phases present of the targets were consis-tent with the TiO2-MgO phase diagram28). Meanwhile, as shown in Fig. 1(b), the XRD patterns of the Mg2+-doped TiO2 thin �lms showed the peaks only from the rutile phase. It should be also noted that a single anatase phase was clearly con�rmed in the undoped-TiO2 thin �lm. This sug-gests that the dissolution of Mg2+ into TiO2 matrix occurs when using the PLD process and this dissolution promotes the formation of the rutile phase, as is the case with Al3+ doping16). For the XRD peaks of the Mg2+-doped TiO2 thin �lms, i.e. (110) at 27.4° and (200) at 39.2°, no clear shift is observed in comparison with those of the undoped ru-tile-type TiO2 thin �lm11). This is attributed to the low Mg2+ concentration and the similarity in the ionic radii between Ti4+ and Mg2+ (rTi4+ = 68 pm and rMg2+ = 65 pm29)).

To con�rm the formation of the rutile phase in the Mg2+-doped TiO2 thin �lms, HT-XRD and Raman analyses were carried out. Figures 2(a) and (b) show the HT-XRD patterns of the 2.7 mol% and 5.5 mol% Mg2+-doped TiO2 thin �lms, respectively. The growth of the rutile phase can be recog-nized at around 550°C. Meanwhile, Ishii et al. have reported

34 A. Ishii, I. Oikawa, M. Imura, T. Kanai and H. Takamura

that the rutile phase does not emerge up to 600°C in the case of amorphous and the anatase-type transparent TiO2 thin �lms11). Figure 2(c) shows the Raman spectra of the un-doped, 2.7 mol% and 5.5 mol% Mg2+-doped TiO2 thin �lms. For the undoped one, a sharp peak at 144 cm−1, which is at-tributed to an Eg mode of the anatase crystal30), is observed with weak peaks attributed to a brookite phase. Meanwhile, for the Mg2+-doped TiO2 thin �lms, two peaks at 445 cm−1 and 608 cm−1 were emerged, which are attributed to the Eg and A1g modes of the rutile crystal31), respectively. This re-sult also supports the formation of single rutile phase in the Mg2+-doped TiO2 thin �lms.

To clarify the phases present and to determine the exact mechanism of the formation of the rutile phase in the Mg2+-doped TiO2 thin �lms, microstructural observation using an electron microscope was carried out. Figure 3 shows the mi-crostructure of the 2.7 mol%Mg2+-doped TiO2 thin �lm de-posited at 350°C. While the weak contrast in the in-plane ADF-STEM image in Fig. 3 (a) may be attributed to differ-

ences in composition, no apparent segregation of the Mg el-ement was observed in the EDS analysis, as can be seen in Fig. 3 (b). This indicates that most of the Mg2+ was incorpo-rated into the TiO2 matrix as the acceptor dopant, which then led to the formation of the rutile phase. Though a phase sep-aration was observed in the case of the 10 mol%Al3+ dop-ing16), almost all of the Mg2+ apparently dissolved presum-ably because the dopant concentration was smaller in the case of Mg2+ doping, and the ionic radius of Mg2+ is closer to that of Ti4+ (rMg2+  =  65 pm, rTi4+  =  68 pm, rAl3+  =  50 pm 29)). In Fig. 3 (c), it can be seen that the crystal grains of the Mg2+-doped TiO2 thin �lm are �ner than those of un-doped-TiO2 thin �lms at ranges of 30–140 nm and 70–200 nm, respectively. This is most likely due to the pinning effect at the grain boundaries by the nanoparticles indicated by arrows in Fig. 3 (d). These nanoparticles are possibly composed of the Mg-rich phase (e.g. MgO, MgTi2O5); how-ever, no reliable compositional difference between the nanoparticles and the matrix was observed. It appears that

Fig. 1　XRD patterns of the undoped, 2.7 mol% and 5.5 mol% Mg2+-doped TiO2 (a) targets and (b) thin �lms deposited at 350°C.

Fig. 2　HT-XRD patterns of (a) 2.7 mol% and (b) 5.5 mol% Mg2+-doped TiO2 thin �lms deposited at 350°C. (c) Raman spectra of the undoped, 2.7 mol% and 5.5 mol% Mg2+-doped TiO2 thin �lms.

35Mg Doping for the Promotion of Rutile Phase Formation in TiO2 Films

most of the Mg2+ was incorporated into the TiO2 matrix as the acceptor dopant, resulting in the promotion of the ru-tile-phase formation in the TiO2 thin �lm.

To further discuss the origin and role of oxygen vacancy, the extinction coef�cient of the 2.7 mol% Mg2+-doped TiO2 thin �lm deposited under P(O2) of 0.5 Pa was measured. Figure 4(a) shows the extinction coef�cient of the 2.7 mol% Mg2+-doped TiO2 thin �lm as deposited and after annealing

at 600°C in air. The as-deposited 2.7 mol% Mg2+-doped TiO2 thin �lm shows a small extinction coef�cient due to oxygen vacancies compensated by electrons, while the �lm after annealing shows a negligibly small extinction coef�-cient in the same wavelength range. This indicates that 1) excess oxygen vacancies are introduced through eq. (1) during deposition under P(O2) of 0.5 Pa, and 2) the excess oxygen vacancies can be oxidized and removed by anneal-

Fig. 3　Microstructure of the 2.7 mol% Mg2+-doped TiO2 thin �lms deposited at 350°C observed by (a) ADF-STEM and (b) its EDS mapping, (c) FE-SEM compared with the undoped-TiO2 thin �lms deposited at 350°C, and (d) TEM.

Fig. 4　(a) Extinction coef�cient of 2.7 mol% Mg2+-doped TiO2 thin �lm deposited at 350°C before/after the annealing up to 600°C. The inset is enlarged view. (b) XRD patterns of undoped-TiO2 thin �lms deposited at 400°C under vacuum and 0.5–7 Pa of oxygen partial pressure. Adapted from Ref. 16) with permission.

36 A. Ishii, I. Oikawa, M. Imura, T. Kanai and H. Takamura

ing under air. The excess oxygen vacancies, however, do not seem to affect the rutile phase formation. Figure 4(b) shows XRD patterns of undoped TiO2 thin �lms deposited at 400°C under various oxygen partial pressure reported by Ishii et al.16) Without the acceptor dopants, the anatase phase was formed at P(O2) of 0.5 Pa. In addition, given that 1) most Mg2+ is dissolved into the TiO2 matrix (Fig. 3), 2) no clear XRD peak shift suggesting interstitial-type defects is ob-served (Fig. 1(b)), and 3) Mg2+ is not compensated by elec-tron hole (as shown later in Fig. 5), the defect equilibrium of

[MgTi] = [V••O ] is most likely as expected, even though fur-

ther study is required for clarifying the defect chemistry in detail.

The refractive index of the TiO2 thin �lms was expected to rise in TiO2 thin �lms with a rutile phase formed by Mg2+ doping. Figure 5 shows the optical properties of the un-doped, 2.7 mol% Mg2+-doped and previously reported 10 mol% Al3+-doped TiO2 thin �lms16) as a function of tem-perature. The 2.7 mol% Mg2+-doped TiO2 thin �lms showed higher n values than the undoped-TiO2 thin �lms, and at 350°C, its high n value of 3.03 is comparable to that of the 10 mol% Al3+-doped TiO2 thin �lms. This is because the formation of the rutile phase was promoted by Mg2+ doping in the same manner as that which occurs with Al3+ doping. The k values of the Mg2+-doped samples are close to zero (less than 0.02) and independent of temperature, which indi-cates that the charge compensation of MgTi is not carried out by h• but by V••O . Note that at 300°C, the Mg2+-doped sample has a higher n value than the Al3+-doped sample. Given that the phases in both samples are identical and the difference of atomic polarizability between Mg2+ and Al3+ is quite small (nMgO ≈  1.7632), nAl2O3

  ≈  1.7933) at λ =  400 nm), the higher n value of the Mg2+-doped sample can likely be attributed to the smaller dopant concentration. This indicates that ru-tile-type TiO2 thin �lms prepared by Mg2+ doping can be ex-pected to superior to those prepared by Al3+ doping in that they will have a higher n value.

The surface of the rutile-type TiO2 thin �lm should be smooth to avoid diffuse re�ection. Figure 6 shows an AFM image of the 2.7 mol% Mg2+-doped TiO2 thin �lms depos-

ited at 350°C. The morphology is smooth, which implies ho-mogeneous formation of the rutile phase16). The arithmetic mean roughness calculated using Fig. 6 is approximately 0.8 nm. This roughness value is much lower than that of the rutile-type TiO2 thin �lm prepared without acceptor doping (≈7 nm11)), and is the same as 10 mol% Al3+-doped ru-tile-type TiO2 thin �lms which suffer negligible optical loss due to diffuse re�ection16).

Based on these results, it can be said that Mg2+ doping has a similar effect to that of Al3+ doping on TiO2 thin �lms pre-pared by PLD. The formation of the rutile phase is pro-moted, a higher n value is achieved even at 350°C, no change in the k value occurs, and the surface is smooth. The 2.7%Mg2+-doped rutile-type TiO2 thin �lm deposited at 350°C showed not only excellent optical properties (n ≈  3.03 and k <  0.02 at λ =  400 nm) but also a smooth surface at ap-proximately 0.8 nm, which is comparable to that of the 10 mol%Al3+-doped rutile-type TiO2 thin �lm deposited at 350°C16). The advantage of Mg2+ doping over Al3+ doping is its effectiveness: the doubly charged acceptor means that the required concentration of the dopant Mg2+ is half that of the dopant Al3+, which minimizes the reduction in the refractive index.

These results suggest that smooth rutile-type TiO2 thin �lms prepared by acceptor doping will contribute to the de-velopment of TiO2-based resistive memory devices. It is well-known that the resistive switching of TiO2 is based on a reversible phase transformation between amorphous-/ana-tase- type TiO2 and the magnéli phase TinO2n−1

34–36), and that the broad dispersion between the resistive switching pa-rameters is problematic, particularly with regard to the set voltage for each set/reset cycles. However, if rutile-type TiO2 thin �lms are applied into the resistive switching ele-ment, a signi�cant reduction in dispersion can be expected since the rutile phase is more thermodynamically stable than the anatase phase and is structurally similar to the magnéli phase37). Furthermore, as reported by Liu et al., the accepter in the TiO2 modi�es the distribution of the oxygen vacancies and enhances the formation and rupture effect of the con-ducting �laments, resulting in uniform set voltage for each of the set/reset cycles38). The resistance switching behavior of acceptor-doped rutile-type TiO2 thin �lms is a topic for

Fig. 6　AFM image of the 2.7 mol% Mg2+-doped TiO2 thin �lms deposited at 350°C.

Fig. 5　Optical properties (at λ =  400 nm) of the undoped, 2.7 mol% Mg2+-doped and 10 mol% Al3+-doped TiO2 thin �lms16) as a function of temperature.

37Mg Doping for the Promotion of Rutile Phase Formation in TiO2 Films

further investigation.

4.　 Conclusion

The effects of Mg2+ doping in the preparation of TiO2 thin �lms were investigated. The phases present, and the micro-structure, with a focus on the Mg2+ distribution were investi-gated, and the optical properties and surface roughness of these �lms were also determined. The formation of the rutile phase was promoted by the 2.7 mol% and 5.5 mol%Mg2+ doping. Negligible segregation of Mg2+ occurred and no no-table change in the extinction coef�cient due to Mg2+ doping was noted, suggesting that Mg2+ worked as the acceptor and induced oxygen vacancies as charge compensation, which resulted in the promotion of the rutile phase formation. The 2.7 mol%Mg2+-doped rutile-type TiO2 thin �lm deposited at 350°C was shown to meet all the critical requirements: high n ≈  3.03, low k <  0.02 (at λ =  400 nm) and smooth rough-ness ≈0.8 nm. These values are comparable to those of the optical-coating-ready 10 mol%Al3+-doped rutile-type TiO2 thin �lm16). Mg2+ doping more effectively promotes the for-mation of the rutile phase than Al3+ doping since Mg2+ is a doubly charged acceptor. This acceptor-doped rutile-type TiO2 thin �lm is expected to contribute to developments in superior optical coatings and resistive memory.

Acknowledgments

We would like to thank Dr. K. Kobayashi for taking the SEM, TEM and STEM images. AI would appreciate the �-nancial support from the Grant-in-Aid for JSPS Research Fellow and the Interdepartmental Doctoral Degree Program for Multi-dimensional Materials Science Leaders in Tohoku University. HT would also like to acknowledge the �nancial support provided by JSPS (26249103).

REFERENCES

1) J.H. Braun, A. Baidins and R.E. Marganski: Prog. Org. Coat. 20 (1992) 105–138.

2) H.A. Tanner and L.B. Lockhart: J. Opt. Soc. Am. 36 (1946) 701–706. 3) T.W. Hickmott: J. Appl. Phys. 33 (1962) 2669–2682. 4) F. Argall: Solid-State Electron. 11 (1968) 535–541. 5) T. Ohzuku, Z. Takehara and S. Yoshizawa: Electrochim. Acta 24

(1979) 219–222. 6) H. Ekström, B. Wickman, M. Gustavsson, P. Hanarp, L. Eurenius, E.

Olsson and G. Lingbergh: Electrochim. Acta 52 (2007) 4239–4245. 7) A. Fujishima and K. Honda: Nature 238 (1972) 37–38. 8) S. Kato and F. Mashio: Kogyo Kagaku Zasshi (J.Chem. Soc. Jpn.) 67

(1964) 1136–1140. 9) A. Fujishima, X. Zhang and D.A. Tryk: Surf. Sci. Rep. 63 (2008)

515–582. 10) F. Bonino, L. Busani, M. Lazzari, M. Manstretta and B. Rivolta: J.

Power Sources 6 (1981) 261–270. 11) A. Ishii, Y. Nakamura, I. Oikawa, A. Kamegawa and H. Takamura:

Appl. Surf. Sci. 347 (2015) 528–534. 12) H.-A. Durand, J.-H. Brimaud, O. Hellman, H. Shibata, S. Sakuragi, Y.

Makita, D. Gesbert and P. Meyrueis: Appl. Surf. Sci. 86 (1995) 122–127.

13) Z. Wang, U. Helmersson and P.-O. Käll: Thin Solid Films 405 (2002) 50–54.

14) M. Mazur, D. Wojcieszak, J. Domaradzki, D. Kaczmarek, S. Song and F. Placido: Opto-Electron. Rev. 21 (2013) 233–238.

15) C.-C. Ting, S.-Y. Chen and D.-M. Liu: J. Appl. Phys. 88 (2000) 4628–4633.

16) A. Ishii, K. Kobayashi, I. Oikawa, A. Kamegawa, M. Imura, T. Kanai and H. Takamura: Appl. Surf. Sci. 412 (2017) 223–229.

17) P.K. Naicker, P.T. Cummings, H. Zhang and J.F. Ban�eld: J. Phys. Chem. B 109 (2005) 15243–15249.

18) H.-S. Yoon and H. Im: Bull. Korean Chem. Soc. 18 (1997) 640–643. 19) K.H.G. Ashbee and R.E. Smallman: Philos. Mag. 7 (1962)

1933–1940. 20) R.N. Ghoshtagore and A.J. Noreika: J. Electrochem. Soc. 117 (1970)

1310–1314. 21) K.N.P. Kumar, K. Keizer and A.-J. Burggraaf: J. Mater. Chem. 3

(1993) 1141–1149. 22) D. Wicaksana, A. Kobayashi and A. Kinbara: J. Vac. Sci. Technol. A

10 (1992) 1479–1482. 23) H. Long, G. Yang, A. Chen, Y. Li and P. Lu: Thin Solid Films 517

(2008) 745–749. 24) E. Le Boulbar, E. Millon, J. Mathias, C. Boulmer-Leborgne, M.

Nistor, F. Gherendi, N. Sbäi and J.B. Quoirin: Appl. Surf. Sci. 257 (2011) 5380–5383.

25) M. Kadoshima, M. Hiratani, Y. Shimamoto, K. Torii, H. Miki, S. Kimura and T. Nabatame: Thin Solid Films 424 (2003) 224–228.

26) G. Hass: Vacuum 2 (1952) 331–345. 27) D.A. Hanaor and C. Sorrell: J. Mater. Sci. 46 (2011) 855–874. 28) P. Villars (Chief Editor): PAULING FILE in Inorganic Solid Phases,

(SpringerMaterials (online database), Springer-Verlag GmbH, Heidelberg, 2016).

29) L. Pauling: The Nature of the Chemical Bond, (Cornell University Press, New York, 1972).

30) T. Ohsaka, F. Izumi and Y. Fujiki: J. Raman Spectrosc. 7 (1978) 321–324.

31) R.J. Betsch, H.L. Park and W.B. White: Mater. Res. Bull. 26 (1991) 613–622.

32) R.E. Stephens and I.H. Malitson: J. Res. Natl. Bur. Stand. 49 (1952) 249–252.

33) I.H. Malitson, F.V. Murphy and W.S. Rodney: J. Opt. Soc. Am. 48 (1958) 72–73.

34) D.H. Kwon, K.M. Kim, J.H. Jang, J.M. Jeon, M.H. Lee, G.H. Kim, X.S. Li, G.S. Park, B. Lee, S. Han, M. Kim and C.S. Hwang: Nat. Nanotechnol. 5 (2010) 148–153.

35) S.J. Park, J.P. Lee, J.S. Jang, H. Rhu, H. Yu, B.Y. You, C.S. Kim, K.J. Kim, Y.J. Cho, S. Baik and W. Lee: Nanotechnology 24 (2013) 295202.

36) K. Szot, M. Rogala, W. Speier, Z. Klusek, A. Besmehn and R. Waser: Nanotechnology 22 (2011) 254001.

37) S. Tominaka, Y. Tsujimoto, Y. Matsushita and K. Yamaura: Angew. Chem. Int. Ed. 50 (2011) 7418–7421.

38) L.F. Liu, J.F. Kang, H. Tang, N. Xu, X. Sun, C. Chen, B. Sun, Y. Wang, X.Y. Liu, X. Zhang and R. Han: Jpn. J. Appl. Phys. 47 (2008) 2701–2703.

38 A. Ishii, I. Oikawa, M. Imura, T. Kanai and H. Takamura