J. Phys. Chem. B 2001, 105, 9191-9195
9191
Surface Orientation of Main and Side Chains of Polyimide Alignment Layer Studied by Near-Edge X-ray Absorption Fine Structure Spectroscopy Takahiro Sakai, Ken Ishikawa, and Hideo Takezoe* Department of Organic and Polymeric Materials, Tokyo Institute of Technology, O-okayama, Meguro-ku, Tokyo 158-8552, Japan
Noritaka Matsuie, Yasushi Yamamoto, Hisao Ishii, and Yukio Ouchi Department of Chemistry, Graduate School of Science, Nagoya UniVersity, Furo-cho, Chikusa-ku, Nagoya 464-8602, Japan
Hiroshi Oji Department of Vacuum UV Photoscience, Institute for Molecular Science, Myodaiji, Okazaki 444-8585, Japan
Kazuhiko Seki Research Center for Materials Science, Nagoya UniVersity, Furo-cho, Chikusa-ku, Nagoya 464-8602, Japan ReceiVed: February 7, 2001; In Final Form: June 8, 2001
The average near-surface orientation of the main and side chains of polyimide (PI) for liquid crystal (LC) alignment has been studied by near-edge X-ray absorption fine structure (NEXAFS) spectroscopy. A detailed analysis of the angular dependences of resonance intensities revealed that rubbing causes the inclination of the main chain with respect to the surface, the angle of which decreases with increasing rubbing strength. This variation was found to agree with the decrease of the LC pretilt angle in the cell fabricated with this PI surface. The relation between the surface tilt of the main chain and the pretilt of LC is almost the same for PI films with side chains. Thus, it is concluded that the main chain alignment is strongly correlated with the pretilt angle even in the side-chain PI alignment layer.
1. Introduction The mechanism of the alignment of liquid crystal (LC) on rubbed polymer surfaces is one of the most important issues from both fundamental and industrial viewpoints. Generally, LC is aligned by rubbing a substrate surface coated with polyimide (PI). This treatment induces a so-called pretilt angle defined as the tilt angle of the liquid crystal with respect to the substrate surface. Since the performance of LC display is affected by the pretilt angle, control of the pretilt angle is very significant. It is well-known that the pretilt angle is strongly influenced by the surface of rubbed polymer. Accordingly, a great deal of effort has been made to study such surface alignment layers. However, there is extensive argument even for the mechanism of surface alignment without any success in clear-cut explanation. In the past, surface-sensitive studies have been carried out using optical second-harmonic generation (SHG) because SHG is a surface-specific phenomenon in nonpolar systems in the bulk. By using this technique, numerous experiments have been made not only for LC monolayers on polymer surfaces1-9 but also polymer alignment layers themselves.10-17 Actually the orientational distribution functions of the SHG-active side chain of polyimide films for LC alignment have been reported.10-14 These studies showed the preferential surface alignment of polyimide side chains along the rubbing direction. However, * Author to whom corresppondence should be addressed. Fax:+81-35734-2876. E-mail:
[email protected].
SHG activity is normally very low for the PI main-chain because of symmetry restriction. Therefore, other techniques are needed for studying main-chain orientation. Near-edge X-ray absorption fine structure (NEXAFS) spectroscopy is a powerful tool for studying the orientation of functional groups of organic molecules.18 The X-ray absorption leads to the emission of electrons, whose escape depth is sufficiently small, so that NEXAFS supplies a surface sensitive tool. The sampling depth of the present experiments can be estimated to be about 10 nm for the total electron yield (TEY) detection mode19 and about 2 nm for the partial electron yield (PEY) detection mode.20 NEXAFS spectroscopy also has polarization dependence for polarized incident X-ray due to the anisotropy of transition dipole moments. NEXAFS spectra of the second row elements (C,N,O,...) are dominated by resonances arising from transitions from the 1s core level to unoccupied molecular orbitals of π* and σ* symmetries. The direction of transition dipole moment depends on the direction of maximum amplitude of the final state orbital on the excited atom. For a carbon double bond, for example, the transition dipole moment of σ* transition is oriented parallel to the direction of the bond, and that of π* transition is oriented perpendicular to it. Thus, the analysis of the angular dependence of each resonance intensity in the spectra gives information of the orientation of molecules. In previous NEXAFS reports by the groups of Ouchi,21,22 Sto¨hr,19,23,24 and Wo¨ll,20 a great deal of effort has been made on the surface studies of various PIs. These studies revealed
10.1021/jp010482g CCC: $20.00 © 2001 American Chemical Society Published on Web 08/25/2001
9192 J. Phys. Chem. B, Vol. 105, No. 38, 2001
Figure 1. (a) Chemical structures of two polyimides with and without side chains; (CBDA-pPD)70(CBDA-CBAB)30 (CP7CC3) polyimide and CBDA-pPD (CP) polyimide. (b) Schematic illustration of a rubbing process. The substrate coated with a thin polyimide film moves underneath a rotating rubbing roller, the surface of which is coated with a cotton velvet cloth. (c) Geometry of the experiment. The sample coordinate system is chosen with the rubbing direction along the x axis and the sample normal along the z axis. The X-ray is incident on the sample at an incidence angle θ from the sample surface.
the preferred molecular orientation at the surface of rubbed polymer films, and concluded that this orientation is the microscopic origin for LC alignment on the surface. Recently, Sto¨hr and co-workers reported that the LC alignment is explained by a model based on maximum directional overlap of the anisotropic charge distributions at rubbed polymer surfaces and LC molecules.24 These studies have been made only on main-chain type PIs that do not contain side chains in the polymer structures. However, little has been investigated on side-chain type PIs in terms of the influence of the presence of side chains. Here, we report the surface orientation of both main chain and side chain under various rubbing conditions by means of NEXAFS measurements. From these results, we discuss the correlation between the surface orientation and the bulk pretilt angle in an LC cell in contact with the rubbed polyimide surfaces. 2. Experimental Section The side-chain type PI sample used was a random copolymer consisting of units with and without a side chain, copoly (1,2,3,4-cyclobutanetetracarboxydiimido-1,4-phenylene)(70) {1,2,3,4-cyclobutanetetracarboxydiimido-1,3-(4-cyanobiphenyl4′-oxy)phenylene}(30), which is referred to as CP7CC3, as shown in Figure 1a. As a reference sample, a main-chain type PI, poly(1,2,3,4-cyclobutanetetracarboxydiimido-1,4-phenylene) referred to as CP, was also used. These polyamic acids were dissolved in N-methyl-2-pyrrolidone (NMP) and spin coated onto 3 × 4 cm2 indium-tin-oxide-coated glass plates for NEXAFS measurements and 1.2 × 1.8 cm2 for pretilt angle measurements. The film thickness was less than 100 nm. They were prebaked at 80 °C for 5 min and baked at 250 °C for 1 h to be completely imidized. We rubbed the PI surface once using
Sakai et al. a commercial rubbing machine (RM-50, EHC Co. Ltd.) with a roller covered with cotton velvet as illustrated in Figure 1b under the following conditions: the translation speed of the sample was 3.6 mm/s; the rotation speed of the roller was 900 rpm; the pile impression was varied between 0.2 mm and 0.5 mm for controlling the rubbing strength. After the rubbing treatment, the glass plates were cut into 3 × 1 cm2 pieces for NEXAFS measurements. The bulk pretilt angle was measured by the crystal rotation method25,26 using antiparallel rubbing cells of 4-n-octyl-4′cyanobiphenyl (5CB) supplied by Merck Japan Ltd. in the nematic phase. The NEXAFS measurements were performed using linearly polarized soft X-ray at beam lines 7A and 11A of the Photon Factory of High Energy Accelerator Research Organization (KEK-PF) in Tsukuba. The spectra were recorded at the K-shell absorption edges of carbon (C) and nitrogen (N) with an energy resolution of better than 500 meV. The data were collected either by total electron yield (TEY) or partial electron yield (PEY) detection. The TEY was obtained by monitoring the drain current of the samples with a Keithley model-617 electrometer. We used a channeltron detector equipped with a retarding voltage grid to use the PEY mode, which probes the surface of only a few nanometers thick. The retarding voltage was -150 V for C K-edge and -250 V for N K-edge. To normalize the signal by the incident photon flux, the signal was divided by the photocurrent from a high-transmission (70%) grid coated in situ with gold. A pre-edge background was then subtracted from the divided spectra and the spectra were normalized to the absorption edge jump, which was arbitrarily set to unity, at far above the K-edge (∼325 eV for C K-edge, ∼425 eV for N K-edge). The latter procedure means that all spectra are normalized to the same number of C or N atoms in the sample, as discussed elsewhere.18 Figure 1c illustrates the experimental geometry. Here we define the sample coordinate system with the x axis along the rubbing direction and the z axis along the sample normal. The X-ray was incident in the x-z plane at an incidence angle θ measured from the surface of the sample. The measurements were performed using a linearly polarized soft X-ray, the electric field vector of which lies in the plane of incidence. The NEXAFS spectra were obtained at various X-ray incidence angles by rotating the sample mounted on a manipulator around the vertical axis to the plane. 3. Results and Discussions Figure 2 shows a NEXAFS spectrum of C K-shell absorption edge at an incidence angle of 90 degrees for an unrubbed CP7CC3 polyimide film using the PEY detection mode. The first three sharp peaks, labeled C1, C2, and C3 at 284.9, 286.3, and 287.7, respectively, are π* resonances corresponding to bonds associated with specific C atoms as indicated in Figure 2. The peaks C1 and C2 originate from the 1s-π* (CdC double bond) transition on C atoms in phenyl ring bonded to H and N, respectively.19,20,23,24 The peak C3 arises from the 1s-π* (CdO double bond) transition on the C atoms in the imide ring.19,20,23,24 Other broad peaks (C4, C5) at higher energies are associated with the 1s-σ* transition.19,20,23,24 Here, particular attention was paid to the absorption peaks at 286.3 eV (C2, π* resonance) and 295.9 eV (C4, σ* resonance). The transition dipole moment of C2 is oriented perpendicular to the phenyl ring, that is, perpendicular to the long axis of a main chain. On the other hand, that of C4 is oriented parallel to the phenyl ring. Angular dependences of these peak intensities for unrubbed and rubbed samples are shown in Figure 3, parts a and b. For
Main and Side Chains of Polyimide Alignment Layer
J. Phys. Chem. B, Vol. 105, No. 38, 2001 9193 TABLE 1: Surface Tilt Angles of Main Chain and Normalized Amplitude of CP7CC3 as a Function of Rubbing Strength (expressed by pile impression) Determined at the C K-shell Absorption Edge (Pretilt angles determined by the crystal rotation method are also shown.) rubbing strength (pile impression/mm)
0
0.20
0.35
0.50
surface tilt angle of main chain γ/degree normalized amplitude (b/a) pretilt angle/degree
0 0.14
7.3 0.15 7.5
5.5 0.18 5.2
4.4 0.17 4.5
The angular dependence of C2, whose transition moment is perpendicular to the ring plane, was fitted to the theoretical function:20,24
I(θ ) ) a + b cos(2(θ - γ ))
Figure 2. Carbon K-edge partial electron yield spectra from an unrubbed CP7CC3 polyimide film. The X-ray incidence angle was 90 degrees. The peaks used for the analysis are C2 and C4 attributed to the 1s-π* and 1s-σ* excitations at carbon atoms specified in the figure.
the unrubbed sample, the angular dependence is symmetrical with respect to 90 degrees, or normal incidence. At 90 degrees, peak intensities of π* transition take a minimum value and intensities of σ* transition take a maximum value. These facts indicate that the phenyl ring of main chain is oriented parallel to the substrate surface on average as shown in Figure 3a. On the other hand, for the rubbed sample, the symmetry axis shifts from 90 degrees to (90 degrees - γ). This shift indicates that the phenyl ring of main chain is no longer parallel to the substrate surface. If we assume a uniaxial distribution of the ring normal orientation, γ means the surface tilt angle between phenyl ring and substrate surface, as shown in Figure 3b. The behavior is essentially the same as that already reported for PI without side chains.23
(1)
The values of surface tilt angle γ determined by the fitting are shown in Table 1 for three rubbing strengths. It was found that γ decreases with increasing rubbing strength and the values of γ coincide with the LC pretilt angles determined by the crystal rotation method for the cells fabricated using the same surface as that used for NEXAFS measurements, as also shown in Table 1. Let us consider the normalized amplitude defined as the value of coefficient b divided by a.20 It is related to the order parameter of the transition dipole moment. If the orientational distribution of the transition dipole is three-dimensionally random, the order parameter is zero. In this case, normalized amplitude is also zero, since b is zero. On the other hand, if all the ring planes are parallel to the surface, even if the main chains orient randomly in the substrate plane, all the transition dipoles orient along the surface normal, then both the order parameter of the transition dipole and the normalized amplitude is 1. The normalized amplitude smaller than 1 indicates the deviation from the uniform orientation, namely the tilting of the ring planes. Thus the normalized amplitude gives a measure of the amount of anisotropy and supplies information of molecular distribution. As rubbing strength becomes stronger, the normalized amplitude becomes larger, as shown in Table 1, suggesting that the orientational distribution of the transition dipole becomes sharper
Figure 3. Angular dependence of the resonance intensities for unrubbed (a) and rubbed (b) CP7CC3 samples. Diamonds and triangles stands for the peak intensities of the π* resonance at 286.3 eV and σ* resonance at 295.9 eV in the PEY spectra.
9194 J. Phys. Chem. B, Vol. 105, No. 38, 2001
Sakai et al. TABLE 2: Assignments of Resonance Peaks in the N K-edge NEXAFS Spectra for CP7CC3 and CP Polyimide Films
Figure 4. Nitrogen K-edge PEY spectra from unrubbed CP7CC3 and CP polyimide films. The X-ray incidence angle was 90 degrees.
with increasing rubbing strength. Here, zero rubbing strength means that the sample is not rubbed. It is known that liquid crystals can align uniaxially under the influence of very small surface anisotropy.27,28 Therefore, even a small normalized amplitude is effective for the alignment, if in-plane anisotropy and/or surface tilt angle exist in the groups that influence the alignment of liquid crystal molecules. Next, let us turn to discuss the results on N K-shell absorption edge. Figure 4 shows NEXAFS spectra of CP7CC3 and CP polyimide films taken in the TEY detection mode. The differ-
peak
photon energy (eV)
final state
N1 N2 N3 N4 N5 N6
398.7 399.5 401.7 402.7 408.2 412.8
π* (C≡Ν) π* (C≡Ν) π* (imide ring) π* (phenyl ring) σ* (C-N) σ* (C-N)
ence of these PIs is whether a cyanobiphenyl side chain exists or not. The full assignments of these resonance peaks are shown in Table 2. By comparing these two spectra, we see that these N1 and N2 peaks assigned to the π* resonaces in the CN triple bond disappeared for CP polyimide without side chains, as expected. The splitting of N1 and N2 peaks is due to the intramolecular interaction, i.e., they are attributed respectively to the π* transition with the transition dipoles perpendicular and parallel to the phenyl plane and both are perpendicular to the side chain axis. The final orbital in the former is conjugated with the π* orbital of the phenyl ring, while the latter is not. We can obtain information about the orientation of the side chain from the N1 and N2 resonance peak, and the orientation of mainchain from the N3 or N4 peak that are attributed to the resonances in the imide and phenyl ring. Figure 5, parts a and b, shows the angular dependence of N1 (at 398.7 eV) and N3 (at 401.7 eV) peak intensities for pile impression of 0.2 and 0.5 mm taken in the PEY detection mode. As mentioned above, the N1 peak is assigned to π* transition on a N atom of CN triple bond, whose transition dipole moment is oriented perpendicular to the long axis of the cyanobiphenyl side chain. On the other hand, the N3 peak is assigned to that on a N of imide ring, of which the transition moment is oriented perpendicular to the imide ring plane. As expected, the symmetrical axis shifts from 90 degrees by about the same amount as in the C K-edge spectra (see Figure 3b) due to the emergence of the surface tilt angle γ. The solid lines are theoretical curves obtained by fitting to eq 1. From the fitting parameters, we can compute the surface tilt angle and normalized amplitude, as shown in Table 3. For the main chain, the
Figure 5. Angular dependence of the resonance intensity for CP7CC3 polyimide films rubbed under the conditions that pile impression is 0.2 mm (a) and 0.5 mm (b). Squares and circles stand for the peak intensities of the π* resonance corresponding to N1 (at 398.7 eV) and N3 (at 401.7 eV) peaks in the PEY spectra.
Main and Side Chains of Polyimide Alignment Layer
J. Phys. Chem. B, Vol. 105, No. 38, 2001 9195
TABLE 3: Surface Tilt Angles of Main Chain and Side Chain of CP7CC3 as a Function of Rubbing Strength (expressed by pile impression) Determined at the N K-Shell Absorption Edge (The normalized amplitudes of the main and side chains are also shown.) rubbing strength (pile impression/mm)
0.20
0.50
surface tilt angle of main chain γ/degree surface tilt angle of side chain γ/degree normalized amplitude (b/a) of main chain normalized amplitude (b/a) of side chain pretilt angle/degree
7.7 25.6 0.11 0.05 7.5
4.9 27.4 0.14 0.04 4.5
values of surface tilt angle γ are about the same as those obtained from the C K-edge PEY spectra (see Table 1). The agreement of the surface tilt angles determined by TEY and PEY detections is reasonable, since the phenyl ring responsible for C K-edge data and the imide ring responsible for N K-edge data are directly connected to each other, and also guarantees that the transition dipole orientation near the surface is uniform. As expected, the normalized amplitude is also approximately the same as that from the C K-edge spectra. The surface tilt angle γ of the main chain decreases with increasing rubbing strength, while γ of the side chain increases. As for the orientation of the side chains, we applied eq 1 to analyze the data, and then obtained a fairly large value of γ, which is much larger than the pretilt angle of 5CB molecules. It is important to note that the bulk pretilt angles measured for LCs in contact with the alignment layers are almost the same as the surface tilt angles of the main-chain parts even for polyimide with side chain (CP7CC3). It is true, however, that the side chains play a role to increase the pretilt angle, as suggested by Sugiyama et al..29 Actually the LC pretilt angles for PI with side chains (CP7CC3) were larger than that for PI without side chains (CP) by a few degrees. It is also true that almost homeotropic alignment is obtained when using PI with side chains in all the units. Using this PI, we confirmed the pretilt angle of the LCs was about 84 degrees under the same condition as that in this report. In this way, it is concluded that the pretilt angle becomes larger by the introduction of side chains, but is more strongly correlated with γ of the main chain at least in the present system, in which the fraction of units with a side chain is 30%. The conclusion concerning the dominant role of the main chain is consistent with the conclusions by FT-IR,30 namely, the LC pretilt angle is strongly correlated with the surface tilt angle of the main chain. 4. Conclusion We have investigated the surface orientation of a polyimide film with a side chain by using surface-sensitive NEXAFS spectroscopy. Rubbing treatment induced the surface tilt angle between the main-chain axis and the rubbing direction and decreases with increasing rubbing strength. It was found that the surface tilt angle of the main chain is strongly correlated with the pretilt angle of molecules even for polyimide with units possessing side chains (CP7CC3) of a fraction of 30%. To further study the effect of side chains, NEXAFS experiments using PI with a variety of side-chain densities are necessary. Acknowledgment. The PI materials used in this study were supplied by Nissan Chemical Co. Ltd. The authors thank Dr.
Y. Kitajima of KEK-PF for his kind help in the measurement of the NEXAFS spectra. We are also grateful to Prof. N. Kosugi of the Institute for Molecular Science for the use of GSCF3 programs for ab initio MO calculations. One of the authors, T.S., is a Research Fellow of the Japan Society for the Promotion of Science. This work was supported in part by a Grant-in-Aid from the Ministry of Science, Education, Sports and Culture (No. 07CE2004) and the Venture Business Laboratory Project “Advanced Nanoprocess Technologies” at Nagoya University. This work was performed under the approval of the Photon Factory Programs Advisory Committee (Nos. 95G372, 97G311, and 99G180). References and Notes (1) Chen, W.; Feller, M. B.; Shen, Y. R. Phys. ReV. Lett. 1989, 63, 2665. (2) Feller, M. B.; Chen, W.; Shen, Y. R. Phys. ReV. 1992, 43, 6778. (3) Barmentlo, M.; van Aerle, N. A. J. M.; Hollering, R. W. J.; Damen, J. P. M. J. Appl. Phys. 1992, 71, 4799. (4) Barmentlo, M.; Hollering, R. W. J.; van Aerle, N. A. J. M. Phys. ReV. A 1992, 46, R4490. (5) Barmentlo, M.; Hollering, R. W. J.; van Aerle, N. A. J. M. Liq. Cryst. 1993, 14, 475. (6) Johannsmann, D.; Zhou, H.; Sonderkaer, P.; Wierenga, H.; Myrvold, B. O.; Shen, Y. R. Phys. ReV. E 1993, 48, 1889. (7) Zhuang, X.; Marrucci, L.; Shen, Y. R. Phys. ReV. Lett. 1994, 73, 1513. (8) Huang, J. Y.; Li, J. S.; Juang, Y. S.; Chen, S. H. Jpn. J. Appl. Phys. 1995, 34, 3163. (9) Shirota, K.; Yaginuma, M.; Sakai, T.; Ishikawa, K.; Takezoe, H.; Fukuda, A. Jpn. J. Appl. Phys. 1996, 35, 2275. (10) Shirota, K.; Ishikawa, K.; Takezoe, H.; Fukuda, A.; Shiibashi, T. Jpn. J. Appl. Phys. 1995, 34, L316. (11) Shirota, K.; Yaginuma, M.; Sakai, T.; Ishikawa, K.; Takezoe, H.; Fukuda, A. Appl. Phys. Lett. 1996, 69, 164. (12) Sakai, T.; Yoo, J. G.; Kinoshita, Y.; Ishikawa, K.; Takezoe, H.; Fukuda, A. Appl. Phys. Lett. 1997, 71, 2274. (13) Yoo, J. G.; Park, B.; Sakai, T.; Kinoshita, Y.; Hoshi, H.; Ishikawa, K.; Takezoe, H. Jpn. J. Appl. Phys. 1998, 37, 4124. (14) Park, B.; Yoo, J. G.; Sakai, T.; Hoshi, H.; Ishikawa, K.; Takezoe, H. Phys. ReV. E 1998, 58, 4624. (15) Meister, R.; Je´roˆme, B. Macromolecules 1999, 32, 480. (16) Sei, M.; Nagayama, H.; Kajikawa, K.; Ishii, H.; Seki, K.; Kondo, K.; Matsumoto, Y.; Ouchi, Y. Jpn. J. Appl. Phys. 1998, 37, 1974. (17) Oh-e, M.; Hong, S. C.; Shen, Y. R. J. Phys. Chem. B 2000, 104, 7455. (18) Sto¨hr, J. NEXAFS Spectroscopy, Vol. 25 of Springer Series in Surface Sciences; Springer: Heidelberg, 1992. (19) Samant, M. G.; Sto¨hr, J.; Brown, H. R.; Russel, T. P.; Sands, J. M.; Kumar, S. K. Macromolecules 1996, 29, 8334. (20) Weiss, K.; Wo¨ll, C.; Bo¨hm, E.; Fiebranz, B.; Forstmann, G.; Peng, B.; Scheumann, V.; Johannsmann, D. Macromolecules 1998, 31, 1930. (21) Ouchi, Y.; Mori, I.; Sei, M.; Ito, E.; Araki, T.; Ishii, H.; Seki, K.; Kondo, K. Physica D 1995, 208/209, 407. (22) Mori, I.; Araki, T.; Ishii, H.; Ouchi, Y.; Seki, K.; Kondo, K. J. Elec. Spectrosc. Relat. Phenom. 1996, 78, 371. (23) Sto¨hr, J.; Samant, M. G.; Cossy-Favre, A.; Dı´az, J.; Momoi, Y.; Odahara, S.; Nagata, T. Macromolecules 1998, 31, 1942. (24) Sto¨hr, J.; Samant, M. G. J. Elec. Spectrosc. Relat. Phenom. 1999, 98/99, 189. (25) Baur, G.; Wittwer, V.; Berreman, D. W. Phys. Lett. A 1976, 56, 142. (26) Scheffer, T. J.; Nehring, J. J. Appl. Phys. 1977, 48, 1783. (27) Ouchi, Y.; Feller, M. B.; Moses, T.; Shen, Y. R. Phys. ReV. Lett. 1992, 68, 3040. (28) Chung, D.-H.; Takanishi, Y.; Ishikawa, K.; Takezoe, H.; Park, B.; Jung, Y.; Hwnag, H. K.; Lee, S.; Han, K. J.; Jang, S. H. Jpn. J. Appl. Phys. 2000, 39, L185. (29) Sugiyama, T.; Kuniyasu, S.; Seo, D. S.; Fukuro, H.; Kobayashi, S. Jpn. J. Appl. Phys. 1990, 29, 2045. (30) Arafune, R.; Sakamoto, K.; Ushioda, S. Appl. Phys. Lett. 1997, 71, 2755.
