Jpn. J. Appl. Phys. Vol. 42 (2003) pp. 7570–7573 Part 1, No. 12, December 2003 #2003 The Japan Society of Applied Physics
Resonant Photoemission Spectroscopy and Theoretical Calculation of the Valence Band Structure in Chromium Aluminum Oxynitride Youngmin C HOI, Hyunju CHANG, Beyong-Hwan RYU, Kijeong K ONG, Jae Do LEE and Kwangsoo N O1 Korea Research Institute of Chemical Technology, 100 Yusong, Taejon 305-600, Korea 1 Korea Advanced Institute of Science and Technology, 373-1 Kusong, Taejon 305-701, Korea (Received May 12, 2003; revised July 14, 2003; accepted August 8, 2003; published December 10, 2003)
The electronic structure of chromium aluminum oxynitride has been investigated using resonant photoemission spectroscopy (RPES) and the discrete variational (DV)-X method. The RPES measurement of the electronic structure around the Cr 2p3=2 absorption edges exhibited significant resonant interference behavior for Cr 3d valence electrons, whereas it exhibited small resonant interference behavior for N 2p valence electrons. Therefore this RPES method can be useful for analyzing the valence band of chromium aluminum oxynitride film. The top of the valence band predominantly consists of Cr 3d and a small amount of N 2p. The difference between the measured photoemission spectra of the valence band and the DV-X calculation of chromium aluminum oxynitride ranges from 0.47 to 3.62 eV. This difference is probaly caused by the Coulomb interactions between the d electrons of chromium and the structure of the amorphous film. Through the experimental and theoretical studies, the valence band structure of chromium aluminum oxynitride came to be understand in detail. [DOI: 10.1143/JJAP.42.7570] KEYWORDS: electronic structure, chromium aluminum oxynitride, DV-X method, resonant photoemission spectroscopy
1.
Introduction
2.
Chromium aluminum oxide has been proposed as a candidate material for the phase-shifting mask (PSM) in deep ultraviolet (DUV) optical lithography,1) but it is difficult to meet the requirements of the transmittance slope without the substitution of nitrogen for chromium aluminum oxide.2) Such optical properties are closely related to the electronic structure of the mask material. Accordingly, understanding the electronic structure of chromium aluminum oxynitride is important in the development of the mask material. In our previous works,3,4) the electronic structure of chromium aluminum oxynitride was investigated by the discrete variational (DV)-X method theoretically and by Xray photoelectron spectroscopy (XPS) experimentally but the variation of valence band with composition was not clear because of the resolution limitations of XPS study. In this study, we precisely measured the valence band of chromium aluminum oxynitride using resonant photoemission spectroscopy (RPES). We used the absorption edge of Cr 2p3=2 as the source energy to obtain the resonant photoemission of the valence band. Resonant photoemission enables us to obtain enhanced intensity and resolution in photoemission study for low-cross-section structures as valence band.5–9) The results of the spectroscopy were compared with the theoretical calculation. The electronic structure was calculated by a first-principles molecular orbital (MO) method. From the crystallographic data, we determined the atomic structure of the crystalline phase. Then, the embedded cluster model within the framework of the DV-X: method was employed to calculate the electronic structure of optimized crystal structures. Several model clusters of different atomic concentrations were chosen to simulate different atomic structures in the crystalline phase.
Experimental
2.1
Sample preparation and resonant photoemission spectroscopy Chromium aluminum oxide and oxynitride films were deposited using a planar circular-type DC magnetron reactive sputtering system with a disc-shaped 10 cm Cr–Al target.10) The Ar, N2 , and O2 gas mixture was injected as a reaction gas into the chamber. The films were deposited under the conditions of substrate temperature of 573 K, power of 40 W and total chamber pressure of 4:7 103 Torr. The target-to-substrate distance was fixed at 132.4 mm. Films were deposited on quartz and Si substrates simultaneously. The detailed conditions of film preparation are described in ref. 10. The film composition was analyzed by a wavelength dispersive spectrometer (WDS). The RPES spectra were measured at the 4B1 beam line in the Pohang Accelerator Light (PAL) source. The source beam energy was calibrated with the Au 4f reference peak at 84 eV.
2.2 Computational details All calculations were performed using a first-principles DV-X: method,11–13) which is widely applicable to molecules and solids. In our calculation, was fixed at 0.7, which was known to be reasonable for the calculation of transition metals.13) In the SCAT code we used, the basis functions were numerical solutions for the atom-like potential.14) To build the atomic structure of chromium aluminum oxynitride, we used the lattice parameters of a continuous series of solid solutions.15) We selected an embedded cluster containing eight metal atoms (Cr or Al) and 33 oxygen atoms, then we substituted nitrogen for oxygen in chromium aluminum oxide to investigate the valence band of chromium aluminum oxynitride. The procedure of electronic structure calculation is well described in our previous work.3,4) 3.
Results and Discussion RPES spectra are compared with XPS spectra for the
E-mail address:
[email protected] 7570
Jpn. J. Appl. Phys. Vol. 42 (2003) Pt. 1, No. 12
7571
(a)
Intensity (arb.units)
XPS
Intensity (arb.units)
Y. CHOI et al.
photon energy (eV) 1253.6 (Mg Kα )
a : AlO b : AlON c : CrAlO d : CrAlON e : CrO f : CrON
e
f
b a
(b)
RPES
d
c
12
10
8
6
4
2
0
Binding energy (eV) Fig. 2. RPES spectra of various chromium aluminum compounds.
photon energy (eV) 396 400 569 574
12
10
8
6
4
2
0
Binding energy (eV) Fig. 1. Photoemission spectra of chromium aluminum oxynitride film; (a) XPS and (b) RPES.
chromium aluminum oxynitride film and the result is shown in Fig. 1. While the valence band top at approximately 2– 3 eV shows a small shoulder in XPS spectra and RPES spectra below 574 eV similarly to XPS spectra, its intensity increases markedly when the photon energy is 574 eV. This enhancement of the valence band top is caused by the resonant effect of the valence band electron at the Cr 2p3=2 (574 eV) absorption edge. The resonant effects of transition metals around 2p thresholds have been reported and discussed with increasing interest in recent years.5–8) The intensity enhancement at the absorption edge relates to the 2p–3d interaction, namely the interaction between a direct photoemission and the direct recombination of an excited state. This means that the normal 3d photoemission 2p6 3s2 3p6 3d4 4s2 + h ! 2p6 3s2 3p6 3d3 4s2 + e and the decay of the core excitation as 2p6 3s2 3p6 3d4 4s2 + h ! 2p5 3s2 3p6 3d5 4s2 ! 2p6 3s2 3p6 3d3 4s2 + e interfere. The direct photoemission from the 3d band produces a continuous photoemission spectrum. In the case of direct recombination, the photon energy of the absorption edge raises an electron from the 2p to the unoccupied 3d band, thus creating a hole in the 2p band and a conduction electron in the 3d band. This electron and hole directly recombine, creating an electron-hole pair (exciton) and liberating energy. The valence electron is ejected by this liberated energy. Because the final states in both processes are the same, the electrons have the same kinetic energy. Thus the intensity of the 3d spectrum is enhanced. There is another possible decay mechanism for the exciton, which is similar to Auger decay. A valence electron recombines with the core hole and the liberated energy ejects
another valence electron into vacuum, leaving the original conduction electron at the Fermi surface. Thus the final state consists of a free electron and two holes in the valence band. This ejected electron has an energy corresponding to the satellite just below the 3d band. Around the 3p threshold, most of the enhancement in photoemission intensity occurs in the region of the satellite, although the valence band emission is also enhanced.16) By contrast, at the 2p threshold no resonance satellites are observed below the valence band, but the main peak of the 3d band exhibits a pronounced resonance behavior. Since the overlap of the 3d wave function with the 2p-hole wave function is much smaller than that with the 3p-hole wave function, the decay of the well-localized discrete excitation state occurs predominantly by way of a direct recombination process, rather than by way of the Auger-type decay.9) It is for this reason that the 2p excitation does not produce a satellite resonance, but rather a 3d valence band photoemission enhancement. Therefore this RPES method can be useful for analyzing the valence of chromium aluminum oxynitride film. Resonant photoemission spectra for the various compositions of chromium aluminum oxynitride are shown in Fig. 2. Incident beam energy was selected at the Cr 2p3=2 absorption edge. In aluminum oxide, the valence band is observed in the energy range between 3 and 11 eV and this corresponds to the O 2p. When nitrogen atom is added to aluminum oxide, a small peak appears in the energy region of 1–3 eV. This is suspected to be the N 2p peak caused by the weak resonance at the Cr 2p3=2 absorption edge. In the case of chromium oxide, the valence band top appears at approximately 1.5 eV and this corresponds to the partially occupied Cr 3d peak caused by the strong resonance effect. The addition of nitrogen atom to chromium oxide changes the valence band peak width to approximately 0.6 eV but does not change the peak position. This broadening originates from the N 2p at approximately 1–3 eV. The top of the valence band in chromium aluminum oxide at the energy of 3.7 eV is observed between O 2p of aluminum oxide and Cr 3d of chromium oxide. When the nitrogen atom is substituted for the oxygen atom in chromium aluminum oxide, the top of the valence band is shifted to the lower binding energy of approximately 0.8 eV. Figure 3 shows the variation of valence band with
7572
Jpn. J. Appl. Phys. Vol. 42 (2003) Pt. 1, No. 12
Y. CHOI et al.
Intensity (arb.units)
Photon energy=574eV
a
b
c
d
e
Cr/Al/O atomic ratio a : 0.00/0.27/0.73 b : 0.08/0.33/0.59 c : 0.16/0.20/0.64 d : 0.29/0.12/0.59 e : 0.61/0.00/0.39
12
10
8
6
4
2
0
Binding energy (eV) Fig. 3. Variation of valence band top with composition change in chromium aluminum oxide.
(a)
AlO
(b)
CrO
O2p
RPES
RPES
O2p
PDOS PDOS
-12
-10
-8
-6
-4
-2
0
(c)
Intensity (arb.units)
AlON
-10
-8
-6
-4
-2
0
(d)
CrON
Cr3d
RPES
O2p
RPES
-12
O2p N2p PDOS
N2p
PDOS
-12
-10
-8
-6
-4
-2
0
(e)
Intensity (arb.units)
CrAlO
-12
-10
-8
-6
-4
-2
(f)
CrAlON
Cr3d
RPES
0
Cr3d RPES O2p
8 O2p
PDOS PDOS
-12
-10
-8
-6
-4
-2
Binding energy (eV)
0
-12
N2p
-10
-8
-6
-4
-2
0
Binding energy (eV)
Fig. 4. Comparison between the measured RPES spectra and calculated PDOS of (a) aluminum oxide, (b) chromium oxide, (c) aluminum oxynitride, (d) chromium oxynitride, (e) chromium aluminum oxide and (f) chromium aluminum oxynitride.
composition in chromium aluminum oxide. An increase in the chromium concentration results in the top of the valence band shifting from the O 2p position in pure aluminum oxide to the Cr 3d position in pure chromium oxide. To compare the RPES results with the calculated PDOS,
Valence band maximum (eV)
Intensity (arb.units)
Cr3d
we superimposed the RPES spectra of the valence band on the corresponding PDOS calculated theoretically. In the aluminum oxide system [Fig. 4(a)], the valence band consisted of the O 2p orbital and both measured and calculated electronic structures are in good agreement. The top of the valence band in the chromium oxide system, as shown in Fig. 4(b), is composed of a partially occupied Cr 3d band. There is a small difference of approximately 0.5 eV between the RPES spectra and PDOS. When the nitrogen is substituted for oxygen in the pure aluminum oxide system, the O 2p band exhibits a shoulder peak at the lower energy side [Fig. 4(c)]. Although there is a difference of approximately 2 eV between RPES spectra and PDOS, the tendency of the shoulder peak is in good agreement between the measured and calculated electronic structures. The chromium oxynitride system exhibits valence band broadening, as shown in Fig. 4(d), caused by N2p mixing with the Cr 3d orbital. In the mixture of chromium oxide and aluminum oxide [Fig. 4(e)], the top of the valence band appears between those of pure chromium oxide and aluminum oxide systems. In Fig. 4(e), the top of the valence band in chromium aluminum oxynitride is shifted to a lower energy approximately 0.8 eV compared with the chromium aluminum oxide system. We can surmise from the above results that chromium addition shifts the position of the valence band to a lower binding energy and nitrogen addition broadens the valence bandwidth. This change of valence band structure is closely related to the transmittance and transmittance slope of chromium aluminum oxynitride, which is useful in PSM application. In the chromium aluminum oxide system, the disagreement between the Cr 3d bands observed by RPES and PDOS is presented in Fig. 5. The top of the valence band increases linearly with aluminum oxide concentration in the measured RPES spectra. However, PDOS does not change in spite of concentration change. This disagreement between the RPES and PDOS measurements suggests the existence of a localization. A possible localization mechanism in chromium aluminum oxide system is the large Coulomb correlation interaction between Cr d electrons. Kang et al. reported that simple addition of the Coulomb interaction term U (approximately 1.5 eV) to the local spin density approximation
6
RPES
4
2 PDOS
0
0.0
Cr2O3
0.2
0.4
0.6
0.8
Al Concentration (mol fraction)
1.0
Al2O3
Fig. 5. Difference in valence band maximum between RPES and PDOS in the chromium aluminum oxide system.
Jpn. J. Appl. Phys. Vol. 42 (2003) Pt. 1, No. 12
(LSDA) calculation makes PDOS of Cr 3d coincide with the RPES spectral peak in Cr-based chalcogenide spinel compounds (Fe0:5 Cu0:5 Cr2 S4 , FCCS).17) This U value is somewhat smaller than those used in LSDA+U calculations for transition metal oxides because the expected U values in metallic transition compounds or in intermetallic transition metal compounds are smaller than those in transition metal oxides.18,19) In the case of chromium oxides, U values are known to be in the range of 3–5 eV according to the type of anion.20) In our chromium aluminum system, Coulomb interaction energy is supposed to change from 1 to 4 eV in accordance with the composition. Nevertheless, this Coulomb correlation interaction cannot explain the entire difference between the experiments and calculations. This disagreement also seems to indicate that the crystal structure used in theoretical calculation does not exactly describe the actual amorphous film. In particular, the neglect of structure relaxation to simplify the model cluster leads to a narrow valence band structure of chromium aluminum oxide in theoretical calculation. We came to be understand the variation of valence band structure with composition from the RPES study and DV-X calculation. Although it is not easy to correlate transmittance with electronic structure because transmittance also depends on film thickness, refractive index and crystal structure, our study provides an understanding of the electronic structure variation of the chromium aluminum oxynitride system with composition, which is closely related to the transmittance property. 4.
Conclusions
To summarize, we have performed resonant photoemission measurements for chromium aluminum oxynitride film and compared them with DV-X calculations. A marked resonant effect of the Cr 3d band has been observed by RPES and can be used to study the valence band of chromium aluminum oxynitride film. It is observed that the top of the valence band is predominantly of the partially occupied Cr 3d, consistent with the DV-X calculation. In chromium aluminum oxide, RPES shows that the top of the valence band changes linearly with aluminum concentration, whereas the calculated PDOS does not. We suspect that this disagreement is due to the Coulomb interaction between the d electrons of chromium and the structure of the amorphous film. When nitrogen is substituted for oxygen in chromium aluminum oxide, the N2p level appears between the O 2p and Cr 3d levels. Thus, the valence band of chromium aluminum oxynitride becomes broader than that of chromium aluminum oxide. Through theoretical calculation and experimental verification, we can elucidate that the transmittance and transmittance slope of chromium aluminum
Y. CHOI et al.
7573
oxynitride film can be controlled properly for PSM application through the modification of the valence band by composition control. Acknowledgments This work was supported by MOST of Korea through the National R&D Project for Nano Science and Technology. The authors thank Professor H. Adachi (Kyoto University) for the use of the DV-X MO program, Dr. Yang-Soo Kim (Korea Advanced Institute of Science and Technology) for helpful discussions. We also thank Dr. Tai-Hee Kang, Mr. Kyu Wook Ihm, Dr. Min Gyu Kim, and Bong Soo Kim of Pohang Accelerator Light (PAL) Institute for spectroscopic analysis. 1) E. Kim, S. Hong, K. Kim, Z. Jiang, D. Kim, S. Lim, S. Woo, Y. Koh and K. No: Appl. Opt. 36 (1997) 7247. 2) E. Kim: Dr. Thesis, Department of Material Science and Engineering, Korea Advanced Institute of Science and Technology, Daejeon, Korea, 2000. 3) Y. Choi, H. Chang, J. D. Lee, E. Kim and K. No: Jpn. J. Appl. Phys. 41 (2002) 5805. 4) H. Chang, Y. Choi and J. D. Lee: Adv. Quantum Chem. 42 (2003) 163. 5) H. Sato, M. Koyama, K. Takada, H. Okuda, K. Shimada, Y. Ueda, J. Ghijsen and M. Taniguchi: J. Electron Spectrosc. Relat. Phenom. 88–91 (1998) 333. 6) M. Weinelt, A. Nilsson, M. Magnuson, T. Wiell, N. Wassdahl, O. Karis, A. Fohlisch and N. Martensson: Phys. Rev. Lett. 78 (1997) 967. 7) T. Kaurila, R. Uhrberg and J. Vayrynen: J. Electron Spectrosc. Relat. Phenom. 88–91 (1998) 399. 8) M. F. Lopez, A. Gutierrez, C. Laubschat and G. Kaindl: Solid State Commun. 94 (1995) 673. 9) S. C. Wu, C. K. C. Lok, J. Sokolov and F. Jona: Phys. Rev. B 39 (1989) 1058. 10) E. Kim, S. Hong, S. Lim, Y. Kim, D. Kim and K. No: Appl. Opt. 37 (1998) 4254. 11) H. Adachi, M. Tsukada and C. Satoko: J. Phys. Soc. Jpn. 45 (1978) 875. 12) T. Tanabe, H. Adachi and S. Imoto: Jpn. J. Appl. Phys. 17 (1978) 49. 13) H. Adachi, S. Shiokawa, M. Tsukada, C. Satoko and Satoru Sugano: J. Phys. Soc. Jpn. 47 (1979) 1528. 14) A. Rosen, D. E. Ellis, H. Adachi and F. W. Averill: J. Chem. Phys. 65 (1976) 3629. 15) M. Watanabe, T. Hirayama, M. Yoshinaka, K. Hirota and O. Yamaguchi: Mater. Res. Bull. 31 (1996) 861. 16) C. Guillot, Y. Ballu, J. Paigne, J. Lecante, K. P. Jain, P. Thiry, R. Pinchaux, Y. Petroff and L. M. Falicov: Phys. Rev. Lett. 39 (1977) 1632. 17) J. S. Kang, S. J. Kim, C. S. Kim, C. G. Olson and B. I. Min: Phys. Rev. B 63 (2001) 144412. 18) M. S. Park, S. K. Kwon, S. J. Youn and B. I. Min: Phys. Rev. B 59 (1999) 10018. 19) V. I. Anisimov, J. Zaanen and O. K. Anderson: Phys. Rev. B 44 (1991) 943. 20) I. Pollini, A. Mosser and J. C. Parlebas: Phys. Rep. 355 (2001) 1.
