1: \documentclass[twocolumn,prl,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
2: %\documentclass[preprint,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
3: %\documentclass[preprint,showpacs,preprintnumbers]{revtex4}
4: \usepackage{graphicx}% Include figure files
5: \usepackage{dcolumn}% Align table columns on decimal point
6: \usepackage{bm}% bold math
7:
8: %\nofiles
9:
10: \begin{document}
11:
12: %\preprint{APS/123-QED}
13: %{\Large{\it DO NOT DISTRIBUTE.}}
14: \title{Direct observation of the mass renormalization in SrVO$_3$ by angle
15: resolved photoemission spectroscopy}
16:
17: \author{T. Yoshida$^1$, K. Tanaka$^1$, H. Yagi$^1$, A. Ino$^2$, H. Eisaki$^3$, A.
18: Fujimori$^1$, and Z.-X. Shen$^4$} \affiliation{$^1$Department of
19: Complexity Science and Engineering and Department of Physics,
20: University of Tokyo, Kashiwa, Chiba 277-8561, Japan}
21: \affiliation{$^2$Hiroshima Synchrotron Radiation Center, Hiroshima
22: University, Kagamiyama 2-313, Higashi-Hiroshima 739-8526, Japan}
23: \affiliation{$^3$National Institute of Advanced Industrial Science
24: and Technology, Tsukuba 305-8568, Japan}
25: \affiliation{$^4$Department of Applied Physics and Stanford
26: Synchrotron Radiation Laboratory, Stanford University, Stanford,
27: CA94305}
28:
29: \date{\today}% It is always \today, today,
30: % but any date may be explicitly specified
31:
32: \begin{abstract}
33: We have performed an angle-resolved photoemission study of the
34: three-dimensional perovskite-type SrVO$_3$. Observed spectral
35: weight distribution of the coherent part in the momentum space
36: shows cylindrical Fermi surfaces consisting of the V 3$d$ $t_{2g}$
37: orbitals as predicted by local-density-approximation (LDA)
38: band-structure calculation. The observed energy dispersion shows a
39: moderately enhanced effective mass compared to the LDA results,
40: corresponding to the effective mass enhancement seen in the
41: thermodynamic properties. Contributions from the bulk and surface
42: electronic structures to the observed spectra are discussed based
43: on model calculations.
44: \end{abstract}
45:
46: \pacs{71.18.+y, 71.20.-b, 71.27.+a, 71.30.+h, 79.60.-i}% PACS, the Physics and Astronomy
47: % Classification Scheme.
48: %\keywords{Suggested keywords}%Use show keys class option if keyword
49: %display desired
50: \maketitle
51: %introduction
52: The effect of electron correlation on the electronic structure of
53: the transition metal oxides (TMO) has been one of the most
54: important and fundamental issues in condensed matter physics
55: \cite{RMPfujimori}. While the high-$T_c$ cuprates and the colossal
56: magneto resistive (CMR) manganites belong to the charge-transfer
57: regime of Zaanen-Allen-Sawatzky diagram \cite{ZSA}, light
58: transition-metal oxides such as perovskite-type Ti and V oxides
59: are prototypical Mott-Hubbard-type systems. The metal-insulator
60: transitions (MITs) in Ti and V oxides can be directly compared
61: with theoretical predictions using the Hubbard model in, e.g.
62: dynamical mean-field theory (DMFT), and are, therefore, ideal
63: model systems to study electron correlation phenomena
64: \cite{Pavarini}. In earlier photoemission studies of
65: perovskite-type Ti and V oxides, such as Ca$_{1-x}$Sr$_x$VO$_3$
66: \cite{inouePES, Morikawa} and La$_{1-x}$Sr$_x$TiO$_3$
67: \cite{Fujimori, Yoshida}, two characteristic structures in the
68: transition-metal $d$ band have been identified. One is the
69: coherent part around the Fermi level ($E_F$) corresponding to the
70: band-like electronic structure, and the other is the incoherent
71: part 1-2 eV away from $E_F$ corresponding to the remnant of the
72: lower Hubbard band (LHB), a behavior which is predicted by DMFT
73: \cite{DMFT}.
74:
75: In the filling-control Mott-Hubbard system
76: La$_{1-x}$Sr$_x$TiO$_3$, a critical mass enhancement toward the
77: MIT has been observed in the electronic specific heats and
78: magnetic susceptibility \cite{Kumagai}, analogous to the mass
79: enhancement scenario developed by Brinkman and Rice \cite{GBR}. A
80: detail analysis of the coherent part of the photoemission spectra
81: in La$_{1-x}$Sr$_x$TiO$_3$ has revealed to some extent enhancement
82: of the effective mass when approaching the MIT \cite{Yoshida},
83: which corresponds to the enhancement of the electronic specific
84: heat coefficients $\gamma$ toward the MIT critical point
85: $x\sim0.06$.
86:
87: On the other hand, Ca$_{1-x}$Sr$_x$VO$_3$ is a bandwidth-control
88: Mott-Hubbard system. The photoemission results on
89: Ca$_{1-x}$Sr$_x$VO$_3$ have suggested that the transition from a
90: correlated metal to the Mott insulating phase is characterized by
91: spectral weight transfer from the coherent to the incoherent parts
92: with decreasing $x$\cite{inouePES}. However, their $\gamma$ does
93: not show such a large change with $x$ and no critical mass
94: enhancement region is reached in this system \cite{inouePRB}.
95: Also, the width of the coherent part does not change appreciably
96: with $x$ while the photoemission spectral intensity at $E_F$
97: increases as $x$ increases. This behavior was interpreted due to a
98: strong momentum dependence of the self-energy \cite{inouePES}.
99: However, a recent ``bulk-sensitive" photoemission study of
100: SrVO$_3$ and CaVO$_3$ using soft X-rays indicated that the
101: spectral weight of the coherent part is larger than that reported
102: in the previous study with low photon energy and that there is no
103: appreciable spectral weight transfer between SrVO$_3$ and CaVO$_3$
104: \cite{sekiyama}. A photoemission study using several photon
105: energies has indicated that the incoherent part includes a certain
106: amount of contributions from the surface \cite{Maiti}. Theoretical
107: studies have shown that the surface electronic structure should
108: have a more strongly renormalized effective mass because of the
109: reduced co-ordination number and hence the reduced bandwidth
110: \cite{Liebsch}.
111:
112: %Importance of direct observation
113: In order to address those fundamental unresolved issues, direct
114: observation of band dispersion by angle-resolved photoemission
115: spectroscopy (ARPES) would provide important information about the
116: mass renormalization. In the present study, we have observed the
117: band dispersion, the Fermi surface and hence the band
118: renormalization in the three-dimensional Mott-Hubbard system
119: SrVO$_3$ by ARPES. Also, we shall address the issue of surface
120: effects by obtained comparison of the spectra with the calculated
121: bulk and surface electronic structures.
122:
123: \begin{figure}
124: \includegraphics[width=9cm]{inc}
125: \caption{\label{incoherent}ARPES spectra for SrVO$_3$ along
126: momentum cut 4 in the Brillouin zone shown in Fig.
127: \ref{nkMapping}. (a) EDC's. (b) Intensity plot in the $E$-$k_y$
128: plane. Dispersive feature within $\sim$0.7 eV of $E_F$ is the
129: coherent part, while broad feature around -1.5 eV is the
130: incoherent part.}
131: \end{figure}
132:
133: \begin{figure}
134: \includegraphics[width=8cm]{nkMapping}
135: \caption{\label{nkMapping}Spectral weight mapping at $E_F$. (a)
136: $k_y$=0 cross-sectional view of the Fermi surfaces (blue curves)
137: and the momentum loci corresponding to the mapping in panels (b)
138: and (c). (b) Mapping for $h\nu$=28.5eV. (c) Mapping for $h\nu$= 24
139: eV. Note that the mapping is projection on the $k_x$-$k_y$ plane.}
140: \end{figure}
141:
142: %Experiment
143: ARPES measurements were performed at Beamline 5-4 of Stanford
144: Synchrotron Radiation Laboratory (SSRL) with a normal incidence
145: monochromator and a Scienta SES-200 electron analyzer. The typical
146: energy and angular resolutions used for the present measurements
147: were about 30 meV and 0.3 degree, respectively. Single crystals of
148: SrVO$_3$ were grown using the travelling-solvent floating zone
149: method. Samples were first aligned by Laue difraction \textit{ex
150: situ}, and cleaved along the cubic (100) surface and measured at a
151: temperature of 15 K in a pressure better than $5 \times 10^{-11}$
152: Torr. We have performed the measurements at photon energies
153: $h\nu$=24 and 28.5 eV. In this paper, the electron momentum is
154: expressed in units of $\pi/a$, where $a \sim 3.84$ \AA, is the
155: lattice constant corresponding to the V-O-V distance. $k_x$ and
156: $k_y$ are the in-plane momenta and $k_z$ are the out-of-plane
157: ones.
158:
159: \begin{figure}
160: \includegraphics[width=9cm]{EDC_Ek}% Here is how to import EPS art
161: \caption{\label{EDC_Ek}(Color) Panels (a) show EDC's corresponding
162: to the cuts in Fig. \ref{nkMapping}. Vertical bars are guides to
163: the eyes indicating the positions of the dispersive features.
164: Panels (b) are intensity plots in $E$-$k$ space of the EDC's in
165: the panels (a). Angle-independent backgrounds have been
166: subtracted. Second derivatives of the EDC's and MDC's are also
167: shown in panels (c) and (d), respectively.}
168: \end{figure}
169:
170: %EDC's including incoherent part
171: First, we show an example of energy distibution curves (EDC's) for
172: SrVO$_3$ in Fig. \ref{incoherent}(a). The coherent part with
173: $\sim$ 0.7 eV of $E_F$ shows a dispersive feature, consistent with
174: the view point that the coherent part corresponds to the
175: band-structure calculation. On the other hand, the incoherent part
176: centered at $\sim$ -1.5 eV does not show a clear dispersion as
177: seen in Fig \ref{incoherent}(b). Background from angle-integrated
178: spectra and the surface effects discussed below may hinder the
179: observation of subtle dispersive features in the incoherent part.
180: In the rest of this paper, we shall focus on the dispersive
181: feature in the coherent part.
182:
183: %Spectral weight mapping
184: Figure \ref{nkMapping} (a) illustrates the $k_y$=0 cross-sectional
185: view of the Fermi surfaces (blue curves) predicted by
186: local-density approximation (LDA) band-calculation
187: \cite{Takegahara} and the loci of electron momenta in the
188: $k_x$-$k_z$ plane for constant photon energies $h\nu$= 24 eV and
189: 28.5 eV. The inner potential of 10 eV has been assumed. Here, the
190: LDA band-structure \cite{Takegahara} to a tight-binding (TB) model
191: consisting of three $t_{2g}$ orbital $d_{xy}$, $d_{yz}$ and
192: $d_{zx}$ of V. The corresponding three bands do not hybridize with
193: each other \cite{TBNN} within the present TB formalism which is
194: the same as that in Ref. \cite{Liebsch}. Figure \ref{nkMapping}
195: (b) and (c) shows spectral weight mapping in the $k_x$-$k_y$
196: momentum space at $E_F$ on the momentum loci in panel (a) for the
197: photon energies $h\nu$=28.5 and 24 eV, respectively. Spectral
198: weight is integrated over 30 meV within $E_F$. As seen in Fig.
199: \ref{nkMapping} (b) and (c), spectral weight for both photon
200: energies are largely confined within the cylindrical Fermi surface
201: extended in the $k_z$ direction. According to panel (a), cuts 2
202: and 4 correspond to momenta near the X point
203: [$\mathbf{k}=(1,0,0)$]. Therefore, the observed spectral weight
204: comes only from one cylindrical Fermi surface referred to as
205: $\gamma$ sheet \cite{inoueDHVA} which arises from the $d_{xy}$
206: orbital. On the other hand, the spectral weight in the first BZ
207: for $h\nu$ =28.5 eV (cut 1) is more intense than that for 24 eV
208: (cut 3). This difference comes from the fact that the momentum
209: locus for cut 1 is within the $\alpha$- and $\beta$ sheets
210: enclosed by the three cylindrical Fermi surfaces \cite{inoueDHVA},
211: while that for cut 3 is out of them. Indeed, energy distribution
212: curves (EDC's) along cut 1 [Fig. \ref{EDC_Ek} (a)] show
213: complicated features suggestive of two energy dispersions.
214:
215: %
216: Figure \ref{EDC_Ek}(a) shows EDC's along the four cuts in Fig.
217: \ref{nkMapping}. Panels in Fig. \ref{EDC_Ek}(b) show spectral
218: weight plots in the energy-momentum ($E$-$k$) space corresponding
219: to each cut. The angle-independent EDC's have been subtracted as
220: the background in these $E$-$k_y$ plots \cite{BG}. The second
221: derivatives of the EDC's and the MDC's are shown in panels (c) and
222: (d), respectively, so that one can visualize the band dispersions
223: in spite of the high backgrounds. While cuts 2 and 4 show single
224: energy dispersions, cut 1, which is around the $\Gamma$ point,
225: shows two band dispersions as in the EDC's in panel (a). This is
226: due to the fact that the momenta of cut 1 are within the $\alpha$
227: and $\beta$ sheets of the Fermi surface.
228:
229: \begin{figure}
230: \includegraphics[width=9cm]{Ek_TB}% Here is how to import EPS art
231: \caption{\label{Ek_TB}(Color) Comparison of spectral weight
232: distributions with calculated band dispersions. (a), (b) and (c)
233: correspond to cuts 1, 2 and 4, respectively. Black dots in (c) are
234: peak positions of MDC's and represents band dispersion.}
235: \end{figure}
236:
237: %Explanation for TB fitting
238: In Fig. \ref{Ek_TB}, we compare the $E$-$k_y$ plots with the
239: energy dispersions expected from the LDA calculation. As seen in
240: the EDC plots of cut 1 and cut 4 in Fig. \ref{EDC_Ek}, there are
241: parabolic energy dispersions with the bottom at $\sim$-0.5 eV,
242: which is shallower than that expected from the LDA calculation
243: $\sim$ -0.9 eV. The two dispersions observed for cut 1 in Fig.
244: \ref{EDC_Ek} may correspond to the dispersions of the $d_{xy}$ and
245: $d_{yz}$ bands, as shown in Fig. \ref{Ek_TB} (a). The dispersion
246: of $d_{zx}$ is not clearly observed in the present plot, probably
247: due to the effect of transition matrix-elements. For cuts 2 and 4
248: [Fig. \ref{Ek_TB} (b) and (c)], since they do not intersect the
249: $d_{yz}$ and $d_{zx}$ bands, we observed only the $d_{xy}$ band.
250:
251: %Discussion of mass renormalization
252: In panel (c), we have derived the band dispersion for cut 4 from
253: MDC peak positions indicated by black dots. By comparing the Fermi
254: velocity of the LDA band structure and that of the present
255: experiment, one can see the overall band narrowing in the measured
256: band dispersion. Near $E_F$, we obtain the mass enhancement factor
257: $m^*/m_b \sim 1.8\pm 0.2$, which is close to the value $m^*/m_b=
258: 1.98$ obtained from the specific heat coefficient $\gamma$
259: \cite{inouePRB}. Since the LDA calculation predicted that the
260: $d_{xy}$ band is highly two-dimensional and isotropic within the
261: $k_x$-$k_y$ plane, a similar mass enhancement is expected on the
262: entire Fermi surface.
263:
264: %Surface states
265: It has been pointed out that the photoemission spectra of this
266: system taken at low photon energies have a significant amount of
267: contributions from surface states, particularly in the incoherent
268: part centered at $\sim$ -1.5 eV \cite{sekiyama, Maiti}. In the
269: photon energy range used in the present work, indeed, the bulk
270: spectra has a smaller spectral weight compared to that of surface,
271: by a factor of $\sim$2 \cite{Maiti}. However, surface spectra tend
272: to have spectral weight in the incoherent part, reducing their
273: spectral weight near $E_F$. Therefore, surface contributions are
274: thought to be relatively small in the coherent part but may not be
275: negligible. In order to distinguish bulk from surface
276: contributions in the present spectra, we shall discuss the
277: character of surface states.
278:
279: Here, we consider several possible origins of surface states. One
280: is a surface reconstruction, which may create two dimensional
281: electronic states on the first layer different from bulk states.
282: Such reconstruction-derived surface bands were observed in the
283: ARPES spectra of Sr$_2$RuO$_4$. In that case, the surface band was
284: symmetric with respect to the ($\pi$,0)-(0,$\pi$) line due to band
285: folding into the new Brillouin zone of the
286: $\sqrt{2}\times\sqrt{2}$ superstructure. Since such a folded band
287: has not been observed in the present results, the effect of
288: surface reconstruction would not exist or would be negligible in
289: the present ARPES spectra.
290:
291: Another possible origin of surface states is the discontinuity of
292: the potential at the surfaces. Figure \ref{surface_states} shows
293: surface-projected density of states and band dispersion states
294: calculated using a tight-binding Green's function formalism
295: \cite{Kalstein} for a semi-infinite system in the same manner as
296: Ref. \cite{Liebsch}. As shown in panels (a) and (b), the surface
297: $d_{xy}$ band does not show an appreciable change from the bulk
298: band because it has the two-dimensional electronic structure
299: within the $x$-$y$ plane. In contrast, spectral weight of the
300: surface $d_{yz, zx}$ states is redistributed between the bulk
301: $d_{yz, zx}$ band and the Fermi level [Fig.
302: \ref{surface_states}(c)], because $k_z$ is no longer a good
303: quantum number. This indicates that we cannot observe clear
304: dispersions of the surface $d_{yz, zx}$ bands in ARPES.
305: Accordingly, the observed dispersion of the $d_{xy}$ band should
306: be similar to that of the bulk $d_{xy}$ band and would therefore
307: represent the bulk electronic states rather than the surface
308: states.
309:
310: %Conclusion
311: In conclusion, we have directly observed the energy dispersions
312: and the Fermi surfaces of SrVO$_3$ by ARPES. The observed
313: effective mass renormalization in the $d_{xy}$ band near $E_F$ is
314: consistent with the moderate mass enhancement in $\gamma$. From
315: the model calculations of the surface electronic structure, we
316: conclude that the observed dispersion of the $d_{xy}$ band
317: represents the bulk electronic structure. The present study has
318: demonstrated that ARPES measurements are useful even for
319: three-dimensional perovskite systems and provide direct
320: information about the band renormalization in Mott-Hubbard system
321: near MIT's.
322:
323: \begin{figure}
324: \includegraphics[width=9cm]{surface_states}
325: \caption{\label{surface_states} Surface-projected electronic
326: states calculated using the semi-infinite tight-binding model. (a)
327: Density of states. (b) Spectral weight distribution of $d_{xy}$
328: states. (c) Spectral weight distribution of $d_{yz}$ states.}
329: \end{figure}
330:
331: %Acknowledgment
332: We are grateful to M. Rozenberg and I. H. Inoue for enlightening
333: discussions and D. H. Lu for technical support. This work was
334: supported by a Grant-in-Aid for Scientific Research ``Invention of
335: Anomalous Quantum Materials" from the Ministry of Education,
336: Science, Culture, Sports and Technology, Japan. SSRL is operated
337: by the Department of Energy's Office of Basic Energy Science,
338: Division of Chemical Sciences and Material Sciences.
339:
340: \begin{thebibliography}{999}
341: \bibitem{RMPfujimori}M. Imada, A. Fujimori, and Y. Tokura, Rev. of Mod. Phys. \textbf{70}, 1039 (1998).
342: \bibitem{ZSA}J. Zaanen, G. A. Sawatzky, and J. W. Allen, Phys. Rev. Lett. \textbf{55}, 418 (1985).
343: \bibitem{Pavarini} E. Pavarini, S. Biermann, A. Poteryaev, A. I. Lichtenstein, A. Georges, and O. K. Andersen, Phys. Rev. Lett. \textbf{92}, 176403 (2004).
344: \bibitem{inouePES} I. H. Inoue, I. Hase, Y. Aiura, A. Fujimori, Y. Haruyama, T. Maruyama, and Y. Nishihara, Phys. Rev. Lett. \textbf{74}, 2539 (1995).
345: \bibitem{Morikawa} K. Morikawa, T. Mizokawa, K. Kobayashi, A. Fujimori, H. Eisaki, S. Uchida, F. Iga, and Y. Nishihara, Phys. Rev. B \textbf{52}, 13711 (1995).
346: \bibitem{Fujimori} A. Fujimori, I. Hase, H. Namatame, Y. Fujishima, Y. Tokura, H. Eisaki, S. Uchida, K. Takegahara, and F. M. F. de Groot, Phys. Rev. Lett. \textbf{69}, 1796 (1992).
347: \bibitem{Yoshida} T. Yoshida, A. Ino, T. Mizokawa, A. Fujimori, Y. Taguchi, T. Katsufuji, and Y. Tokura, Europhys. Lett. \textbf{59}, 258 (2002).
348: \bibitem{DMFT} A. Georges, G. Kotliar, W. Krauth, and M. J. Rozenberg, Rev. Mod. Phys. \textbf{68}, 13 (1996).
349: \bibitem{Kumagai}K. Kumagai, T. Suzuki, Y. Taguchi, Y. Okada, Y. Fujishima, and Y. Tokura, Phys. Rev. B \textbf{48}, 7636 (1993).
350: \bibitem{GBR}M. C. Gutzwiller, Phys. Rev. A \textbf{137}, A1726
351: (1965).; W. F. Brinkmann and T. M. Rice, Phys. Rev. B \textbf{2}
352: 4302 (1970).
353: \bibitem{inouePRB} I. H. Inoue, O. Goto, H. Makino, N. E. Hussey, and M. Ishikawa, Phys. Rev. B
354: \textbf{58}, 4372 (1998).
355: \bibitem{sekiyama} A. Sekiyama, H. Fujiwara, S. Imada, S. Suga, H. Eisaki, S. I. Uchida, K. Takegahara, H. Harima, Y. Saitoh, I. A. Nekrasov, G. Keller, D. E. Kondakov, A. V. Kozhevnikov, Th. Pruschke, K. Held, D. Vollhardt, and V. I. Anisimov, Phys. Rev. Lett. \textbf{93}, 156402 (2004).
356: \bibitem{Maiti} K. Maiti, P. Mahadevan, and D. D. Sarma, Phys. Rev. Lett. \textbf{80}, 2885 (1998).
357: \bibitem{Liebsch} A. Liebsch, Phys. Rev. Lett. \textbf{90}, 096401 (2003).
358: \bibitem{Takegahara}K. Takegahara, J. Elec. Spectrosc. Relat. Phenom. \textbf{66}, 303 (1994).
359: \bibitem{TBNN} Distant-neighbor hopping couses hybridization
360: between the three $t_{2g}$ bands. However, this has small effect
361: and may be a negligible in the present analysis.
362: \bibitem{inoueDHVA} I. H. Inoue, C. Bergemann, I. Hase, and S. R. Julian, Phys. Rev. Lett. \textbf{88}, 236403 (2002).
363: \bibitem{BG} For the present analysis, we have used EDC's with no dispersive feature
364: outside of the cylindrical Fermi surface as a background. The high
365: background seen in the EDC's may come from the angle-integrated
366: spectral weight due to disorder or inperfection of the irregularly
367: cleaved or fractured part of the surfaces. Nonetheless, we could
368: observe clear structures with energy dispersions on the
369: background. We consider that the dispersive features come from the
370: well cleaved part of the sample surface.
371: \bibitem{Kalstein} D. Kalkstein, P. Soven, Surf. Science \textbf{26}, 85 (1971).
372: \end{thebibliography}
373: \end{document}
374: