1: \documentclass[prl,superscriptaddress,showpacs,twocolumn]{revtex4}

2: 

3: \usepackage{graphicx}

4: 

5: \def\figwidth{8cm}

6: 

7: \def\iomn{i\omega_n}

8: 

9: \newcommand{\mr}[1]{{{\mathrm{#1}}}}

10: 

11: \begin{document}

12: 

13: \title{Observation of Hubbard Bands in $\gamma$-Manganese}

14: 

15: \author{S. Biermann$^{\ast}$}

16: 

17: %\email{biermann@lpt.ens.fr}

18: 

19: \affiliation{Institut f{\"u}r Festk{\"o}rperforschung, 

20: Forschungszentrum J{\"u}lich, D-52425 J{\"u}lich, Germany }

21: 

22: \author{A. Dallmeyer}

23: 

24: \affiliation{Institut f{\"u}r Festk{\"o}rperforschung, 

25: Forschungszentrum J{\"u}lich, D-52425 J{\"u}lich, Germany }

26: 

27: \author{C. Carbone}

28: 

29: \affiliation{Institut f{\"u}r Festk{\"o}rperforschung, 

30: Forschungszentrum J{\"u}lich, D-52425 J{\"u}lich, Germany }

31: 

32: \author{W. Eberhardt}

33: 

34: \affiliation{Institut f{\"u}r Festk{\"o}rperforschung, 

35: Forschungszentrum J{\"u}lich, D-52425 J{\"u}lich, Germany }

36: 

37: \author{C. Pampuch}

38: 

39: \affiliation{BESSY, Albert-Einstein-Stra{\ss}e 15, D-12489 Berlin, Germany}

40: 

41: \author{O. Rader}

42: 

43: \affiliation{BESSY, Albert-Einstein-Stra{\ss}e 15, D-12489 Berlin, Germany}

44: 

45: \author{M. I. Katsnelson}

46: 

47: \affiliation{University of Nijmegen, NL-6525 ED Nijmegen, The Netherlands}

48: \affiliation{Institute of Metal Physics, 620219 Ekaterinburg, Russia} 

49: 

50: \author{A. I. Lichtenstein}

51: 

52: %\email{A.Lichtenstein@sci.kun.nl}

53: 

54: \affiliation{University of Nijmegen, NL-6525 ED Nijmegen, The Netherlands}

55: 

56: \date{\today}

57: 

58: \begin{abstract}

59: 

60: We present angle-resolved photoemission spectra of

61: the $\gamma$-phase of manganese as well as a theoretical

62: analysis using a recently developed approach that combines

63: density functional and dynamical mean field methods (LDA+DMFT). 

64: The comparison of experimental data and theoretical predictions allows

65: us to identify effects of the Coulomb correlations, namely the presence

66: of broad and undispersive Hubbard bands in this system.

67: \end{abstract}

68: 

69: \pacs{71.20.Be,79.60.-i,71.15.Qe}

70: \maketitle

71: 

72: The electronic theory of metals is based on the concept of

73: quasiparticles, elementary excitations in the many-electron

74: system that show a one-to-one correspondence with

75: non-interacting electrons.

76: They are characterized by a

77: dispersion law describing the dependence of their energy on a

78: quasimomentum, which can be measured by angle-resolved

79: photoemission spectroscopy (ARPES) \cite{kevan}. 

80: Hubbard showed for the first time,

81: that strong electronic correlations can destroy this picture and

82: result in the formation of so-called Hubbard bands of essentially

83: many-body nature \cite{hubbard_III}. 

84: This concept is crucial for modern theories

85: of strongly correlated electron systems 

86: \cite{gkkr}. 

87: The formation of Hubbard bands

88: takes place, e.g., in many transition metal-oxide compounds,

89: which thus have to be viewed

90: as Mott insulators or doped Mott insulators 

91: \cite{mott}.

92: Transition metals represent another class of systems where  

93: many-body effects are important (see \cite{LKK}

94: and Refs. therein). However, according to common belief, they are

95: moderately correlated systems and normal Fermi liquids.

96: 

97: Electronic spectra of transition metals

98: have been probed intensively by angle-resolved photoemission.

99: Copper with its filled d-band was the first metal to be investigated

100: thoroughly by this technique and the results were in excellent

101: agreement with band structure calculations 

102: \cite{thiry_photoemission, knapp_cu_arpes}. The same

103: technique, however, showed substantial deviations when applied

104: to Ni and provided evidence for many-body behavior, such as the

105: famous $6$ eV satellite 

106: \cite{huefner_photoemission_ni, guillot}.

107: The quasiparticle damping in iron can

108: be as large as 30 $\%$ of the binding energy 

109: \cite{lda++_fe,himpsel_transition_metals}. 

110: Correlation effects are indeed important for metals with partially 

111: filled 3d bands and should be taken into account for an adequate 

112: description of ARPES spectra.

113: Nevertheless, the main part of the spectral density in Fe is related 

114: to usual quasiparticles, and the spectral weight of the satellite in 

115: Ni amounts to only 20 $\%$

116: \cite{himpsel_transition_metals}.

117: 

118: Investigations of an extended Hubbard model show that

119: correlation effects are strongest for half-filled d-bands \cite{zein}.

120: Normally the geometrical frustrations in crystals (such as in the

121: fcc-lattice) further enhance electronic correlations \cite{gkkr} 

122: so that one of the best candidates among the transition metals 

123: for the search of strong correlation effects is the fcc-($\gamma$) 

124: phase of manganese. 

125: It is an example of a very strongly frustrated magnetic

126: system; according to band-structure calculations \cite{moruzzi} the

127: antiferromagnetic ground state of $\gamma$-Mn lies extremely close to

128: the boundary of the non-magnetic phase. Moreover, an

129: anomalously low value of the bulk modulus \cite{guillermet} might be

130: considered as a first experimental hint of strong electronic

131: correlations.

132: 

133: The physical properties of bulk $\gamma$-Mn are hardly accessible in

134: the experiment, since the $\gamma$-phase is only stable at temperatures

135: between 1368 K and 1406 K, where it shows paramagnetic

136: behavior. Thin films of $\gamma$-Mn, however, can be stabilized by

137: epitaxial growth on Cu$_3$Au(100) \cite{schirmer}, which has an interatomic

138: spacing ($2.65 \AA$) very close to the interatomic spacing of Mn-rich

139: alloys ($2.60-2.68 \AA$). Schirmer et al. have shown that Cu$_3$Au(100)

140: supports layer-by-layer growth at room temperature up to

141: coverages of 20 monolayers (ML) \cite{schirmer}. A low-energy electron

142: diffraction (LEED) I(V) analysis revealed that the Mn films adopt

143: the in-plane spacing of the Cu$_3$Au(100) substrate and a

144: comparatively large tetragonal distortion of the fcc-lattice. For the

145: inner layers of a 16 ML Mn film, this distortion amounts to -6$\%$,

146: whereas the surface-subsurface distance is very close to the

147: Cu$_3$Au value.

148: 

149: We have used angle-resolved photoemission on the undulator

150: beamline TGM-5 and on the TGM-1 beamline at BESSY to probe

151: the electronic states in $\gamma$-Mn. The Cu$_3$Au(100) substrate was

152: prepared by repeated cycles of Ne$^{+}$ sputtering and annealing,

153: until a very good LEED pattern with sharp diffraction spots and a

154: low background intensity confirmed a high degree of structural

155: order. The base pressure of 

156: $2\times10^{-10}$ mbar rose to $7\times10^{-10}$ mbar

157: as Mn was deposited by electron beam evaporation. To avoid

158: interdiffusion of Cu and Au, the onset of which was determined to

159: be above room temperature \cite{schirmer} we used to keep the sample at

160: room temperature during the Mn deposition and the photoemission

161: measurements. The high quality of our Mn samples was routinely

162: verified by means of LEED and Auger spectroscopy.

163: 

164: Angle-resolved photoemission measures the electron spectral

165: density $A({\bf k},E)$ as a function of the quasimomentum $\bf k$ and the

166: energy $E$ multiplied by the Fermi distribution function $f(E)$ 

167: \cite{kevan}. For

168: a given electron emission angle corresponding to a given $\bf k$ the

169: spectral density usually has a well-defined maximum as a function

170: of $E$ that determines the quasiparticle dispersion $E({\bf k})$ for the

171: occupied part of the electronic bands. The experimental data (Fig.~1a) 

172: obtained for $\gamma$-Mn at a photon energy of $34$ eV and for

173: different electron emission angles, however, are characterized by two

174: striking features. These are a weakly dispersive quasiparticle band

175: near the Fermi level $E_F$ and a broad and almost {\bf k}-independent

176: maximum at approximately $2.5$ eV below $E_F$%

177: \footnote{The $2.5$ eV feature is insensitive to 

178: oxygen adsoption and therefore

179: not related to surface-localized states.}. 

180: These structures lack

181: a significant dispersion also in spectra taken in normal electron

182: emission for photon energies from $14$ to $70$ eV (Fig.~2). This

183: behavior clearly distinguishes $\gamma$-Mn from other transition

184: metals investigated with angle-resolved photoemission, which

185: are strongly dispersive \cite{rader}. 

186: 

187: \begin{figure}

188: \centerline{\includegraphics[width=9cm,height=10cm]{Fig1.ps}}

189: \caption{\label{fig1}

190: Experimental photoemission spectra taken at a photon energy

191: of $34$ eV for different electron emission angles (a) in comparison

192: with the spectral function $A({\bf k},\omega)$ of $\gamma$-Mn as 

193: calculated within the LDA+DMFT approach (b). The {$\bf k$}-values 

194: corresponding to the experimental data vary approximately

195: between the $\Gamma$ and the $L$ point in the Brillouin zone,

196: binding energies are measured with respect to the Fermi energy.

197: Colors from white, yellow, orange, red, brown to black denote

198: increasing intensities. 

199: The blue lines in (b) give the LDA band structure.} 

200: \end{figure}

201: 

202: \begin{figure}

203: \centerline{\includegraphics[width=12cm,height=13cm]{Fig2.ps}}

204: \caption{\label{fig2}

205: ARPES spectra of a $17$-monolayer

206: $\gamma$-Mn film taken in normal emission at photon energies

207: of $14$ to $70 eV$. The lack of dispersion distinguishes $\gamma$-Mn

208: from other transition metals. Note that the spectral changes

209: from $48 eV$ to $52 eV$ are due to resonant transitions between

210: $3p$ and $3d$ states.

211: } 

212: \end{figure}

213: 

214: These data cannot be understood in the framework of a standard

215: quasiparticle picture, since first-principles calculations of the band

216: structure for different magnetic phases of $\gamma$-Mn show an energy

217: dispersion of more than $1.5$ eV \cite{crockford}. 

218: Instead, the overall shape of the

219: experimental spectra is very close to that of the Hubbard model on

220: the metallic side of the Mott transition with a quasiparticle band

221: near the Fermi level and a broad Hubbard band below $E_F$ \cite{georges82}. 

222: 

223: To test this hypothesis we have carried out first-principle (LDA+DMFT) 

224: calculations \cite{ani1,lda++} of the electronic

225: structure of $\gamma$-Mn that include correlation effects in a local but

226: fully dynamical approximation for the electron self-energy. 

227: In this approach a realistic description of the delocalized s and p electrons

228: within the local density approximation (LDA) is supplemented by a term

229: describing the partially localized nature of the $d$-states. 

230: The Hamiltonian thus reads

231: \begin{eqnarray}\label{LDAUham}

232: H &=& H^{LDA}

233: +\frac 12\sum_{imm^{\prime }\sigma }U_{mm^{\prime }}^in_{im\sigma

234: }n_{im^{\prime }-\sigma }  \nonumber \\

235: &+&\frac 12\sum_{im\neq m^{\prime }\sigma }(U_{mm^{\prime }}^i-J_{mm^{\prime

236: }}^i)n_{im\sigma }n_{im^{\prime }\sigma },

237: \end{eqnarray}

238: where $a_{im\sigma}^{+}$ [$a_{im\sigma }$]

239: creates [destroys] an electron with spin

240: $\sigma$ in state $m$ at site $i$ and 

241: $n_{im\sigma }=a_{im\sigma}^{+} a_{im\sigma }$

242: is the corresponding number operator.

243: $U_{mm^{\prime }}$ and $J_{mm^{\prime}}$ are the direct

244: and exchange term of the screened Coulomb interaction:

245: $U_{mm^{\prime }} =\langle mm^{\prime }|V_{scr}({\bf r-r}^{\prime

246: })|mm^{\prime }\rangle$

247: and 

248: $J_{mm^{\prime }} =\langle mm^{\prime }|V_{scr}({\bf r-r}^{\prime

249: })|m^{\prime }m \rangle$, which can be expressed in terms of the

250: average Coulomb and exchange interaction parameters $U$ and $J$,

251: the values of which are known ($U\sim 3$eV, $J\sim 0.9$ eV)%

252: \footnote{We also performed calculations for other values of $U$ ($4$

253:   eV, $5$ eV). Higher $U$ values slightly shift the general features but do not

254:   lead to a qualitatively different behavior.}

255: \footnote{Moreover, we use the fact that for transition metals

256:   the ratio of the Slater integrals $F_2/F_4$ is to a good approximation

257:   constant and equals $0.0625$. Cf. V.~I.~Anisimov, F.~Aryasetiawan,

258:   A.~I.~Lichtenstein, J.~Phys.: Condens.~Matter, {\bf 9} 767 (1997)} 

259: .

260: We use an LDA-LMTO \cite{andersen1} effective Hamiltonian $H^{LDA}$,

261: corrected for double counting of the Coulomb energy of the

262: $d$ states in the usual way \cite{lda++}.

263: In Eq.(\ref{LDAUham}) the sums run over the $3d$

264: states only, whereas in the LDA Hamiltonian, $4s$, $3d$ and $4p$ states

265: are included.

266: The Coulomb interaction term is treated within the

267: dynamical mean-field theory (DMFT) approach, which

268: is the most efficient local approach: it

269: reduces an original many-body lattice problem to the solution of

270: an effective quantum impurity model in a self-consistent electron

271: bath \cite{gkkr}. 

272: This multiband impurity problem has been solved by a

273: numerically exact quantum Monte Carlo scheme %\cite{lda++} 

274: based on the

275: algorithm of Hirsch and Fye \cite{hirsch}.

276: Using 64 or 128 slices in imaginary time allows to reliably access

277: temperatures down to $T \sim 500K$. This is still higher than

278: in the experiment; however, test calculations show that 

279: higher temperatures only result in a slight smoothening

280: of the spectra, therefore justifying the comparison with

281: the experimental data at lower temperature.

282: Typically, about $10^5$ QMC sweeps and 10 to 15 DMFT iterations

283: are sufficient to reach convergence.

284: The main quantities that we calculate and analyze are: a) the

285: local Green's function, b) the {\bf $k$}-resolved local Green's function

286: \begin{eqnarray}\label{integralgleichung}

287: \hat{G}({\bf k}, \tau) = %%%%%%\sum_{{\bf k}} 

288: \frac{1}{\beta} \sum_n e^{-i \omega_n \tau}

289: \left(i \omega_n + \mu - \hat{H}^{LDA}({\bf k}) - 

290: \hat{\Sigma}(i \omega) \right)^{-1}

291: \end{eqnarray}

292: where $\omega_n$ are the Matsubara frequencies corresponding to

293: the inverse temperature $\beta$.

294: Inversion of the spectral representations of these functions 

295: by means of 

296: a Maximum Entropy scheme \cite{jarrell}

297: yields the density of states (DOS) $\rho(\omega)$

298: and the spectral function $A({\bf k},\omega)$.

299: To our knowledge these calculations are the first ones that determine

300: the ${\bf k}$-dependence 

301: of the spectral density for a material with $d$-states from 

302: LDA+DMFT with a realistic five-band Coulomb vertex.

303: 

304: \begin{figure}

305: \centerline{\includegraphics[width=\figwidth,height=10cm]{Fig3.ps}}

306: \caption{\label{fig3}

307: Upper panel: 

308: density of states of $\gamma$-Mn as calculated within LDA+DMFT.

309: The ``three-peak structure'' with the two broad Hubbard bands (HB)

310: and a narrow quasiparticle (QP) Kondo resonance at the Fermi level

311: (solid line) is typical of strongly correlated systems. The 

312: calculated photoemission spectrum (dashed line), i.e. the

313: density of states multiplied with the Fermi function and broadened

314: with the experimental resolution, shows reasonable agreement with

315: the experimental spectra Fig.~1a and Fig.~2.

316: Lower panel: ${\bf k}$-resolved density of states [arbitrary units]

317: as calculated within LDA+DMFT.

318: The different curves correspond to {\bf $k$}-points between

319: the $\Gamma$ and the $L$-point.

320: }

321: \end{figure}

322: 

323: The results are shown in Fig.~1b.

324: For a given k-point there are two energy regions that carry the main

325: part of the spectral weight: one narrow quasi-particle (QP) 

326: feature near the Fermi

327: level and a very broad Hubbard band at about $-2.5$ eV. 

328: Given the facts that (i) the experiments are done at a somewhat lower

329: temperature than the calculations, that (ii) we do not take into

330: account matrix elements for interpreting the photoemission data

331: and that (iii) using the Maximum Entropy scheme for determining the

332: spectral function, a quantity {\it not} directly measured within

333: the Quantum Monte Carlo simulations, introduces a further approximation,

334: the theoretical spectral function agrees reasonably well with the

335: experimental data (Fig.~1a). 

336: Also plotted in Fig.~1b are the Kohn-Sham eigenvalues taken from

337: the LDA calculation. The absence of LDA bands in the energy 

338: region carrying most of the spectral weight in this {$\bf k$}-space

339: direction is striking and underlines the necessity of a proper

340: many body treatment as done in LDA+DMFT. Note that assuming

341: antiferromagnetic order (of the type detailed below) would slightly

342: shift the LDA bands. However, the antiferromagnetic LDA band

343: structure displays a dispersion of more than $2$ eV and could thus

344: not explain the undispersive photoemission feature.

345: 

346: The calculated ({$\bf k$}-integrated and {$\bf k$}-resolved) 

347: density of states curves

348: (Fig.~3) demonstrate a characteristic "three-peak structure", with

349: two broad Hubbard bands and a narrow quasiparticle Kondo

350: resonance at the Fermi level which is typical of strongly correlated

351: electron systems \cite{gkkr}. The quasiparticle peak at the Fermi level and

352: the lower Hubbard band are seen in the present ARPES spectra; 

353: in k-unresolved (BIS) measurements \cite{speier}

354: a broad peak has been observed at $1.4$ eV. To identify this peak

355: with the upper Hubbard band (located at $1.2$ eV in our calculations)

356: one should prove the dispersionless nature of this peak. We have

357: checked that all these incoherent features do not depend on the

358: directions in {$\bf k$}-space used in our calculations. For the above

359: reasons we believe that $\gamma$-Mn belongs to the class of strongly

360: correlated materials and that the ARPES

361: data can be considered as the first observation of Hubbard bands

362: in a transition metal. 

363: As discussed above, correlation effects are 

364: indeed observed in other transition metals,

365: e.g. the Ni satellite or some broadening of the quasi-particle

366: bands in Fe. Still, even if the mechanism leading to these features

367: is of the same origin, their spectral weight is not comparable to

368: the weight of the Hubbard bands in $\gamma$-manganese.

369: 

370: The energy scale associated with the correlation effects 

371: that lead to the formation of the Hubbard bands ($\sim U$) 

372: is much larger than that of the magnetic interactions.

373: Therefore the observed effects are not very sensitive to long-range

374: magnetic order. 

375: We have carried out the electronic structure

376: calculations for both the paramagnetic and the antiferromagnetic

377: structure with wave vector Q=($\pi$,0,0), which is typical of

378: $\gamma$-Mn-based alloys \cite{fishman}. 

379: The magnetic ordering changes the electron spectrum

380: little in comparison with the nonmagnetic case. However, in

381: comparison with the results of standard band theory \cite{moruzzi}, the

382: correlation effects stabilize the antiferromagnetic structure leading

383: to a magnetic moment of about 2.9 $\mu_B$. 

384: 

385: According to the present results, $\gamma$-Mn can be considered a

386: unique case of a strongly correlated transition metal. An even

387: larger correlation would transform the system to a Mott insulator

388: where every atomic multiplet forms its own narrow but dispersive

389: Hubbard band \cite{hubbard_III,mott}.

390: On the other hand, in most metals correlations are

391: small enough for the quasiparticles to be

392: well-defined in the whole energy region and usual band theory

393: gives a reasonable description of the energy dispersion.

394: Note that the correlation strength and bandwidth

395: have almost the same magnitude for all $3d$ metals.

396: $\gamma$-Mn is

397: probably an exceptional case among the transition elements due to

398: the half-filled d-band and geometric frustrations in the

399: fcc-structure.

400: 

401: In conclusion, our ARPES data for the $\gamma$-phase

402: of manganese and their theoretical analysis 

403: by means of LDA+DMFT, an approach that accounts

404: not only for band structure effects on the LDA level but also

405: allows for a full description of local effects of strong 

406: Coulomb correlations, provide evidence for the formation of

407: Hubbard bands in metallic manganese. This is a qualitatively

408: new aspect in the physics of transition metals.

409: 

410: \begin{acknowledgments}

411: Acknowledgements: 

412: This research has been supported by

413: a grant of supercomputing time at

414: NIC J{\"u}lich and by Netherlands Organization 

415: for Scientific Research (NWO project 047-008-16).

416: 

417: $\ast$ Present address: 

418: LPS, CNRS-UMR 8502, UPS B\^at. 510, 91405 Orsay, France

419: 

420: \end{acknowledgments}

421: 

422: \begin{thebibliography}{28}

423: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi

424: \expandafter\ifx\csname bibnamefont\endcsname\relax

425:   \def\bibnamefont#1{#1}\fi

426: \expandafter\ifx\csname bibfnamefont\endcsname\relax

427:   \def\bibfnamefont#1{#1}\fi

428: \expandafter\ifx\csname citenamefont\endcsname\relax

429:   \def\citenamefont#1{#1}\fi

430: \expandafter\ifx\csname url\endcsname\relax

431:   \def\url#1{\texttt{#1}}\fi

432: \expandafter\ifx\csname urlprefix\endcsname\relax\def\urlprefix{URL }\fi

433: \providecommand{\bibinfo}[2]{#2}

434: \providecommand{\eprint}[2][]{\url{#2}}

435: 

436: \bibitem[{\citenamefont{Kevan}(1992)}]{kevan}

437: \bibinfo{author}{\bibfnamefont{S.~D.} \bibnamefont{Kevan}},

438:   \emph{\bibinfo{title}{Angle-resolved photoemission: Theory and current

439:   applications}}, vol.~\bibinfo{volume}{74} (\bibinfo{publisher}{Elsevier},

440:   \bibinfo{address}{Amsterdam}, \bibinfo{year}{1992}).

441: 

442: \bibitem[{\citenamefont{Hubbard}(1964)}]{hubbard_III}

443: \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Hubbard}},

444:   \bibinfo{journal}{Proc. R. Soc. A} \textbf{\bibinfo{volume}{281}},

445:   \bibinfo{pages}{401} (\bibinfo{year}{1964}).

446: 

447: \bibitem[{\citenamefont{Georges et~al.}(1996)\citenamefont{Georges et al.}}]{gkkr}

448: \bibinfo{author}{\bibfnamefont{For a review see A.}~\bibnamefont{Georges et al.}},

449:   \bibinfo{journal}{Rev. Mod. Phys.} \textbf{\bibinfo{volume}{68}},

450:   \bibinfo{pages}{13} (\bibinfo{year}{1996}).

451: 

452: \bibitem[{\citenamefont{Mott}(1974)}]{mott}

453: \bibinfo{author}{\bibfnamefont{N.~F.} \bibnamefont{Mott}},

454:   \emph{\bibinfo{title}{Metal-Insulator Transitions}}

455:   (\bibinfo{publisher}{Taylor and Francis}, \bibinfo{address}{London},

456:   \bibinfo{year}{1974}).

457: 

458: \bibitem[{\citenamefont{Lichtenstein et~al.}(2001)\citenamefont{Lichtenstein,

459:   Katsnelson, and Kotliar}}]{LKK}

460: \bibinfo{author}{\bibfnamefont{A.~I.} \bibnamefont{Lichtenstein}},

461:   \bibinfo{author}{\bibfnamefont{M.~I.} \bibnamefont{Katsnelson}},

462:   \bibnamefont{and} \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Kotliar}},

463:   \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{87}},

464:   \bibinfo{pages}{067205} (\bibinfo{year}{2001}).

465: 

466: \bibitem[{\citenamefont{Thiry et~al.}(1979)}]{thiry_photoemission}

467: \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Thiry}} 

468:   \bibinfo{author}{\bibnamefont{et~al.}}, \bibinfo{journal}{Phys. Rev. Lett.}

469:   \textbf{\bibinfo{volume}{43}}, \bibinfo{pages}{82} (\bibinfo{year}{1979}).

470: 

471: \bibitem[{\citenamefont{Knapp et~al.}(1979)\citenamefont{Knapp, Himpsel, and

472:   Eastman}}]{knapp_cu_arpes}

473: \bibinfo{author}{\bibfnamefont{J.~A.} \bibnamefont{Knapp}},

474:   \bibinfo{author}{\bibfnamefont{F.~J.} \bibnamefont{Himpsel}},

475:   \bibnamefont{and} \bibinfo{author}{\bibfnamefont{D.~E.}

476:   \bibnamefont{Eastman}}, \bibinfo{journal}{Phys. Rev. B}

477:   \textbf{\bibinfo{volume}{19}}, \bibinfo{pages}{4952} (\bibinfo{year}{1979}).

478: 

479: \bibitem[{\citenamefont{H{\"u}fner and

480:   Wertheim}(1975)}]{huefner_photoemission_ni}

481: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{H{\"u}fner}} \bibnamefont{and}

482:   \bibinfo{author}{\bibfnamefont{G.~K.} \bibnamefont{Wertheim}},

483:   \bibinfo{journal}{Phys. Lett. A} \textbf{\bibinfo{volume}{51}},

484:   \bibinfo{pages}{299} (\bibinfo{year}{1975}).

485: 

486: \bibitem[{\citenamefont{Guillot et~al.}(1979)\citenamefont{Guillot et al.}}]{guillot}

487: \bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Guillot et al.}},

488:   \bibinfo{journal}{Phys. Rev. Lett.}

489:   \textbf{\bibinfo{volume}{39}}, \bibinfo{pages}{1632} (\bibinfo{year}{1977}).

490: 

491: \bibitem[{\citenamefont{Katsnelson and Lichtenstein}(1999)}]{lda++_fe}

492: \bibinfo{author}{\bibfnamefont{M.~I.} \bibnamefont{Katsnelson}}

493:   \bibnamefont{and} \bibinfo{author}{\bibfnamefont{A.~I.}

494:   \bibnamefont{Lichtenstein}}, \bibinfo{journal}{J.~Phys.: Condens.~Matter}

495:   \textbf{\bibinfo{volume}{11}}, \bibinfo{pages}{1037} (\bibinfo{year}{1999}).

496: 

497: \bibitem[{\citenamefont{Himpsel et~al.}(1981)\citenamefont{Himpsel, Heimann,

498:   and Eastman}}]{himpsel_transition_metals}

499: \bibinfo{author}{\bibfnamefont{F.~J.} \bibnamefont{Himpsel}},

500:   \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Heimann}}, \bibnamefont{and}

501:   \bibinfo{author}{\bibfnamefont{D.~E.} \bibnamefont{Eastman}},

502:   \bibinfo{journal}{J. Applied Phys.} \textbf{\bibinfo{volume}{52}},

503:   \bibinfo{pages}{1658} (\bibinfo{year}{1981}).

504: 

505: \bibitem[{\citenamefont{Zein}(1995)}]{zein}

506: \bibinfo{author}{\bibfnamefont{N.~E.} \bibnamefont{Zein}},

507:   \bibinfo{journal}{Phys.~Rev.~B} \textbf{\bibinfo{volume}{52}},

508:   \bibinfo{pages}{11813} (\bibinfo{year}{1995}).

509: 

510: \bibitem[{\citenamefont{Moruzzi et~al.}(1989)\citenamefont{Moruzzi, Marcus, and

511:   K{\"u}bler}}]{moruzzi}

512: \bibinfo{author}{\bibfnamefont{V.~L.} \bibnamefont{Moruzzi}},

513:   \bibinfo{author}{\bibfnamefont{P.~M.} \bibnamefont{Marcus}},

514:   \bibnamefont{and} \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Kubler}},

515:   \bibinfo{journal}{Phys.~Rev.~B} \textbf{\bibinfo{volume}{39}},

516:   \bibinfo{pages}{6957} (\bibinfo{year}{1989}).

517: 

518: \bibitem[{\citenamefont{Guillermet and Grimvall}(1989)}]{guillermet}

519: \bibinfo{author}{\bibfnamefont{A.~F.} \bibnamefont{Guillermet}}

520:   \bibnamefont{and} \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Grimvall}},

521:   \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{40}},

522:   \bibinfo{pages}{1521} (\bibinfo{year}{1989}).

523: 

524: \bibitem[{\citenamefont{Schirmer et~al.}(1999)\citenamefont{Schirmer et al.}}]{schirmer}

525: \bibinfo{author}{\bibfnamefont{B.}~\bibnamefont{Schirmer et al.}},

526:   \bibinfo{journal}{Phys.~Rev.~B} \textbf{\bibinfo{volume}{60}},

527:   \bibinfo{pages}{5895} (\bibinfo{year}{1999}).

528: 

529: %\bibitem[\citenamefont{16}]{surface_state}

530: % \bibinfo{note}{The 2.5 eV feature is insensitive to 

531: %oxygen adsorption and therefore

532: %not related to surface-localized states.}

533: 

534: \bibitem[{\citenamefont{Rader and Gudat}(1999)}]{rader}

535: \bibinfo{author}{\bibfnamefont{O.}~\bibnamefont{Rader}} \bibnamefont{and}

536:   \bibinfo{author}{\bibfnamefont{W.}~\bibnamefont{Gudat}},

537:   \bibinfo{volume}{Landolt-B{\"o}rnstein vol.III 23~C2} (\bibinfo{publisher}{Springer Verlag},

538:   \bibinfo{address}{Berlin}, \bibinfo{year}{1999}).

539: 

540: \bibitem[{\citenamefont{Crockford et~al.}(1991)\citenamefont{Crockford, Bird,

541:   and Long}}]{crockford}

542: \bibinfo{author}{\bibfnamefont{D.~J.} \bibnamefont{Crockford}},

543:   \bibinfo{author}{\bibfnamefont{D.~M.} \bibnamefont{Bird}}, \bibnamefont{and}

544:   \bibinfo{author}{\bibfnamefont{M.~W.} \bibnamefont{Long}},

545:   \bibinfo{journal}{J.~Phys.: Cond.~Matt.} \textbf{\bibinfo{volume}{3}},

546:   \bibinfo{pages}{8665} (\bibinfo{year}{1991}).

547: 

548: \bibitem[{\citenamefont{Georges and Kotliar}(1992)}]{georges82}

549: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Georges}} \bibnamefont{and}

550:   \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Kotliar}},

551:   \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{45}},

552:   \bibinfo{pages}{6479} (\bibinfo{year}{1992}).

553: 

554: \bibitem[{\citenamefont{Anisimov et~al.}(1997)\citenamefont{Anisimov et al.}}]{ani1}

555: \bibinfo{author}{\bibfnamefont{V.~I.} \bibnamefont{Anisimov et al.}},

556:   \bibinfo{journal}{J.~Phys.: Condens.~Matter} \textbf{\bibinfo{volume}{9}},

557:   \bibinfo{pages}{7359} (\bibinfo{year}{1997}).

558: 

559: \bibitem[{\citenamefont{Lichtenstein and Katsnelson}(1998)}]{lda++}

560: \bibinfo{author}{\bibfnamefont{A.~I.} \bibnamefont{Lichtenstein}}

561:   \bibnamefont{and} \bibinfo{author}{\bibfnamefont{M.~I.}

562:   \bibnamefont{Katsnelson}}, \bibinfo{journal}{Phys.~Rev.~B}

563:   \textbf{\bibinfo{volume}{57}}, \bibinfo{pages}{6884} (\bibinfo{year}{1998}).

564: 

565: \bibitem[{\citenamefont{Andersen}(1975)}]{andersen1}

566: \bibinfo{author}{\bibfnamefont{O.~K.} \bibnamefont{Andersen}},

567:   \bibinfo{journal}{Phys.~Rev.~B} \textbf{\bibinfo{volume}{12}},

568:   \bibinfo{pages}{3060} (\bibinfo{year}{1975}).

569: 

570: \bibitem[{\citenamefont{Hirsch and Fye}(1986)}]{hirsch}

571: \bibinfo{author}{\bibfnamefont{J.~E.} \bibnamefont{Hirsch}} \bibnamefont{and}

572:   \bibinfo{author}{\bibfnamefont{R.~M.} \bibnamefont{Fye}},

573:   \bibinfo{journal}{Phys.~Rev.~Lett.} \textbf{\bibinfo{volume}{56}},

574:   \bibinfo{pages}{2521} (\bibinfo{year}{1986}).

575: 

576: \bibitem[{\citenamefont{Jarrell and Gubernatis}(1996)}]{jarrell}

577: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Jarrell}} \bibnamefont{and}

578:   \bibinfo{author}{\bibfnamefont{J.~E.} \bibnamefont{Gubernatis}},

579:   \bibinfo{journal}{Phys.~Rep.} \textbf{\bibinfo{volume}{269}},

580:   \bibinfo{pages}{133} (\bibinfo{year}{1996}).

581: 

582: \bibitem[{\citenamefont{et~al.}(1984)}]{speier}

583: \bibinfo{author}{\bibfnamefont{W.~Speier} \bibnamefont{et~al.}},

584:   \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{30}},

585:   \bibinfo{pages}{6921} (\bibinfo{year}{1984}).

586: 

587: \bibitem[{\citenamefont{Fishman and Liu}(1999)}]{fishman}

588: \bibinfo{author}{\bibfnamefont{R.~S.} \bibnamefont{Fishman}} \bibnamefont{and}

589:   \bibinfo{author}{\bibfnamefont{S.~H.} \bibnamefont{Liu}},

590:   \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{59}},

591:   \bibinfo{pages}{8681} (\bibinfo{year}{1999}).

592: 

593: \end{thebibliography}

594: 

595: \end{document}

596: 

597: