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

2: 

3: \usepackage{graphicx,color,subfig,amsmath,bm,amssymb}

4: 

5: \begin{document}

6: 

7: \title{Strong electron-phonon coupling of the Fe breathing mode of

8:   LaO$_{1-x}$F$_x$FeAs} 

9: 

10: \author{Helmut Eschrig}

11: 

12: \email{h.eschrig@ifw-dresden.de}

13: \homepage{http://www.ifw-dresden.de/institutes/itf/members/helmut}

14: 

15: \affiliation{IFW Dresden, PO Box 270116, D-0111171 Dresden, Germany}

16: 

17: \begin{abstract}

18:   The electron-phonon coupling of LaO$_{1-x}$F$_x$FeAs is

19:   re-investigated on the basis of density functional theory in local

20:   density approximation. The implications of the $(\pi,\pi)$ nesting of

21:   the Fermi surfaces are carefully studied and found to lead to a

22:   non-standard electron-phonon coupling of the corresponding Fe in-plane

23:   breathing mode, which might make a strong contribution to the high

24:   superconducting transition temperature. The semi-metallic behavior of

25:   the undoped material is also further illuminated.

26: \end{abstract}

27: 

28: \pacs{74.25.Jb,74.25.Kc,74.70.Dd,71.20.-b}

29: \keywords{FeAsLaO$_{1-x}$F$_x$, superconductor, electron-phonon

30:   coupling, Fermi surface nesting}

31: 

32: \maketitle

33: 

34: 

35: The recently discovered \cite{kamihara08.1} layered superconductor

36: LaO$_{1-x}$F$_x$FeAs, $0.03 < x < 0.13$, attracts at present

37: considerable interest due to its comparably high critical temperature

38: $T_c \approx 40$K and possibly high upper critical field $H_{c2} \approx

39: 50$Tesla, due to its similarities to the cuprates, and due to many

40: unusual properties. One peculiarity of all density functional based

41: electronic structure calculations

42: \cite{singh08.1,xu08.1,boeri08.1,mazin08.1,cao08.1,ma08.1} is that,

43: although there is strong Fe-$3d$ As-$4p$ covalency resulting in a total

44: width of the corresponding band complex of about 7eV, in a window of

45: about 2.5 eV around and below the Fermi level the electronic states are

46: of nearly pure Fe-$3d$ character (Fig.~\ref{fig:f1}). Mazin \textit{et

47:   al.}\ \cite{mazin08.1} pointed out that there is a strong Fermi

48: surface (FS) nesting in the undoped material with $q=(\pi,\pi,0)$

49: nesting wave vector between the two hole cylinders around the tetragonal

50: axis $\Gamma-Z$ and the two electron cylinders around the axis $M-A$

51: shifted by $q$ against $\Gamma-Z$ (Fig.~\ref{fig:f5}). This nesting led

52: those and several other authors

53: \cite{cao08.1,ma08.1,kuroki08.1,dong08.1} speculate on an important role

54: of antiferromagnetic correlations with magnetic in-plane superstructure

55: vector $(\pi,\pi)$, an apparently well justified scenario.

56: Electron-phonon (e-p) coupling obtained on the basis of density

57: functional perturbation theory \cite{singh08.1,boeri08.1} yielded only a

58: weak coupling quite not enough to explain the high $T_c$.

59: 

60: However, as is well known \cite{kohn59,migdal58} FS nesting also leads

61: to phonon anomalies, and the corresponding e-p coupling may not be

62: correctly described in lowest order perturbation theory which disregards

63: the degeneracy. Moreover, since metallicity and superconductivity take

64: place in the Fe layers of this quasi-two-dimensional tetragonal material

65: only (Fig.~\ref{fig:f2}), an average of linearized e-p coupling over the

66: whole structure might be misleading.  In the present text we report

67: results of investigation of these points which might be equally

68: important compared to the magnetic correlations to understand the

69: material. It is indeed found that e-p coupling may be quite strong. 

70: 

71: High precision density functional calculations were performed with the

72: local density approximation functional of \cite{perdew.92} using the

73: full-potential local-orbital code \cite{e134} in the version FPLO7-28

74: \cite{FPLO} with its default orbital settings. The experimental lattice

75: constants of undoped LaOFeAs, $a=4.03552$\AA, $c=8.7393$\AA, were used,

76: and the two Wyckoff parameters were optimized to be $z_{As}=0.365,\;

77: z_{La}=0.1435$.

78: 

79: \begin{figure}

80:   \centering

81:   \includegraphics[height=0.9\columnwidth,angle=-90,clip=]{f1.eps}

82:   \caption{(color online) Total and component decomposed densities of

83:     states of LaOFeAs.} 

84:   \label{fig:f1}

85: \end{figure}

86: 

87: \begin{figure}

88:   \centering

89:   \includegraphics[height=\columnwidth,angle=-90,clip=]{f5.eps}

90:   \caption{Bands close to the Fermi level ($\varepsilon_{\text{F}} = 0$)

91:     in the P4/nmm zone.}

92:   \label{fig:f5}

93: \end{figure}

94: 

95: The phonon mode relevant in this respect is the in-plane breathing mode

96: of the Fe square lattice (checkerboard pattern where every other white

97: square is shrunk, Fig.~\ref{fig:f3}, left part). A static displacement

98: of Fe according to this breathing mode reduces the P4/nmm symmetry of

99: the lattice to P4mm by doubling the unit cell and removing the inversion

100: from the space group P4/nmm. (The center of inversion was half-way

101: between two neighboring Fe sites.) Doubling the unit cell of the lattice

102: means folding down the unit cell of the reciprocal lattice, the

103: Brillouin zone. The original point $M: \bm q = (\pi,\pi,0)$ of the

104: P4/nmm lattice (Fig.~\ref{fig:f5}) is folded onto $\Gamma: \bm q' =

105: (0,0,0)$ of the P4mm lattice; the folded band structure from

106: Fig.~\ref{fig:f5} is shown on Fig.~\ref{fig:f6}. For symmetry reasons,

107: at $\bm q=(\pi,\pi,q_3)$ the Fe-mode shown on Fig.~\ref{fig:f3} could

108: only mix with a corresponding oxygen breathing mode; both are, however,

109: expected to interact only very weakly, and we neglect this mixing. The

110: total energy change due to a static displacement (frozen phonon) is

111: shown on Fig.~\ref{fig:f3}, right part.  Despite the FS nesting the mode

112: potential is quite harmonic, the reason for this quadratic dependence on

113: the Fe displacement $u_{\text{Fe}}$ is given below. It corresponds to a

114: frequency of $\omega_{\text{Fe}} = 29.4$meV in good agreement with the

115: upper limit of the metal mode energies obtained in \cite{singh08.1}.

116: Note that there are two equivalent such modes of the Fe atoms as a

117: rotation of the displacement vector by $90^\circ$ in the left part of

118: Fig.~\ref{fig:f3} corresponds just to an in-plane shift of the whole

119: P4mm lattice by $(1/2,0,0)$ which together with a reflection of the

120: $c$-axis is a glide symmetry element of the original P4/nmm lattice.

121: 

122: \begin{figure}

123:   \centering

124:   \includegraphics[width=0.6\columnwidth,clip=]{f2.eps}

125:   \caption{(color online) The P4mm unit cell (Fe breathing mode) of

126:     LaOFeAs.}

127:   \label{fig:f2}

128: \end{figure}

129: 

130: \begin{figure}

131:   \centering

132:   \mbox{

133:   \includegraphics[width=0.3\columnwidth,clip=]{f3.eps}

134:   }

135:   \raisebox{0.35\columnwidth}{

136:   \includegraphics[height=0.55\columnwidth,angle=-90,clip=]{f4.eps}

137:   }

138:   \caption{Left:Displacement pattern of the Fe square lattice for the Fe

139:     breathing mode. The P4mm unit cell is shown (thick lines) and the

140:     P4/nmm unit cell (dashed lines, its center of inversion for the full

141:     structure is half-way between neighboring Fe sites).

142:     Right: Total energy per two formula units vs. Fe displacement of the

143:     Fe breathing mode.}

144:   \label{fig:f3}

145: \end{figure}

146: 

147: \begin{figure}

148:   \centering

149:   \includegraphics[height=\columnwidth,angle=-90,clip=]{f6n.eps}

150:   \caption{(color online) Band deformations due to the Fe breathing mode

151:     in the P4mm zone. The upward dispersing band on the line $\Gamma -

152:     M$ is two-fold degenerate. The inset on the right shows the zoomed

153:     detail from the line $M-\Gamma$.}

154:   \label{fig:f6}

155: \end{figure}

156: 

157: On Fig.~\ref{fig:f6} the bands corresponding to the undistorted lattice

158: and to two static mode amplitudes, respectively, are shown in the

159: downfolded P4mm zone. The common Fermi level is again

160: $\varepsilon_{\text{F}}=0$. Now, the nesting for $\bm q'=0$ is seen as

161: band crossings very close to the Fermi level (observe the energy scale

162: of the figure) on the lines $\Gamma-M$ (and $\Gamma-X$) and $Z-R$ (the

163: latter for $q'_3=\pi$; the new point $M$ of the folded zone corresponds

164: to point $X$ on Fig.~\ref{fig:f5}.) From these bands as function of the

165: displacement $u_{\text{Fe}}$ of the Fe atoms, deformation potentials are

166: derived.  Care is, however, needed since deformation potentials are

167: touchy entities \cite{khan.84}. First of all, they have to be determined

168: from a common Fermi level. In the present case, moreover, reversion of

169: $u_{\text{Fe}}$ is equivalent to an in-plane shift of the whole P4mm

170: lattice by $(1,1,0)$ which is a lattice vector of the original P4/nmm

171: lattice, and hence changes in band energies $\Delta\varepsilon_{\bm k

172:   n}$ must be even in $u_{\text{Fe}}$.  Consequently, well away from

173: band crossings there is no first order deformation potential,

174: $\Delta\varepsilon_{\bm k n} \sim u^2_{\text{Fe}}$. At crossing of

175: folded bands according to the symmetry of the lattice with

176: $u_{\text{Fe}} \neq 0$ present (Fig.~\ref{fig:f6}), gaps open with

177: $\Delta\varepsilon_{\text{gap}} \sim |u_{\text{Fe}}|$. In the vicinity

178: of those crossings one has

179: \begin{equation}

180:   \label{eq:1}

181:   \Delta\varepsilon_{\bm k n} \approx 

182:   \pm\sqrt{(v\Delta\bm k)^2 + (I_{\text{Fe}} u_{\text{Fe}})^2}

183:   \mp |v\Delta\bm k|,

184: \end{equation}

185: where $v$ is the average of the two band velocities at the crossing,

186: $\Delta\bm k$ is the deviation from the crossing point, and

187: $I_{\text{Fe}} u_{\text{Fe}}$ is the deformation potential at the

188: gap:

189: \begin{equation}

190:   \label{eq:2}

191:   \Delta\varepsilon_{\text{gap}} \approx 2I_{\text{Fe}}u_{\text{Fe}}.

192: \end{equation}

193: For the displacement corresponding to one phonon per unit cell with

194: energy $\omega_{\text{Fe}}, \; \langle u^2_{\text{Fe}}\rangle =

195: (2M_{\text{Fe}}\omega_{\text{Fe}})^{-1/2}$ one has $I_{\text{Fe}}\langle

196: u^2_{\text{Fe}}\rangle^{1/2} \approx 20$meV$ \approx

197: 0.7\omega_{\text{Fe}}$. By expanding the square root of (\ref{eq:1}) for

198: $|v\Delta\bm k| = 20$meV, approximately the position of the FSs of the

199: fluor doped material (at lower doping level compared to

200: Fig.~\ref{fig:f8}, where 12.5 p.c. fluor were considered to limit the

201: computational effort), an FS-averaged e-p coupling

202: constant

203: \begin{equation}

204:   \label{eq:3}

205:   g = \overline{g_{\bm kn,\bm k'n'}} = \frac{1}{2}\,

206:   \frac{I^2_{\text{Fe}} \langle u^2_{\text{Fe}}\rangle}{20\text{meV}}

207:   \approx 10\text{meV} \approx 

208:   I_{2,\text{Fe}}\langle u^2_{\text{Fe}}\rangle

209: \end{equation}

210: is obtained, where $I_{2,\text{Fe}}$ is the corresponding second order

211: deformation potential. Of course, whenever a phonon mode connects two

212: points on the FS, this gaping with lattice displacement happens.

213: However, normally it happens just for those two points without

214: statistical weight. The peculiarity here is that, as a consequence of the

215: nesting, for one mode it happens in the vicinity of the whole FS in a

216: layer of thickness $\sim\omega_\nu$ in energy (cf. Fig.~\ref{fig:f7}).

217: 

218: In the standard theory the e-p coupling strength $\lambda = Z(0) - 1 =

219: i(\partial\Sigma/\partial\varepsilon)_{\varepsilon = 0}$ is obtained

220: from the e-p self-energy

221: \begin{equation}

222:   \label{eq:4}

223:   \Sigma(\varepsilon) = \sum_{n\nu}\int\frac{d\omega}{2\pi}

224:   \int\frac{d^3k}{(2\pi)^3} g^2 

225:   \frac{2\omega_\nu}{\omega^2 + \omega^2_\nu} 

226:   G_n(\varepsilon-\omega,\bm k)

227: \end{equation}

228: ($n$ being the band index and $\nu$ the phonon mode index) by treating

229: the phonon quantities as nearly $\bm k$-independent and using the $\bm

230: k$-integrated electron Green's function

231: \begin{multline}

232:   \label{eq:5}

233:   \sum_n \int\frac{d^3k}{(2\pi)^3} G_n(\varepsilon - \omega,\bm k) =\\

234:   = N(0) \int_{-\varepsilon_c}^{\varepsilon_c} d\varepsilon_c 

235:   \frac{1}{i(\varepsilon - \omega) - \varepsilon_c} =\\

236:   = -i\pi N(0)\text{sign}(\varepsilon - \omega) + O(1/\varepsilon_c),

237: \end{multline}

238: where $N(0)$ is the electron density of states (DOS) at the Fermi level,

239: and the high-energy cutoff $\varepsilon_c$ is considered negligible

240: small (of the order of Migdal's parameter). Then, the derivative of sign

241: yields a delta function and the $\omega$-integration is readily done

242: yielding $\lambda = \sum_\nu 2N(0)g^2/\omega_\nu$.

243: 

244: \begin{figure}

245:   \centering

246:   \includegraphics[height=\columnwidth,angle=-90,clip=]{f7.eps}

247:   \caption{(color online) Deformation of the total DOS due to the Fe

248:     breathing mode (states per two formula units).} 

249:   \label{fig:f7}

250: \end{figure}

251: 

252: In the present case, which is not subject to Migdal's theorem, this

253: approach is not allowed since the renormalized electron Green's

254: functions have a low energy structure at $\eta_c \approx 20$meV, the

255: distance of the gap edges from Fermi level, which causes additional

256: terms in (\ref{eq:5}). Moreover, Eq.~(\ref{eq:3}) represents a

257: two-phonon vertex (four-leg vertex), and the corresponding contribution

258: to the electron self-energy has two phonon propagators connecting the

259: two vertices and an additional $\omega$-integration in the self-energy,

260: and without the situation with the gaps one would again be left with

261: $\lambda \approx \sum_\nu N(0)g^2/\omega_\nu$, where only $g$ is now the

262: coupling strength (\ref{eq:3}) of the four-leg vertex, which in the

263: present case is of the same order of magnitude as the usual $g$ of an

264: ordinary three leg vertex. However, due to the gap structure all

265: frequency integrations are entangled, and resonances of the type

266: $1/(\omega_\nu - \eta_c)$ may appear in the course of self-consistency

267: for $G_n$ \cite{newns.07}. A quantitative theory for this case is

268: complicated and not yet well developed. Non of the approximate

269: assumptions leading to (\ref{eq:4}) and (\ref{eq:5}) can be maintained.

270: So far one could only speculate on the magnitude of the effects.

271: 

272: Now, the nesting is considered in more detail. A closer look on

273: Fig.~\ref{fig:f6} reveals that a displacement $u_{\text{Fe}} =

274: 0.0645$\AA, which does not much exceed the average displacement of one

275: phonon per mode, already removes two of the four FSs of the in-plane

276: bands (Fe-$3d_{xz}$ and Fe-$3d_{yz}$ orbital character) by opening a gap

277: of about 75meV. Only an electron FS of the shape of a cylindric shell

278: around the tetragonal axis, quite isotropic perpendicular to it,

279: survives (and the hole pocked around point $Z$ of the band of

280: Fe-$3d_{z^2}$ orbital character). Just for illustration the effect of a

281: (not quite realistic) two times larger displacement is also plotted on

282: Fig.~\ref{fig:f6} for which the material becomes a semi-metal with only

283: two tiny FS pockets, an electron ring and a hole pocket, both around

284: $Z$. Even slightly larger displacements would make it a true

285: semiconductor. This is reflected in the behavior of the total DOS shown

286: on Fig~\ref{fig:f7} (where the normalization is for the unit cell of the

287: undistorted P4/nmm structure). One important point is that the nesting

288: is only approximate and, more importantly, that at a given energy the

289: bands which were originally electron or hole bands move in the same

290: direction. Since the band crossings are not exactly at Fermi level and

291: in e-p coupling theory the Fermi level has to be kept fixed

292: \cite{khan.84}, the considered e-p coupling is transmitted to all bands

293: and is also transmitted to the band of Fe-$3d_{z^2}$ character. These

294: findings (together with the similar magnetic peculiarities

295: \cite{mazin08.1}) are capable of explaining many of the unusual

296: electronic properties of the undoped material.

297: 

298: Finally, the effect of fluor doping for oxygen is studied by replacing

299: one of eight oxygen atoms with a fluor atom in an ordered way. This

300: reduces the symmetry of the lattice further to the orthorhombic Pmm2

301: group with a four times larger unit cell, square in plane, compared to

302: the P4/nmm structure. The corresponding bands (again folded, into the

303: Pmm2 zone, which however does not results in new shifts of the FSs which

304: are already close enough to the central line $\Gamma - Z$) and total DOS

305: are shown on Figs.~\ref{fig:f8} and \ref{fig:f9}. As can be seen by

306: comparing Figs.~\ref{fig:f6} and \ref{fig:f7} with Figs.~\ref{fig:f8}

307: and \ref{fig:f9}, one fluor atom donates just one electron without

308: changing the Fe derived $t_{2g}$-bands $(d_{xz},\;d_{yz})$ in the

309: vicinity of the Fermi level.  Therefore, also no disorder effect of

310: fluor doping is expected in the relevant electronic structure. Moreover,

311: the Fe-$3d_{z^2}$ derived bands, which shift a little bit due to the

312: change of layer charges, are now completely filled (one of them

313: producing the DOS peak at about -150meV), and the corresponding hole

314: pocket around $Z$ has gone, leaving only the four slightly deformed

315: cylindrical FSs, two larger electron cylinders around the $\Gamma-Z$

316: axis and two smaller hole cylinders around the $M-A$ axis of the

317: original P4/NMM structure (cf.\ Fig.~\ref{fig:f5}).

318: 

319: \begin{figure}

320:   \centering

321:   \includegraphics[height=\columnwidth,angle=-90,clip=]{f8.eps}

322:   \caption{Bands of LaO$_{0.875}$F$_{0.125}$FeAs in the Pmm2 zone.}

323:   \label{fig:f8}

324: \end{figure}

325: 

326: \begin{figure}

327:   \centering

328:   \includegraphics[height=0.8\columnwidth,angle=-90,clip=]{f9.eps}

329:   \caption{Total DOS of ordered LaO$_{0.875}$F$_{0.125}$FeAs (states per

330:     two formula units).}

331:   \label{fig:f9}

332: \end{figure}

333: 

334: In summary, we have shown that the e-p coupling in LaO$_{1-x}$F$_x$FeAs

335: is indeed very strong due to the Fe in-plane breathing mode, and this

336: cannot be treated in standard e-p theory of superconductivity. With

337: respect to this breathing mode, the material rather compares with

338: MgB$_2$, although the details are quite different, and the magnetic

339: couplings already discussed in the literature must also be taken

340: seriously into account. In that respect, the material may have much in

341: common with the cuprates. A principally similar but in detail quite

342: different case is a breathing mode of the copper square lattice in

343: connection with the approximate $(\pi,\pi)$-nesting \cite{newns.07}. It

344: is intriguing hence, whether LaOFeAs could provide some bridge between

345: the other two cases, MgB$_2$ and the cuprates. The undoped material

346: appears to be a bad metal with a great complexity of very strong

347: couplings, e-p and magnetic.  Doping (potentially both electron or hole)

348: moves LaOFeAs away from this messy coupling situation into a still

349: strong coupled but good metallic regime.  It is to be expected from

350: former \cite{mazin08.1} and the present analysis that LaOFeAs will be

351: more symmetric with respect to both electron and hole doping than the

352: cuprates.

353: 

354: 

355: \begin{acknowledgments}

356:   Helpful discussion with Klaus Koepernik, Stefan-Ludwig Drechsler, B.

357:   B\"uchner and in particular with M. Eschrig are gratefully

358:   acknowledged. I.I. Mazin pointed out to me a substantial error in an

359:   earlier version of the paper.

360: \end{acknowledgments}

361: 

362: \bibliography{../../bib/sc.bib,../../bib/part.bib,../../bib/eschrig.bib,%

363: ../../bib/dft.bib}

364: 

365: \end{document}

366: 

367: %%% Local Variables: 

368: %%% mode: latex

369: %%% TeX-master: t

370: %%% End: 

371: