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

2:  

3: \usepackage{amsmath}

4: \usepackage{bm}

5: \usepackage{graphicx}

6: 

7: \newcommand{\cuo}{CuO$_2$}

8: \newcommand{\bp}{\ensuremath{\mathbf{p}}}

9: \newcommand{\eps}{\ensuremath{\varepsilon}}

10: \newcommand{\Eref}[1]{Eq.~(\ref{#1})}

11: \newcommand{\Fref}[1]{Fig.~\ref{#1}}

12: 

13: 

14: \begin{document}

15: 

16: 

17: \title{The $3d$-to-$4s$-by-$2p$ highway to superconductivity in cuprates}

18: 

19: \author{T.~M.~Mishonov} 

20: %

21: \affiliation{Laboratorium voor Vaste-Stoffysica en Magnetisme,

22: Katholieke Universiteit Leuven, Celestijnenlaan 200~D, B-3001 Leuven,

23: Belgium}

24: %

25: \affiliation{Faculty of Physics, Sofia University ``St. Kliment

26: Ohridski'', 5~J.~Bourchier~Blvd., 1164~Sofia, Bulgaria}

27: %

28: \author{J.~O.~Indekeu}

29: %

30: \affiliation{Laboratorium voor Vaste-Stoffysica en Magnetisme,

31: Katholieke Universiteit Leuven, Celestijnenlaan 200~D, B-3001 Leuven,

32: Belgium}

33: %

34: \author{E.~S.~Penev}

35: %

36: \affiliation{Faculty of Physics, Sofia University ``St. Kliment

37: Ohridski'', 5~J.~Bourchier~Blvd., 1164~Sofia, Bulgaria}

38: 

39: 

40: \date{\today}

41: 

42: \begin{abstract}

43: High-temperature superconductors are nowadays found in great variety

44: and hold technological promise. It is still an unsolved mystery that

45: the critical temperature $T_c$ of the basic cuprates is so high. The

46: answer might well be hidden in a conventional corner of theoretical

47: physics, overlooked in the recent hunt for exotic explanations of new

48: effects in these materials.  A forgotten intra-atomic $s$-$d$

49: two-electron exchange in the Cu atom is found to provide a strong

50: ($\sim$~eV) electron pairing interaction. A Bardeen-Cooper-Schrieffer

51: approach can explain the main experimental observations and predict

52: the correct $d_{x^2- y^2}$ symmetry of the gap.

53: \end{abstract}

54: 

55: \pacs{}

56: 

57: \maketitle

58: 

59: The discovery of high-temperature superconductivity~\cite{Bednorz:86}

60: in cuprates and the subsequent ``research rush'' have led to the

61: appearance of about 100,000 papers to

62: date~\cite{Ginzburg:00}. Virtually every fundamental process known in

63: condensed matter physics was probed as a possible mechanism of this

64: phenomenon. Nevertheless, none of the theoretical efforts resulted in

65: a coherent picture~\cite{Ginzburg:00}. For the conventional

66: superconductors the mechanism was known to be the interaction between

67: electrons and crystal-lattice vibrations, but the development of its

68: theory lagged behind the experimental findings.  The case of cuprate

69: high-$T_c$ superconductivity appears to be opposite: we do not

70: convincingly know which mechanism is to be incorporated in the

71: traditional Bardeen-Cooper-Schrieffer (BCS) theory~\cite{BCS:57}. Thus

72: the path to high-$T_c$ superconductivity in cuprates, perhaps

73: carefully hidden or well-forgotten, has turned into one of the

74: long-standing mysteries in physical science.

75: 

76: In contrast with all previous proposals, we advance the exchange of

77: two electrons between the $4s$ and $3d_{x^2-y^2}$ shells of the Cu

78: atom as the origin of high-$T_c$ superconductivity in the layered

79: cuprates and show that the basic spectroscopic, thermodynamic and

80: kinetic experiments can be explained by it. This process has been

81: familiar since the dawn of quantum mechanics~\cite{Heisenberg:28} and

82: is known to be responsible for the magnetism of transition metals. Its

83: energetic characteristic is referred to as the $s$-$d$ exchange

84: amplitude (or integral) $J_{sd}.$ The gist of this two-electron

85: correlation is sketched in \Fref{fig:1}. The question then naturally

86: arises as to how this exchange process can trigger pairing between

87: electrons with opposite momenta and spins, the so-called Cooper pairs.

88: 

89: The underlying idea is analogous to Heitler and London's

90: reasoning~\cite{Heitler:27} which led to their celebrated model for

91: the valence bonding of the H$_2$ molecule: The strong electron

92: correlation brought about by the ubiquitous Coulomb repulsion lowers

93: the energy when the two electrons have opposite spins (the singlet

94: state). In quantum mechanics, according to the Hellmann-Feynman

95: theorem, this energy decrease ``drives'' the attractive interatomic

96: force. Reverting to the Cu atom, \Fref{fig:1}, one can think of the

97: $s$-$d$ exchange as an ``intra-atomic valence bond''. We shall further

98: argue that it is the specific crystal structure of the cuprates that

99: renders this intensive process macroscopically observable,

100: ``disguised'' as the phenomenon of superconductivity for temperatures

101: lower than some critical value $T < T_c.$

102: 

103: \begin{figure}[t]

104: \includegraphics[width=\columnwidth]{./fig1.eps}

105: \caption{Pairing two-electron exchange amplitude $J_{sd}$

106: ``hidden'' in the Cu atom.  (a) Classical Bohr-Sommerfeld

107: representation of the $s$-$d$ two-electron exchange process. The inset

108: shows how the Coulomb scattering leads to an effective electron spin

109: exchange.  (b) Electron charge distribution for Cu~$4s$ and Cu~$3d$

110: orbitals: the dashed line marks the Cu-O distance in the \cuo\ plane.

111: \label{fig:1}

112: }

113: \end{figure}

114: 

115: Superconducting cuprates have as their main structural detail the

116: copper-oxygen plane shown in \Fref{fig:2}~(a). It has previously been

117: demonstrated~\cite{Mishonov:00} that its normal-phase electronic

118: properties can be understood on the basis of Bloch's tight-binding

119: formalism (chemists would call it an extended H\"uckel method) applied

120: to the set of atomic orbitals indicated in \Fref{fig:2}~(b).  This

121: plane behaves much like a two-dimensional metal. Electronic Bloch

122: states of quasi-momentum $\bp=(p_x,p_y) \in (0, 2\pi)$ are described

123: by a four-component tight-binding wave function $\psi_{\bp} =

124: (D_{\bp},S_{\bp},X_{\bp},Y_{\bp}).$ At $T=0$ these states are occupied

125: up to the Fermi energy $E_F.$ The corresponding energy level is

126: comprised in a single conduction band $\eps_{\bp}$ of dominant

127: Cu~$3d_{x^2-y^2}$ character, \Fref{fig:3}~(a). Thus, the amplitude for

128: a conduction electron to be Cu~$4s$ is $S_{\bp}.$ $D_{\bp}$

129: corresponds to Cu~$3d_{x^2- y^2},$ $X_{\bp}$ to O~$2p_x$ and $Y_{\bp}$

130: to O~$2p_y$, \Fref{fig:2}~(b).

131: 

132: The remarkable success of the \cuo\ plane in mediating high-$T_c$

133: superconductivity is due to the following reasons. (i) The

134: quasi-two-dimensional $d$-band $\eps_{\bp}$ created by $p$-$d$

135: hybridization is relatively narrow and the density of states rather

136: high. The wide $s$-band resulting mainly from $s$-$p$ hybridization is

137: empty, which is why up to now the $s$-$d$ model has not been

138: applied. (ii) In order for the $s$-$d$ exchange process to become a

139: pairing mechanism the $s$- and $d$-levels must be close. A virtual

140: population of the $s$-level by band hybridization is needed in order

141: to make the $J_{sd}$ amplitude operative. Indeed, the conduction

142: $d$-band is the result of an $s$-$p$-$d$ hybridization in the \cuo\

143: plane. Another favorable factor for hybridization is the proximity of

144: the O~$2p$ and Cu~$3d_{x^2-y^2}$ levels. As the $3d$ and $4s$ states

145: are orthogonal the hybridization requires a go-between, the O~$2p$

146: orbital. Hence this theory can be nicknamed ``$3d$-to-$4s$-by-$2p$

147: highway to high-$T_c$ superconductivity in cuprates''.

148: 

149: \begin{figure}

150: \includegraphics[width=\columnwidth]{./fig2.eps}

151: \caption{The $3d$-to-$4s$-by-$2p$ highway in the \cuo\ plane. (a) The

152: shaded square is the \cuo\ unit cell indexed by

153: $\mathbf{n}=(n_x,n_y),\, n_{x,y}=0,\pm1,\pm2,\dots$. (b) A single

154: electron hops from the $3d$ atomic orbital to $2p_x$ with amplitude

155: $t_{pd}$, contained in $\hat{H}_0$. From $2p_x$ to $2p_y$ the

156: hopping amplitude is $t_{pp}$, and from there to $4s$ with amplitude

157: $t_{sp}$. Correlated hopping of two electrons in opposite directions

158: between $3d$ and $4s$ creates the pairing. The exchange integral

159: $J_{sd}$ from \protect\Fref{fig:1}~(a) is depicted as a double

160: arrow. \label{fig:2} }

161: \end{figure}

162: 

163: 

164: The foregoing discussion can be put in mathematical form. The central

165: entity is the many-particle Hamiltonian $\hat{H} = \hat{H}_0 + {\hat{

166: H}}_{\mathrm{int}}.$ Here $\hat{H}_0$ describes~\cite{Mishonov:00}

167: self-consistent motion of independent Fermi particles in

168: Bloch-H\"uckel approximation, and ${\hat{H}}_{\mathrm{int}}$ specifies

169: the interaction. Following the seminal works by Schubin and

170: Wonsowsky~\cite{Schubin:34}, Vonsovskii~\cite{Vonsovskii:46}, and

171: Zener~\cite{Zener:51}, cf. the Feynman lectures on

172: physics~\cite{Feynman:66}, we set up the $s$-$d$ pairing Hamiltonian

173: for the \cuo\ plane

174: %

175: \begin{equation}

176:  \hat{H}_{\mathrm{int}} = -J_{sd} \sum_{\mathbf{n},\alpha,\beta}

177:  \hat{S}^{\dag}_{{\bf n}\alpha}\hat{D}^{\dag}_{{\bf n}\beta}

178:  \hat{S}_{{\bf n}\beta}\hat{D}_{{\bf n}\alpha},\quad\!

179:  \alpha,\beta =\, \uparrow,\downarrow.

180: \label{eq:Hint}

181: \end{equation}

182: 

183: This Hamiltonian is written in the language of second quantization but

184: one can easily trace the connection with the classical Bohr-Sommerfeld

185: picture, \Fref{fig:1}~(a). $\hat{D}_{{\bf n}\alpha}$ is the Fermi

186: operator for annihilating a $3d_{x^2-y^2}$ electron with

187: spin projection $\alpha$ in the {\bf n}th unit cell. The electron is

188: then recreated without spin flip in the $4s$ state by the Fermi

189: creation operator $\hat{S}^{\dag}_{{\bf n}\alpha}$.  Simultaneously,

190: another electron with spin $\beta$ makes a reverse hop, from $4s$ to

191: $3d_{x^2-y^2}$ in the same Cu atom. Thus, the

192: $\hat{S}^{\dag}\hat{D}^{\dag}\hat{S}\hat{D}$ product in

193: Eq.~\ref{eq:Hint} represents the two-electron exchange drawn as a

194: double arrow in \Fref{fig:2}~(b) and shown in \Fref{fig:1}~(a).

195: 

196: A model which incorporates a Hamiltonian of the form $\hat{H} =

197: \hat{H}_0 + {\hat{H}}_{\mathrm{int}}$ implies the validity of the

198: Fermi-liquid picture for the cuprates. That is, we assert that an

199: electron is merely a Fermi quasiparticle and not a composite object

200: displaying spin-charge separation. Two key

201: experiments~\cite{Feng:01,Proust:02} have shown that this is

202: correct. The long-sought bilayer splitting in the electronic structure

203: of the cuprates was observed~\cite{Feng:01} for overdoped

204: Bi$_2$Sr$_2$CaCu$_2$O$_{8+\delta}$ (BSCCO). Furthermore, the overdoped

205: Tl$_2$Ba$_2$CuO$_{6+\delta}$ was found~\cite{Proust:02} to obey the

206: 150-years-old Wiedemann-Franz law to a remarkable accuracy of the

207: Sommerfeld's value for the Lorentz ratio.  Let us stress also that the

208: linear temperature dependence of the normal-state resistivity is an

209: intrinsic property~\cite{Mishonov2:00} of the ``layered'' electron gas

210: and cannot be used as an argument in favor of non-Fermi-liquid

211: behavior.

212: 

213: We now apply the traditional BCS approach~\cite{BCS:57} to the

214: Hamiltonian $\hat{H}$ and derive an equation for the basic

215: characteristic of superconductors---the energy gap $\Delta_{\bp}(T)$

216: (the energy needed to break up a Cooper pair is $2\Delta$).  The

217: local, intra-atomic character of the pairing interaction tremendously

218: simplifies the BCS gap equation and the solution has the plain form

219: %

220: \begin{equation}

221:  \Delta_{\bp}(T) = \Xi(T)\,\chi_{\bp},\qquad  \chi_{\bp} \equiv S_{\bp}D_{\bp}.

222: \label{eq:gap}

223: \end{equation}

224: %

225: The order parameter $\Xi(T),$ which satisfies the standard BCS gap

226: equation, is maximal at $T=0$ and vanishes for $T\rightarrow T_c.$ The

227: $S_{\bp}D_{\bp}$ product is the amplitude of the $s$-$d$ hybridization

228: of the conduction band. In this theory it determines the momentum

229: dependence of the superconducting gap and is shown in

230: \Fref{fig:3}~(b). Note that $\chi_{\bp}$ exhibits the desired $d$-wave

231: symmetry which is now well established

232: experimentally~\cite{Tsuei:00}. It is a direct consequence of the

233: electron band hybridization. The mediator O~$2p$ orbitals transmit the

234: symmetry of the Cu~$3d_{x^2 - y^2}$ orbital to the momentum dependence

235: of the Bloch wave amplitude $S_{\bp}$ and the superconducting gap

236: $\Delta_{\bp}(T)$. This is why the pairing state has $d$-wave symmetry

237: directly coming from the Cu~$3d_{x^2-y^2}$ orbital.

238: %

239: \begin{figure}[t]

240: \includegraphics[width=0.69\columnwidth]{./fig3a.eps}\\

241: \includegraphics[width=0.69\columnwidth]{./fig3b.eps}\\

242: \includegraphics[width=0.69\columnwidth]{./fig3c.eps}

243: \caption{Electronic properties of the superconducting \cuo\ plane.

244: (a) Conduction band energy $\eps_{\bp}$ as a function of the

245: quasi-momentum $\bp.$ The red contour corresponding to the Fermi

246: energy, $\eps_{\bp} = E_F,$ is in excellent agreement with the

247: ARPES data~\protect\cite{Schwaller:95}. (b) Momentum dependence of the

248: gap-anisotropy function $\chi_{\bp}$ within the $s$-$d$ model.  The

249: functional values along the Fermi contour are indicated by a green

250: line.  (c) Superconducting gap at zero temperature $\Delta_{\bp}$

251: (green line) according to our analytical result~\protect\Eref{eq:gap},

252: plotted along the Fermi contour (red line). The ARPES

253: data~\protect\cite{Mesot:99} for BSCCO are given as prisms with sizes

254: corresponding to the experimental error bars. The gap function along

255: the Fermi contour has the same qualitative behavior and symmetry as

256: the Cu~$3d_{x^2-y^2}$ electron wave function along the circular orbit

257: sketched in \protect\Fref{fig:1}~(a).

258: \label{fig:3} }

259: \end{figure}

260: 

261: The angle-resolved photoemission spectroscopy (ARPES) provides a

262: straightforward method for measuring $\Delta_{\bp}$. We have fitted

263: our theoretical result Eq.~\ref{eq:gap} to the recent precise ARPES

264: data~\cite{Mesot:99} for BSCCO and the outcome is given in

265: \Fref{fig:3}~(c). It is the first demonstration that the theory

266: provides a satisfactory description of overdoped cuprates.

267: 

268: We mentioned some normal state

269: properties~\cite{Feng:01,Proust:02,Mishonov2:00} of overdoped cuprates

270: which are governed by conventional physics.  The point to be stressed

271: now is that not only the normal phase but even the superconducting

272: phase at optimal doping, for which the $T_c$ is the highest, can be

273: explained by conventional theory, since there is overwhelming evidence

274: for the applicability of the Landau quasiparticle picture.

275: 

276: The quantitative analysis of the thermal conductivity~\cite{Chiao:00}

277: of optimally doped Bi$_2$Sr$_2$CaCu$_2$O$_8$ gave values for the Fermi

278: velocity $v_F$ and derivative of the gap function with

279: respect to the momentum component tangential to the Fermi contour,

280: $v_{\Delta}=\partial \Delta/\partial p_l,$ within $10\%$ of the values

281: found in independent ARPES experiments, as emphasized in a recent

282: review~\cite{Orenstein:00}. Furthermore, measurements of heat

283: transport, specific heat, and superfluid density through the

284: penetration depth also strongly support the existence of low-energy

285: Fermi excitations of the conventional BCS ground state for a $d$-wave

286: superconductor. Moreover, the quality of the quasiparticles is

287: astonishing since their mean-free path in some high-quality crystals

288: can reach 2500~\AA\ at 15~K, as reviewed by Lee~\cite{Lee:97}.

289: 

290: Our calculations consist of solving the gap equation in the framework

291: of the BCS theory using linear combination of atomic orbitals for the

292: electron band structure. In contrast with theoretical explanations

293: advanced so far our results indicate that the extraordinarily large

294: pairing amplitude together with the correct gap symmetry cannot be

295: explained by any interatomic or even interlayer exchange

296: interaction. Instead it must be due to the powerful intra-atomic

297: two-electron exchange process, which incidentally is responsible for

298: the magnetism of the transition metals as well. 

299: %

300: The superconductivity of the optimally doped cuprates, most important

301: for practical applications, can now be understood. The pairing

302: amplitude we uncover is so strong that the high $T_c$ is robust to

303: many other new effects discovered in optimally doped cuprates,

304: including the existence of a weak pseudo-gap~\cite{Arko:99}, the

305: appearance of stripes~\cite{Howald:02,Cho:02}, and the confluence of

306: superconductivity and magnetism~\cite{Ramires:99}.

307: %

308: Although these new effects can be considered to make the physics of

309: cuprates revolutionary, and require novel theoretical ideas for their

310: explanation, our point is that all these effects are irrelevant to the

311: main issue of explaining the high $T_c$ of high-$T_c$

312: superconductors~\cite{Anderson:00}.

313: 

314: The proposed model contains energies and hopping parameters which have

315: to be determined by fits to experimental data or by first-principles

316: calculations. Predicting thermodynamic properties close to $T_c$

317: constitutes a stringest test of the model.  For example, let us recall

318: that the BCS result for the jump of the specific heat in isotropic-gap

319: superconductors, $\Delta C$, at $T=T_c$, relative to the specific heat

320: of the normal phase $C_n$, is given by $\Delta

321: C/C_n=12/(7\zeta(3))\approx 1.43$ ($\zeta$ is the Riemann zeta

322: function). For the overdoped samples we find that this ratio is

323: renormalized by the gap anisotropy according to the relation

324: %

325: \begin{equation}

326: \frac{\Delta C}{C_n} =1.43 \,\frac{\langle|\Delta_{\bp}|^2\rangle^2

327:  }{

328:    \langle 1 \rangle\langle|\Delta_{\bp}|^4\rangle

329:  }.

330: \end{equation}

331: %

332: Here $\langle\dots\rangle$ denotes averaging over the Fermi surface

333: and $\langle1\rangle$ is the density of states. This simple expression

334: provides another direct means for experimental verification of the

335: model's validity for overdoped samples.

336: 

337: As a fundamental microscopic process the two-electron $s$-$d$ exchange

338: has been known in the physics of magnetism long

339: before~\cite{Schubin:34,Vonsovskii:46,Zener:51} the advent of the BCS

340: theory~\cite{BCS:57}. So it is not surprising that an exchange-driven

341: superconductivity was attempted shortly after~\cite{Akhiezer:59} the

342: BCS theory. Reventuring this idea we have shown how the two paradigms

343: ($s$-$d$ \& BCS) can be reconciled to obtain a coherent traditional

344: theory of high-$T_c$ superconductivity in the

345: cuprates. Magnetism of transition metals and high-$T_c$

346: superconductivity of cuprates seem to be two faces of the same

347: ubiquitous two-electron exchange amplitude.

348: 

349: \begin{acknowledgments}

350: This work was supported by the Flemish Programmes VIS, IUAP and GOA.

351: \end{acknowledgments}

352: 

353: \input{./prl.bbl}

354: 

355: \bibliographystyle{apsrev}

356: 

357: \end{document}