1: \documentclass[10pt,aps,prb,showpacs,superscriptaddress,floatfix,

2: twocolumn]{revtex4}

3: \usepackage{amsmath}

4: \usepackage{citesort}

5: \usepackage{graphicx}

6: \usepackage{dcolumn}

7: %\renewcommand\baselinestretch{1.6}

8: \begin{document}

9: \title{Optical sum in Nearly Antiferromagnetic Fermi Liquid Model}

10: \author{E. Schachinger}

11: \email{schachinger@itp.tu-graz.ac.at}

12: \homepage{www.itp.tu-graz.ac.at/~ewald}

13: \affiliation{Institute of Theoretical and Computational Physics\\

14: Graz University of Technology, A-8010 Graz, Austria}

15: \author{J.P. Carbotte}

16: \affiliation{Department of Physics and Astronomy, McMaster University,\\

17: Hamilton, Ontario, L8S 4M1 Canada}

18: \date{\today}

19: \begin{abstract}

20: We calculate the optical sum (OS) and the kinetic energy (KE) for a

21: tight binding band in the Nearly Antiferromagnetic Fermi Liquid

22: (NAFFL) model which has had some success in describing the electronic

23: structure of the high $T_c$ cuprates. The interactions among electrons

24: due to the exchange

25: of spin fluctuations profoundly change the probability of occupation

26: $(n_{{\bf k},\sigma})$ of states of momentum {\bf k} and spin $\sigma$

27: which is the central quantity in the calculations of OS and KE. Normal and

28: superconducting states are considered as a function of temperature.

29: Both integrals are found to depend importantly on interactions

30: and an independent electron model is inadequate.

31: \end{abstract}

32: \pacs{74.20.Mn 74.25.Gz 74.72.-h}

33: \maketitle

34: \newpage

35: \section{Introduction}

36: 

37: The high $T_c$ oxides fall in the category of highly correlated systems.

38: A manifestation of this fact is that in the

39: underdoped regime there exists a pseudogap.\cite{r1,r2,r3,r4,r5}

40: Precisely how it is to be described

41: remains controversial.\cite{r5,r6,r7,r23a,r8,r9,r10,r11,r12,r13}

42: In certain theories it is closely related to

43: superconducting correlations,\cite{r5,r6,r7,r23a} and the

44: superconducting transition temperature $T_c$ is the temperature at which

45: phase coherence

46: is lost. In other theories the

47: pseudogap has its origin in completely different correlations

48: and is a manifestation of a

49: competing order such as a $d$-density wave (DDW).\cite{r8,r9,r10,r11,%

50: r12,r13} In any case interactions among charge carriers play an

51: important role in such systems and cannot be ignored in any realistic

52: approach to their properties even around optimum doping which is the

53: case of interest here.

54: 

55: While many aspects of the superconducting state can be understood

56: qualitatively on the basis of extensions of BCS theory,

57: particularly around optimum doping,

58: the search

59: for essential differences\cite{r14} has remained an important

60: avenue of investigation. In particular the idea of kinetic energy

61: as opposed to potential energy driven superconductivity

62: (i.e.: the kinetic energy is reduced at the transition temperature)

63: and its relation to the OS has

64: recently been given serious consideration in

65: theory\cite{r14,r15,r16,r17,r18,r19,r20,r21,r22,r23}

66: and experimentally.\cite{r24,r25,r26,r27,r28,r29}

67: An issue of importance is the relation of the

68: kinetic energy (KE) to the optical sum (OS). A recent paper\cite{r28}

69: has provided some insight into this relationship and has given a

70: comparison with experiment for the temperature variation of the OS

71: in the normal state, its change at $T_c$, and its further

72: evolution in the superconducting state. Correlations beyond

73: BCS pairing were not considered, however.

74: There have also been several recent studies of the

75: temperature dependence of the OS, experimentally for the system

76: LA$_{2-x}$Sr$_x$CuO$_4$, Ref.~\onlinecite{r42a}, and, theoretically,

77: using other models.\cite{r42b,r42c,r42d}

78: 

79: Within a BCS model, the condensation is potential energy driven and,

80: in fact, the KE increases because the probability of occupation

81: of the state {\bf k} $(n_{\bf k})$ becomes smeared at the Fermi

82: energy by the opening of the gap. This translates into a decrease in the

83: OS as compared with its normal state value. But this is opposite to

84: the behavior obtained experimentally in Ref.~\onlinecite{r26}.

85: However, as mentioned, the theoretical discussion of Ref.~\onlinecite{r28}

86: is based on a non interacting model in the normal state and includes

87: pairing correlations in the superconducting state only at the level

88: of BCS. More sophisticated formulations of the theory of

89: superconductivity could give different results.

90: Also, interactions can significantly modify the results even

91: in the normal state, as has been demonstrated recently by

92: Knigavko {\it et al.}\cite{r30} in a simplified model in which

93: the charge carriers are coupled to a single Einstein mode. While

94: only the normal state was considered, it was found that

95: boson hardening (softening) results in an increase (decrease)

96: in the OS and that interactions

97: play an important role in determining temperature dependences.

98: 

99: In this paper we study a tight binding model with an emphasis on the

100: effect interactions can have on the OS, particularly on its

101: temperature dependence and its relationship to the KE. Here,

102: the interactions among the charge carriers are treated in the

103: Nearly Antiferromagnetic Fermi Liquid (NAFFL) model.\cite{r31}

104: A review of its main properties and successes is given in

105: Ref.~\onlinecite{r31}. The model is phenomenological and falls into

106: the general class of boson exchange models where the interaction

107: between electrons proceeds through the exchange of spin

108: fluctuations.\cite{r32} The imaginary part of the spin susceptibility

109: replaces the phonon propagator of the classic electron-phonon

110: Eliashberg theory. The spin susceptibility could be calculated

111: from microscopic theory. More usually, however, it is fit to

112: experimental data, specifically to NMR in Ref.~\onlinecite{r32}.

113: The basic idea is that the doped metallic cuprates are near an

114: antiferromagnetic phase boundary and that coupling to spin

115: fluctuations is therefore strong.

116: While there is no consensus as to the validity of such a model when

117: applied to the oxides the NAFFL model has been widely discussed and

118: has had considerable

119: successes particularly in correlating

120: superconducting properties,\cite{r33,r34,r35,r35a,r35b,r35c} and has

121: been used to

122: give a detailed description of the optical properties\cite{r35,r36,%

123: r37,r38,r39,r40,r43a} in the high $T_c$ oxides at optimum

124: doping. In any case, the NAFFL model\cite{r41}

125: provides a convenient and specific

126: framework within which  we can study the effect of correlations on the

127: OS and on related properties.

128: 

129: In Sec.~\ref{sec:2} we summarize the basic equations that are needed to

130: compute the optical sum integral as well as the KE. They are a set

131: of three coupled generalized Eliashberg equations written for any

132: momentum {\bf k} in the two dimensional CuO$_2$ Brillouin zone.

133: They involve renormalized Matsubara frequencies, the renormalized

134: quasiparticle energies as well as the superconducting energy gap.

135: The interaction

136: between the charge carriers is mediated by the exchange of spin

137: fluctuations and involve the spin susceptibility. Fast Fourier

138: transforms (FFT) provide solutions which give a $d$-wave gap as

139: observed in experiments. In Sec.~\ref{sec:3} we apply our solutions to

140: evaluate the probability of occupation of the state of momentum

141: {\bf k} and spin $\sigma$ $(n_{{\bf k},\sigma})$ from which the OS

142: and the KE follow. Their temperature dependence is studied. Comparison

143: with the non interacting case is made and it is concluded that

144: interactions can profoundly modify results. Different behaviors

145: can result depending on the choice of microscopic parameters. In

146: Sec.~\ref{sec:4} we consider explicitly the superconducting state. Here,

147: again, generalized Eliashberg equations give variations with $T$

148: which are considerably different from earlier BCS  results.

149: When the electron-exchange boson interaction is taken as

150: temperature independent and independent of state, the OS is lower

151: in the superconducting than in the normal state. Nevertheless,

152: it can keep increasing with decreasing temperature in contrast

153: to BCS where it was found to decrease. This increase can be traced to the

154: underlying temperature dependence of the OS which depends on

155: details of the band structure and interactions involved, for

156: example on the model spin susceptibility. Further, if we take account

157: of the low energy gaping of the spin susceptibility

158: which is brought

159: about by the superconducting transition (a process which is not

160: operative in the normal state) the KE can be further decreased

161: and there is an additional increase in the OS which can effectively

162: increase faster than it does in the normal state. In this case

163: the KE in the superconducting state with low energy gaping of the

164: spin susceptibility can be less

165: than in the normal state without low energy gaping.

166: An important conclusion

167: of our work is that the observation of a faster increase in the OS with

168: decreasing temperature in the superconducting state than in the

169: normal state cannot unambiguously be taken

170: to be an indication of kinetic energy driven superconductivity

171: in contrast to a recent claim by van der Marel {\it et al.}\cite{r28}

172: In Sec.~\ref{sec:5} we provide a brief conclusion.

173: 

174: We use units in which $\hbar = c = 1$ throughout this paper.

175: 

176: \section{Formalism}

177: \label{sec:2}

178: 

179: In the NAFFL model the interaction between holes proceeds through the

180: exchange of spin fluctuations and the spin susceptibility

181: $\chi({\bf q},\omega)$ plays a central role. The three Eliashberg

182: equations for the renormalized frequencies $\tilde{\omega}({\bf k},i\omega_n)$,

183: the energy renormalization $\xi({\bf k},i\omega_n)$ and the pairing energy

184: $\phi({\bf k},i\omega_n)$ as a function of momentum {\bf k} in the

185: two dimensional CuO$_2$ Brillouin zone, and of fermionic Matsubara

186: frequencies $i\omega_n = i\pi T(2n+1),\,n=0, \pm 1, \pm 2,\ldots$

187: and the temperature $T$ are\cite{r35a,r35b,r35c}

188: \begin{subequations}

189: \label{eq:1}

190: \begin{eqnarray}

191:   \tilde{\omega}({\bf k},i\omega_n) &=& \omega_n+

192:   T\sum\limits_m\sum\limits_{{\bf k}'}

193:   \lambda_{SF}({\bf k}-{\bf k}',i\omega_n-i\omega_m)¸\nonumber\\

194:   &&\times

195:   \frac{\tilde{\omega}({\bf k}',i\omega_m)}{D({\bf k}',i\omega_m)},

196:   \label{eq:1a}\\

197:   \xi({\bf k},i\omega_n) &=&

198:   -T\sum\limits_m\sum\limits_{{\bf k}'}

199:   \lambda_{SF}({\bf k}-{\bf k}',i\omega_n-i\omega_m)\nonumber\\

200:   &&\times

201:   \frac{\epsilon_{{\bf k}'}+\xi({\bf k}',i\omega_m)}{D({\bf k}',i\omega_m)},

202:   \label{eq:1b}\\

203:   \phi({\bf k},i\omega_n) &=&

204:   -T\sum\limits_m\sum\limits_{{\bf k}'}

205:   \lambda_{SF}({\bf k}-{\bf k}',i\omega_n-i\omega_m)\nonumber\\

206:   &&\times

207:   \frac{\phi({\bf k}',i\omega_m)}{D({\bf k}',i\omega_m)}.

208:   \label{eq:1c}

209: \end{eqnarray}

210: \end{subequations}

211: In Eqs.~\eqref{eq:1} $D({\bf k},i\omega_n)$ is given by

212: \begin{equation}

213:   \label{eq:2}

214:   D({\bf k},i\omega_n) = \tilde{\omega}^2({\bf k},i\omega_n)+

215:   \left[\epsilon_{\bf k}+\xi({\bf k},i\omega_n)\right]^2+

216:   \phi^2({\bf k},i\omega_n),

217: \end{equation}

218: and $\epsilon_{\bf k}$ is the charge carrier dispersion relation. In a

219: tight binding model without the inclusion of the coupling to the

220: spin fluctuations it is given by:

221: \begin{eqnarray}

222:   \epsilon_{\bf k} &=& -2\left\{t\left[\cos(ak_x)+\cos(ak_y)\right]

223:   \right.\nonumber\\ &&-\left.

224:   2t'\cos(ak_x)\cos(ak_y)\right\}-\mu^\ast.

225:   \label{eq:3}

226: \end{eqnarray}

227: Here, $a$ is the lattice parameter in the copper oxide plane,

228: $t$ the nearest neighbor hopping,

229: $t'$ the next nearest neighbor hopping, and $\mu^\ast$ the chemical

230: potential. We will discuss within this

231: context, two tight binding models with the parameters given in

232: Table~\ref{tab:1}. The corresponding Fermi surface is presented in

233: Fig.~\ref{fig:2}a for model A and in Fig.~\ref{fig:2}b for model B.

234: These figures define also the various points in the Brillouin zone.

235: The dotted lines indicate the antiferromagnetic Brillouin zone

236: boundary. For model A, the Fermi surface crosses the anti-ferromagnetic

237: \begin{table}[tp]

238: \caption{\label{tab:1}The two tight binding models used within this

239: paper. Model A corresponds to the tight binding model discussed by

240: van der Marel {\it et al.}\protect{\cite{r28}} $t$ and $t'$ are

241: given in meV, the critical temperature $T_c$ in K, and the filling

242: $\langle n\rangle$ is defined in Eq.~\protect{\eqref{eq:9}}.

243: }

244: \begin{ruledtabular}

245: \begin{tabular}{ldddd}

246: Model & t & t' & \langle n\rangle & T_c\\

247: \hline

248: A & 148.8 & 40.9 & 0.425 & 90\\

249: B & 100.0 & 16.0 & 0.4 & 100\\

250: \end{tabular}

251: \end{ruledtabular}

252: \end{table}

253: Brillouin zone around $X$ and symmetry related points. These points

254: are `hot spots' for which Fermi surface to Fermi surface transitions

255: are possible with momentum transfer $(\pi/a,\pi/a)$ (nesting

256: property). Model A has

257: been used previously in Ref.~\onlinecite{r28} to discuss the OS and

258: KE in Bi$_2$Sr$_2$Cu$_2$O$_{8+\delta}$ (BSCCO) assuming no interactions

259: between the electrons in the

260: normal state. Therefore, it is natural to employ this same model

261: in the present study which extends the previous calculations to

262: include exchange of spin fluctuations between charge carriers in

263: the NAFFL model. The second model, Model B, was chosen to contrast

264: with the first. It has no hot spots and its Fermi surface is closed

265: around the $\Gamma$-point in contrast to Model A which has a Fermi surface

266: which is closed around the $M$-point. It also has a smaller value of

267: $t$ which, on its own, would imply a smaller absolute value of KE.

268: These differences in band parameters lead, as we shall see, to

269: some differences in KE and optical sum at $T=0$ in the

270: normal state.

271: 

272: In the phenomenological model of Millis {\it et al.}\cite{r31,r32} and

273: Monthoux {\it et al.}\cite{r41} (MMP-model) the magnetic susceptibility

274: was fit to NMR data and its imaginary part takes on the form

275: \begin{equation}

276:   \label{eq:4}

277:   \Im{\rm m}\chi_{MMP}({\bf q},\omega) =

278:   \frac{\chi_{\bf Q}\,(\omega/\omega_{SF})}{\left[1+\zeta^2({\bf q}-{\bf Q})^2

279:   \right]^2+(\omega/\omega_{SF})^2},

280: \end{equation}

281: where $\chi_{\bf Q}$ is the static susceptibility, {\bf Q} is the

282: commensurate antiferromagnetic wave vector $(\pi/a,\pi/a)$

283: in the upper right hand quadrant of the CuO$_2$-plane Brillouin zone

284: and symmetry related points. $\zeta$

285: \begin{figure}[tp]

286: \vspace*{-1cm}

287:   \includegraphics[width=9cm]{vmFig1a.eps}\\

288: \vspace*{-1cm}

289:   \includegraphics[width=9cm]{vmFig1b.eps}

290: \vspace*{-10mm}

291:   \caption{The Fermi surface for the non interacting system (dashed line)

292: and the system with interaction (solid line). a) model A of Table~\ref{tab:1},

293: b) for model B.}

294:   \label{fig:2}

295: \end{figure}

296: is the magnetic coherence length, and $\omega_{SF}$ a characteristic

297: spin fluctuation frequency. We set $\zeta = 2.5\,a$ throughout this

298: paper and various values for $\omega_{SF}$ are investigated.

299: 

300: The kernel $\lambda_{SF}({\bf q},i\nu_{n-m})$ in Eqs.~\eqref{eq:1}

301: with momentum transfer ${\bf q} = {\bf k}-{\bf k}'$ and the

302: bosonic Matsubara frequency $i\nu_{n-m} = i\omega_n-i\omega_m$ is

303: given as

304: \begin{equation}

305:   \label{eq:5}

306:   \lambda_{SF}({\bf q},i\nu_n) = \frac{g^2\chi_{\bf Q}}

307:   {1+\zeta^2({\bf q}-{\bf Q})^2+(\vert\nu_n\vert/\omega_{SF})},

308: \end{equation}

309: with $g^2\chi_{\bf Q}$ adjusted to get the desired value of the critical

310: temperature $T_c$ for a certain value of $\omega_{SF}$ from the

311: solution of the linearized Eqs.~\eqref{eq:1}. This defines the model.

312: 

313: The aim of this paper is to investigate the effect interactions have on

314: the OS defined as

315: \begin{equation}

316:   \label{eq:6}

317:   %\frac{\pi\, e^2}{\hbar^2}

318:    \pi\,e^2\,I_\sigma = 

319:   \int\limits_{-\Omega}^\Omega\!d\omega\,\Re{\rm e}\sigma_{xx}(\omega) =

320:   \frac{\pi\,e^2}{V}\sum\limits_{{\bf k},\sigma}

321:   n_{{\bf k},\sigma}\frac{\partial^2\epsilon_{\bf k}}{\partial k_x^2},

322: \end{equation}

323: where $e$ is the charge on the electron, $V$ the volume, and

324: $n_{{\bf k},\sigma}$ is the probability of occupation of a state of

325: momentum {\bf k} and spin $\sigma$. Finally, $\sigma_{xx}(\omega)$ is

326: the optical conductivity. The integral in Eq.~\eqref{eq:6} is to be

327: taken over the single band with $\Omega$, the upper limit in the

328: integral of Eq.~\eqref{eq:6}, large enough to include all

329: possible transitions in that band. We are also interested

330: in the relationship between the OS and the kinetic energy. By

331: definition

332: \begin{equation}

333:   \label{eq:7}

334:   I_{\rm KE} = \left\langle H_{\rm KE}\right\rangle = \frac{a^2}{V}

335:   \sum\limits_{{\bf k},\sigma} n_{{\bf k},\sigma}\epsilon_{\bf k}.

336: \end{equation}

337: We will see that, to a good approximation $I_\sigma$ and

338: $-I_{\rm KE}$ are nearly proportional to each other. Thus,

339: \begin{equation}

340:   \label{eq:8}

341:   \rho_L \propto \frac{1}{\pi e^2}\int\limits_{-\Omega}^\Omega\!

342:   d\omega\,\Re{\rm e}\,\sigma_{xx}(\omega) \approx -\frac{1}{2}\left\langle

343:   H_{\rm KE}\right\rangle.

344: \end{equation}

345: holds approximately. (An equal sign would be appropriate if the

346: dispersion relation \eqref{eq:3} contained only nearest neighbor

347: interaction, i.e.: $t'=0$.) Here, $\rho_L$ is the experimentally

348: determined value of the OS $(I_\sigma)$.

349: 

350: Interactions have a profound effect on the probability of occupation of

351: the state $\vert{\bf k},\sigma\rangle$ which would be one or zero for

352: occupied and unoccupied states respectively in the non interacting case.

353: In Fig.~\ref{fig:1} we show results for $n_{\bf k}$, along

354: certain selected

355: \begin{figure}[tp]

356: \vspace*{-3mm}

357:   \includegraphics[width=9cm]{vmFig2.eps}

358:   \caption{The occupation number $n_{\bf k}$ for both spin states for

359: selected directions in the CuO$_2$ Brillouin zone. Model A of

360: Table~\protect{\ref{tab:1}} was used.

361: Top frame: the non interacting case. Center frame: The interacting

362: case at a temperature $T=20\,$K. We show normal state (solid line)

363: and superconducting state (dashed line) results. Bottom frame: The

364: temperature influence on the normal state $n_{\bf k}$ for

365: $T=20\,$K (solid line) and $T=150\,$K (dashed line). }

366:   \label{fig:1}

367: \end{figure}

368: directions in the CuO$_2$ Brillouin zone. In all cases Model A of

369: Table~\ref{tab:1} is used for the

370: electronic dispersion \eqref{eq:3} with $\mu^\ast$ adjusted to the

371: required filling which is is defined as

372: \begin{equation}

373:   \label{eq:9}

374:   \langle n\rangle = \frac{1}{2}-\sum\limits_{\bf k}\sum\limits_{n\ge 0}

375:   \frac{\epsilon_{\bf k}+\xi({\bf k},i\omega_n)}

376:   {\tilde{\omega}^2({\bf k},i\omega_n)+\left[\epsilon_{\bf k}+

377:   \xi({\bf k},i\omega_n)\right]^2+\phi^2({\bf k},i\omega_n)},

378: \end{equation}

379: and the charge carrier spin fluctuation strength

380: $g^2\chi_{\bf Q}$ is adjusted to get a $T_c = 90\,$K for the

381: superconducting state. In the top frame of Fig.~\ref{fig:1} we show

382: $n_{\bf k}$ in the non interacting case as we go from $\Gamma$ to $X$

383: and from $X$ to $M$ in the Brillouin zone with the Fermi surface defined

384: as the value of {\bf k} at which $n_{\bf k}$ jumps from one to zero.

385: It is obvious from Fig.~\ref{fig:2}a (dashed line) that

386: the path $\overline{XM}$ crosses the Fermi surface. A second

387: crossing of the Fermi surface can be observed along the path from $M$

388: to $\Gamma$.

389: 

390: The center frame of Fig.~\ref{fig:1} shows $n_{\bf k}$

391: when interactions are taken into account. (The corresponding Fermi surface

392: is shown as the solid line in Fig.~\ref{fig:2}a.) We see a drastic

393: difference in the value of $n_{\bf k}$ which is now of the order

394: 0.9 at the $\Gamma$ point indicating that the effect of the

395: interaction is very significant even in the center of the Brillouin zone.

396: Also, $n_{\bf k}$ is of the order 0.1 outside the non interacting

397: Fermi surface where $n_{\bf k}\approx 0$ for the non interacting case.

398: The solid line applies to the normal state at $T=20\,$K and the dashed

399: line to the superconducting state at the same temperature. On comparing

400: these two cases we see that the transition to the superconducting state

401: depletes even further the non interacting sea in the sense that it further

402: reduces $n_{\bf k}$ at ${\bf k}$s below the Fermi surface and

403: correspondingly increases the probability of occupation of states right

404: above the Fermi surface beyond the effect of interactions in the

405: normal state. It is to

406: be noted, however, that the onset of superconductivity results in rather

407: modest changes in $n_{\bf k}$ as compared to the difference between

408: interactions and no interactions in the normal state.

409: 

410: The bottom frame of Fig.~\ref{fig:1} gives results in the normal state

411: but compares two temperatures, namely $T=20\,$K (solid line) and

412: $T=150\,$K (dashed line). Comparison with the middle frame shows that

413: increasing the temperature has roughly the same qualitative effect

414: on $n_{\bf k}$ as does the transition to the superconducting state. In

415: both cases, the KE given by Eq.~\eqref{eq:7} increases because the states

416: of lower $\epsilon_{\bf k}$ get depleted while states with higher

417: $\epsilon_{\bf k}$ are occupied with increasing probability. This will

418: also hold for the OS according to Eq.~\eqref{eq:6} which will now

419: depend on interactions and on temperature.

420: 

421: \section{Results for the optical sum in the normal state}

422: \label{sec:3}

423: 

424: In Fig.~\ref{fig:4} we show results for the

425: \begin{figure}[tp]

426: \vspace*{-5mm}

427:   \includegraphics[width=9cm]{vmFig3.eps}

428:   \caption{Optical sum $I_\sigma$ and kinetic energy, $-I_{\rm KE}/2$,

429: as a function of $T^2$. Solid circles and squares are for the non

430: interacting case while solid up-triangles and solid down-triangles

431: include interactions.

432: The top frame applies to

433: Model A of Table~\protect{\ref{tab:1}} and $\omega_{SF}=82\,$meV was

434: used. Here, the interacting and non interacting cases show similar

435: temperature dependencies. The bottom frame is for

436: Model B of Table~\protect{\ref{tab:1}} with an MMP-model

437: $\omega_{SF}=10\,$meV. Note the difference in temperature dependence

438: between interacting and non interacting case.

439: }

440:   \label{fig:4}

441: \end{figure}

442: optical sum $(I_\sigma)$ Eq.~\eqref{eq:6} and compare with the kinetic

443: energy $(I_{\rm KE})$, Eq.~\eqref{eq:7}. The top frame is based on

444: Model A and the bottom frame on Model B of Table~\ref{tab:1}.

445: The solid squares and circles are $-I_{\rm KE}/2$

446: and $I_\sigma$ respectively in the free tight binding case, i.e.:

447: no interactions, plotted as a function of the square of the temperature

448: for the normal state. (A $128\times 128$ sampling of the

449: ${\bf k}$-space, $ak_x, ak_y\in [0,\pi]$, was used but going to

450: a $256\times 256$ sampling did not influence the results.)

451: In both models variation with $T$ over the range $0$ to

452: $200\,$K is small (of order 1\% for Model A and 2\% for Model B)

453: as was also found in the work of

454: Molegraaf {\it et al.}\cite{r26} Also, the two integrals

455: $(I_\sigma$, $-I_{\rm KE}/2)$ track each

456: other closely even though they are not equal in magnitude.

457: (They would be equal for $t'=0$. We tried other Fermi surfaces, even

458: one with perfect nesting, i.e.: $t'=0$ and $\langle n\rangle = 0.5$,

459: and found no qualitative changes.) These results are for

460: comparison with results indicated by up/down-triangles

461: which include interactions in the NAFFL model.

462: The magnitude of both, the optical

463: sum integral and the kinetic energy has been changed considerably by

464: the interactions although the order remains the same, i.e.: $-I_{\rm KE}/2$

465: is greater than $I_\sigma$ and, again, they track each other.

466: More importantly, for the discussion here,

467: the temperature variation has been changed. Both integrals now

468: show variation of the order 8 to $9\%$. Also the dependence on

469: $T^2$ is not linear at small values of $T^2$. Model A shows similar

470: behavior for small values of $\omega_{SF}$.

471: It is clear that any estimate

472: based on the independent particle tight binding model is unreliable.

473: It is, however,

474: possible to chose specific parameters in the MMP-model which show

475: variations in the interacting case that are much closer to the non

476: interacting case. This is illustrated in the top frame of Fig.~\ref{fig:4}.

477: Here we used Model A of Table~\ref{tab:1}.

478: Again, results with and without interaction

479: are compared and both show little temperature variation. To get

480: this we used $\omega_{SF}=82\,$meV in our MMP form of Eq.~\eqref{eq:4}

481: without a change in the magnetic

482: coherence length. Both results, with and without interaction, show

483: little variation with temperature.

484: What this shows is that the magnitude as well as

485: the temperature variation of KE and of OS depends significantly on the

486: parameters used to characterize their electronic structure,

487: particularly the spin susceptibility.

488: 

489: We have done additional calculations for Model A with

490: $\omega_{SF} = 40,\,20,$ and $10\,$meV.

491: In all cases the change in KE due to interactions at $T=0$

492: increases with decreasing values of $\omega_{SF}$. In particular,

493: it changes at $T=0$ by 5.3\% when compared with the non interacting

494: case, for $\omega_{SF}=82\,$meV and by 15.8\% for $\omega_{SF}=10\,$meV.

495: The corresponding temperature changes from $T=0$ to $T=200\,$K are

496: roughly a factor of 5 smaller, more precisely, they are

497: 0.7\% and 3.4\% respectively. 

498: Thus, a change in KE at $T=0$ due

499: to interactions also implies a corresponding change in temperature

500: dependence with both changes tracking each other.

501: For Model B the change in KE due to interactions is 34\% for

502: $\omega_{SF} = 10\,$meV with a 8.7\% increase in KE from $T=0$

503: to $T=200\,$K. These variations are about a factor of two larger

504: than for the equivalent case of Model A with a comparable value

505: of $\omega_{SF}$. Despite the fact that the two models

506: represent very different band structures $I_\sigma$ and

507: $-I_{KE}/2$ show essentially the same qualitative features in their

508: temperature and $\omega_{SF}$ dependence. However, the quantitative

509: differences are important.

510: 

511: \section{Results for the optical sum in the superconducting state}

512: \label{sec:4}

513: 

514: Results for the superconducting state are illustrated in Fig.~\ref{fig:5}

515: \begin{figure}[tp]

516:   \includegraphics[width=9cm]{vmFig4.eps}

517:   \caption{Comparison of normal and superconducting state for the

518: optical sum and the kinetic energy. The top frame applies to

519: Model A of Table~\ref{tab:1} and $\omega_{SF}=82\,$meV and the

520: bottom frame is for the band structure Model B of Table~\ref{tab:1} and

521: $\omega_{SF} = 10\,$meV.

522: }

523:   \label{fig:5}

524: \end{figure}

525: which has two frames. The top frame applies to the band structure

526: Model A of Table~\ref{tab:1} and is for $\omega_{SF}=82\,$meV

527: as in the top frame of Fig.~\ref{fig:4}. We have also included

528: 2\% impurities in the

529: unitary limit but this serves mainly to illustrate that impurities

530: introduce no qualitative differences into our results. We see that,

531: as we expect, superconductivity reduces the optical integral

532: (open triangles) as compared with its normal state (solid

533: triangles) value at the same temperature.

534: This reduction is small. For the top frame which shows the least

535: temperature dependence, the KE integral shows a reduction of about

536: 0.25\% below its normal state value which can be compared with the

537: results shown in the bottom frame of Fig.~6 of Ref.~\onlinecite{r29}

538: where the difference is 0.2\% in their BCS calculations. On the

539: other hand, in the bottom frame of our Fig.~\ref{fig:5} for Model

540: B the reduction is about 0.8\% (four times larger). This shows

541: that the Eliashberg results depend on band structure as well as on

542: the details of the interactions involved, in particular on the

543: value of $\omega_{SF}$. In the BCS limit the increase in KE

544: normalized to the absolute value of the condensation energy is

545: given by the formula $\left[\ln\left(\frac{\omega_D}{T_c}\right)%

546: -0.38\right]$ for both $s$- and $d$-wave superconductors. Here

547: $\omega_D$ is the Debye energy. This shows a strong dependence on

548: $\omega_D/T_c$. The formula itself, however, is valid only for

549: $\omega_D/T_c\gg 1$ and cannot be used to understand our Eliashberg

550: results. The NAFFL model includes interactions which,

551: as we have seen, change importantly the probability of occupation

552: $n_{\bf k}$ and consequently the optical integral as well as the

553: kinetic energy. For the parameters

554: of Model B and $\omega_{SF} = 10\,$meV we find

555: that $I_\sigma$ and $-I_{\rm KE}/2$ can keep increasing with

556: decreasing temperature in the superconducting state

557: (bottom frame of Fig.~\ref{fig:5}). The open squares and

558: triangles (superconducting state) are below their solid

559: counterparts (normal state) but still keep growing as the temperatures

560: is reduced. This does not indicate an exotic mechanism

561: but comes directly from a

562: generalization of Eliashberg theory that includes anisotropy in the band

563: structure and, more importantly, the interaction due to coupling to

564: the spin fluctuations.

565: 

566: In Fig.~\ref{fig:6} we show additional results where the OS is seen to

567: \begin{figure}[tp]

568:   \includegraphics[width=9cm]{vmFig5.eps}

569:   \caption{The optical sum as a function of the square of the temperature

570: for the band structure Model A of Table~\protect{\ref{tab:1}}

571: with different values of $\omega_{SF}$. Also in

572: one case a low frequency cutoff is applied to the spin susceptibility.

573: Note the significance of the $T^2$-variation on the value of

574: $\omega_{SF}$. The solid line indicates experimental normal

575: state results reported by Molegraaf {\it et al.}\protect{\cite{r26}}

576: }

577:   \label{fig:6}

578: \end{figure}

579: increase even more rapidly, with reduced temperature below the onset

580: of superconductivity than it does in the normal state above $T_c$.

581: What is shown is $\rho_L$ which is $I_\sigma$ or $-I_{\rm KE}/2$

582: scaled to agree with experiment as discussed below; either,

583: $I_\sigma$ or $-I_{\rm KE}/2$, will do since these differ

584: mainly by a different scaling factor. Only

585: the OS is considered but the KE integral follows the same trend and

586: therefore the optical measurement can again be used to get information on

587: KE and its variation with temperature.

588: 

589: While in obtaining Fig.~\ref{fig:6} we applied Model A which

590: was used by van der Marel {\it et al.}\cite{r28} to describe

591: their optimally doped and underdoped BSCCO samples,

592: we now vary, in contrast to the top frame of Fig.~\ref{fig:5},

593: the value of $\omega_{SF}$ used in the MMP-model for

594: the spin susceptibility Eq.~\eqref{eq:4}. Results are presented

595: for $\omega_{SF}=20\,$meV (dashed line), $10\,$meV (dotted line),

596: $13\,$meV (solid squares) for the normal state, and

597: solid triangles for the

598: superconducting state. Also shown as the thick solid line are the

599: experimental results of Ref.~\onlinecite{r26} for their optimally

600: doped sample. We have scaled our

601: theoretical results to agree with experiment at $T=120\,$K. We first

602: note that varying $\omega_{SF}$ in the normal state can strongly

603: influence the temperature dependence obtained for $\rho_L$ (in meV).

604:  The value

605: of $\omega_{SF}=13\,$meV was chosen from a best fit in the region

606: $120\,{\rm K}\le T\le 200\,$K. The scaling factor required to get

607: agreement with this data is approximately 2. (When interactions are

608: neglected, as in Ref.~\onlinecite{r28}, the scaling factor is 

609: approximately 1.5.)

610: To reduce this

611: discrepancy, the value of $t$ would need to be increased but this

612: would also decrease the sensitivity of $I_\sigma(T)$ to temperature

613: variations and, thus, $\omega_{SF}$ would need to be adjusted as well.

614: To prove this we employ results of local density

615: approximation calculations by Markievicz {\it et al.}\cite{r42}

616: which suggest significantly bigger values for $t$ in BSCCO with

617: a Fermi surface which is little different from the one presented

618: in Fig.~\ref{fig:1}a. Using the dispersion relation, Eq.~(3),

619: and the parameter values of Table I of Ref.~\onlinecite{r42}

620: we find $I_\sigma(T=0) = 267.34\,$meV for the non-interacting system,

621: well above the value of $\rho_L(T=0) \approx 171.53\,$meV as has been

622: extrapolated from the normal state experimental data of

623: Ref.~\onlinecite{r26}. In order to reproduce the experimentally

624: observed temperature dependence $\rho_L(T)/\rho_L(T=0)$ we have

625: to introduce interactions and a value $\omega_{SF} = 8\,$meV is

626: found to give excellent agreement of $I_\sigma(T)/I_\sigma(T=0)$

627: with experiment. In this case $I_\sigma(T=0) = 230.34\,$meV, still

628: well above the experimental value and a down-scaling of

629: $I_\sigma(T)$ by a factor of 0.745 is required to achieve agreement

630: with experiment. Ultimately, a dispersion relation somewhere

631: between Model A and the one reported by Markievicz {\it et al.}\cite{r42}

632: and an $\omega_{SF}$ between $8$ and $13\,$meV will result in

633: a $I_\sigma(T)$ from theory which agrees with the experimental

634: $\rho_L(T)$ without scaling.

635: However, our main aim is not to treat a specific case but to

636: understand better the role interactions between the charge carriers

637: can play in the OS. Interactions introduce a new energy scale into

638: the problem, namely $\omega_{SF}$ for the NAFFL model. This

639: energy scale is additional to the chemical potential or the hopping

640: parameter $t$. With the values of the microscopic

641: parameters associated with the NAFFL model just described,

642: we proceed to compute

643: the OS for temperatures at and below $T_c$. The solid squares give

644: the continuation of the normal state curve and are presented for

645: comparison with the solid triangles which are the equivalent results

646: in the superconducting state. Again, superconducting state results

647: fall below the normal state ones but they are seen to, nevertheless,

648: increase with decreasing $T$. This occurs even if an Eliashberg

649: formulation is

650: used which represents a generalization of BCS theory and, in that

651: sense, is not exotic. The mechanism is the exchange of

652: antiferromagnetic spin fluctuations.

653: On comparing the top frame of Fig.~\ref{fig:5} with Fig.~\ref{fig:6},

654: we note that whether or not the superconducting state results keep increasing

655: with decreasing temperature is, for a given band structure, governed

656: by the value of $\omega_{SF}$.

657: 

658: In Fig.~\ref{fig:6} we show additional results, open squares for a

659: model normal state and open triangles for the superconducting

660: state. Now, there is a further dramatic increase in the OS both in the

661: normal and the superconducting state

662: as compared to its value in the normal state at $T=T_c=90\,$K. A

663: detailed explanation of how these results were arrived at is required.

664: In their analysis of optical data Carbotte {\it et al.}\cite{r34}

665: found that the spin fluctuations themselves are modified when the

666: superconducting state sets in. To carry out their analysis these

667: authors used a simplified version of our Eliashberg Eqs.~\eqref{eq:1}

668: which follows when the sum over {\bf k} is changed to an energy

669: integral as well as an angular average and the energy integral is done

670: analytically in a constant density of states approximation for an

671: infinite band model with interactions pinned to the Fermi surface.

672: Here we have been more realistic but what is important for us in

673: the work of Ref.~\onlinecite{r34} is that they find that the spin

674: fluctuation spectrum is gaped at low energies, or at

675: the very least loses intensity

676: and a spin resonance or peak forms at higher energy. This readjustment

677: in the spin susceptibility is not unexpected

678: and is a characteristic that should

679: be seen in any electronic mechanism for superconductivity.\cite{r37,%

680: r38,r39,r40,r43a,r41,r44,r45,r46,r47,r48}

681: Details are not important for the present discussion beyond the fact

682: that some adjustment of the spin susceptibility

683: $\Im{\rm m}\chi({\bf q},\omega)$ at small $\omega$ is expected,

684: which weakens the inelastic scattering.

685: Here we simply use the same low $\omega$ $[\omega_c(T)]$ cutoff 

686: applied to Eq.~\eqref{eq:4} which was determined by

687: E.~Schachinger {\it et al.},\cite{r43a} through consideration of

688: microwave data. Another approach would be to calculate the low

689: energy gaping of the spin susceptibility

690: from first principles but this would go beyond the scope of this

691: work and would introduce additional uncertainties.

692: The temperature dependence of $\omega_c(T)$

693: follows the temperature dependence of the superconducting gap

694: with a maximum value of $24\,$meV.

695: Application of this cutoff in

696: otherwise standard Eliashberg calculations based on our

697: Eqs.~\eqref{eq:1} yield the open triangles (superconducting state)

698: and open squares (normal state) of Fig.~\ref{fig:6}. The physics

699: underlying these curves has been made clear in a simple model

700: recently studied by Knigavko {\it et al.}\cite{r30} These authors

701: studied a model in which the charge carriers are coupled to a single

702: Einstein mode of unspecified origin. What they found was that stiffening

703: of this mode decreases the kinetic energy and hence increases the OS.

704: This is precisely the same mechanism that is operative in Fig.~\ref{fig:6}.

705: By applying a low frequency cutoff to the spin fluctuations in our

706: MMP-model we are decreasing the KE. This decrease in KE, present in the

707: underlying normal state below $T_c$, compensates for the increase in

708: KE intrinsic to the superconducting transition which results from the opening

709: of the superconducting gap. We note, however, that in our formulation

710: the OS, at any given temperature, is always below (although not very

711: much) its normal state value at this same temperature calculated with the

712: spectrum with a low frequency cutoff (open triangles). But this cutoff

713: is only operative in the superconducting state and is responsible for

714: making the open triangles fall above the solid squares.

715: The kinetic energy in the superconducting state with low

716: frequency cutoff is now less than the normal state kinetic energy

717: without cutoff.

718: Including the feedback mechanism of the formation of the

719: superconducting state on the spin susceptibility itself has the

720: net effect, at zero temperature (where it is largest),

721: of changing the sign of the KE

722: contribution to the condensation energy from that in BCS.

723: 

724: \section{Conclusion}

725: \label{sec:5}

726: 

727: We have used the Nearly Antiferromagnetic Fermi Liquid model to study

728: the effect of interactions on the optical sum and on the kinetic energy

729: in tight binding bands. Comparison of normal state results with

730: equivalent results when interactions are neglected showed that

731: temperature variations can be strongly affected by details of the

732: microscopic parameters involved in the spin fluctuation exchange

733: mechanism. Behaviors are possible which

734: can be quite different from the non interacting independent

735: particle model. Comparison with normal state

736: experimental data proves that the tight binding model of non

737: interacting particles is certainly not adequate to describe

738: properly the temperature dependence of the optical sum. (This

739: has also been observed by Benfatto {\it et al.}\cite{r42b}) Taking

740: into account interactions between the charge carriers makes the

741: tight binding model a viable model for the analysis of the

742: temperature dependence of the normal state optical sum. This

743: was demonstrated for the particular case of BSCCO and a particle

744: interaction modeled on the NAFFL. Other models,

745: like the one presented by Toschi {\it et al.}\cite{r42c} are also

746: capable to reproduce the temperature dependence

747: $I_\sigma(T)/I_\sigma(T=0)$ but they lack agreement with the

748: value of the optical sum at zero temperature.

749: 

750: When superconductivity is considered within an

751: Eliashberg formalism, the superconducting gap has $d$-wave

752: symmetry as a function of momentum in the two dimensional

753: Brillouin zone. The optical sum is found to decrease

754: with decreasing temperature for some range of parameters characterizing

755: the spin susceptibility but can also increase. This increase

756: cannot necessarily be interpreted as kinetic energy driven superconductivity.

757: In fact, in all cases considered, the optical sum is always lower,

758: at a given temperature in the superconducting state, than it is

759: in the corresponding normal state but, in some cases not by much.

760: Correspondingly, the kinetic energy is increased in the

761: superconducting state. What makes the optical sum and KE integral

762: continue to go up (in some cases) with decreasing temperature

763: is the fact that the interactions themselves introduce a

764: temperature dependence in the underlying normal state.

765: 

766: The results just described were obtained for a fixed

767: (i.e.: temperature independent) value

768: of the spin susceptibility. If

769: we consider the possibility that the spin fluctuation spectrum may itself

770: be modified\cite{r33,r34,r35,r44,r45,r46,r47,r48} by the onset of

771: superconductivity

772: and by temperature, even larger increases in the optical sum with

773: decreasing temperature can be obtained.

774: It is widely recognized that a generic feature of an electronic

775: mechanism of superconductivity is the possible gaping of the

776: excitation spectrum itself at small energies due to the opening

777: of the superconducting gap. This leads to a weakening of

778: interactions at small $\omega$ and to the so called collapse of

779: the inelastic scattering rate\cite{r44,r45,r46,r47,r48} which manifests

780: itself as a large peak in the temperature dependence of the

781: microwave conductivity. The weakening of the interaction

782: in the superconducting state through the opening of a low energy

783: gap in the spin susceptibility

784: corresponds to a decrease in KE in the superconducting state which can,

785: in the case considered, more than compensate for the intrinsic

786: increase that accompanies the formation of Cooper pairs

787: and, consequently, the OS rises with a larger slope in the

788: superconducting state than in the normal state just above $T_c$.

789: At $T_c$ there is no low energy gaping of the spin susceptibility and,

790: therefore, the mechanism for

791: KE reduction just described is not operative. In this sense

792: our model does not describe KE driven superconductivity.

793: 

794: \section*{Acknowledgment}

795:  

796: Research supported by the Natural Sciences and Engineering

797: Research Council of Canada (NSERC) and by the Canadian

798: Institute for Advanced Research (CIAR).

799: 

800: \begin{thebibliography}{99}

801: \bibitem{r1} T.~Timusk and B.~Statt, Rep.~Prog.~Phys. {\bf 62}, 61

802: (1999) and references therein.

803: \bibitem{r2} H.~Ding, T.~Yokoya, J.C.~Campuzano, T.~Takahashi,

804: M.~Randeria, M.R.~Norman, T.~Mochiku, K.~Kadowaki, and

805: J.~Giapintzakis, Nature (London) {\bf 382}, 51 (1996).

806: \bibitem{r3} A.G.~Loeser, Z.-X.~Shen, D.S.~Dessau, D.S.~Marshall,

807: C.H.~Park, P.~Fournier, and A.~Kapitulnik, Science {\bf 273}, 325 (1996).

808: \bibitem{r4} J.C.~Campuzano, M.R.~Norman, and M.~Randeria, in

809: {\it The Physics of Superconductivity: Conventional and Unconventional},

810: edited by K.H.~Bennemann and J.B.~Ketterson (Springer, Berlin, 2003)

811: Vol. 2, p. 167.

812: \bibitem{r5} Ch.~Renner, B.~Revaz, J.-Y.~Genoud, K.~Kadowaki, and

813: O.~Fischer, \prl {\bf 80}, 149 (1998).

814: \bibitem{r6} V.J.~Emery and S.A.~Kivelson, Nature (London) {\bf 374},

815: 134 (1995).

816: \bibitem{r7} E.~Carlson, V.J.~Emery, S.A.~Kivelson, and D.~Orgard, in

817: {\it The Physics of Superconductivity: Conventional and Unconventional},

818: edited by K.H.~Bennemann and J.B.~Ketterson (Springer, Berlin, 2003)

819: Vol. 2, p. 275.

820: \bibitem{r23a} T.~Eckl, W.~Hanke, and E.~Arrigoni, \prb {\bf 68},

821: 014505 (2003).

822: \bibitem{r8} Q.~Chen, I.~Kosztin, B.~Jank\'{o}, and K.~Levin, \prl

823: {\bf 81}, 4708 (1998).

824: \bibitem{r9} S.~Chakravarty, R.B.~Laughlin, D.K.~Morr, and

825: C.~Nayak, \prb {\bf 63}, 094503 (2001).

826: \bibitem{r10} B.~D\`{o}ra, A.~Virosztek, and K.~Maki, \prb

827: {\bf 65}, 155119 (2002), K.~Maki, B.~D\`{o}ra, M.~Kartsovnik,

828: A.~Virosztek, B.~Korin-Hamzic, and M.~Basletic, \prl {\bf 90},

829: 256402 (2003).

830: \bibitem{r11} S.~Chakravarty, H.-Y.~Kee, and C.~Nayak, Int.\ J.\ Mod.\

831: Phys.\ B {\bf 15}, 2901 (2001).

832: \bibitem{r12} J.-X.~Zhu, W.~Kim, C.S.~Ting, and J.P.~Carbotte,

833: \prl {\bf 87}, 197001 (2001).

834: \bibitem{r13} X.~Yang and C.~Nayak, \prb {\bf 65}, 064523 (2002).

835: \bibitem{r14} P.W.~Anderson, in {\it The Theory of Superconductivity

836: in the High-$T_c$ Cuprates\/} (Princeton University Press, Princeton,

837: 1998).

838: \bibitem{r15} J.E.~Hirsch, Physica C {\bf 199}, 305 (1992); Physica C

839: {\bf 201}, 347 (1992).

840: \bibitem{r16} J.E.~Hirsch and F.~Marsiglio, \prb {\bf 62}, 15131 (2000).

841: \bibitem{r17} J.E.~Hirsch, Science {\bf 295}, 2226 (2002).

842: \bibitem{r18} Wonkee Kim and J.P.~Carbotte, \prb {\bf 64}, 104501 (2001);

843: \prb {\bf 61}, R11886 (2000); \prb {\bf 63}, 140505(R) (2001).

844: \bibitem{r19} S.~Chakravarty, Euro. Phys. J. B {\bf 5}, 337 (1998).

845: \bibitem{r20} M.R.~Norman and C.~P\'epin, \prb {\bf 66}, 100506(R) (2002).

846: \bibitem{r21} M.R.~Norman and C.~P\'epin, Rep.\ Prog.\ Phys. {\bf 66},

847: 1547 (2003).

848: \bibitem{r22} L.~Benfatto, S.G.~Sharapov, and H.~Beck, Eur. Phys. J.

849: {\bf 39}, 469 (2004).

850: \bibitem{r23} A.E.~Karakozov, E.G.~Maksimov, and O.V.~Dologov,

851: Solid State Comm. {\bf 124}, 119 (2002).

852: \bibitem{r24} D.N.~Basov, S.I.~Woods, A.S.~Katz, E.J.~Singley, R.C.~Dynes,

853: M.~Xu, D.G.~Hinks, C.C.~Homes, and M.~Strongin, Science {\bf 283}, 49 (1999).

854: \bibitem{r25} A.S.~Katz, S.I.~Woods, E.J.~Singley, T.W.~Li,

855: M.~Xu, D.G.~Hinks, R.C.~Dynes, and D.N.~Basov, \prb {\bf 61}, 5930 (2000).

856: \bibitem{r26} H.J.A.~Molegraaf, C.~Presura, D.~van der Marel, P.H.~Kes,

857: and M.~Li, Science {\bf 295}, 22 (2002).

858: \bibitem{r27} A.F.~Santander-Syro, R.P.M.S.~Lobo, W.~Bontemps,

859: Z.~Konstantinovic, Z.Z.~Li, and H.~ Raffy, Europhys. Lett. {\bf 62},

860: 568 (2003).

861: \bibitem{r28} D.~van der Marel, H.J.A.~Molegraaf, C.~Presura, and

862: L.~Santoso, in {\it Concepts in Electron Correlations}, edited by

863: A.~Hewson and V.~Zlatic (Kluwer, 2003) p. 7.

864: \bibitem{r42a}M.~Ortolani, P.~Calvani, and S.~Lupi, \prl {\bf 94},

865: 067002 (2005).

866: \bibitem{r42b}B.L.~Benefatto, S.~Sharapov, N.~Andrenacci, and

867: H.~Beck, \prb {\bf 71}, 104511 (2005).

868: \bibitem{r42c}A.~Toschi, M.~Capone, M.~Ortoloani, S.~Lupi, P.~Calvani,

869: and C.~Castellani, cond-mat/0502528 (unpublished).

870: \bibitem{r42d}A.J.~Millis, A.~Zimmers, R.P.S.M.~Lobo, and N.~Bontemps,

871: cond-mat/0411172 (unpublished).

872: \bibitem{r29} A.V.~Boris, N.N.~Kovalera, O.V.~Dolgov, T.~Holden,

873: C.T.~Jin, B.~Keimer, and C.~Bernhard, Science {\bf 304}, 708 (2004).

874: \bibitem{r30} A.~Knigavko, J.P.~Carbotte, and F.~Marsiglio,

875: \prb {\bf 70}, 224501 (2004).

876: \bibitem{r31} A.V.~Chubukov, D.~Pines, and J.~Schmalian, in

877: {\it The Physics of Superconductivity: Conventional and Unconventional},

878: edited by K.H.~Bennemann and J.B.~Ketterson (Springer-Verlag, Berlin, 2003),

879: Vol. 2, p. 495.

880: \bibitem{r32}A.J.~Millis, H.~Monien, and D.~Pines,

881: \prb {\bf 42}, 167 (1990).

882: \bibitem{r33} E.~Schachinger, J.P.~Carbotte, and D.N.~Basov,

883: Europhys.~Lett. {\bf 54}, 380 (2001).

884: \bibitem{r34}J.P.~Carbotte, E.~Schachinger, and D.N.~Basov,

885: Nature (London) {\bf 401}, 354 (1999).

886: \bibitem{r35} E.~Schachinger and J.P.~Carbotte, \prb {\bf 62}, 9054 (2000).

887: \bibitem{r35a}D.~Branch and J.P.~Carbotte, Can.\ J.\ Phys. {\bf 77},

888: 531 (1999); J.\ Superconductivity {\bf 12}, 667 (1999);

889: {\bf 13}, 535 (2000).

890: \bibitem{r35b}P.~Monthoux and D.~Pines, \prb {\bf 47}, 6069 (1993);

891: {\bf 49}, 4261 (1994).

892: \bibitem{r35c}A.~Abanov, A.V.Chubukov, and J.~Schmalian,

893: Adv. in Physics {\bf 52}, 119 (2003).

894: \bibitem{r36} E.~Schachinger, J.J.~Tu, and J.P.~Carbotte. \prb {\bf 67},

895: 214508 (2003).

896: \bibitem{r37} E.~Schachinger and J.P.~Carbotte, \prb {\bf 64}, 094501

897: (2001); \prb {\bf 65}, 064514 (2002).

898: \bibitem{r38}A.~Abanov, A.V.~Chubukov, and J.~Schmalian, \prb {\bf 63},

899: 180510 (2001); J.\ Electron Spectrosc. {\bf 117}, 129 (2001).

900: \bibitem{r39} J.P.~Carbotte and E.~Schachinger, \prb {\bf 69}, 224501

901: (2004).

902: \bibitem{r40} E.~Schachinger and J.P.~Carbotte, 

903: in: {\it Models and Methods of High-TC Superconductivity: some Frontal

904: Aspects}, edited by J.K.~Srivastava and S.M.~Rao,

905: (Nova Science, Hauppauge, NY, 2003),  Vol. II, p. 73.

906: \bibitem{r43a}E.~Schachinger, J.P.~Carbotte, and F.~Marsiglio,

907: \prb {\bf 56}, 2738 (1997).

908: \bibitem{r41} P.~Monthoux, A.V.~Balatsky, and D.~Pines, \prl {\bf 67},

909: 3448 (1991); \prb {\bf 46}, 14 803 (1992).

910: \bibitem{r42}R.S.~Markiewicz, S.~Sahrakorpi, M.~Lindroos, Hsin Lin,

911: and A.~Bansil, cond-mat/0503064 (unpublished) and references therein.

912: \bibitem{r44}M.C.~Nuss, P.M.~Mankiewich, N.L. O'Malley, E.H.~Westwick,

913: and P.B.~Littlewood, \prl {\bf 66}, 3305 (1991).

914: \bibitem{r45}D.B.~Romero, C.D.~Porter, D.B.~Tanner, L.~Forro, D.~Mandrus,

915: L.~Mihaly, C.C.~Carr, and G.P.~Williams, \prl {\bf 68}, 1590 (1992).

916: \bibitem{r46}D.A.~Bonn, P.~Dosanjh, R.~Liang, and W.N.~Hardy, \prl

917: {\bf 68}, 2390 (1992).

918: \bibitem{r47}A.~Hosseini, R.~Harris, S.~Kamal, P.~Dosanjh, J.~Preston,

919: R.~Liang, W.N.~Hardy, and D.A.~Bonn, \prb {\bf 60}, 1349 (1999).

920: \bibitem{r48}E.J.~Nicol and J.P.~Carbotte, \prb {\bf 44}, R7741 (1991).

921: \end{thebibliography}

922: \end{document}

923: 

924: %%% Local Variables: 

925: %%% mode: latex

926: %%% TeX-master: t

927: %%% End: 

928: