cond-mat0405239/z.tex
1: % INJ
2: 
3: \documentclass[aps,superscriptaddress,showpacs,twocolumn]{revtex4}
4: %\documentclass[aps,prb,superscriptaddress,showpacs,twocolumn]{revtex4}
5: %\documentclass[aps,superscriptaddress,showpacs,preprint]{revtex4}
6: \usepackage{graphicx}
7: \usepackage{color}
8: \usepackage{amssymb}
9: \usepackage{amsmath}
10: 
11: \newcommand{\bk}{{\bf k}}
12: \newcommand{\bq}{{\bf q}}
13: \newcommand{\bQ}{{\bf Q}}
14: \newcommand{\bp}{{\bf p}}
15: \newcommand{\bx}{{\bf x}}
16: \newcommand{\br}{{\bf r}}
17: \newcommand{\Li}{{\mathop{\rm{Li}}\nolimits}}
18: \renewcommand{\Im}{{\mathop{\rm{Im}}\nolimits\,}}
19: \renewcommand{\Re}{{\mathop{\rm{Re}}\nolimits\,}}
20: \newcommand{\sgn}{{\mathop{\rm{sgn}}\nolimits\,}}
21: \newcommand{\Tr}{{\mathop{\rm{Tr}}\nolimits\,}}
22: \newcommand{\EF}{E_{\mathrm{F}}}
23: \newcommand{\kB}{k_{\mathrm{B}}}
24: \newcommand{\Green}{{\mathcal G}}
25: \newcommand{\dSC}{{\mathrm{dSC}}}
26: \newcommand{\dPG}{{\mathrm{dPG}}}
27: \newcommand{\dDW}{{\mathrm{dDW}}}
28: \newcommand{\Ret}{{\mathrm{R}}}
29: \newcommand{\Tau}{T_\tau}
30: 
31: 
32: \begin{document}
33: 
34: \title{Linear response theory around a localized impurity in the
35:    pseudogap regime of an anisotropic superconductor: precursor pairing
36:    \emph{vs} the $d$-density-wave scenario}
37: 
38: 
39: \author{N. Andrenacci}
40: \email{Natascia.Andrenacci@unine.ch}
41: \affiliation{Institut de Physique, Universit\'e de Neuch\^atel,
42:    CH-2000 Neuch\^atel, Switzerland} 
43: \author{G. G. N. Angilella}
44: \email{Giuseppe.Angilella@ct.infn.it}
45: \affiliation{Dipartimento di Fisica e Astronomia, Universit\`a di
46:    Catania,\\ and Istituto Nazionale per la Fisica della Materia,
47:    UdR di Catania,\\ Via S. Sofia, 64, I-95123 Catania, Italy}
48: \author{H. Beck}
49: \affiliation{Institut de Physique, Universit\'e de Neuch\^atel,
50:    CH-2000 Neuch\^atel, Switzerland} 
51: \author{R. Pucci}
52: \affiliation{Dipartimento di Fisica e Astronomia, Universit\`a di
53:    Catania,\\ and Istituto Nazionale per la Fisica della Materia,
54:    UdR di Catania,\\ Via S. Sofia, 64, I-95123 Catania, Italy}
55: 
56: \date{\today}
57: 
58: 
59: \begin{abstract}
60: \medskip
61: We derive the polarizability of an electron system in (i) the
62:    superconducting phase, with $d$-wave symmetry, (ii) the pseudogap
63:    regime, within the precursor pairing scenario, and (iii) the
64:    $d$-density-wave (dDW) state, characterized by a $d$-wave hidden order
65:    parameter, but no pairing.
66: Such a calculation is motivated by the recent proposals that
67:    imaging the effects of an isolated impurity may distinguish between
68:    precursor pairing and dDW order in the pseudogap regime of the
69:    high-$T_c$ superconductors.
70: In all three cases, the wave-vector dependence of the polarizability
71:    is characterized by an azymuthal modulation, consistent with the
72:    $d$-wave symmetry of the underlying state.
73: However, only the dDW result shows the fingerprints of nesting, with
74:    nesting wave-vector $\bQ=(\pi,\pi)$, albeit imperfect, due to a
75:    nonzero value of the hopping ratio $t^\prime /t$ in the band
76:    dispersion relation.
77: As a consequence of nesting, the presence of hole pockets is also
78:    exhibited by the $(\bq,\omega)$ dependence of the retarded
79:    polarizability.
80: \\
81: \pacs{% 
82: 74.25.Jb,
83: % Electronic structure
84: 73.20.Hb,
85: %Impurity and defect levels; energy states of adsorbed species
86: 74.20.-z
87: %Theories and models of superconducting state
88: }
89: \end{abstract} 
90: 
91: \maketitle
92: 
93: \section{Introduction}
94: 
95: Imaging of the electronic properties around an isolated
96:    nonmagnetic impurity such as Zn in the high-$T_c$ superconductors
97:    (HTS) has provided direct evidence of the unconventional nature of
98:    the superconducting state in the cuprates, and in particular of the
99:    $d$-wave symmetry of its order parameter below the critical
100:    temperature $T_c$ \cite{Pan:99,Hudson:99,Hudson:01,Pan:01}.
101: In the underdoped regime of the HTS, various models have
102:    been proposed to describe the pseudogap state above $T_c$.
103: 
104: Several experimental results provide substantial evidence of a pseudogap
105:    opening at the Fermi level in underdoped cuprates for $T_c < T <
106:    T^\ast$, even though no unique definition of the characteristic
107:    temperature $T^\ast$ is possible, as it generally depends
108:    on the actual experimental technique employed
109:    (see Refs.~\cite{Randeria:97-2,Timusk:99} for a review, and
110:    refs. therein). 
111: Also, the doping dependence of $T^\ast$ is still a matter of controversy
112:    \cite{Tallon:01}.
113: Owing to its $d$-wave symmetry, the pseudogap has been naturally
114:    interpreted in terms of precursor superconducting pairing.
115: In particular, the pseudogap has been associated to phase fluctuations
116:    of the order parameter above $T_c$ \cite{Emery:95a} (see
117:    Ref.~\cite{Loktev:01} for a review).
118: Within this precursor pairing scenario, the phase diagram of the HTS
119:    can be described as a crossover from Bose-Einstein
120:    condensation (in the underdoped regime) to BCS superconductivity
121:    (in the overdoped regime)
122:    \cite{Loktev:01,Randeria:95,Andrenacci:99,Strinati:00}. 
123: 
124: Recently, it has been proposed that many properties of the pseudogap
125:    regime may be explained within the framework of the so-called
126:    $d$-wave-density scenario (dDW)
127:    \cite{Chakravarty:01,Chakravarty:01a,Chakravarty:00}.
128: This is based on the idea that the pseudogap regime be characterized
129:    by a fully developed order parameter, at variance with the
130:    precursor pairing scenario, where a fluctuating order parameter is
131:    postulated.
132: The dDW state is an ordered state of unconventional kind, and is
133:    usually associated 
134:    with staggered orbital currents in the CuO$_2$ square lattice of
135:    the HTS \cite{Affleck:88,Marston:89,Kotliar:88,Schulz:89}.
136: Much attention has been recently devoted to show the consistency of
137:    the dDW scenario with several experimental properties of the HTS
138:    \cite{Chakravarty:01a}.
139: These include transport properties, such as the electrical and thermal
140:    conductivities \cite{Yang:02,Sharapov:03} and the Hall effect
141:    \cite{Chakravarty:02,Balakirev:03}, thermodynamic properties
142:    \cite{Kee:02,Wu:02}, time symmetry breaking \cite{Kaminski:02}, and
143:    angular resolved photoemission spectroscopy (ARPES)
144:    \cite{Chakravarty:03a}.
145: The possible occurrence of a dDW state in microscopic models of
146:    correlated electrons has been checked in ladder networks
147:    \cite{Marston:02}.
148: 
149: It has been recently proposed that
150:    direct imaging of the local density of states (LDOS) around an
151:    isolated impurity by means of scanning tunneling microscopy (STM)
152:    could help understanding the nature of the `normal' state in the
153:    pseudogap regime \cite{Zhu:01a,Wang:02,Morr:02,MoellerAndersen:03}.
154: The idea that an anisotropic superconducting gap should give rise to
155:    directly observable spatial features in the tunneling conductance
156:    near an impurity was suggested by Byers \emph{et al.}
157:    \cite{Byers:93}, whereas earlier studies \cite{Choi:90} had
158:    considered perturbations of the order parameter to occur within a
159:    distance of the order of the coherence length $\xi$ around an
160:    impurity.
161: Later, it was shown that an isolated impurity in a $d$-wave
162:    superconductor produces virtual bound states close to the Fermi
163:    level, in the nearly unitary limit \cite{Balatsky:95}.
164: Such a quasi-bound state should appear as a pronounced peak near the
165:    Fermi level in the LDOS at the impurity site \cite{Salkola:96}, as
166:    is indeed observed in Bi-2212 \cite{Pan:99} and YBCO
167:    \cite{Iavarone:02}.
168: 
169: In the normal state, the frequency-dependent LDOS at the nearest and
170:    next-nearest neighbor sites, with respect to the impurity site,
171:    should contain fingerprints of whether the pseudogap regime is
172:    characterized by precursor pairing \cite{Kruis:01} or dDW order
173:    \cite{Wang:02,Morr:02}. 
174: This is due to the fact that while pairing above $T_c$ without phase
175:    coherence is a precursor of Cooper pairing, and therefore of
176:    spontaneous breaking of U(1) gauge invariance, the dDW state can be
177:    thought as being characterized by the spontaneous breaking of
178:    particle-hole symmetry, in the same way as a charge density wave
179:    breaks pseudospin SU(2) symmetry \cite{Shen:97}.
180: The LDOS around a nonmagnetic impurity in both the dSC, the dDW and
181:    the competing dSC+dDW phases in the underdoped regime has been
182:    actually calculated \emph{e.g.} by Zhu \emph{et al.}
183:    \cite{Zhu:01a}.
184: 
185: In this context, a complementary information is that provided by the
186:    polarizability $F^\Ret (\bq,\omega)$ of the system, which gives a
187:    measure of the linear response of the charge density to an impurity
188:    potential.
189: In the case of $d$-wave superconductors, it has been demonstrated that
190:    the anisotropic dependence of the superconducting order parameter
191:    on the wave-vector $\bq$ gives rise to a clover-like azymuthal
192:    modulation of $F^\Ret (\bq,\omega)$ along the Fermi line for a 2D
193:    system \cite{Angilella:02f}.
194: 
195: These patterns in the $\bq$ dependence of $F^\Ret (\bq,\omega)$ are
196:    here confirmed also for a more realistic band for the cuprates.
197: In addition to that, the dDW result also shows fingerprints of the
198:    $\bQ=(\pi,\pi)$ nesting properties of such a state.
199: 
200: The paper is organized as follows.
201: In Sec.~\ref{sec:dSCdPG}, we review the expression of the
202:    polarizability for the $d$-wave superconducting state (dSC) and
203:    derive that of the $d$-wave pseudogap regime, within the precursor
204:    pairing scenario (dPG).
205: In Sec.~\ref{sec:dDW}, we derive the polarizability for the dDW state.
206: By allowing nonzero values of the hopping ratio $t^\prime /t$ in the
207:    dispersion relation \cite{Pavarini:01,Angilella:01,Angilella:03g},
208:    we will explicitly consider the case in which perfect nesting is
209:    destroyed.
210: Such a case is relevant for the study of the dDW state, given its
211:    particle-hole character.
212: In Sec.~\ref{sec:dSCdDW}, we consider the competition of
213:    dDW order with an subdominant dSC state in the underdoping regime.
214: In Sec.~\ref{sec:numerical}, we present our numerical results for the
215:    polarizability in the dSC, dPG, and dDW states, both in the static
216:    limit and as a function of frequency.
217: We eventually summarize and make some concluding remarks in
218:    Sec.~\ref{sec:conclusions}.
219: 
220: 
221: 
222: \section{Linear response function in the \lowercase{d}SC and
223:    \lowercase{d}PG states} 
224: \label{sec:dSCdPG}
225: 
226: Within linear response theory, the displaced charge density
227:    $\delta\rho(\br)$ by a scattering potential $V(\br)$ in the Born
228:    approximation is given by
229: \begin{equation}
230: \delta\rho(\br) = \int  V(\br^\prime ) F^\Ret (\br-\br^\prime
231:    ,\EF ) d\br^\prime ,
232: \label{eq:lr}
233: \end{equation}
234: which implicitly defines the linear response function $F^\Ret
235:    (\br,\EF)$ at the Fermi energy $\EF$.
236: Here and in the following we set the elementary charge $e=1$.
237: Its relevance in establishing the electronic structure of isolated
238:    impurities in normal metals and alloys has been earlier emphasized
239:    by Stoddart \emph{et al.} \cite{Stoddart:69,Jones:73-2}.
240: In momentum space, Eq.~(\ref{eq:lr}) readily translates into $\delta
241:    \rho(\bq) = V(\bq) F^\Ret (\bq,\EF)$, showing that, for a highly
242:    localized scattering potential in real space [$V(\br)=V_0
243:    \delta(\br)$, say], the Fourier transform $\delta\rho(\bq)$ of the
244:    displaced charge is simply proportional to $F^\Ret (\bq,\EF)$.
245: 
246: In the presence of superconducting pairing, the generalization of the
247:    linear response function is given by the density-density
248:    correlation function (polarizability) \cite{Prange:63}: 
249: \begin{eqnarray}
250: F(\bq,i\omega_\nu ) &=& \Tr \frac{1}{\beta} \sum_{\omega_n} \frac{1}{N} \sum_\bk
251:    \tau_3 \Green (\bk,i\omega_n ) \nonumber\\
252: &&\times \, \tau_3 \Green (\bk-\bq,i\omega_n
253:    -i\omega_\nu ) 
254: \label{eq:Prange}
255: \end{eqnarray}
256: where $\Green (\bk,i\omega_n )$ is the matrix Green's function in
257:    Nambu notation, $\beta=T^{-1}$ is the inverse temperature,
258:    $\omega_\nu = 2\nu\pi T$ is a bosonic Matsubara frequency, $\tau_i$ are the
259:    Pauli matrices in spinor space, the summations are performed over the $N$
260:    wave-vectors $\bk$ of
261:    the first Brillouin zone (1BZ) and all fermionic Matsubara
262:    frequencies $\omega_n = (2n+1)\pi T$, and the trace is over the
263:    spin indices.
264: Here and below we shall use units such that $\hbar = \kB
265:    = 1$ and lattice spacing $a=1$.
266: The retarded polarizability is defined as usual in terms of the
267:    analytic continuation as $F^\Ret (\bq,\omega) =
268:    F(\bq,i\omega_\nu \mapsto \omega+i0^+ )$.
269: In the normal state, Equation~(\ref{eq:Prange}) correctly reduces to
270:    the Lindhard function for the polarizability of a free electron gas
271:    \cite{Mahan:90}. 
272: 
273: In the following, by specifying the functional form of $\Green$ in
274:    the case of pairing with and without phase coherence, we will in turn
275:    derive the explicit expression for $F$ in the superconducting phase
276:    with a $d$-wave order parameter,
277:    and in the pseudogap regime, characterized by fluctuating
278:    $d$-wave order (precursor pairing scenario).
279: 
280: \subsection{Superconducting phase}
281: 
282: We assume the following BCS-like Hamiltonian:
283: \begin{equation}
284: H_\dSC = \sum_{\bk s} \xi_\bk c^\dag_{\bk s}
285:    c_{\bk s} + \sum_{\bk\bk^\prime} V_{\bk\bk^\prime}
286:    c^\dag_{\bk\uparrow} c^\dag_{-\bk\downarrow} c_{-\bk^\prime
287:    \downarrow} c_{\bk^\prime \uparrow} ,
288: \label{eq:HdSC}
289: \end{equation}
290: where $c^\dag_{\bk s}$ ($c_{\bk s}$) is a creation (annihilation)
291:    operator for an electron in the state with wave-vector $\bk$ and
292:    spin projection $s\in\{\uparrow,\downarrow\}$ along a specified
293:    direction, and $\xi_\bk = \epsilon_\bk -\mu$, with $\epsilon_\bk$
294:    the single-particle dispersion relation: 
295: \begin{equation}
296: \epsilon_\bk = -2t(\cos k_x + \cos k_y ) + 4t^\prime \cos k_x \cos k_y
297:    ,
298: \label{eq:disp}
299: \end{equation}
300: where $t=0.3$~eV, $t^\prime /t =0.3$ are tight binding
301:    hopping parameters 
302:    appropriate for the cuprate superconductors, and $\mu$ the chemical
303:    potential.
304: In Eq.~(\ref{eq:HdSC}), $V_{\bk\bk^\prime}$ is a model
305:    potential, which we assume to be separable and attractive in the
306:    $d_{x^2 -y^2}$-wave channel: $V_{\bk\bk^\prime} = -\lambda g_\bk
307:    g_{\bk^\prime}$, with $g_\bk = \frac{1}{2} (\cos k_x - \cos k_y )$
308:    and $\lambda>0$.
309: Under these assumptions, the Hamiltonian, Eq.~(\ref{eq:HdSC}), is
310:    characterized by a nonzero superconducting order parameter $\langle
311:    c_{\bk\uparrow} c_{-\bk\downarrow} \rangle$, leading to a nonzero
312:    $d$-wave mean-field gap $\Delta_\bk = \Delta_\circ g_\bk$ below the
313:    critical temperature $T_c$.
314: 
315: Making use of the explicit expression for the matrix Green's function
316:    $\Green_\dSC$ in the superconducting state \cite{AGD}:
317: \begin{equation}
318: \Green_\dSC (\bk,i\omega_n ) = \frac{i\omega_n \tau_0
319:    +\xi_\bk
320:    \tau_3 +\Delta_\bk \tau_1}{(i\omega_n )^2 - E_\bk^2 },
321: \label{eq:GdSC}
322: \end{equation}
323: with $E_\bk = (\xi_\bk^2 + \Delta_\bk^2 )^{1/2}$ the
324:    upper branch of the superconducting spectrum and $\tau_0$ the
325:    identity matrix in spin space, and performing the
326:    trace over spin indices
327:    and the summation over the internal frequency \cite{Mahan:90} in
328:    Eq.~(\ref{eq:Prange}), 
329:    we obtain the linear response function for a $d$-wave
330:    superconducting state \cite{Prange:63}:
331: \begin{widetext}
332: \begin{eqnarray}
333: F_\dSC (\bq,i\omega_\nu ) &=&
334: \frac{1}{N} \sum_\bk \left[
335: (u_\bk u_{\bk-\bq} - v_\bk v_{\bk-\bq} )^2 \left(
336: \frac{f(E_\bk ) - f(E_{\bk-\bq} )}{E_\bk -E_{\bk-\bq} -i\omega_\nu}
337: + \mathrm{H.c.} \right) \right.\nonumber \\
338: &&\left. +
339: (u_{\bk-\bq} v_\bk + u_\bk v_{\bk-\bq} )^2 \left(
340: \frac{f(E_\bk ) + f(E_{\bk-\bq} )-1}{E_\bk +E_{\bk-\bq} -i\omega_\nu}
341: + \mathrm{H.c.} \right) \right],
342: \label{eq:FdSC}
343: \end{eqnarray}
344: \end{widetext}
345: where $u_\bk^2 =
346:    \frac{1}{2} [1+ \xi_\bk /E_\bk ]$, $v_\bk^2 = 1-u_\bk^2$
347:    are the usual coherence factors of BCS theory, and
348:    $f(\epsilon)=(1+e^{\beta\epsilon} )^{-1}$ is the 
349:    Fermi function at temperature $T$.
350: In the limit of zero external frequency and $T\to0$,
351:    Eq.~(\ref{eq:FdSC}) reduces to the static polarizability studied in
352:    Ref.~\cite{Angilella:02f} for a $d$-wave superconductor.
353: 
354: 
355: 
356: \subsection{Pseudogap regime, within the precursor pairing scenario}
357: 
358: In the pseudogap regime, for $T_c < T <T^\ast$, within the precursor
359:    pairing scenario \cite{Loktev:01}, one assumes the existence of
360:    Cooper pairs characterized by a `binding energy' $\Delta_\bk$
361:    having the same symmetry of the true 
362:    superconducting gap below $T_c$, but no phase coherence.
363: In other words, no true off-diagonal long range order develops, and
364:    one rather speaks of a `fluctuating' order \cite{Emery:95a}.
365: This means that the quasiparticle spectrum $E_\bk = (\xi_\bk^2 +
366:    \Delta_\bk^2 )^{1/2}$ is still characterized by a pseudogap
367:    $\Delta_\bk = \Delta_\circ g_\bk$ opening at the Fermi energy with
368:    $d$-wave symmetry, but now without phase coherence.
369: Therefore, the diagonal elements of the matrix Green's function
370:    $\Green_\dPG$ coincide with those of its superconducting
371:    counterpart, Eq.~(\ref{eq:GdSC}), while the off-diagonal, anomalous
372:    elements are null:
373: \begin{equation}
374: \Green_\dPG (\bk,i\omega_n ) = \frac{i\omega_n \tau_0
375:    +\xi_\bk \tau_3 }{(i\omega_n )^2 - E_\bk^2 }.
376: \label{eq:GdPG}
377: \end{equation}
378: The effects due to a finite lifetime of the precursor Cooper
379:    pairs can be mimicked by adding a finite imaginary energy linewidth
380:    $i\Gamma$ to the dispersion relation entering Eq.~(\ref{eq:GdPG}),
381:    or by substituting the spectral functions associated with the
382:    quasiparticle states with `broadened' ones, as discussed in
383:    Appendix~\ref{app:lifetime}.
384: The relation between the two approaches and with analytical
385:    continuation has been discussed in Appendix~A of
386:    Ref.~\cite{Andrenacci:03}.
387: 
388: Within this precursor pairing scenario, Equation~(\ref{eq:Prange}) in the
389:    pseudogap regime then reads:
390: \begin{widetext}
391: \begin{eqnarray}
392: F_\dPG (\bq,i\omega_\nu ) &=&
393: \frac{1}{N} \sum_\bk \left[
394: (u_\bk^2 u_{\bk-\bq}^2 + v_\bk^2 v_{\bk-\bq}^2 ) \left(
395: \frac{f(E_\bk ) - f(E_{\bk-\bq} )}{E_\bk -E_{\bk-\bq} -i\omega_\nu}
396: + \mathrm{H.c.} \right) \right.\nonumber \\
397: &&\left. +
398: (u_{\bk-\bq}^2 v_\bk^2 + u_\bk^2 v_{\bk-\bq}^2 ) \left(
399: \frac{f(E_\bk ) + f(E_{\bk-\bq} )-1}{E_\bk +E_{\bk-\bq} -i\omega_\nu}
400: + \mathrm{H.c.} \right) \right],
401: \label{eq:FdPG}
402: \end{eqnarray}
403: \end{widetext}
404: where we are implicitly assuming $T_c < T < T^\ast$.
405: 
406: 
407: \section{Linear response function in the \lowercase{d}DW state} 
408: \label{sec:dDW}
409: 
410: The mean-field Hamiltonian for the $d$-density-wave state is
411:    \cite{Chakravarty:01}:
412: \begin{equation}
413: H_\dDW = \sum_{\bk s} [\xi_\bk c^\dag_{\bk s}
414:    c_{\bk s} + i D_\bk c^\dag_{\bk s} c_{\bk+\bQ s} ],
415: \label{eq:HdDW}
416: \end{equation}
417: where the summation is here restricted to all wave-vectors $\bk$ belonging
418:    to the first Brillouin zone, $\bQ = (\pi,\pi )$ is the dDW ordering
419:    wave-vector, and $D_\bk = D_\circ g_\bk$ is the dDW order
420:    parameter.
421: As anticipated above, the dDW state is characterized by a broken
422:    symmetry and a well-developed order parameter, at variance with the
423:    precursor pairing scenario of the pseudogap regime.
424: Such a state is associated to staggered orbital currents circulating
425:    with alternating sense in the neighboring plaquettes of the
426:    underlying square lattice.
427: As a result, the unit cell in real space is doubled, and the Brillouin
428:    zone is correspondingly halved.
429: At variance with other `density waves', the dDW order is characterized
430:    not by charge or spin modulations, but rather by current
431:    modulations.
432: 
433: The nonzero, singlet order parameter $\Phi_\bQ$ breaks pseudospin
434:    invariance in the particle-hole space:
435: \begin{equation}
436: \langle c^\dag_{\bk+\bQ s} c_{\bk s^\prime} \rangle = i \Phi_\bQ g_\bk
437:    \delta_{ss^\prime} .
438: \label{eq:dDWop}
439: \end{equation}
440: Whereas it possesses $d$-wave symmetry, as expected, its imaginary
441:    value leads to the breaking of a relatively large number of
442:    symmetries, such as time reversal, parity, translation by a lattice
443:    spacing, and rotation by $\pi/2$, although the product of any two
444:    of these is preserved (see Ref.~\cite{Sharapov:03} for a detailed
445:    analysis).
446: 
447: Introducing the spinor $\Psi_{\bk s}^\dag = (c^\dag_{\bk s} \,\,
448:    c^\dag_{\bk+\bQ s} )$, the dDW Hamiltonian, Eq.~(\ref{eq:HdDW}), can be
449:    conveniently rewritten as \cite{Yang:02,Sharapov:03}:
450: \begin{equation}
451: H_\dDW = {\sum_{\bk s}}^\prime \Psi_{\bk s}^\dag [(\epsilon_\bk^+
452:    -\mu ) \tau_0 + \epsilon_\bk^- \tau_3 + D_\bk \tau_1 ] \Psi_{\bk s} ,
453: \end{equation}
454: where $\epsilon_\bk^\pm = \frac{1}{2} (\epsilon_\bk \pm
455:    \epsilon_{\bk+\bQ} )$, and the prime restricts the summation over
456:    wave-vectors $\bk$ belonging to the reduced (`magnetic') Brillouin
457:    zone only.
458: Notice that $\epsilon^\pm_{\bk+\bQ} = \mp\epsilon_\bk^\pm$.
459: Correspondingly, the matrix Green's function at the imaginary time
460:    $\tau$ can be defined as $\Green_\dDW (\bk,\tau) = -\langle \Tau
461:    \Psi_{\bk s} (\tau) \Psi^\dag_{\bk s} (0) \rangle$, whose inverse reads
462:    \cite{Morr:02,Sharapov:03}: 
463: \begin{equation}
464: \Green_\dDW^{-1} (\bk,i\omega_n ) =
465: \begin{pmatrix}
466: i\omega_n -\xi_\bk & iD_\bk \\
467: -iD_\bk & i\omega_n -\xi_{\bk+\bQ} 
468: \end{pmatrix}.
469: \label{eq:iGdDW}
470: \end{equation}
471: In the case of perfect nesting ($t^\prime = 0$) for the dispersion
472:    relation, Eq.~(\ref{eq:disp}), Sharapov \emph{et al.}
473:    \cite{Sharapov:03} explicitly find
474: \begin{equation}
475: \Green_\dDW (\bk,i\omega_n ) = \frac{(i\omega_n + \mu)\tau_0 +
476:    \epsilon_\bk \tau_3 -D_\bk \tau_2}{(i\omega_n + \mu)^2 -
477:    \epsilon_\bk^2 - D_\bk^2} ,
478: \label{eq:GdDWSharapov}
479: \end{equation}
480: to be compared and contrasted with Eq.~(\ref{eq:GdSC}) for the
481:    superconducting phase.
482: Notice, in particular, the different way the chemical potential $\mu$
483:    enters the two expressions.
484: 
485: In the general case ($t^\prime \neq 0$), perfect nesting is lost, and
486:    we have to refer to the general form of $\Green_\dDW^{-1}$,
487:    Eq.~(\ref{eq:GdDW}).
488: One finds:
489: \begin{eqnarray}
490: \Green_\dDW (\bk,i\omega_n ) &=&
491: \frac{1}{(i\omega_n -E_\bk^+ )(i\omega_n -E_\bk^- )}
492:    \nonumber \\
493: &&\times\,\begin{pmatrix}
494: i\omega_n -\xi_{\bk+\bQ} & -iD_\bk \\
495: iD_\bk & i\omega_n -\xi_\bk
496: \end{pmatrix},
497: \label{eq:GdDW}
498: \end{eqnarray}
499: where $E_\bk^\pm = -\mu + \epsilon_\bk^+ \pm \sqrt{(\epsilon_\bk^- )^2
500:    +D_\bk^2}$ are the two branches of the quasiparticle spectrum
501:    obtained by diagonalizing Eq.~(\ref{eq:HdDW}) \cite{Yang:02}.
502: Notice that $E_{\bk+\bQ}^\pm = E^\pm_\bk$.
503: 
504: In the limit $t^\prime =0$, Eq.~(\ref{eq:GdDW}) correctly reduces to
505:    Eq.~(\ref{eq:GdDWSharapov}), even though it is not straightforward
506:    to express Eq.~(\ref{eq:GdDW}) in the same compact matrix notation.
507: In the limit of perfect nesting ($t^\prime =0$), the dispersion
508:    relation, Eq.~(\ref{eq:disp}), is antisymmetric with respect to
509:    particle-hole conjugation, $\epsilon_{\bk+\bQ} = -\epsilon_\bk$.
510: As a result, $E_\bk^\pm = -\mu \pm (\epsilon_\bk^2 + D_\bk^2 )^{1/2}$,
511:    which is to be contrasted with the quasiparticle spectrum of the
512:    superconducting state or the pseudogap state within the precursor
513:    pairing scenario, $\pm E_\bk = \pm [(\epsilon_\bk -\mu)^2 +
514:    \Delta_\bk^2 ]^{1/2}$.
515: The difference comes again from the fact that the Bogoliubov
516:    excitations in the dSC and the dPG states are Cooper pairs, while
517:    the dDW ordered state is characterized by particle-hole mixture
518:    \cite{Kee:02,MoellerAndersen:03}. 
519: 
520: The polarizability in the dDW state is derived in
521:    Appendix~\ref{app:poldDW}.
522: We just quote here the final result, which can be cast in compact
523:    matrix notation as:
524: \begin{eqnarray}
525: F_\dDW (\bq,i\omega_\nu ) &=&
526: \Tr \frac{1}{\beta} \sum_{\omega_n} \frac{1}{N} {\sum_\bk}^\prime
527: \kappa \Green_\dDW (\bk,i\omega_n )
528: \nonumber\\
529: &&\times \, \kappa \Green_\dDW (\bk-\bq,i\omega_n -i\omega_\nu ),
530: \label{eq:PrangedDW}
531: \end{eqnarray}
532: where now $\kappa = \tau_0 + \tau_1$, and $\Green_\dDW$ is
533:    the matrix Green's function for the dDW state, Eq.~(\ref{eq:GdDW}).
534: Performing the frequency summation \cite{Mahan:90}, one eventually
535:    finds:
536: \begin{equation}
537: F_\dDW (\bq,i\omega_\nu ) = \frac{1}{N} {\sum_\bk}^\prime
538: \sum_{i,j=\pm} \frac{f(E_\bk^i ) - f(E_{\bk-\bq}^j )}{E_\bk^i -
539:    E_{\bk-\bq}^j -i\omega_\nu} .
540: \label{eq:FdDW}
541: \end{equation}
542: 
543: 
544: \section{Competition between \lowercase{d}\uppercase{SC} and
545:    \lowercase{d}\uppercase{DW} orders}
546: \label{sec:dSCdDW}
547: 
548: In the underdoped regime, it has been predicted on phenomenological
549:    grounds that the dDW order should compete with a subdominant dSC phase
550:    \cite{Chakravarty:01}.
551: This has been confirmed by model calculations at the mean-field level
552:    \cite{Zhu:01,Wu:02}, showing that indeed an existing broken
553:    symmetry of dDW kind at high temperature suppresses that critical
554:    temperature for the subdominant dSC ordered phase.
555: Recently, the competition between dDW and dSC orders has been shown to
556:    be in agreement with the unusual $T$-dependence of the restricted
557:    optical sum rule, as observed in the underdoped HTS
558:    \cite{Benfatto:03}.
559: 
560: In order to take into account for the competition between the dSC and dDW
561:    orders at finite temperature, one has to separately consider the
562:    electron states within the two inequivalent halves of the Brillouin
563:    zone.
564: Therefore, it is convenient to make use of the 4-components Nambu
565:    spinor $\Psi_\bk^\dag \equiv (\Psi_{\bk\uparrow}^\dag ~~
566:    \Psi_{-\bk\downarrow}^\top )$, or explicitly:
567: \begin{equation}
568: \Psi_\bk = \begin{pmatrix}
569: c_{\bk \uparrow} \\
570: c_{\bk+\bQ  \uparrow} \\
571: c^\dag_{-\bk \downarrow} \\
572: c^\dag_{-\bk-\bQ \downarrow}
573: \end{pmatrix} .
574: \end{equation}
575: At the mean-field level, the Hamiltonian for the competing dSC and
576:    dDW phases thus reads
577: \begin{equation}
578: H_{\dSC + \dDW} = {\sum_\bk}^\prime \Psi_\bk^\dag \hat{H}_\bk \Psi_\bk ,
579: \end{equation}
580: where $\hat{H}_\bk$ is the $4\times 4$ Hermitean matrix defined by
581: \begin{equation}
582: \hat{H}_\bk = \begin{pmatrix}
583: \xi_\bk & iD_\bk & \Delta_\bk & 0 \\
584: -iD_\bk & \xi_{\bk+\bQ} & 0 & -\Delta_\bk \\
585: \Delta_\bk^\ast & 0 & -\xi_{-\bk} & iD_{-\bk} \\
586: 0 & -\Delta^\ast_\bk & -iD_{-\bk} & - \xi_{-\bk-\bQ} 
587: \end{pmatrix},
588: \label{eq:Hmatrix}
589: \end{equation}
590: with real eigenvalues $E_{\bk i}$ (here, $i=0,\ldots 3$) given by
591: \begin{eqnarray}
592: E^2_{\bk i} &=& \Delta_\bk^2 + D_\bk^2 - \xi_\bk \xi_{\bk + \bQ}
593: \nonumber\\
594: \!\!\!\!\!
595: &&+ (\xi_\bk + \xi_{\bk + \bQ} ) 
596: \left[
597: \epsilon_\bk^+ -\mu \pm \sqrt{(\epsilon_\bk^- )^2 +D_\bk^2}
598: \right],
599: \label{eq:Eki}
600: \end{eqnarray}
601: and orthonormal eigenvectors $|\bk i\rangle$ (for given $\bk$ in
602:    the reduced 1BZ).
603: It may be straightforwardly checked that $E_{\bk i}$ reduces to the dSC
604:    superconducting spectrum, $\pm E_\bk$, and to the dDW quasiparticle
605:    dispersion 
606:    relations, $E_\bk^\pm$, in the limits $D_\circ =0$ (pure dSC) and
607:    $\Delta_\circ = 0$ (pure dDW), respectively, when the halving of
608:    the 1BZ is removed.
609: 
610: In the particle-hole symmetric case ($t^\prime = 0$), making use of
611:    the nesting properties described in Sec.~\ref{sec:dDW},
612:    Eq.~(\ref{eq:Eki}) simplifies as:
613: \begin{equation}
614: E_{\bk i} = \pm \sqrt{\Delta_\bk^2 + \left( \mu \pm
615:    \sqrt{\epsilon_\bk^2 + D^2_\bk } \right)^2} .
616: \end{equation}
617: For $\mu=0$, the four branches of the spectrum degenerate into the Dirac cone
618:    \cite{Lee:97,Lee:97a}
619: \begin{equation}
620: E_{\bk,0\equiv 2,1\equiv 3} = \pm \sqrt{\epsilon_\bk^2 + \Delta_\bk^2
621:    + D^2_\bk} ,
622: \label{eq:cones}
623: \end{equation}
624: thus showing that the two gaps have the same role, \emph{i.e.} the
625:    system may be equivalently described as a dDW or a dSC
626:    superconductor, with a $d$-wave gap $\sqrt{\Delta_\bk^2 + D^2_\bk}$
627:    in either case.
628: Either a non-zero hopping ratio ($t^\prime /t \neq 0$) or a hole-doping away
629:    from half-filling ($\mu\neq 0$ when $t^\prime = 0$) destroys this
630:    particular symmetry, and one has to resort to the eigenvalues
631:    $E_{\bk i}$, Eq.~(\ref{eq:Eki}).
632: 
633: In order to obtain the Green's functions in the dSC+dDW case, it is
634:    useful to introduce the $4\times 4$ matrices ($i,j=0,\ldots 3$)
635: \begin{equation}
636: \Gamma_{ij} = \tau_i \otimes \tau_j ,
637: \end{equation}
638: whose algebra is given by
639: \begin{equation}
640: \Gamma_{ij} \Gamma_{lm} = i \varepsilon_{ijk} i \varepsilon_{lmn}
641:    \Gamma_{kn} ,
642: \end{equation}
643: where $\varepsilon_{ijk}$ is the totally antisymmetric Levi-Civita
644:    tensor, and
645: \begin{equation}
646: \Gamma_{33}^\pm = \frac{1}{2} (\Gamma_{30} \pm \Gamma_{33} ).
647: \end{equation}
648: Then the matrix Hamiltonian Eq.~(\ref{eq:Hmatrix}) takes the form
649: \begin{equation}
650: \hat{H}_\bk = \epsilon_\bk \Gamma_{33}^+ + \epsilon_{\bk+\bQ}
651:    \Gamma_{33}^- - \mu \Gamma_{30} - D_\bk \Gamma_{02} + \Delta_\bk
652:    \Gamma_{13} ,
653: \end{equation}
654: whence the inverse Green's function (now a $4\times 4$ matrix in Nambu
655:    space) straightforwardly follows as 
656: \begin{equation}
657: \Green^{-1}_{\dSC+\dDW} (\bk ,i\omega_n ) =
658: i\omega_n \Gamma_{00} - \hat{H}_\bk .
659: \end{equation}
660: As in the dDW case, Eq.~(\ref{eq:PrangedDW}) (see also
661:    Appendix~\ref{app:poldDW}), the linear response function in the
662:    dSC+dDW case can be given a compact matrix form as
663: \begin{eqnarray}
664: F_{\dSC+\dDW} (\bq,i\omega_\nu ) &=&
665: \Tr \frac{1}{\beta} \sum_{\omega_n} \frac{1}{N} {\sum_\bk}^\prime
666: \hat{\kappa} \Green_{\dSC+\dDW} (\bk,i\omega_n )
667: \nonumber\\
668: &&\!\!\!\!\!\!\!
669: \times \, \hat{\kappa} \Green_{\dSC+\dDW} (\bk-\bq,i\omega_n -i\omega_\nu ),
670: \label{eq:PrangedSCdDW}
671: \end{eqnarray}
672: where now the vertex matrix in the $4\times 4$ Nambu spinor space is
673:    given by
674: \begin{equation}
675: \hat{\kappa} = \tau_3 \otimes \kappa = \begin{pmatrix} \tau_0 + \tau_1 & 0
676:    \\
677: 0 & -\tau_0 -\tau_1 \end{pmatrix} .
678: \end{equation}
679: Finally, it can be shown that Eq.~(\ref{eq:PrangedSCdDW}) also admits
680:    the following spectral decomposition, analogous to
681:    Eq.~(\ref{eq:FdDW}):
682: \begin{eqnarray}
683: F_{\dSC+\dDW} (\bq,i\omega_\nu ) &=& \nonumber \\
684: &&\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!
685: \frac{1}{N} {\sum_\bk}^\prime
686: \sum_{i,j} \frac{f(E_{\bk i} ) - f(E_{\bk-\bq j}
687:    )}{E_{\bk i} - E_{\bk-\bq j} -i\omega_\nu} u_{ij}
688:    (\bk,\bq) , \nonumber \\
689: \label{eq:FdSCdDW}
690: \end{eqnarray}
691: where $u_{ij} (\bk,\bq) = \Tr (\hat{\kappa} P_{\bk i} \hat{\kappa}
692:    P_{\bk-\bq j} )$, and $P_{\bk i} =
693:    |\bk i\rangle\langle\bk i|$ is the 
694:    orthonormal projector operator on the $i$ eigenstate of the
695:    matrix Hamiltonian, Eq.~(\ref{eq:Hmatrix}).
696: 
697: 
698: \section{Numerical results and discussion}
699: \label{sec:numerical}
700: 
701: We have evaluated numerically the polarizability for the dPG and the
702:    dDW cases, Eqs.~(\ref{eq:FdPG}) and (\ref{eq:FdDW}), and
703:    for the mixed dSC+dDW case, Eq.~(\ref{eq:FdSCdDW}), as a function of
704:    the relevant variables. 
705: Our numerical results for the pure dSC case turn out to be
706:    very similar to 
707:    the dPG case (at least over the range of variables considered
708:    below), and will not be shown here.
709: In the dPG and in the pure dDW cases,
710: we adopt the following set of parameters, which are believed to be
711:    relevant for the cuprate superconductors: $t=0.3$~eV, $t^\prime /t
712:    = 0.3$, $\mu=-t$, corresponding to a hole-like Fermi line and a
713:    hole doping $\sim 14.3\%$, $\Delta_\circ = D_\circ = 0.06$~eV 
714:    in the dPG and in the dDW cases, respectively
715:    \cite{Chakravarty:03a}, and $T=100$~K. 
716: 
717: 
718: 
719: \begin{figure}[t]
720: \begin{center}
721: \begin{minipage}[c]{0.49\columnwidth}
722: \includegraphics[height=\textwidth,angle=-90]{lrpg_pgplot_01.ps}
723: \end{minipage}
724: \begin{minipage}[c]{0.49\columnwidth}
725: \includegraphics[height=1.45\textwidth,angle=-90]{ddw0.ps}
726: \end{minipage}\\
727: \begin{minipage}[c]{0.49\columnwidth}
728: \begin{flushleft}
729: (a)
730: \end{flushleft}
731: \end{minipage}
732: \begin{minipage}[c]{0.49\columnwidth}
733: \begin{flushleft}
734: (b)
735: \end{flushleft}
736: \end{minipage}
737: \end{center}
738: \caption{Static polarizability for the dPG state in momentum
739:    space, $F_\dPG(\bq,0)$, Eq.~(\protect\ref{eq:FdPG}) [in eV$^{-1}$,
740:    panel (a)], for $\Delta_\circ = 0.06$~eV, $T=100$~K, and $\mu=-t =
741:    -0.3$~eV, corresponding to the hole-like 
742:    Fermi line in the normal state shown in panel (b).
743: Panel (b) also reports the special points $\Gamma=(0,0)$,
744:    $X=(\pi,0)$, $M=(\pi,\pi)$, along with the points $A$ and $B$ where
745:    the Fermi line $\xi_\bk = 0$ intersects the
746:    symmetry contour $\Gamma$--$X$--$M$--$\Gamma$.
747: }
748: \label{fig:qdPG}
749: \end{figure}
750: 
751: \begin{figure}[t]
752: \begin{center}
753: \begin{minipage}[c]{0.49\columnwidth}
754: \includegraphics[height=\textwidth,angle=-90]{lrddw_pgplot_01.ps}
755: \end{minipage}
756: \begin{minipage}[c]{0.49\columnwidth}
757: \includegraphics[height=1.45\textwidth,angle=-90]{ddw2.ps}
758: \end{minipage}\\
759: \begin{minipage}[c]{0.49\columnwidth}
760: \begin{flushleft}
761: (a)
762: \end{flushleft}
763: \end{minipage}
764: \begin{minipage}[c]{0.49\columnwidth}
765: \begin{flushleft}
766: (b)
767: \end{flushleft}
768: \end{minipage}
769: \end{center}
770: \caption{Static polarizability for the pure dDW state in momentum
771:    space, $F_\dDW(\bq,0)$, Eq.~(\protect\ref{eq:FdDW}) [in eV$^{-1}$,
772:    panel (a)], for $D_\circ = 0.06$~eV, $T=100$~K, and
773:    $\mu=-t=-0.3$~eV, corresponding to the hole pockets 
774:    $E_\bk^- = 0$ shown in panel (b).
775: Panel (b) also reports the special points $\Gamma=(0,0)$,
776:    $X=(\pi,0)$, $M=(\pi,\pi)$, along with the points $B_1$ and $B_2$ where
777:    the line $E_\bk^- = 0$ intersects the
778:    symmetry contour $\Gamma$--$X$--$M$--$\Gamma$.
779: }
780: \label{fig:qdDW}
781: \end{figure}
782: 
783: 
784: \subsection{Zero external frequency}
785: \label{ssec:static}
786: 
787: In order to make contact with earlier work \cite{Angilella:02f}, we
788:    first consider the case of zero external (bosonic) frequency,
789:    $\omega_\nu = 0$, in the time-ordered polarizabilities,
790:    Eqs.~(\ref{eq:FdPG}) and (\ref{eq:FdDW}).
791: 
792: Our numerical results for the wave-vector dependence of $F(\bq,0)$
793:    over the 1BZ in the dPG and in the pure dDW cases are shown in
794:    Figs.~\ref{fig:qdPG} and \ref{fig:qdDW}, respectively.
795: As a result of the $d$-wave symmetry of both the pseudogap within the
796:    precursor pairing scenario, and of the dDW order parameter, $F(\bq,0)$
797:    is characterized by a four-lobed pattern or azymuthal modulation
798:    \cite{Angilella:02f}.
799: However, the dDW case is also characterized by the presence of `hole
800:    pockets', centered around $\bQ/2 = (\pi/2,\pi/2)$ and symmetry related
801:    points, due to the (albeit 
802:    imperfect) nesting properties of the dDW state, with nesting vector
803:    $\bQ=(\pi,\pi)$ (see Fig.~\ref{fig:qdDW}b).
804: Such a feature is reflected in the $\bq$ dependence of $F_\dDW
805:    (\bq,0)$, which is characterized by local maxima at the hole
806:    pockets, for the value $\mu=-t$ of the chemical potential
807:    considered here.
808: (Other values of the chemical potentials give rise to analogous
809:    features, which are absent in the dPG case.)
810: 
811: Fig.~\ref{fig:qdSCdDW} shows our numerical results for the static
812:    polarizability $F(\bq,0)$ in the mixed dSC+dDW case,
813:    Eq.~(\ref{eq:FdSCdDW}).
814: Representative values of the amplitudes of the dSC and dDW order
815:    parameters have been taken as in Ref.~\cite{Zhu:01}, \emph{viz.}
816:    $\Delta_\circ = 0.1 t = 0.03$~eV, 
817:    $D_\circ = 0.08 t = 0.024$~eV, 
818:    at $T=0.01t\simeq 35$~K,
819: for a particle-like Fermi line in the underdoped regime ($t=0.3$~eV,
820:    $t^\prime /t = 0.3$, $\mu = -0.2016$~eV).
821: Panel (b) of Fig.~\ref{fig:qdSCdDW} shows the contour plot of the
822:    eigenvalue spectrum $E_{\bk i}$, Eq.~(\ref{eq:Eki}).
823: The latter is characterized by pronounced minima near the hot spots at
824:    $\bQ /2$, which evolve into cone-like nodes in the limit of
825:    pure dSC ($D_\circ = 0$), or in the very special case $t^\prime =
826:    \mu =0$ [see Eq.~(\ref{eq:cones})].
827: Accordingly, Fig.~\ref{fig:qdSCdDW}a for the static polarizability
828:    $F(\bq,0)$ over the 1BZ is characterized by local maxima at the hot
829:    spots centered around $\bQ/2$, as is the case in the pure dDW case
830:    (cf. Fig.~\ref{fig:qdDW}a).
831: Whereas the precise behavior of the static polarizability $F(\bq,0)$
832:    is of course determined by the actual amount of dSC+dDW mixing at a
833:    given temperature and doping, we can conclude that a sizeable dDW
834:    component manifests itself through the appearance of hole pockets
835:    centered around $\bQ/2$ in the $\bq$-dependence of $F(\bq,0)$, also
836:    in the presence of a dSC condensate.
837: 
838: We have next evaluated the spatial dependence of $F(\br,0)$ (not
839:    shown), by Fourier transforming $F(\bq,0)$ to real space.
840: While $F(\br,0)$ is characterized by Friedel-like oscillations as
841:    $|\br|$ increases from the impurity site, as expected
842:    \cite{Zhu:01,Angilella:02f}, these radial, damped oscillations are
843:    superimposed by an azymuthal modulation, due to
844:    the $d$-wave symmetry of the normal state, both in the dPG and in
845:    the dDW cases.
846: As a consequence, $F(\br,0)$ is characterized by a checkerboard pattern,
847:    closely related to the symmetry of the underlying square lattice,
848:    with local 
849:    maxima on the nearest neighbor and local minima on the next-nearest
850:    neighbor sites.
851: Since these features are common to both the dPG and dDW cases, the
852:    spatial dependence of the charge density oscillations is not
853:    directly helpful in distinguishing between the dPG and dDW states.
854: However, real-space and wave-vector dependences of several quantities
855:    of interest for STM studies can be easily connected by means of
856:    Fourier transform scanning tunneling microscopy (FT-STM) techniques
857:    (see, \emph{e.g.}, Ref.~\cite{Capriotti:03}, and refs. therein).
858: Such a technique has been proved very effective in detecting
859:    large-amplitude Friedel oscillations of the electron density on the
860:    Be(0001) \cite{Sprunger:97,Petersen:98} and Be(10$\bar{1}$0)
861:    surfaces \cite{Briner:98}, and has been recently discussed in
862:    connection with experimental probes of fluctuating stripes in the
863:    HTS \cite{Kivelson:03}.
864: In particular, FT-STM
865:    experiments \cite{Sprunger:97,Petersen:98,Briner:98} have evidenced
866:    the role of correlation and reduced dimensionality in establishing
867:    such `giant' Friedel oscillations in the electron density.
868: 
869: 
870: 
871: \begin{figure}[t]
872: \begin{center}
873: \begin{minipage}[c]{0.49\columnwidth}
874: \includegraphics[height=\textwidth,angle=-90]{lrmixedr_pgplot_01.ps}
875: \end{minipage}
876: \begin{minipage}[c]{0.49\columnwidth}
877: \includegraphics[height=1.45\textwidth,angle=-90]{bands_mixed.ps}
878: \end{minipage}\\
879: \begin{minipage}[c]{0.49\columnwidth}
880: \begin{flushleft}
881: (a)
882: \end{flushleft}
883: \end{minipage}
884: \begin{minipage}[c]{0.49\columnwidth}
885: \begin{flushleft}
886: (b)
887: \end{flushleft}
888: \end{minipage}
889: \end{center}
890: \caption{Static polarizability for the mixed dSC+dDW state
891:    in momentum space, $F_{\dSC+\dDW} (\bq,0)$,
892:    Eq.~(\protect\ref{eq:FdSCdDW}) [in eV$^{-1}$, panel (a)], for
893:    $\Delta_0 = 0.1t = 0.03$~eV, $D_\circ = 0.08t=0.024$~eV, $T=0.01
894:    t\simeq 35$~K \protect\cite{Zhu:01}, and $\mu=-0.2016$~eV,
895:    corresponding to a 
896:    particle-like Fermi line closed around the $\Gamma$ point
897:    (underdoped regime).
898: Panel (b) shows the contour plots of the eigenvalue spectrum
899:    $E_{\bk i}$, Eq.~\protect\ref{eq:Eki}.
900: }
901: \label{fig:qdSCdDW}
902: \end{figure}
903: 
904: 
905: 
906: 
907: \subsection{Frequency dependence}
908: \label{ssec:dynamic}
909: 
910: We have next evaluated the frequency dependence of the
911:    retarded polarizabilities, in both the dPG and the dDW cases.
912: Figures~\ref{fig:opg} and \ref{fig:oddw} show our numerical results
913:    for the $\omega$ dependence of $F^\Ret_\dPG
914:    (\bq,\omega)$ and $F^\Ret_\dDW (\bq,\omega)$, respectively.
915: Each curve refers to either the real or the imaginary part of $F^\Ret
916:    (\bq,\omega)$ as a function of 
917:    $\omega$, for a fixed value of wave-vector $\bq$ along the symmetry
918:    contour $\Gamma$--$X$--$M$--$\Gamma$ in the 1BZ (see 
919:    Figs.~\ref{fig:qdPG}b and \ref{fig:qdDW} for its definition).
920: While $\bq$ runs along such contour, the Fermi line $\xi_\bk =0$ is
921:    traversed twice (once at point $A$ along $X$--$M$ and once at $B$ along
922:    $M$--$\Gamma$, for the hole-like Fermi line considered here; see
923:    Fig.~\ref{fig:qdPG}b), while the
924:    hole-pocket contour defined by $E_\bk^- = 0$ is traversed twice
925:    along the $M$--$\Gamma$ line (points $B_1 \equiv B$ and $B_2$ in
926:    Fig.~\ref{fig:qdDW}b).
927: 
928: As a consequence of the summation over $\bk$ in either the full or
929:    reduced BZ in Eqs.~(\ref{eq:FdPG}) and (\ref{eq:FdDW}),
930:    respectively, $\Re F^\Ret (\bq,\omega)$ is an even function of
931:    $\omega$, while $\Im F^\Ret (\bq,\omega)$ is an odd function of
932:    $\omega$ in both the dPG and dDW cases.
933: Therefore, the different contributions of particle and hole states in
934:    the two cases is averaged out, and no asymmetric peaks in the
935:    $\omega$ dependence of such quantities are to be expected in the
936:    dDW case, as is the case for the local density of states
937:    \cite{Morr:02,MoellerAndersen:03}.
938: 
939: On the other hand, the existence of hole pockets centered around
940:    $\bQ/2$ in the dDW state is clearly responsible for the
941:    different $\omega$ dependence of $\Re F^\Ret_\dPG$
942:    (Fig.~\ref{fig:opg}a) versus $\Re F^\Ret_\dDW$
943:    (Fig.~\ref{fig:oddw}a), say, as $\bq$ runs along the 
944:    $\Gamma$--$X$--$M$--$\Gamma$ contours.
945: While $\Re F^\Ret_\dPG$ is characterized by a single relative maximum
946:    for $\omega>0$ for all wave-vectors $\bq$ under consideration, $\Re
947:    F^\Ret_\dDW$ possesses two relative maxima (or a relative maximum
948:    and a shoulder) for $\omega>0$.
949: These two maxima tend to merge into a single one for $B_2 \prec \bq
950:    \prec B_1$, \emph{i.e.} inside the hole pocket, and for $\bq\approx
951:    A$, \emph{i.e.} at the intersection of the free-particle Fermi line
952:    with the $X$--$M$ side (Fig.~\ref{fig:oddw}a).
953: Likewise, the single relative maximum for $\omega>0$ in $\Re
954:    F^\Ret_\dDW$ shifts towards larger frequencies as $\bq$ runs from
955:    $\Gamma$ to $X$, is `diffracted' at $A$ along the Fermi line as
956:    $\bq$ runs from $X$ to $M$, and `bounces back' at $B$, again along
957:    the Fermi line, as $\bq$ runs from $M$ back to $\Gamma$.
958: A similar analysis may be performed for $\Im F^\Ret$ in the two cases
959:    (Figs.~\ref{fig:opg}b and \ref{fig:oddw}b).
960: 
961: As in the static limit, the competition of a sizeable dDW order parameter
962:    with an underlying dSC condensate does not give rise to
963:    qualitatively different results in the $\omega$-dependence of the
964:    polarizability, with respect to the pure dDW case.
965: 
966: One may conclude that, in both the dPG and dDW cases, the evolution
967:    with $\bq$ of the features in $\omega$ dependence of $F^\Ret
968:    (\bq,\omega)$ are closely related to the location of wave-vector
969:    $\bq$ with respect to the Fermi line, and may therefore serve to
970:    indicate the presence of hole-pockets, as is the case for the dDW
971:    state.
972: 
973: \begin{figure}[t]
974: \begin{center}
975: \begin{minipage}[c]{0.9\columnwidth}
976: \includegraphics[height=\columnwidth,angle=-90]{ocnt_repg_01.ps}
977: \end{minipage}
978: \begin{minipage}[c]{\columnwidth}
979: \begin{flushleft}
980: (a)
981: \end{flushleft}
982: \end{minipage}
983: \begin{minipage}[c]{0.9\columnwidth}
984: \includegraphics[height=\columnwidth,angle=-90]{ocnt_impg_01.ps}
985: \end{minipage}
986: \begin{minipage}[c]{\columnwidth}
987: \begin{flushleft}
988: (b)
989: \end{flushleft}
990: \end{minipage}
991: \end{center}
992: \caption{%
993: Frequency dependence of the real [panel (a)] and imaginary parts
994:    [panel (b)] of the retarded
995:    polarizability, $F^\Ret (\bq,\omega)$, in the
996:    dPG case, for wave-vector $\bq$ varying along a symmetry contour
997:    $\Gamma$--$X$--$M$--$\Gamma$ in the 1BZ (see
998:    Fig.~\protect\ref{fig:qdPG}b).
999: All curves have been shifted vertically for clarity, by an amount
1000:    proportional to the path length from $\Gamma$ to actual wave-vector
1001:    $\bq$ along such symmetry contour (see right scale).
1002: Dotted line is the zero axis for $F^\Ret (0,\omega)$.
1003: All other parameters as in Fig.~\protect\ref{fig:qdPG}.
1004: }
1005: \label{fig:opg}
1006: \end{figure}
1007: 
1008: \begin{figure}[t]
1009: \begin{center}
1010: \begin{minipage}[c]{0.9\columnwidth}
1011: \includegraphics[height=\columnwidth,angle=-90]{ocnt_reddw_01.ps}
1012: \end{minipage}
1013: \begin{minipage}[c]{\columnwidth}
1014: \begin{flushleft}
1015: (a)
1016: \end{flushleft}
1017: \end{minipage}
1018: \begin{minipage}[c]{0.9\columnwidth}
1019: \includegraphics[height=\columnwidth,angle=-90]{ocnt_imddw_01.ps}
1020: \end{minipage}
1021: \begin{minipage}[c]{\columnwidth}
1022: \begin{flushleft}
1023: (b)
1024: \end{flushleft}
1025: \end{minipage}
1026: \end{center}
1027: \caption{%
1028: Same as Fig.~\protect\ref{fig:opg}, but for the dDW case.
1029: Special points in the 1BZ are as in Fig.~\protect\ref{fig:qdDW}b.
1030: }
1031: \label{fig:oddw}
1032: \end{figure}
1033: 
1034: \section{Conclusions}
1035: \label{sec:conclusions}
1036: 
1037: Motivated by recent STM experiments around a localized impurity in the
1038:    HTS, we have derived the polarizability (density-density
1039:    correlation function) $F(\bq,\omega)$ for the pseudogap phase, both
1040:    in the precursor pairing scenario and in the $d$-density-wave
1041:    scenario.
1042: Expressions for the same function have been derived also in
1043:    the underdoped regime, characterized by competing dSC+dDW orders.
1044: 
1045: In the static limit (here defined as the limit of zero external
1046:    frequency for the time-ordered correlation function), the $\bq$
1047:    dependence of $F(\bq,0)$ reflects the $d$-wave symmetry of the
1048:    precursor pairing `pseudogap' or of the dDW order parameter, with an
1049:    azymuthal modulation consistent with a clover-like pattern, as
1050:    expected also for a superconductor with an isotropic band
1051:    \cite{Angilella:02f}.
1052: However, at variance to the dPG case, the $\bq$ dependence of the
1053:    static polarizability in the dDW state clearly exhibits the
1054:    presence of hole pockets, due to the (albeit imperfect) nesting
1055:    properties of the dDW state, with nesting vector $\bQ=(\pi,\pi)$.
1056: Qualitatively similar results to the pure dDW case are obtained also
1057:    in the mixed dSC+dDW, thus showing that hole pockets are a
1058:    distinctive feature of dDW order.
1059: Such a behavior is confirmed by the $\br$ dependence of the static
1060:    polarizability in real space.
1061: A detailed comparison with experimental data for the $\br$-dependence
1062:    of the charge density displacement would of course require a much
1063:    more detailed knowledge of the $\bq$ dependence of the impurity
1064:    potential, here crudely approximated with an $s$-wave Dirac
1065:    $\delta$-function.
1066: In particular, the presence of higher momentum harmonics in the
1067:    impurity potential may break the $d$-wave symmetry of the possible
1068:    correlated or ordered states (dPG, dSC, dDW) here studied.
1069: Also, an extension of the present Born approximation for the impurity
1070:    perturbation, \emph{e.g.} to the $T$-matrix formalism, would afford
1071:    a more reliable comparison with experimental results.
1072: 
1073: 
1074: An analysis of the frequency dependence of the retarded polarizability
1075:    $F^\Ret (\bq,\omega)$ reveals that the $\bq$ evolution of the
1076:    features (local maxima or shoulders) in the $\omega$ dependence of
1077:    this function is closely connected with the relative position of
1078:    wave-vector $\bq$ with respect to the Fermi line, and is
1079:    therefore sensitive to the possible presence of hole
1080:    pockets, as is the case for the dDW state. 
1081: 
1082: \begin{acknowledgments}
1083: We are indebted with P. Castorina, J. O. Fj\ae{}restad, F. E. Leys,
1084:    V. M. Loktev, N. H. March, M. Salluzzo, S. G. Sharapov, F. Siringo,
1085:    D. Zappal\`a for stimulating discussions and correspondence.
1086: \end{acknowledgments}
1087: 
1088: \appendix
1089: 
1090: \section{Finite lifetime effects}
1091: \label{app:lifetime}
1092: 
1093: In order to take into account for finite lifetime effects on the
1094:    linear response function for the pseudogap regime within the
1095:    precursor pairing scenario, we write the diagonal elements of the
1096:    matrix Green's function as
1097: \begin{equation}
1098: \Green_{ii} (\bk, i\omega_n ) = \frac{1}{2\pi} \int d\omega
1099:    \frac{A_{ii} (\bk,\omega )}{i\omega_n - \omega} ,
1100: \end{equation}
1101: where $A_{11} (\bk,\omega) = 2\pi u_\bk^2 \delta(E_\bk -\omega)$,
1102:    $A_{22} (\bk,\omega) = 2\pi v^2_\bk \delta(E_\bk + \omega)$ are the
1103:    appropriate spectral functions for BCS theory.
1104: 
1105: A finite energy linewidth $\Gamma$ can be attached to the energy state $E_\bk$
1106:    by replacing the $\delta$-functions in the spectral functions
1107:    $A_{ii}$ with broader ones, \emph{e.g.} a Lorentzian function
1108:    $a(\omega) = \frac{1}{\pi} \Gamma/(\omega^2 + \Gamma^2 )$.
1109: Setting
1110: \begin{equation}
1111: A(\bk,\omega) = 2\pi [u_\bk^2 a(E_\bk -\omega) + v_\bk^2 a(E_\bk +
1112:    \omega)] ,
1113: \end{equation}
1114: in the static limit one obtains:
1115: \begin{eqnarray}
1116: F^\Ret_\dPG (\bq,\omega_{\mathrm{ext}}=0) &=&
1117: -\int  \frac{f(\omega)-f(\omega^\prime
1118:    )}{\omega-\omega^\prime} \nonumber\\
1119: &&\times \, \phi(\bq,\omega,\omega^\prime ) \, d\omega \, d\omega^\prime,
1120: \end{eqnarray}
1121: where
1122: \begin{equation}
1123: \phi(\bq,\omega,\omega^\prime ) = \frac{1}{(2\pi)^2} \frac{1}{N}
1124:    \sum_\bk A(\bk,\omega) A(\bk-\bq,\omega^\prime ).
1125: \end{equation}
1126: 
1127: \section{Polarizability for the \lowercase{d}DW state}
1128: \label{app:poldDW}
1129: 
1130: In order to derive the analog of the polarizability,
1131:    Eq.~(\ref{eq:Prange}), for the dDW state, we start with considering
1132:    the density-density correlation function:
1133: \begin{equation}
1134: F(\bq,\tau) = -\langle \Tau \rho(\bq,\tau) \rho(-\bq,0) \rangle,
1135: \end{equation}
1136: where $\rho(\bq,\tau)=\sum_{\bk s} c^\dag_{\bq s} (\tau) c_{\bk+\bq s}
1137:    (\tau) $ is the electron density operator, and $T_\tau$ denotes
1138:    ordering with respect to the imaginary time $\tau$.
1139: Application of Wick's theorem then yields
1140: \begin{eqnarray}
1141: F(\bq,\tau) &=& \sum_{\genfrac{}{}{0pt}{2}{\bk\bk^\prime}{s s^\prime}}
1142: \langle \Tau c_{\bk+\bq s} (\tau) c^\dag_{\bk^\prime s^\prime} (0)
1143:    \rangle
1144: \langle \Tau c_{\bk^\prime -\bq s^\prime} (0) c^\dag_{\bk s} (\tau)
1145:    \rangle \nonumber\\
1146: &&-\langle \rho(\bq,0)\rangle \langle \rho(-\bq,0)\rangle ,
1147: \label{eq:app:derdDW:1}
1148: \end{eqnarray}
1149: the last term being a constant with respect to $\tau$, which can be
1150:    neglected in Fourier transforming to the Matsubara frequency
1151:    domain.
1152: In the dDW state, the contributions of terms like Eq.~(\ref{eq:dDWop})
1153:    should be explicitly considered.
1154: Therefore, we make use of the identity
1155: \begin{equation}
1156: \sum_\bk f_\bk = {\sum_\bk}^\prime (f_\bk + f_{\bk + \bQ} ),
1157: \end{equation}
1158: for the summations on both $\bk$ and $\bk^\prime$ in
1159:    Eq.~(\ref{eq:app:derdDW:1}), where the prime 
1160:    restricts the summation to wave-vectors $\bk$ belonging to 
1161:    the reduced (magnetic) Brillouin zone.
1162: After Fourier transforming to the Matsubara frequency domain, one
1163:    eventually has:
1164: \begin{eqnarray}
1165: F_\dDW (\bq,i\omega_\nu ) &=&
1166: \frac{1}{\beta} \sum_{\omega_n} \frac{1}{N} {\sum_\bk}^\prime
1167: [
1168: \Green_{11} (\bk,i\omega_n )
1169: +
1170: \Green_{12} (\bk,i\omega_n )
1171: \nonumber\\
1172: &&\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!
1173: +
1174: \Green_{21} (\bk,i\omega_n )
1175: +
1176: \Green_{22} (\bk,i\omega_n )
1177: ]\nonumber\\
1178: &&\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\times
1179:    \, [
1180: \Green_{11} (\bk-\bq,i\omega_n -i\omega_\nu )
1181: +
1182: \Green_{12} (\bk-\bq,i\omega_n -i\omega_\nu )
1183: \nonumber\\
1184: && \!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!
1185: +
1186: \Green_{21} (\bk-\bq,i\omega_n -i\omega_\nu )
1187: +
1188: \Green_{22} (\bk-\bq,i\omega_n -i\omega_\nu )
1189: ],\nonumber\\
1190: \label{eq:app:FdDW}
1191: \end{eqnarray}
1192: where $\Green_{ij}$ are the entries of $\Green_\dDW
1193:    (\bk,i\omega_n )$ in Eq.~(\ref{eq:GdDW}).
1194: The last expression can then be cast into the compact matrix form,
1195:    Eq.~(\ref{eq:PrangedDW}), by introducing the constant auxiliary matrix
1196:    $\kappa=\tau_0 + \tau_1 = \begin{pmatrix} 1 & 1 \\ 1 & 1
1197:    \end{pmatrix}$.
1198: Equation~(\ref{eq:app:FdDW}) simplifies further, by observing that
1199:    $\Green_{12} = -\Green_{21}$ and that $\Green_{11} + \Green_{22} =
1200:    (i\omega_n - E_\bk^+ )^{-1} + (i\omega_n - E_\bk^- )^{-1}$.
1201: 
1202: 
1203: 
1204: \begin{small}
1205: \bibliographystyle{apsrev}
1206: \bibliography{a,b,c,d,e,f,g,h,i,j,k,l,m,n,o,p,q,r,s,t,u,v,w,x,y,z,zzproceedings,Angilella}
1207: \end{small}
1208: 
1209: \end{document}
1210: