Nonadiabatic breakdown and pairing in high-Tc compounds

The electron-phonon interaction plays a fundamental role on the superconducting and normal state properties of all the high-Tc materials, from cuprates to fullerenes. Another common element of these compounds is in addition the extremely small Fermi energy EF, which is comparable with the range ω...

Повний опис

Збережено в:
Бібліографічні деталі
Опубліковано в: :Физика низких температур
Дата:2006
Автори: Pietronero, L., Cappelluti, E.
Формат: Стаття
Мова:English
Опубліковано: Фізико-технічний інститут низьких температур ім. Б.І. Вєркіна НАН України 2006
Теми:
Онлайн доступ:https://nasplib.isofts.kiev.ua/handle/123456789/120187
Теги: Додати тег
Немає тегів, Будьте першим, хто поставить тег для цього запису!
Назва журналу:Digital Library of Periodicals of National Academy of Sciences of Ukraine
Цитувати:Nonadiabatic breakdown and pairing in high-Tc compounds / L. Pietronero, E. Cappelluti // Физика низких температур. — 2006. — Т. 32, № 4-5. — С. 455–478. — Бібліогр.: 141 назв. — англ.

Репозитарії

Digital Library of Periodicals of National Academy of Sciences of Ukraine
id nasplib_isofts_kiev_ua-123456789-120187
record_format dspace
spelling Pietronero, L.
Cappelluti, E.
2017-06-11T12:07:28Z
2017-06-11T12:07:28Z
2006
Nonadiabatic breakdown and pairing in high-Tc compounds / L. Pietronero, E. Cappelluti // Физика низких температур. — 2006. — Т. 32, № 4-5. — С. 455–478. — Бібліогр.: 141 назв. — англ.
0132-6414
PACS: 74.10.+v, 63.20.Kr
https://nasplib.isofts.kiev.ua/handle/123456789/120187
The electron-phonon interaction plays a fundamental role on the superconducting and normal state properties of all the high-Tc materials, from cuprates to fullerenes. Another common element of these compounds is in addition the extremely small Fermi energy EF, which is comparable with the range ωph of the phonon frequencies. In the situation the adiabatic principle ωph/EF 1, on which the standard theory of the electron-phonon interaction and of the superconductivity relies, breaks down. In this contribution we discuss the physical consequences of the breakdown of the adiabatic assumption, with special interest on the superconducting properties. We review the microscopic derivation of the nonadiabatic theory of the electron-phonon coupling which explicitly takes into account higher order electron-phonon scattering not included in the conventional picture. Within this context we discuss also the role of the repulsive electron-electron correlation and the specific phenomenology of cuprates and fullerides.
The authors acknowledge fruitful collaborations on this subject with C. Grimaldi, S. Strassler, P. Benedetti, M. Scattoni, P. Paci, M. Botti, L. Boeri, S. Ciuchi and G.B. Bachelet. We also acknowledge financial support from the MIUR projects COFIN03 and FIRB RBAU017S8R.
en
Фізико-технічний інститут низьких температур ім. Б.І. Вєркіна НАН України
Физика низких температур
General Aspects
Nonadiabatic breakdown and pairing in high-Tc compounds
Article
published earlier
institution Digital Library of Periodicals of National Academy of Sciences of Ukraine
collection DSpace DC
title Nonadiabatic breakdown and pairing in high-Tc compounds
spellingShingle Nonadiabatic breakdown and pairing in high-Tc compounds
Pietronero, L.
Cappelluti, E.
General Aspects
title_short Nonadiabatic breakdown and pairing in high-Tc compounds
title_full Nonadiabatic breakdown and pairing in high-Tc compounds
title_fullStr Nonadiabatic breakdown and pairing in high-Tc compounds
title_full_unstemmed Nonadiabatic breakdown and pairing in high-Tc compounds
title_sort nonadiabatic breakdown and pairing in high-tc compounds
author Pietronero, L.
Cappelluti, E.
author_facet Pietronero, L.
Cappelluti, E.
topic General Aspects
topic_facet General Aspects
publishDate 2006
language English
container_title Физика низких температур
publisher Фізико-технічний інститут низьких температур ім. Б.І. Вєркіна НАН України
format Article
description The electron-phonon interaction plays a fundamental role on the superconducting and normal state properties of all the high-Tc materials, from cuprates to fullerenes. Another common element of these compounds is in addition the extremely small Fermi energy EF, which is comparable with the range ωph of the phonon frequencies. In the situation the adiabatic principle ωph/EF 1, on which the standard theory of the electron-phonon interaction and of the superconductivity relies, breaks down. In this contribution we discuss the physical consequences of the breakdown of the adiabatic assumption, with special interest on the superconducting properties. We review the microscopic derivation of the nonadiabatic theory of the electron-phonon coupling which explicitly takes into account higher order electron-phonon scattering not included in the conventional picture. Within this context we discuss also the role of the repulsive electron-electron correlation and the specific phenomenology of cuprates and fullerides.
issn 0132-6414
url https://nasplib.isofts.kiev.ua/handle/123456789/120187
citation_txt Nonadiabatic breakdown and pairing in high-Tc compounds / L. Pietronero, E. Cappelluti // Физика низких температур. — 2006. — Т. 32, № 4-5. — С. 455–478. — Бібліогр.: 141 назв. — англ.
work_keys_str_mv AT pietronerol nonadiabaticbreakdownandpairinginhightccompounds
AT cappellutie nonadiabaticbreakdownandpairinginhightccompounds
first_indexed 2025-11-25T20:40:24Z
last_indexed 2025-11-25T20:40:24Z
_version_ 1850526236623765504
fulltext Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5, p. 455–478 Nonadiabatic breakdown and pairing in high-Tc compounds L. Pietronero and E. Cappelluti Dipartimento di Fisica, Universit� «La Sapienza», P. le A. Moro 2, 00185 Rome, Italy INFM-CNR, SMC-Istituto dei Sistemi Complessi, CNR, v. dei Taurini 19, 00185 Rome, Italy E-mail: Emmanuele.Cappelluti@roma1.infn.it Received August 25, 2005 The electron-phonon interaction plays a fundamental role on the superconducting and normal state properties of all the high-Tc materials, from cuprates to fullerenes. Another common element of these compounds is in addition the extremely small Fermi energy EF, which is comparable with the range �ph of the phonon frequencies. In the situation the adiabatic principle �ph/EF �� 1, on which the standard theory of the electron-phonon interaction and of the superconductivity relies, breaks down. In this contribution we discuss the physical consequences of the breakdown of the adiabatic assumption, with special interest on the superconducting properties. We review the mic- roscopic derivation of the nonadiabatic theory of the electron-phonon coupling which explicitly takes into account higher order electron-phonon scattering not included in the conventional pic- ture. Within this context we discuss also the role of the repulsive electron-electron correlation and the specific phenomenology of cuprates and fullerides. PACS: 74.10.+v, 63.20.Kr Keywords: high-Tc superconductivity, electron-phonon interaction, fullerenes. 1. Introduction For many years the concept of superconductivity has been strictly associated with the electron-phonon interaction. One of the fingerprints of the phonon-based Migdal—Eliashberg (ME) theory in conventional low-temperature superconductors has been in fact the prediction and observation of many peculiar features which are a direct evidence of a phonon mediated superconductivity, for instance the isotope effect �Tc on the critical temperature, the ex- traction of the electron-phonon (el-ph) coupling func- tion � �2F( ) from tunneling experiments, phonon anomalies occurring at temperature T Tc� , etc. [1]. The belief that superconductivity was intimately re- lated to an electron-phonon pairing was so strong that a semi-empirical upper limit for the critical tempera- ture Tc max � 20–25 K was thought to be valid before the occurrence of lattice instabilities [2,3], in agree- ment with the maximum Tc � 23 K achieved in Nb3Sn. This phonon-based scenario was shaken in 1986 by the discovery of high-Tc superconductivity in copper oxides [4] with Tc’s up to 140 K, well above the em- pirical limit Tc max � 20–25 K. In addition, the isotope effect on the critical temperature at the optimal dop- ing �opt in cuprates was found to be unconventionally small �Tc � 01. [5], suggesting a nonphonon mediated mechanism. Following this perspective a large amount of work has been devoted in the two last decades to the study of purely electronic models to explain the high-Tc superconductivity. In the recent years however, the evidence of an im- portant role of the electron-phonon interaction on many properties of the normal and superconducting state has been increasing. On one hand, the small value of �Tc turned out to be a peculiarity of the opti- mal doping, whereas in the underdoped region �Tc could be significantly larger, even higher than the BCS limit �Tc � 0 5. [6,7]. On the other hand a re- markable isotope effect on the zero temperature Lon- don penetration depth �L( )0 has been observed both in the nearly optimal and in the underdoped regime © L. Pietronero and E. Cappelluti, 2006 [8–10]. The finite isotope shift on �L( )0 has been re- lated to an isotope effect on the effective electron mass m* through the relation �L sn /m( ) *0 � , where n s is the superfluid density. The observation of a finite isotope effect on �L( )0 or on m* is highly puzzling since these quantities are expected to show strictly zero isotope effect in the conventional elec- tron-phonon framework. The report of a finite isotope effect on m* can be thus regarded not only as an indi- cation of an important role of the electron-phonon coupling, but also as an evidence of the unconven- tional nature of the el-ph interaction [11–15]. Further support to a significant electron-phonon coupling in cuprates comes from angle-resolved photoemission spectroscopy (ARPES). The kink of the electronic dis- persion observed by these measurements was indeed claimed to be of phononic origin, since it shows a neg- ligible dependence on doping, on temperature and on the angle direction along the Fermi surface [16]. Moreover an isotope shift, not only of the kink itself, but also of the high energy electronic dispersion, points out once more a dominant but still not well un- derstood role of the electron-phonon interaction [17]. Motivated by this experimental scenario, there is nowadays a revamping interest about the unconven- tional role of the electron-phonon interaction in cuprates and its relation with superconductivity. The observation of high-Tc superconductivity with critical temperatures up to Tc max � 40 K in fullerenes [18–21] and in the recently discovered MgB2 [22], where the phononic origin of the superconducting pairing is widely accepted, points out that a phonon-based mechanisms can be actually a route for high-Tc super- conductivity, and it suggests a common mechanism for all these compounds. An useful insight on this issue is provided by the so called Uemura’s plot [23,24] (Fig. 1) which shows that high-Tc and exotic superconductors (cuprates, fullerenes, but also heavy fermions, perovskites and Chevrel phases) are all characterized by a small den- sity of charge carriers n, which is usually parametrized in terms of the Fermi energy EF . This feature suggests that a positive role can be played by the small carrier density within an unique unconventional framework relevant for all these superconductors. This hypothesis could look quite puzzling because in conventional sys- tems a low charge carrier density is usually considered detrimental for superconductivity since it decreases the number of metallic charges available for the Coo- per pairing and it decreases the dynamical screening of the electron-electron Coulomb repulsion. In the following we shall identify in the small Fermi energy EF induced by low values of n the miss- ing link between small carrier density and high-Tc su- perconductivity. More precisely we shall see that when the electronic Fermi energy EF is small enough to be comparable with the characteristic energy of a generic boson mediator (�ph for the phononic case) one of the basic assumption of the conventional Migdal—Eliashberg theory, the adiabatic principle �ph �� EF , breaks down [25]. This situation calls for a generalization of the ME theory to explicitly include nonadiabatic effects in the same spirit as self-energy renormalization is taken into account in the ME the- ory itself with respect to the BCS one [26]. The review is structured as the following: in Sec. 2 we introduce the adiabatic problem and we discuss the implications arising from the failure of the adiabatic assumption in terms of the quantum field theory. In Sec. 3 we present the generalized theory of supercon- ductivity in the nonadiabatic regime which takes into account the onset of nonadiabatic channels of interac- tion, while the relevance of the nonadiabatic scatter- ing and the peculiar features of specific high-Tc mate- rials, as fullerenes and cuprates, will be discussed in Sec. 4 within the context of the nonadiabatic elec- tron-phonon theory. 2. Breakdown of Migdal’s theorem 2.1. The nonadiabatic hypothesis One of the most popular principles in solid state physics is the adiabatic assumption which is on the ba- sis of Born—Oppenheimer approximation [27]. Fun- damental element of this approximation in metallic systems is the observation that the electron energy scale, parametrized by the Fermi energy EF , is some orders of magnitude (typically 103–104) larger than the lattice dynamics energy ruled by the phonon fre- 456 Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 L. Pietronero and E. Cappelluti ❉❉❉ ❉ ❉❉ ❄ ❄ ✦ Low Tc MgAlB2 , MgCB2❉ A15 BKBO A3 C60❄ BEDT✦ Chevrel ✜ ✩ ✩✩ ✩ ✩ La214 Y123 Bi2223 TMTSF✜ Heavy Fermions✩ 2 –1 0 1 1 2 3 4 5 E [K]F Zn Al Sn Nb 10 10 10 10 T [K ] c 10 10 10 10 10 Fig. 1. Re-elaboration of Tc vs. EF plot after Refs. 23, 24 including magnesium diboride alloys. quencies �ph. This situation leads to the decoupling of the electron and lattice dynamics providing a suitable approximation scheme where the complex many-body problem becomes now affordable. A powerful theoretical tool to deal with interacting electron systems is the quantum field theory which can be conveniently cast in terms of Feynman’s dia- grams [28,29]. The diagrammatic Feynman’s approach results particularly useful when nonperturbative schemes are needed, as in the case of superconductiv- ity, and in identifying classes of diagrams associated with a particular physical property. Feynman’s theory provides indeed an elegant way to generalize the BCS theory in the strong coupling Migdal—Eliashberg re- gime. From a generic point of view all the physical prop- erties of an interacting electron system could be com- puted from the knowledge of the electron Green’s function G or, in an equivalent way, of the electron self-energy �, which is related to G through the Dyson’s equation [28]: G( , ) [ ( ) ( , )] .k k k� � �� �� 1 (1) The self-energy itself can be in its turn expressed as a functional of the electron Green’s function G, of the effective electron-electron potential V, and of the so called vertex function �: � � �� [ , , ]G V . In the elec- tron-phonon case here considered, or for any elec- tron-electron interaction mediated by a generic boson, the potential V can be directly related to the phonon (boson) propagator D, so that � � �� [ , , ].G D (2) In the normal state Eq. (2) has the particular simple form: � �( ) ( ) ( ) ( ) ( , ),k dk g k k D k k G k k k k� � � � � � � � (3) where the indexes k and k� comprehend both frequen- cies and momenta and g is the electron-phonon matrix element. A diagrammatic expression of Eq. (3) is shown in Fig. 2, where the solid line represents the electron propagator G, the wavy line the phonon Green’s function D, and the filled circles the elec- tron-phonon vertex function �. The complex many- body nature of the problem is thus hidden in the un- known quantity � which in principle does not have an analytical expression. It is often useful to split the to- tal vertex function into a zero-order constant term plus a vertex correction function denoted as P: �( , ) ( )[ ( , )] ,k k k g k k P k k k� � � � � � 1 (4) where P contains all the higher order interaction pro- cesses and it is constituted by an infinite set of dia- grams. The evaluation of the electronic self-energy, as ex- pressed in Eq. (3), still constitutes a formidable task which cannot be analytically performed. A huge sim- plification to this aim was provided in the late 50ies by the so-called Migdal’s theorem [25] which showed that, in the (adiabatic) limit �ph/EF � 0, the vertex correction function P scales as P EF � � �ph . (5) In common metals, as above discussed, we have �ph/EF � �10 3–10 4� . The total vertex function can be thus safely replaced by its lowest order (constant) term, � � 1 signalizing that the electron-phonon ver- tex function is not renormalized by strong-coupling effects in the adiabatic regime. In this framework the normal state self-energy simply reads thus: �( ) ( ) ( ) ( ).k dk g k k D k k G k� � � � �� 2 (6) In similar way a self-consistent equation for the superconducting order parameter � can be derived un- der the assumption of Migdal’s theorem validity [30]: � �( ) ( ) ( ) ( ) ( ) ( ).k dk g k k D k k G k G k k� � � � � � �� 2 (7) Equations (6) and (7), whose diagrammatic re- presentation is shown in Fig. 3, define a closed set Nonadiabatic breakdown and pairing in high-Tc compounds Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 457 ����� Fig. 2. Diagrammatic expression of the electron-phonon self-energy. The solid line represents the electron Green’s function, the wary line the phonon propagator and the filled circles the electron-phonon vertex function. � a � � b� � Fig. 3. Self-energy (a) and superconducting pairing (b) of an electron-phonon system in conventional Migdal— Eliashberg framework. of equations provided the phonon spectral function is known. They are thus the basis of the Migdal— Eliashberg theory of the electron-phonon interaction which was successful to describe many properties of common metal and low-temperature superconductors. For instance the normal state electron-phonon self- energy at low energy simply reads: � � �� , (8) and the coherent part of the Green’s function can be written as G Z /Z ( , ) ( ) ,k k � � � 1 1 (9) where Z is the electron-phonon renormalization factor Z � 1 �. Migdal—Eliashberg theory leads thus to an effective renormalization by the factor 1 � of many physical quantities of the system, as the Fermi veloci- ty v v /F F * ( )� 1 � , the electron mass m m* ( )� 1 � or the specific heat �� � �� ( ) ( )1 2 0 32N / [27]. Similar renormalization arguments permit to un- derstand in a qualitative way the generalization from the BCS equation for Tc, Tc � � �� � �� � �ph exp , 1 (10) to the McMillan’s formula Tc � � � � � � �� � � �� �ph exp . ( ) ( . ) , 104 1 1 0 62 (11) valid in strong coupling regime. Taking into account the renormalization of the quasi-particle density of states at the Fermi level N N*( ) ( )( )0 0 1� � as well as of the electron-phonon matrix elements g g/* ( )� 1 � leads indeed to an effective elec- tron-phonon coupling �� � �� /( )1 which, once plugged in Eq. (10), gives: Tc � � �� � �� � � �ph exp , 1 (12) which is nothing else that the strong coupling McMillan’s formula in the absence of Coulomb repul- sion �*. It is interesting to note that at the adiabatic ME level all the normal state quantities (renormalized Fermi velocity vF * , electronic mass m*, specific heat ��, quasi-particle spectral weight Z, as well as the reduced critical temperature t T /c c� �ph, depend on the electron-phonon interaction only through the parameter � which is independent of the atomic mass Mat . The isotope coefficient � A d A/d M� log atlog on the normal state pro- perties is thus expected to be zero while �Tc � 0 5. . We would like to make clear once more that all the above results, commonly discussed within the frame- work of the standard theory of electron-phonon inter- action, strongly rely on the adiabatic assumption. When this latter breaks down, a different theory with qualitative different phenomenology is expected. In order to give just the flavor of the possible deep changes on the physical properties, let us schematize the total electron-phonon vertex function � in non- adiabatic regime as [31] � � � � � � � �g EF 1 � �ph , (13) where the (nonadiabatic) nonzero-order vertex pro- cesses were taken into account according with Eq. (5). Equation (13) points out that a sizable renormalization of the electron-phonon vertex func- tion is expected in nonadiabatic systems. The effec- tive phonon mediated electron-electron pairing �* is thus modified in a schematic way in the form � � �� �* ( ) ( )� 1 12 ph/E /F , and the correspond- ing expression for Tc reads [31]: T /E c F � � � � � � � � � � � � �� ph ph exp ( ) . 1 1 2 (14) In the extreme nonadiabatic case �ph/EF � 1 and assuming «normal» values � � 0 4. , �ph � 700 K, Eq. (14) would thus easily give Tc � 135 K whereas Eq. (12) would predict Tc � 24 K. We would like to make clear once more that this simple minded gener- alization should be considered only as indicative since Eq. (5) estimates only the largest magnitude of first order vertex diagrams, not their effective size or sign. In order to determine in a more quantitative way how the onset of nonadiabatic channels on interaction af- fects the electron-phonon phenomenology, a more de- tailed analysis is needed. 2.2. The vertex function beyond Migdal’s theorem From the above discussion it is clear that a rigorous way of evaluating the nonadiabatic effects is in princi- ple not available since it would imply the full exact solution of the electron-phonon many-body problem. In this situation it is thus of the highest importance to identify physical processes responsible for the non- adiabatic phenomenology and to individuate the more appropriate approaches to deal with them. Along this view the polaronic picture [32–36] and the nonadiabatic theory of superconductivity [37–39], which we propose for the high-Tc compounds, describe two distinct different contexts although they arise from the same electron-phonon interaction [15]. This difference is reflected in the different theoretical tool 458 Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 L. Pietronero and E. Cappelluti employed in the two cases. A perturbation approach is for instance clearly inadequate to study the polaronic state where ground-state properties are radically dif- ferent from Fermi liquid ones [36,40]. Nontrans- lationally invariant states are thus preferably used to account for polaron localization [41,42]. On the other hand a perturbative way appears a natural approach when nonadiabatic effects are expected to modify the Fermi liquid properties in a new nonadiabatic regime without destroying its metallic character [37–39]. In this perspective we believe that a perturbation ap- proach sustained by a diagrammatic representation can provide an useful insight on the relevant physical processes in nonadiabatic systems and a qualitative in- vestigation of their properties. On this basis a quanti- tative description is therefore beyond the main pur- poses of our work. In generalizing the strong coupling theory of Migdal—Eliashberg in nonadiabatic regime, a natural starting point is suggested by Migdal’s theorem itself. A controlled perturbative quantity is indeed identified in Eq. (5) by the size of the vertex correction ��ph/EF , which can be therefore used as a small ex- pansion parameter. First step of any perturbation analysis of nonadiabatic effects is the explicit evalua- tion of the first order vertex function P, diagrammati- cally represented in Fig. 4. Its analytical expression reads [25,26,37]: P T g D G n m l n l m ( , ; , ) | ( )| ( ) ( , k k k p k p k p � � � � � ��� � � � � 2 � � �l n lG) ( , ),p (15) where (k,�n), (k�,�m) and (p,�l) are, respectively, the momenta and energies of the incoming, outcoming and internal electrons in Matsubara notations, G and D are the electron and phonon propagators and g g( , ) ( )k p k p� is the electron-phonon matrix ele- ment. The evaluation of the first order vertex diagram beyond Migdal’s theorem [Eq. (15)] and its inclusion in systems of electrons scattering with high-energy bosons was previously addressed in the early 80ies by Grabowski and Sham in the context of plas- mon based superconductivity [43]. They estimated the vertex function P in the static limit lim lim ( , )q q� �0 0� �P where q k k� � and � � �� m n are the exchanged momentum and fre- quency, respectively. In this limit they found that the nonadiabatic electron-phonon interferences described by Eq. (15) disfavor the effective electron-boson at- traction, namely lim lim ( , )q q� � �0 0 0� �P . Similar results are found in later studies based on a local ap- proximation, where the momentum dependence of the vertex function is neglected in favor of averages over the energy variables [44–50]. On this basis the inclu- sion of vertex diagrams was assumed to be in any case negative with respect to the superconducting pairing. A full momentum dependent analysis shows however that these results are just due to a partial analysis and an effective enhancement of the electron-phonon in- teraction can actually be induced when the complex momentum structure of the vertex function is properly taken into account [37–39]. In order to evidence this point it is useful to derive an analytical expression of the vertex function which, although approximated, can be used as a guideline to discuss the general momentum-frequency structure. To this aim we consider for the moment the most simplified electron-phonon model containing however all the essential ingredients to the problem. In par- ticular we consider the first order vertex function with bare electron and phonon propagators. Electrons are assumed to interact with a dispersionless Ein- stein phonon with energy �0 through a momen- tum independent electron-phonon matrix element g g( )k p � . A half-filled constant density of states will be in addition considered N N( )0 0� [ ] � �W W where the half-bandwidthW represents the Fermi energy W EF� . Under these assumptions Eq. (15) reads thus [26,37]: P g T i i n m n ll m l ( , ; , ) ( ) [ k k p � � � � ��� � � � � � � � � 2 1 2 0 0 2 0 2 2 � i in l� � ( )][ ( )] . k p k p (16) To obtain an analytical expression the electronic expres- sion can be further linearized around the Fermi level Nonadiabatic breakdown and pairing in high-Tc compounds Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 459 k' k – k' p p + k – k' k – p k Fig. 4. Feynman’s representation of the first order dia- gram appearing in the nonadiabatic regime. ��( ) ( ) ( ) ( ) | | cosp k k p v k k p q � � � � � p Fv where � is the angle between p and q k k� � , assuming all the electronic momenta p, k, k� to lie on the Fermi surface. The first order vertex diagram results thus to be simply function of the exchanged momentum, P Pn m n m( , ; , ) ( ; , )k k q� �� � � � : P N g T dn m n ll E E F F ( ; , ) ( ) ( ) q � � � � � � � � � � � � � � 2 0 2 0 0 2 0 2 2 1 1 2 2� d i i i E Q im l n F l cos ( cos )( ) , � � � � � � (17) where E v kF F F� and Q kF� | |/q 2 . It can be shown that the main frequency dependence of the vertex function is only through the exchanged frequency � �n m . We can therefore set an external frequency to zero �n � 0 so that the exchanged frequency � reads just � � � �� �m n m . We can now finally de- rive the following analytic expression for P in the limit of a small Q expansion valid for small ex- changed phonon momenta [26,37]: P Q E Q E F F ( , )� � � � � � � � � � � � �� � � � � � �� 0 0 0 2 arctan arctan � � � � � � � � � � � � � � �� � � arctan arctan 2 2 2 E Q E Q E Q F F F � � � � � � � � � � � � � � � � � � � � � � � � ( )[( ) ] [( ) ] 0 0 2 2 0 2 2 2 2E E E F F F � �� . (18) Note that Eq. (18) provides an analytical derivation of Migdal’s theorem by evaluating the limit lim�0 0/EF P� for generic values of Q and �. We ob- tain: lim ( , ) | | � � � � � � �0 0 0 02 2 2 /E FF P Q E Q Q � � ! "" # $ %% � � arctan� � � � , (19) which explicitly shows that the first order vertex dia- gram scales as �0/EF in the adiabatic limit. Equation (18) will be now used as basis to discuss the complex momentum-frequency dependence of the vertex function. In particular we are interested in de- termining a possible increase or decrease of the effec- tive electron-phonon pairing as due to the onset of nonadiabatic effects. This concept can be quantita- tively related to the sign of the vertex correction P, where a positive sign (P & 0) leads to an enhancement of the effective electron-phonon coupling and a nega- tive sign (P � 0) to a reduction [51]. Positive and negative regions of the first order vertex function P are plotted in Fig. 5 in the momen- tum frequency Q—� space for an adiabatic ratio �0 0 5/EF � . . We can see that the total sign of P in not a priori defined but it depends strongly on the specific values of Q and �. The complex momen- tum-frequency structure of the vertex function P can be characterized by its static and dynamic limits, re- spectively, P s and Pd [37,51]. It can be shown in full generality that: P P Q P P Q s Q d Q � � � & � � � � lim lim ( , ) , lim lim ( , ) , 0 0 0 0 0 0 � � � � (20) (21) signalizing a nonanalytic point of the vertex function P in (Q � 0, � � 0). We can now fully understand how the negative sign found in Ref. 43 does not represents the total structure of the vertex diagram but only a limit (static) case. The evaluation of the effects of the nonadiabatic vertex diagrams on the electron-phonon coupling appears thus much more complex and it will depend in general on the specific momentum-fre- quency region actually probed by the electron-phonon scattering in a particular material. On one hand, in systems with negligible momentum- frequency dependence of the electron-phonon inter- action we can expect for instance that the wholeQ—� space will be effectively probed. Positive and negative parts of the vertex diagram in this case almost cancel out with a slight dominance of the negative sign. A re- sulting small reduction of the electron-phonon pairing is thus expected, as confirmed by the analysis in Refs. 44–50. On the other hand materials characte- 460 Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 L. Pietronero and E. Cappelluti 0.2 0.4 0.6 0.8 1.0 � � 0 0 0.2 0.4 0.6 0.8 Q V P < 0 V P > 0 Fig. 5. Sign of the vertex function in the Q—� space for a nonadiabatic system (�0 05/EF � . ). rized by a net predominance of forward (small q) scat- tering would mainly probe the smallQ positive region of P leading to an effective enhancement of the elec- tron-phonon pairing [37–39]. The crucial role of the electron-phonon momentum structure appears now evident from the above discus- sion. In this context, approaches as the dynamic mean- field theory (DMFT) or numerical calculations on small clusters are expected to underestimate the im- portance of the q dependence and are not suitable for a correct evaluation of the nonadiabatic effects. A useful tool to study in a controlled way the role of possible q (Q) momentum selection in the non- adiabatic framework is the introduction of a fictitious cut-off qc (Q q / kc c F� 2 ) which selects small | |q � qc momenta [37–39]. In next sections we shall related on a more compelling ground the cut-off qc to the micro- scopic properties of the system, e.g., the electronic correlation [52–57]. The electron-phonon matrix ele- ments can be thus modelled as [38]: | ( )| ( ) | | .g f q g qc cq q2 2� �' ( (22) The prefactor f qc( ) can be properly chosen in order to have a total electron-phonon coupling � indepen- dent of qc, so that the effect of the small momenta se- lection on the vertex function can be separated from the total phase space reduction which would affect �. The prefactor f qc( ) depends thus in principle on the spatial dimension and on peculiar characteristics of the model. For a isotropic system in three dimensions we would have for instance [38]: �| ( )| ,g Q g Q Q Q c c 2 2 2 � � (23) where we have expressed the exchanged momentum | |q in the dimensionless variable Q kF� | |/q 2 . Equa- tion (23) ensures that the Fermi surface average of the phonon mediated electron-electron interaction is constant ) * �| ( )|g Q gFS 2 2. In this way the electron- phonon coupling constant � and the cut-off parameter Qc can be considered as two independent free vari- ables. On the basis provided by Fig. 5 and by Eq. (18) we can now qualitatively investigate the role of a small momenta selection in the context of a nonadiabatic electron-phonon interaction. This issue will be parametrized as function of the cut-off Qc, where Qc � 1 represents a shapeless electron-phonon interac- tion in the momentum space andQc �� 1 a marked for- ward scattering predominance which could be repre- sentative of strongly correlated systems [52–57]. In order to evaluate the effect of the inclusion of the first order vertex function P in a generic phonon mediated property (electronic self-energy, superconducting pairing) we consider a weighted average of P Q( , )� in the whole momentum-frequency space, P Qc av ( ) [37]. Weighting factors will be the momentum dependence electron-phonon matrix elements as given by Eq. (23), which would take into account a possible small momentum selection, and a simple Lorentzian in frequency � � �0 2 0 2 2/( ) to simulate the presence of a phonon propagator with characteristic energy �0 which would be always connected to the elec- tron-phonon vertex function. The dependence of the weighted average vertex function on the adiabatic ratio �0/EF is shown in Fig. 6 for different values of the cut-off Qc. Note the strong enhancement for small values ofQc and the fact that a significant magnitude of P Qc av ( ) is already ap- preciable for relatively small values of �0/EF . According to this preliminary qualitative analysis we can then identify a physical regime characterized by a finite adiabatic ratio �0 0/EF + and small mo- mentum phonon scattering (Qc � 1) where the non- adiabatic effects induced by the breakdown of Mig- dal’s theorem could lead to a remarkable enhancement of the electron-phonon pairing. Guided by these re- sults a rigorous generalization of the theory of super- conductivity in nonadiabatic regime [38,39] will be presented in the next section. We are going to see that the enhancement of the superconducting pairing sug- gested by the above discussion is actually confirmed by a numerical accurate analysis. The normal and superconducting state phenomenology of the non- adiabatic systems will be also investigated. Nonadiabatic breakdown and pairing in high-Tc compounds Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 461 0 0.2 0.4 0.6 0.8 1.0 � 0 /E F 0 0.1 0.2 0.3 0.4 P av (Q c ) Q c = 0.1 Qc = 0.9 Fig. 6. Momentum-frequency average of the vertex dia- gram P Qc av( ) as function of the adiabatic parameter �0/EF for different values of the momentum cut-off Qc (from the top to the bottom): Qc � 0.1, 0.3, 0.5, 0.7, 0.9, and for � � 1. 3. Nonadiabatic theory of superconductivity and normal state In Sec. 2 we have briefly mentioned the possibility to build a perturbative theory in regime of strong cou- pling � , 1 on the basis of a small �0/EF expansion. The most straightforward way to achieve this aim is the use of the functional formalism based on the Baym–Kadanoff technique [58,59] to derive a con- serving scheme which links the superconducting and normal state properties to a self-consistent evaluation of the normal state self-energy in nonadiabatic regime. In terms of Feynman’s diagrams a convenient starting point is the skeleton formalism where each graphical element represents fully renormalized quantities [28,29]. 3.1. One-particle self-energy In this framework the diagrammatic generalization of the electron-phonon self-energy at the first order in a �0/EF expansion is depicted in Fig. 7. Retaining only the first term on the right side is equivalent to the so called noncrossing approximation (NCA) where nonadiabaticity would be taken into account only through finite bandwidth effects. More impor- tance is attached to the second term on the right side which explicitly contains the first order vertex dia- gram arising from the breakdown of Migdal’s theo- rem. Its analytic expression can be written in the com- pact form as [11,38,39]: �( , ) ( )| ( )| [ ( , ; , ) , k k k k k k � � �n m n m n m T D g P W W � � � � � � � 2 1 ] ( , ) ,G Wmk� (24) where we have introduced the renormalized Matsu- bara frequencies iW in n n� � ��( ), D is the generic phonon propagator coupled to the electrons through the electron-phonon Eliashberg function � �2F( ) [27,60,61]: D F dn m n m ( ) ( ) ( ) ,� � � �� � � � � � � � 2 2 2 2 (25) and P is the first order vertex function: P W W T g D G W n m l n l l ( , ; , ) | ( )| ( ) ( , , k k k p p k k p � � � � � � � 2 � � n m lG W� ) ( , ).p (26) In an isotropic system the angular dependence of observable quantities is negligible and it can be dropped. We can therefore replace the self-energy in Eq. (24) and the whole electron-phonon matrix ele- ments in Eq. (26), including the vertex corrections, with their averages over the Fermi surface [11,38,39]: � � �( , ) ( , ) ( ),k k� � �n n FS n� ) * � (27) | ( )| [ ( , ; , )] [ ( , ; , )] g P W W P W W n m n m FS k k k k k k � � � � ) � * 2 1 1 � � g P Q W Wc n m 2 1[ ( ; , )] . (28) Equations (24), (27) and (28) define a self-consis- tent expression for the self-energy which can be writ- ten in the compact form: W T P Q W W D E W n n m c n m n m F M � � � ! "" �� � � � 2 1[ ( ; , )] ( ) arctan # $ %% , (29) and which can be numerically solved once the mo- mentum averaged vertex function P Q W Wc n m( ; , ) is known. An explicit expression of P Q W Wc n m( ; , ) can be derived within the same approximations employed in Sec. 2. However, because of the self-consistent renormalization of the internal Green’s functions in Eq. (26), we cannot obtain now an analytical form for P Q W Wc n m( ; , ) since it would involve a sum over the Matsubara frequencies. A longsome derivation of P Q W Wc n m( ; , ), which is essentially based on a small Qc expansion, can be found in Refs. 62, 63 where we refer for more details. We report here therefore only the final result: 462 Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 L. Pietronero and E. Cappelluti ��� Fig. 7. Nonadiabatic electron-phonon self-energy including the first order vertex diagram arising from the breakdown of Migdal’s theorem. P Q W W T D B n m l A n m l B n m l c n m l n l( ; , ) ( ) ( , , ) ( , , ) ( , , ) � �� � � [ ] ( ) W W E Q l l n m F c � ! " " " � � 2 2 22 � ! " " # $ % % � � � � 1 4 1 1 2 1 42 2 2E Q W W E Q W W F c l l n m F c l l n ln m ! " " # $ % % � � � � � � � � � � - . / 0 / 1 2 / 3 / # $ % % %% 2 1 2 , (30) where A n m l W W E W E Wl l n m F l F l ( , , ) ( )� ! "" # $ %% � � � arctan arctan n m� ! "" # $ %% � � � � � � , (31) B n m l W W E W E W E E W l l n m F l n m F l n m F F ( , , ) ( ) [ ] � � � � � � � 2 2 2 2 l n m� � 2 . (32) Equations (29)–(32) define now a closed set of equations which determine in an unambiguous way all the one-particle properties of a nonadiabatic system. The nonadiabatic equations are valid for any generic electron-phonon spectrum described by the Eliashberg function � �2F( ). They will be thus the basis for inves- tigating the characteristic features in the normal state of nonadiabatic systems. For practical purposes we shall consider in the following, unless better specified, a single Einstein phonon spectrum, which reproduces however all the relevant nonadiabatic characteristics. As we are going to see, interesting new physics is in- duced by the onset of new channels of interaction in the nonadiabatic regime. A natural quantity to be investigated in this frame- work is the effective electronic mass m* which is renormalized by the electron-phonon interaction. Hallmark result of the conventional Migdal—Eliash- berg theory is the simple relation between the ef- fective mass m* and the unrenormalized one m: m m* ( )� 1 � [27]. Straightforward consequence of this relation is the absence of any isotope effect on m*, since the electron-phonon coupling constant � can be shown not to depend on the atomic masses. This pre- diction is indeed verified in all common metals. The effective mass m* in our context can be easily obtained from Eqs. (29)–(32) as a simple limit of the renormalization function Z i i /in n n( ) ( )� � �� 1 � [11]: m m Z i i ii n i n nn n * lim ( ) lim ( ) .� � � � � � � � � �� � � � �0 0 1 � (33) In Fig. 8 we plot the quantity Z i n( )� , calculated in nonadiabatic regime (�0 0 2/EF � . ) by including the first order vertex diagram as in Fig. 7, as function of �n for � � 10. [11]. A marked dip around �n � 0 is observed in the nonadiabatic theory (solid lines) which is absent when the vertex diagram is omitted (dashed line, noncrossing approximation). This fea- ture can be understood by considering that at �n � 0 the electron mass renormalization factor Z is mostly modified by the static limit of the vertex function, lim ( ; , )W W c n mm n P Q W W� , which was previously shown to be negative. Another interesting feature appearing in the non- adiabatic regime is an effective dependence of the elec- tronic mass m* on the adiabatic ratio and conse- quently on the atomic mass Mat . This concept leads thus to a generalization of the standard Mig- dal—Eliashberg expression for m* in nonadiabatic systems: m mf Em F * , ,*� ! "" # $ %%� �0 (34) Nonadiabatic breakdown and pairing in high-Tc compounds Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 463 0 0.5 1.0 1.5 2.0 2.5 �n �0 1.1 1.3 1.5 1.7 1.9 Z (i � n ) Q c = 0.1 Q c = 0.5 Fig. 8. Renormalization function Z i n( )� for �0 02/EF � . and � � 10. . Solid lines: nonadiabatic theory with Qc � 01 02 05. , . , , .� . Dashed line: noncrossing approxima- tion with no vertex diagram. where lim ( , )*� � � � 0 0 0 1/E m FF f /E� � . The ex- plicit dependence of the effective electronic mass on the adiabatic ratio is reflected in a finite isotope effect on m*. In Fig. 9 we report the isotope coef- ficient � m* as function of the adiabatic parameter �0/EF [11]. Solid lines correspond to the nonadia- batic theory for different values of Qc, dashed line to the noncrossing approximation where nonadiabaticity was retained only through finite bandwidth effects. A finite negative isotope effect is recovered in the nonadiabatic regime, mainly due to the onset of nonadiabatic channels of interaction (compared with the dashed line). These results were later confirmed by DMFT calculations [15]. Note moreover that � m* does not show in the whole region of �0/EF a crucial dependence on Qc. This is again due to the particular limit of the vertex function probed by the electronic mass m* which is not significantly affected by the small | |q selection. The prediction of a finite and nega- tive isotope effect on m* is of the highest importance with respect to the experimental observation of an isotope effect on the penetration depth �L( )0 in cuprates which is indeed thought to stem from a cor- responding isotope effect on the electronic mass m*. 3.2. Superconducting instability The study of the nonadiabatic effects on the super- conducting instability and on the resulting critical temperature Tc appears of fundamental importance in the light of the specific properties of the high-Tc ma- terials, that we have seen to present small Fermi ener- gies, and of the previous qualitative discussion in Sec. 2 which suggested that an enhancement of the superconducting pairing is achievable in the nonadiabatic regime under favorable conditions like the predominance of forward scattering [37]. The formal derivation of the nonadiabatic theory of superconductivity follows essentially the same proce- dure developed for the normal state. The explicit equations can be determined by the requirement of a conserving theory consistent with the self-energy � de- picted in Fig. 7. Technical steps are the introduction of an external field h� coupled with the superconduct- ing condensate � , ) *c c† † and the functional deriva- tive of � with respect to h� in the spirit of the Baym—Kadanoff formalism. The practical derivation is accomplished in the clearest way in terms of the diagrammatic representa- tion. The graphical expression of the superconducting instability equation is thus obtained by replacing in all the possible combinations one of the electronic lines of Fig. 7 with an anomalous propagator which can be in its turn expressed as function of the su- perconducting order parameter � and of two normal state propagators with (k, ,�n � ) and ( , , ) 4k �n quantum numbers [38,39], respectively. We obtain then a self-consistent equation for the superconduct- ing order parameter � as diagrammatically shown in Fig. 10. We can see that the failure of Migdal’s theorem gives rise to two vertex diagrams, which have been al- ready widely discussed, and to one so-called «cross» term [38,39]. In similar way with the normal state self-energy, we can write down the explicit self-con- sistent equation for �: � � � �( , ) ( )| ( )| ( , ; , , k k k k k k n n m m n m T D g P W W � � � � � � � 2 1 2[ �) ( , ; , ) ( , ) ( , ) ( , ) , ] � � � � � � C W W G W G W n m m m m k k k k k� � (35) where the cross function C is defined as: 464 Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 L. Pietronero and E. Cappelluti � 0 /EF 0 0.2 0.4 0.6 0.8 1.0 0 –0.1 –0.2 –0.3 Q c = 0.6 Q c = 0.2 � m Fig. 9. Isotope coefficient �m* on the effective electronic mass m* calculated for � � 1. Solid lines: nonadiabatic the- ory for Qc � 02 03 06. , . , , .� . Dashed line: noncrossing ap- proximation with no vertex diagrams. � � �� � � ��� Fig. 10. Self-consistent equation for the superconducting order parameter � in the nonadiabatic theory. C W W T g g D D n m l n l ( , ; , ) | ( )| | ( )| ( ) ( , k k k p p k p � � � � � � 2 2 � � � �l m l n m lG W G W � � � � � ) ( , ) ( , ) .p k k p (36) For an isotropic superconductivity (we shall dis- cuss later d-wave symmetry), we can also average the superconducting order parameter � on the Fermi sur- face. We end up with the resulting equation for the superconducting instability in Matsubara frequencies: � � � � � � n c c n m m n m c n m m T P Q W W D C Q W W W � � � �2 1 2[ ( ; , )] ( ) ( ; , ) m F m E W arctan ! "" # $ %% . (37) The explicit derivation of the cross function can be carried on in similar way with the vertex P to ob- tain [62]: C Q W W T D D B n m l c n m c l n l l m( ; , ) ( ) ( ) ( , , ) � � � �� � � � �2 2 arctan 4 2E Q W W A n m l B n m F c l l n m ! " " # $ % % � - . / 0/ � � � ( , , ) ( , ,l W W E Q W W l l n m F c l l n m )[ ] [ ] , 1 2 / 3/ � � � � 2 2 22 (38) where the functions A n m l( , , ) and B n m l( , , ) are the same ones defined in Eqs. (31), (32). Equation (37), together with Eq. (29), evaluated at T Tc� , defines the nonadiabatic theory of supercon- ductivity which can be employed for numerical calcu- lations. We are now in the position to investigate with a full numerical solution the role of the opening of nonadiabatic pairing channels on the superconducting critical temperature. In Fig. 11 we show the critical temperature Tc as function of the adiabatic parameter �0/EF for differ- ent values of the dimensionless momentum cut-off Qc and for � � 0 7. . The case �0 0/EF � corresponds to the Migdal’s limit. Interestingly, nonadiabatic effects can affect in different ways the superconducting pairing, resulting in an increase or reduction of Tc depending on the microscopic details of the system (the ratio �0 0/EF � , the degree of the electronic correlation Qc, ...). We can note however that the intuitive idea that strong electronic correlations (small Qc’s) can fa- vor superconductivity by selecting positive regions of the vertex function is sustained by the numerical calculations [38,39]. In Fig. 11 a marked enhance- ment of Tc with respect to the conventional Migdal—Eliashberg theory is shown for relatively small values of Qc. We can also easily compute from Eqs. (29), (37) the isotope coefficient on Tc. The results are shown in Fig. 12. We can note some interesting features. First of all a reduced isotope effect �Tc � 0 5. can be found as result of the nonadiabatic pairing (note that in the present analysis no Coulomb repulsion was taken into account so that should be strictly equal to 0 5. in the Migdal—Eliashberg framework). At the same time an isotope coefficient �Tc & 0 5. can be also observed, which is a quite unconventional feature. It is interest- Nonadiabatic breakdown and pairing in high-Tc compounds Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 465 0.1 0.2 0.3 0.4 0.5 0� /EF 0 0.05 0.10 0.15 0.20 0.25 T c /� 0 cQ = 0.1 cQ = 0.9 � = 0.7 Fig. 11. Superconducting critical temperature Tc in the nonadiabatic theory as function of the ratio �0/EF for � � 07. and different values of Qc (from the top to the bot- tom line): Qc � 01 03 05 07 09. , . , . , . , . . 0 0.1 0.2 0.3 0.4 0.5 �0 /E F 0.2 0.3 0.4 0.5 0.6 0.7 � T c cQ = 0.1 cQ = 0.9 � = 0.7 Fig. 12. Isotope coefficient �Tc on Tc in the nonadiabatic theory as function of the adiabatic ratio. Same values of � and Qc as in previous figure. Smaller values of Qc corre- spond to lines with steeper initial slope. ing to note also that the stronger variations of �Tc are predicted for small Qc’s, i.e., when higher Tc’s are re- covered. The Nambu formalism permits to generalize the nonadiabatic theory of superconductivity also below the critical temperature Tc to evaluate zero tempera- ture quantities as for instance the superconducting gap 5. The formal derivation is not difficult but quite tedious and we refer to Ref. 64 for technical details. An important point is that the vertex and cross func- tions, as well the corresponding nondiagonal quanti- ties in the Nambu space, need to be evaluated in the presence of the superconducting gap which partially reduces the nonanaliticity of these function at the (q � 0, � � 0) point. The moment-frequency structure of the vertex function in the superconducting state is report in Fig. 13 which shows that the net result of the opening of the superconducting gap is to reduce the positive region of the vertex function and to increase the negative one. A direct consequence of this effect is that the effective superconducting pairing in the superconducting state at T � 0 is smaller than the one at T Tc� , so that the ratio 25/Tc, for a given �, is smaller in the nonadiabatic framework than in ME theory [64]. The scenario here outlined shows the breakdown of the conventional picture of the superconducting and normal state properties in nonadiabatic systems. In particular it is clear that the inclusion of higher order vertex terms arising from the violation of Migdal’s theorem does not lead to an effective renormalization of the electron-phonon coupling � but defines a new physical regime where conventional microscopic pa- rameters (�, �0, ...) can give rise to unconventional features [65]. This is evident for instance by looking at the isotope effects on Tc and on m* which would be �Tc � 0 5. and � m* � 0, respectively, if just a renor- malization of � would be involved. This is clearly shown also by the comparison between Figs. 9 and 12 where it is clear that the intrinsic dependence on the microscopic parameters (�, �0,Qc, ...) can be very dif- ferent according which physical quantity is consi- dered. Additional evidence of the breakdown of the con- ventional phenomenology in nonadiabatic supercon- ductors is provided by the anomalous dependence of the critical temperature on nonmagnetic impurities [62]. A well-known theorem in the conventional the- ory of superconductivity states indeed that scattering from disorder or nonmagnetic impurities should have a strictly zero effect on Tc in isotropic s-wave supercon- ductors in adiabatic regime [66]. A violation of this theorem however has been experimentally observed in fullerides and in electron-doped s-wave cuprates. The nonadiabatic theory of superconductivity provides a natural framework to explain this anomalous behavior as shown in Fig. 14 where we show the strong reduc- tion of the critical temperature Tc as function of the impurity scattering rate � [62]. In agreement with the previous data on Tc, m*, �Tc , � m* , this unconven- tional reduction is more marked for small Qc’s where nonadiabatic effects are shown to be stronger. 466 Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 L. Pietronero and E. Cappelluti 0 1 2 �/�0 0 1 –1 2 P (Q , � ) 0 1 2 3 �/�0 Q = 0 Q = 1.0 � = 0 Q = 0 Q = 1.0 � = 0.1 Fig. 13. Frequency structure of the vertex function for different values of the exchanged momentum: (from top line to the bottom) Q � 0 02 04 10, . , . , , .� . The adiabatic pa- rameter is here set �0 02/EF � . . Left panel refers to the normal state (� � 0), right panel to the superconducting state (� � 01 0. � ). Filled circles mark the static and dy- namic limits in the normal and superconducting state. 0 0.2 0.4 0.6 0.8 1.0 /� 0 0.05 0.08 0.11 0.14 0.17 0.20 T c / � 0 � = 0.7 Q = 0.1c Q = 0.3c Q = 0.5c �0 F/E = 0.2 Fig. 14. Critical temperature as a function of the impurity scattering rate � for different values of Qc in the nonadiabatic theory. The dashed line corresponds to the noncrossing approximation with no vertex contribution. 3.3. Phenomenology in normal state: the Pauli spin susceptibility In the previous section we have stressed the impor- tance to predict electron-phonon anomalous effects in the nonadiabatic scenario which are not expected within the ME theory and which can be employed as direct experimental tests. One of them was the the ef- fective electron mass m* which at the ME level de- pends only on � but not on the phonon frequencies so that no isotope effect should appear. Along this line the Pauli susceptibility 6P is another promising quan- tity to evidence nonadiabatic electron-phonon effects since it is expected to be completely unaffected by the electron-phonon coupling within the conventional Migdal—Eliashberg theory [27]. Following the Baym—Kadanoff formalism [58,59] one can write a general expression for 6P : 6 � �P B n n nT G W W� �2 2 2 k k k , ( , ) ( , ),� (39) where �B is the Bohr magneton and �� is the total spin vertex function. It can be shown that the elec- tron-phonon interaction in the Green’s functions G and in the spin vertex �� cancels out in the adiabatic limit since the electron-phonon self-energy � ep in the presence of a magnetic field H has nonzero contribu- tions only at a nonadiabatic level [67–69]: lim ( ). H ep FHO /E � � 0 0� � (40) Hence, any evidence of electron-phonon effects on the Pauli spin susceptibility would be a direct proof of a nonadiabatic electron-phonon coupling [63,69]. The proper inclusion of the electron-phonon inter- action in Eq. (39) requires the evaluation of the spin vertex function �� in nonadiabatic regime [63]. This can be done following the diagrammatic scheme previ- ously discussed for the two particle superconducting response. The pictorial expression of the spin vertex �� including electron-electron exchange interaction and the nonadiabatic electron-phonon coupling is shown in Fig. 15. Explicit expressions for the vertex and the cross diagrams appearing in Fig. 15 can be derived in similar way as in the Cooper pairing case, by using the vertex and cross functions defined in Eqs. (30), (38). Numerical calculations can be there- fore performed by the self-consistent solution of the Fig. 15 and of the frequency renormalization equation. In Fig. 16 we plot the total spin susceptibility (electron-electron + electron-phonon scattering) as function of the electron-phonon coupling and of the adiabatic parameter �0/EF for zero temperature [63]. Dashed lines are the results obtained within the non- crossing approximation without electron-phonon ver- tex diagrams [69], while solid lines are the data for the fully vertex corrected theory [63]. For this latter case we show the results for different values of Qc. The first main results of Fig. 16 is that the inclu- sion of the electron-phonon coupling, in the non- adiabatic regime, yields a sensible reduction of 6 with respect to the pure electronic spin susceptibility 6 ee . As expected this effect vanishes as � � 0 (right panel) or �0 0/EF � ( left panel). Note that both the non- crossing approximation and the vertex corrected the- ory yield a similar reduction. This is quite different Nonadiabatic breakdown and pairing in high-Tc compounds Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 467 �� ���� ����� Fig. 15. Diagrammatic representation of the spin vertex function �� in nonadiabatic regime. Wavy lines represents the electron-phonon interaction, dashed ones the elec- tron-electron Coulomb repulsion. 0 0.4 0.8 � 0 /E F 0.5 0.6 0.7 0.8 0.9 1.0 � /� ee � = 0.7 0 0.4 0.8 � � 0 /E F = 0.7 Fig. 16. Spin susceptibility as function of the adiabatic parameter �0/EF and of the electron-phonon coupling �. The total Pauli spin susceptibility is normalized with re- spect to the purely electronic one ee with a Stoner factor N I( ) .0 04� . Dashed lines represent the spin susceptibility in the noncrossing approximation, solid lines are the non- adiabatic theory with vertex diagram (from lower to upper line: Qc � 01 03 05 07 09. , . , . , . , . ). from the situation encountered in the superconducting pairing channel, where the effect of the vertex dia- grams is much stronger and highly dependent on Qc [38,39]. A more clear signature of the nonadiabatic effects is in addition provided by the isotope dependence of the Pauli spin susceptibility. In Fig. 17 we report the nu- merical calculations of the isotope coefficient � 6 6 �� � �d /d M / d /dlog log ( ) log logat 1 2 0 as function of the adiabatic ratio �0/EF and of � [63,69]. A finite and negative isotope effect is thus predicted with a strong dependence onQc and � in the intermediate nonadiabatic regime. The observation of the isotope effect, which would be absent in metals fulfilling the Migdal—Eliashberg framework, could be therefore an additional and stringent evidence for a nonadiabatic electron-phonon coupling. 3.4. Photoemission and real axis analysis In Secs. 3.1–3.3 we have briefly discussed the nonadiabatic effects on some one- and two-particle normal state properties. This analysis was simplified by the fact that all these quantities (the effective mass m*, the critical temperature Tc, the zero temperature superconducting gap 5, the Pauli spin susceptibility 6P) are «thermodynamical» properties, meaning that they are static quantities which can be determined within the context of the imaginary frequency Matsubara space. On the other hand, the anomalous electron-phonon and isotope effects reported in cuprates by the photoemission spectroscopy give rise to a deep interest about the role of nonadiabaticity on frequency-dependent spectroscopic quantities. This task is however made quite difficult by the need of an analytical continuation from imaginary to real axis frequencies, which in the mean-field-like ME theory results to be relatively simple but which in the nonadiabatic vertex corrected context becomes very hard. In this section we summarize thus only non- adiabatic anomalies which arise from finite band ef- fects neglecting for the moment the explicit inclusion of the vertex diagrams. We shall see however that the photoemission phenomenology presents already at this level many interesting features. In particular we see that the following fundamental el-ph properties [27,70] are no longer valid when EF , �0: i) the el-ph self-energy �( )� does not renormalize the electronic dispersion for � much larger than the phonon energy scale �0; ii) impurity scattering affects only the imaginary part of the self-energy but not the real part, and hence not the electronic dispersion; iii) different channels of electron scattering (pho- nons, impurities, ...) sum linearly in the self-energy. The small Fermi energy effects on the real axis due to the finite bandwidth can be conveniently dealt with by means a proper generalization of the Marsig- lio—Schossmann—Carbotte technique [71], which in finite bandwidth systems and in the presence of impu- rity scattering reads [72]: � � ( ) ( ) ( ) ( ) , ( ) i iT i i i T n m n m m n m � � � � 7 � �7 � � � � � � � � � 2 2 2 ( , ) ( ) ( ) [ ( ) ( )] ( ) � � 7 � � � 7 � �7 m m d F N f � � � � �� � � 8 8 8 8 8 2 ( ) , ( ) ( )[ ( ) ( )] ( ) � � � � 7 � �7 � 8 8 8 8 8 �� � � � � � �� � �d F N f2 � �( ) ,� (41) (42) (43) where N x( ) and f x( ) are the Bose and Fermi distribu- tion functions, respectively, �2F( )8 is the el-ph Eliashberg function, � is the impurity scattering rate. Moreover � �( ) ( ) [ ]z d F / z� �� � � 8 8 82 (z complex number), � � � � � �� � ( , ) Im ( )m mi , and 7 � � � ( ) ( ) ,m F m m E Z � � � � � � �arctan 7 � � � � � � � � � � � � � � � ( ) ln [ ( )] [ ( )] [ ( )] [ 1 2 2 2 2 E Z Z E Z Z F F � � � � � � � � �( )] , � 2 � � � � � � � � � � � � � � � � �� � � � � ( ) ( ) ( ) ( arctan arctan E Z Z E ZF F ) ( ) , � �Z� � � �� � 468 Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 L. Pietronero and E. Cappelluti 0 0.4 0.8 � 0 /E F 0 –0.04 –0.08 –0.12 –0.16 � � = 0.7 0 0.4 0.8 � �0 /E F = 0.7 Fig. 17. Isotope effect on the spin susceptibility as a func- tion of �0/EF (� � 07. , N I( ) .0 04� ) and as a function of � (�0 07/EF � . , N I( ) .0 04� ). Solid lines and dashed lines as in the previous figure. where Z z z /z( ) ( )� 1 � . In Fig. 18,a we plot the real and imaginary part of the self-energy for an el-ph Einstein model with EF � 4 0� and � � 1 in the presence of impurity scat- tering. Let us first the nonadiabatic self-energy for EF � 4 0� with the ME one (EF � 9) in the � � 0 limit (dashed line). Note that in this latter case the real part of the self-energy is always negative implying that the effective electronic band Ek is always less steep than the bare one k : Ek k� for any energy. In addition the low-energy part of ��( )� is just �� ,( )� ��, which gives the well-known renormali- zation of the electronic dispersion E /k k� �( )1 close to the Fermi level [27,70], while for � �& 0 the ��( )� goes rapidly to zero implying that no significant renormalization effect is expected. Note also that the magnitude of ���( )� is a monotonously increasing function with � and saturates for � �: 0. The presence of a Fermi energy of the same order of the phonon frequencies gives rise to a number of quali- tative new features. Most important here we would like to signalize [72]: i) ���( )� is no longer a monoto- nous function of �, but when � becomes roughly �� EF the imaginary part of the self-energy starts to decrease in modulus and it goes quite rapidly to zero. This is easily understandable in small Fermi energy systems if one considers that for � && EF there are no electronic states into which an electron with energy � could decay within an energy window , �0. Another interesting feature is indeed the large positive hump of the real part of the self-energy which occurs in corre- spondence of the drop of the imaginary part and it scales with EF . In particular we note that, in contrast with the case EF � 9, for finite EF the real part of the self-energy ��( )� becomes positive in a large range of energy for � �� 2 0. Finally, by looking at the effect of the impurity scattering, we note that the presence of two sources (impurities and phonons) of scattering does not sum linearly neither in ��( )� neither in ���( )� . These anomalous features have an important impact on the renormalized electronic dispersion ob- tained by E Ek k k � � � ( ) 0 which corresponds in ARPES measurements to the dispersion inferred by the momentum distribution curves (MDC). As shown in Fig. 18,b the positive part of ��( )� implies an «anti-renormalization» of the electron band, namely Ek k& . This new feature extends up to an energy scale which does not depend on �0 but only on EF , while its magnitude depends on el-ph parameters as � or EF itself. In such a situation the high-energy part Ek & �0 of the experimental electronic dispersion [16,73] does not represent anymore the bare band k but it expected to show a steeper behavior than k . In addition the dependence of the electronic dispersion on the impurity scattering shows that the high-energy part of Ek is still highly dependent on the microscopic details not only of the electron-phonon scattering but also of any other source of electronic scattering, as for instance here impurities. This consideration could ac- count for the strong dependence of the high-energy part of the ARPES data on the hole doping � in Nonadiabatic breakdown and pairing in high-Tc compounds Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 469 –2 –1 0 1 2 � ‘( ) � 0 2 4 6 8 10 � –3 –2 –1 0 � ''( ) � a –4 –3 –2 –1 0 � k –5 –4 –3 –2 –1 0 E k b Fig. 18. Panel a: Real and imaginary part of for a Ein- stein phonon mode with � � 1 and EF � 4 0� in the pres- ence of impurity scattering. Solid lines corresponds to (up- per panel: from bottom to the top; lower panel: from top to the bottom): � �/ 0 0 02 04 10� , . , . , , .� where � is the im- purity scattering rate. Energy quantities are expressed in units of �0. The dashed line in the upper panel is the real part of the self-energy in the adiabatic infinite bandwidth limit EF � �. Panel b: renormalized electron dispersion corresponding (from left to the right) to panel a. The dashed line represents the adiabatic limit. cuprates, where the electron-electron interaction is ex- pected to be strongly dependent on � [73]. 4. Nonadiabatic superconductivity in fullerides and cuprates In the previous section we have introduced the gen- eral equations of the superconducting and normal state theories of nonadiabatic systems. A number of different properties have been explicitly evaluated in simple models in order to point out the anomalous nonadiabatic features arising from the opening of new electron-phonon interaction channels related to the vertex diagrams. While the nonadiabatic theory above discussed would apply in full generality in a wide va- riety of nonadiabatic systems, specific features of dif- ferent materials would depend of course on specific microscopical and materials-science details. In the course of our discussion we have already linked the predictions of the nonadiabatic theory with some un- conventional experimental findings in cuprates and fullerides. In this last part we would like thus to ad- dress in more details the specific nonadiabatic super- conducting properties of these compounds. In regards with the nonadiabatic features, fullerides represent a very important test since, due to the extremely high frequencies of the intra-molecular phonon spectrum and to the weak inter-molecular electron hopping [74–76], the nonadiabatic ratio �ph/EF is significantly large, �ph/EF � 0.4–0.8 [74,75]. On the other hand, in cuprates the two-di- mensional character of the electronic band and the closeness of the Fermi level to a logarithmic Van Hove singularity [77] permit to investigate additional non- adiabatic effects triggered by the flatness of the elec- tronic bands near the Van Hove saddle point. In both these materials the large local Hubbard repulsion U compared to the small Fermi energy EF gives rise to strong correlation effects. 4.1. Correlation effects on the electron-phonon scattering The interplay between the electronic correlation and the electron-phonon interaction plays a funda- mental role within the context of the nonadiabatic theory of the electron-phonon interaction. To better understand this issue we remind that the nonadiabatic effects, especially on the superconducting pairing, de- pend significantly on the sign of the vertex processes. In particular we have shown that a predominance of forward scattering with small momentum | |q would lead to an enhancement of the effective electron- phonon coupling by selecting positive regions of the vertex function [37–39]. From the physical point of view this situation has been argued in literature to be naturally realized in systems with a strong degree of electronic correlation [52–57]. The effects of strong electronic correlations on the one-particle properties has been already studied in ex- tensive way in the literature [78–81]. Common wis- dom describes the evolution from a free-like system to a correlated one in term of coherent and incoherent parts of the Green’s functionG( , )k � . In the context of a Fermi liquid picture the coherent part of G can be written as [78–80,82]: G Z Zcoh ( , ) ( ) ,k k � � � (44) where the reduced quasi-particle weight Z which renormalizes also the quasi-particle dispersion is due to the electronic correlation and it depends on the mi- croscopic parameters of the system. In the context of Hubbard model for instance Z Z U n� ( , ), where U is the on-site Hubbard repulsion and n is the electron density per site. One-particle correlation effects can be thus parametrized in terms of Z where Z , 1 repre- sents a weakly correlated system and Z � 0 a strong correlation regime. For Z � 0 a metal-insulator transi- tion is expected where quasi-particle weight vanishes [78–82]. Concerning the interplay between the elec- tron-electron correlation and the electron-phonon in- teraction a crucial interest is paid to the two-particle response functions. In particular, since phonons are directly coupled to the electronic charge, a primary role is played by the charge density response. Work in this direction has been mainly based on analytic tools, as slave-boson techniques [52,57,83–86] or Hubbard X-operator formalism [53–55], recently supported by quantum Monte Carlo calculations [56]. Object of in- terest has been the effect of the electronic correlation on the phonon mediated electron-electron interaction � �( , )q q or, equivalently, on the electron-phonon ma- trix element g( )q which, in the absence of correlation, can be assumed to be shapeless g g0( )q � . The two quantities are simply connected by � �( , )q q � � | ( )| ( )g D qq 2 � , where D q( )� is the phonon propa- gator. The evolution of | ( )|g q 2 on the degree of electronic correlation has been widely studied. In Refs. 52–57 it was shown that a prominent peak arises for small | |q scattering by increasing the electronic correlation while large | |q momenta are strongly suppressed. In the strong correlation regime this peaked structure in | |q can be in good approximation described by a Lorentzian: 470 Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 L. Pietronero and E. Cappelluti | ( )| | | ,g gq q 2 2 2 2 1 � �; (45) where ; �1 represents the inverse correlation length which can be identified with the cut-off parameter qc introduced in Sec. 2. The small momenta selection be- comes more and more pronounced by approaching the metal-insulator transition and for Z � 0, qc � 0 [52–57]. The concept of correlation length can be useful to get a physical understanding of the small | |q selection. Let us consider an electron on the site i in a un- correlated system. In good approximation it interacts with the other electrons only through their mean den- sity. The space positions of the single electrons are thus independent each others (Fig. 19,a) and any spa- tial charge modulation with wave vector q probes the same equal electronic response. Things are different in correlated systems. In this case indeed the position of one electron at the site i is correlated with the position of the other electrons within a radius ; (Fig. 19,b) [79]. The internal dy- namics within a size ; is thus frozen out and charge modulations with wavelength l � ; are prevented. In the momentum space this is reflected in a predomi- nance of small momenta | |q � �; 1 and in a suppression of the large ones | |q & �; 1. We can now understand the empirical relation be- tween the high critical temperature superconductivity and the strong electronic correlation on the basis of the nonadiabatic scenario. Nonadiabatic effects and a strong electronic correlation are natural by-products of small Fermi energy systems when EF becomes com- parable with both the phonon and the Hubbard energy scales. The first feature (nonadiabaticity) gives rise to new interaction channels which can enhance or sup- press the electron-phonon coupling depending on mi- croscopic details [37–39]; the second one (electronic correlation, small | |q momenta) will select the positive part of the vertex processes and as a consequence it will favor the resulting electron-phonon coupling en- hancement due to the nonadiabatic effects. According this picture strongly correlated systems appear as na- tural good candidate for high-Tc superconductivity within the context of the nonadiabatic theory [39]. The constructive interplay between nonadiabatic elec- tron-phonon interaction and electronic correlation, and the role of the modulation of the small-q scatter- ing as function of the correlation degree, are pointed out in the most remarkable way in the fullerene com- pounds and in cuprates. 4.2. Fullerenes Superconductivity in fullerenes is commonly asso- ciated with A C3 60 compounds [74,75]. Pristine C60 is a band insulator where the highest molecular orbitals hu are completely filled [74,75]. Chemical intercala- tion with alkali atoms, which are completely ionized, provides additional charges and permits to dope these systems. Band theory would thus predict a metallic character up to n � 6 electrons for buckyball when the lowest unoccupied molecular orbitals t u1 are expected to be completely filled. The actual phase diagram vs. n is however quite more complex, and a narrow metallic regime is found only close to n � 3 in A C3 60 com- pounds [87]. Present understanding of this anomalous phase diagram is not at all exhaustive and certainly it needs the proper inclusion of the electronic correlation and of anomalous lattice features. It is interesting to note however that superconductivity appears only in the metallic regime. This observation suggests that the metallic character is a fundamental requirement for superconductivity in these systems. In spite of many characteristic properties of C60 compounds which make them unlikely candidates for high-Tc superconductivity in the Migdal—Eliashberg framework (strong electron-electron repulsion, signi- ficant electronic correlation, low-carrier density), su- perconductivity in A3C60 has been often regarded as conventional [88]. The high critical temperatures (Tc � 33 K in RbCs C2 60, Tc � 40 K in Cs C3 60 under pressure) are thus attributed to extremely favorable microscopic values: a particularly strong electron- phonon coupling � , 1 (LDA calculations estimate � � 0.5–1) and high-energy phonon frequencies �ph � 1000–1500 K [89–91]. These features should thus compensate the relatively small density of states and the strong electron-electron Coulomb repulsion �* .� 0 4 [74,92,93]. The Jahn—Teller nature of the electron-phonon interaction is suggested in addition to further reduce the effect of the Coulomb repulsion [94]. In this situation the critical temperatures of the Nonadiabatic breakdown and pairing in high-Tc compounds Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 471 e electron – e ba r n(r) r n(r) – electron correlateduncorrelated � Fig. 19. Schematic picture of uncorrelated (a) and corre- lated (b) electrons. Correlated systems are characterized by the correlation hole (depicted in panel b) which sur- rounds each electron and prevent charge fluctuations with wavelength less than . C60 compounds are close to maximum values theoreti- cally achievable. On a microscopic ground the standard Mig- dal—Eliashberg theory is however intrinsically incon- sistent with respect to the adiabatic problem. C60 compounds are indeed characterized by a set of very narrow bands with typical EF � 0 25. eV, for both electron and hole doping, whereas phonon mode ener- gies range up to �ph � 0 2. eV [95,96]. The breakdown of the nonadiabatic hypothesis is shown in the most re- markable way in Fig. 20 where the «adiabatic phase diagram» (�ph/EF vs. �) of fullerenes, obtained by different numerical calculations, is drawn. In this sit- uation the nonadiabatic theory outlined in the previ- ous section is the unavoidably starting point of any realistic description of superconductivity in these ma- terials. In addition we can expect that the interplay between the strong electronic correlation and the nonadiabatic electron-phonon coupling would en- hance the superconducting pairing. A significant indi- cation about the relevance of a nonadiabatic effects is given by the observed reduction of Tc upon induced disorder [99]. This feature, as we have previously seen, can be considered one of the hallmarks of a nonadiabatic pairing in s-wave superconductors [62]. To illustrate the role of the opening of nonadia- batic channels in fullerenes, the concrete example of Rb3C60 can be useful since for this compound best experimental data are available. Recent measure- ments have indeed determined with the highest degree of accuracy the carbon isotope coefficient on Tc, �Tc � 0 21. [100], which, together with the large criti- cal temperature Tc � 30 K, can be used of basis of analysis to test both Migdal—Eliashberg and the nonadiabatic theories. In Fig. 21 we show the numeri- cal solutions of both the adiabatic Migdal—Eliash- berg and the nonadiabatic theory constrained to repro- duce the experimental values Tc � 30 K and � � 0 21. with an Einstein phonon �ph. For given values of �ph lying in the physical range of the fullerene phonon spectrum �ph � 2300 K a extremely large electron- phonon coupling � � 1–4 is required in ME theory (filled squares), in contrast with local density approx- imation results which find � � 1. On the other hand, the same experimental data are fitted in the non- adiabatic theory (open triangles) with much more rea- sonable values of � in good agreement with the theo- retical calculations [95]. We would like to stress that, within the context of the nonadiabatic theory of superconductivity, the high values of the critical temperatures in fullerides are not related to some particularly strong elec- tron-phonon coupling rather more to the onset of higher order diagrams scaling with �ph/EF . This per- spective has interesting consequences in regards with the optimization of the superconducting properties in these and in other nonadiabatic superconductors. In particular, since the electron-phonon coupling is ex- pected to be in the weak-intermediate regime � , 0.5–1, the fullerenes are likely to be far from lat- tice instabilities, which are mostly due to the close- ness of the Mott—Hubbard metal-insulator transi- 472 Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 L. Pietronero and E. Cappelluti 0.8 0.6 0.4 0.2 0 1 2 KC8 Al Sn Nb Pb Hg Nb Sn3 �ph F/E � A C3 60 Fig. 20. Phase diagram of conventional superconductors compared with the fullerene compounds in the space de- fined by the electron-phonon coupling � and the adiabatic parameter �ph/EF. Data for the A C3 60 family where taken from DFT, tight-binding, and ab initio calculations [89–91,97,98] by using standard values for the density of states N EF( ) = 10 states/(eV-spin-C60) and for the Fermi energy EF � 025. eV. 0.5 1.0 1.5 2.0 2.5 3.0 3.5 � 0 500 1000 1500 2000 � p h [K ] ME theory nonadiabatic theory 0 0.2 0.4 0.6 � � Fig. 21. Phonon frequency �ph (lower panel) and Cou- lomb pseudopotential �* (upper panel) as functions of the electron-phonon coupling �. Both the ME (filled squares) and the nonadiabatic (open triangles) equations have been solved in order to reproduce Tc � 30 K and � � 021. . tion, so that Tc could reasonably be further increased in fullerides upon increasing the electron-phonon coup- ling itself. A widely used way to control the electron-phonon coupling in fullerides is by tuning the lattice spacing a. This is usually done in the A C3 60 by varying the al- kali atoms A with different ionic sizes [74]. Since the phonon and electron-phonon properties are mainly ruled by intra-molecular properties, the main effect of the lattice spacing is of tuning electronic quantities like the density of states N( )0 and the Fermi energy EF . Larger lattice parameters a correspond thus to high density of states and to higher Tc’s. Another con- venient way to obtain large lattice spacing in fullerides is by means of ammonia intercalation. Although at the present ammonia intercalation in ( )NH3 xA C3 60 (x � 1) is usually found to suppress su- perconductivity, it is important to note that the reduc- tion of Tc is always accompanied by the reduction of the symmetry of the system, global (long-range antiferromagnetism, lattice distorsion) or local one (crystal field splitting) [101]. According our view such reductions of symmetry are more likely to be re- lated to the electron-electron correlations than to the electron-phonon coupling. In this perspective, we sug- gest that the electron-phonon coupling and thus the critical temperature could be significantly further in- creased in ammonia intercalated compounds once these reductions of symmetry are prevented. The high critical temperature Tc � 30 K of (NH A C3 4 3 60) [102], where the full electron and lattice symmetries of the A C3 60 are restored, seems to confirm this pic- ture. 4.3. Copper oxides In comparison with the C60 compounds the cuprate family presents a much more complex phenomenology as it is testified by the richness of anomalous features in the T vs. � phase diagram, schematically depicted in Fig. 22. Undoubtedly the revealing of different phases in cuprates is made easier than in fullerenes by the fact that the electronic filling in copper oxides can be tuned in a continuous way by stoichiometric doping, whereas C60 compounds become rapidly insulators as soon as the electron concentration per buckyball moves away from x � 3 (half-filling). In addition, al- though correlation effects are certainly important in fullerenes, the A C3 60 compounds lie on the metallic side of the Brinkmann—Rice transition while cuprates at half-filling are Mott insulator with long range antiferromagnetic order. The appearing of a variety of exotic phases close at half-filling is not surprising since in this regime the quasi-particle kinetic energy of the electrons is highly suppressed so that it can be easily overwhelmed by different long-range or short- range orderings. At the same time, the smallness of the kinetic energy scale, parametrized by the Fermi energy EF , gives rise unavoidably also to nonadiabatic ef- fects when compared with the phonon energy scale. As a general rule, the «bell-shape» profile of the superconducting phase as function of the hole doping is commonly regarded in a twofold way from the scien- tific community [103]. According the first point of view, the onset of some long- or short-range ordering (AF fluctuations, stripes, spin glasses, pseudogap, CDW quantum critical point) is the active principle for the high-Tc superconductivity [104–114]. Along this perspective the disappearing of these features and the restoring of normal metal properties in the overdoping region leads to a reduction of Tc. The above scenario is reversed according the second point of view which regards the suppression of Tc in the underdoping regime as arising from the competition of the superconducting phase with other different kinds of ordering [115–118]. In this scenario more emphasis is thus paid to the overdoped region where other ac- tors detrimental for the superconductivity are absent and it should be easier to identify the mechanism un- derlying the superconducting pairing. The nonadiabatic theory of superconductivity per- mits to understand in a very natural way the bell- shape of the superconducting phase diagram [119]. The active principle in this context is the onset of the Nonadiabatic breakdown and pairing in high-Tc compounds Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 473 T overdopedunderdoped spin glass antiferromagnetic spin fluctuations stripes (static) anomalous metal pseudogap T (dynamic) stripes optimally doped Fermi liquid nearly ( T , a > 1) a ( T , a < 1) a � AF SC ( T)� � Fig. 22. Schematic phase diagram of the copper oxides compounds in the temperature vs. doping space. nonadiabatic channels of interaction. A crucial tuning role however is also played by the electronic correla- tion which promotes small q scattering and selects the positive (attractive) parts of the vertex diagrams. Moving from the overdoped to the underdoped region we expect thus the electronic correlation to be higher, the small q selection more effective and the effective superconducting pairing stronger. This trend, which is essentially based on the doping dependence of the par- ticle-particle Cooper interaction, has however to com- pete with the reduction of the one-particle spectral weight, which roughly scales with � in the low doping regime [78,79,82]. The phase diagram resulting from the competition of these two kinds of effects can be conveniently discussed in terms of the linearized superconducting kernel in the nonadiabatic regime [119], Eq. (37), which can be rewritten in a simplified way as: � � � � � n c m m m F m T Z Z P Z C W E W � � ! "" # $ %%[ ] ,1 2 arctan (46) where the prefactor Z in front of the superconducting kernel and of the vertex and cross diagrams arises from the quasi-particle spectral weight reduction ac- cording Eq. (44), and where we neglect for sake of simplicity the frequency dependence of the nonadia- batic contributions P, C. In this way, we can roughly see the total electron-phonon coupling as the product of two terms: an effective electron-phonon coupling of ME theory renormalized by the electronic correla- tion, �ME , and the enhancement due to nonadiabatic vertex and cross (vc) diagrams �vc: � � � � � � � � eff � � � ME vc ME vc c c Z Z P Q Z C Q , , ( ) ( ) .1 2 The schematic behavior of these quantities as a func- tion of the hole doping � is shown in the upper panel of Fig. 23 [119]. The physics behind the �-dependence of �ME can be easily related to the loss of spectral weight approaching the metal-insulator transition for � � 0. This effect, which is present also in �vc, is however in that case competing with the enhance- ment of the effective coupling due to P Qc( ) and C Qc( ) which will be maximum and positive close to half-filling (where Qc � 0) and negative at high dopings. The interplay between these two effects will give rise to a maximum of �vc, and hence of �eff , somewhere in the small doping region where the com- petition between the spectral weight loss and the pos- itive nonadiabatic effects is stronger (see upper and middle panels in Fig. 23). We can now also consider the effect of the residual Morel—Anderson-like repulsion �; first of all, we ob- serve that the reduction of spectral weight will lead to an effective repulsion � �eff � Z . Superconductivity will be possible only when the net electron-phonon at- traction overcomes the repulsion term: � �eff eff & 0 (see lower panel of Fig. 23) [119]. The resulting total coupling is expected to exhibit a bell-shape which is mostly due to the �-dependence of the nonadiabatic factor �vc. It is interesting to note two things. First, in 474 Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 L. Pietronero and E. Cappelluti SC �0 �vc = 1 �vc ME � SC eff� eff� eff � eff � – Fig. 23. Graphical sketch of the different contributions to the effective superconducting coupling. Top panel: the coupling function �ME is mainly determined by the coher- ent spectral weight, and it exhibits a monotonous growing behavior as a function of doping. The vertex factor �vc tends to enhance the effective coupling at low doping and to depress it at high doping. Middle panel: the total ef- fective electron-phonon coupling � � �eff � ME vc has a maxi- mum at some finite value of �; when the effective Morel— Anderson pseudopotential is subtracted, superconductivity is suppressed at high doping. Lower panel: resulting phase diagram for superconductivity: superconductivity is only possible in a finite region of phase space (gray region), where � �eff eff� is positive. the extreme case � �ME � eff , where no superconduc- tivity would be predicted in the whole � range by the conventional ME theory, we could expect finite Tc in a small � region, due to purely nonadiabatic effects � � � �eff eff� &ME vc . Secondly, it is clear that within the ME framework a net attractive interaction in the Cooper channel at a certain doping �, which corre- sponds to � �ME & eff , would imply a superconducting order also at larger � since the two quantities � �ME, eff scale in the same way � Z; on the other hand, in the nonadiabatic theory superconductivity, Tc is expected to be limited to some maximum value of doping, due to the negative contribution of the nonadiabatic diagrams P andC at large � (large Qc’s). We would like to stress that the above phase dia- gram was drawn keeping in mind a minimal scenario where the only actors were the enhancement of the particle-particle interaction due to the nonadiabatic terms and the reduction of the one particle spectral weight. It is understood on the other hand that an ex- haustive description of the different features of the ex- perimental phase diagram, including also the normal state, involves unavoidably the taking into account of a series of other different factors. Without entering in details, we would like just to mention several features of the cuprates which appear unconventional according the Migdal—Eliashberg theory but which can be naturally accounted for within the nonadiabatic context. We have previously already mentioned the onset of unconventional isotope effects both on Tc, with isotopic coefficient �Tc � 0.2–0.8 either larger or smaller than the BCS limit, and on the effective electronic mass m*, with � m* � 0. Both these features are indeed experimen- tally observed in cuprates [6–10]. Another apparently puzzling issue in cuprates regards the symmetry of the superconducting order parameter. While the d-wave symmetry seems well assessed in hole doped systems [120], the debate is still open about the symmetry in electron doped materials, where some indications sug- gest a transition from d- to s- or to other anomalous symmetries as function of doping, maybe also in the hole-doped compounds [121–126]. The s-wave symme- try is on the other hand well accepted in fullerenes. In this situation it seems difficult to reconcile the differ- ent phenomenology of these different systems with a unique superconducting mechanism. In hole-doped cuprates the d-wave symmetry is often discussed to rule out an electron-phonon mechanism. This view is however based on the assumption of an isotropic elec- tron-phonon scattering | ( )|g g2 2q � , which is signifi- cantly violated in correlated systems as discussed in Sec. 4.1. In this situation, the momentum structure of the attractive electron-phonon interaction + the elec- tron-electron repulsion can give rise to different sym- metry of the superconducting order parameter, from s- to d-wave, depending on the momentum cut-off | |q c [55,127,128], so that the electron-phonon interaction results to be a valid candidate for explaining the ob- servation of different symmetries within the context a unique pairing mechanism. As a final remark we would like to mention the report of the linear behavior in temperature of the resistivity in cuprates at optimal doping [129]. This strict linear dependence has also been discussed as an evidence of a nonphononic scat- tering. However, specific studies show that the linear behavior <( )T T, can be also naturally explained within the context of an electron-phonon interaction by taking into account the Van Hove singularity ef- fects [130]. Note that the flatness of the electronic band associated with the Van Hove singularity is ex- pected to give rise unavoidably to nonadiabatic effects [131]. 5. Conclusion The interest about nonadiabatic effects in high-Tc superconductors arises from the experimental obser- vation of a small energy scale associate with the elec- tronic dynamics, characterized by the Fermi energy EF . In the high-Tc materials the Fermi energy is comparable with the phonon frequency scale �ph (�ph � EF). In this situation the adiabatic assump- tion of Migdal’s theorem (�ph/EF � 0), on the basis of the conventional picture of electron-phonon inter- action in metals, breaks down. A novel approach, which takes explicitly into account the nonadiabatic effects in this new regime, is thus unavoidably re- quired. In the present contribution we have shown how a nonadiabatic theory of the superconductivity and of the normal state can naturally account for the anoma- lous phenomenology of various high-Tc compounds, where we have focused on fullerides and copper ox- ides. In these compounds, in particular, the origin of the high critical temperature stems out from the inter- play between strong electronic correlation and elec- tron-phonon coupling in the new nonadiabatic regime where new channels of electron-phonon scattering, due to the breakdown of Migdal’s theorem, are opera- tive. The nonadiabatic theory of superconductivity provides thus an unifying scenario for different kinds of «exotic» superconductors. Within this framework it is also interesting to mention the specific case of MgB2, where the origin of the nonadiabatic effects is slightly different. In this compounds, indeed, band structure calculations predict a Fermi energy EF � 0.4–0.5 [132–134], to be compared with the Nonadiabatic breakdown and pairing in high-Tc compounds Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 475 relevant phonon frequencies �ph � 70–80 meV [133–137]. This simple analysis would thus predict a nonadiabatic ratio �ph/EF � 0.1–0.2, which would locate MgB2 in the weakly nonadiabatic regime [138,139]. A peculiar feature of this system is however that the electron-phonon coupling is mainly concen- trated in only one-phonon mode, which is charac- terized by a very strong deformation potential I [135–137]. In this context new nonadiabatic effects are thus triggered by the quantum lattice fluctuations [140,141] which lead to modifications of the band structure of the same order of the Fermi energy itself I u EF) * �2 1 2/ , where ) *u2 1 2/ is the root means square of the lattice displacements. Work is at the mo- ment in progress to formalize this new kind of break- down of the adiabatic assumption. We think that fur- ther research on this field would provide, for this class of materials as well as in fullerides and cuprates, new routes for the optimization of the superconducting properties of the existing materials and for the search of new high-Tc compounds based on a nonadiabatic type of pairing. Acknowledgments The authors acknowledge fruitful collaborations on this subject with C. Grimaldi, S. Str�ssler, P. Bene- detti, M. Scattoni, P. Paci, M. Botti, L. Boeri, S. Ciuchi and G.B. Bachelet. We also acknowledge fi- nancial support from the MIUR projects COFIN03 and FIRB RBAU017S8R. 1. For a review on conventional low-Tc metals see: Super- conductivity, R.D. Parks (ed.), Dekker, New York (1969). 2. P.W. Anderson and C.C. Yu, in: Highlights of Con- densed Matter Theory, Proc. of the Int. School of Physics «E. Fermi», LXXXIX, F. Bassani and M. Tosi (eds.), North-Holland, New York (1985). 3. P.B. Allen and B. Mitrovic, in: Solid State Physics, v. 37, H. Ehrenreich, D. Turnbull, and F. Seitz (eds.), Academic Press, New York (1982). 4. J.G. Bednorz and K.A. M�ller, Z. Phys. B64, 189 (1986). 5. L.C. Bourne, M.F. Crommie, A. Zettl, H.C. zur Loye, S.W. Keller, K.J. Leary, A.M. Stacy, K.J. Chang, and M.L. Cohen, Phys. Rev. Lett. 58, 2337 (1987). 6. M.K. Crawford, W.E. Farneth, E.M. McCarron, R.L. Harlow, and E.H. Moudden, Science 250, 1390 (1990). 7. J.P. Franck, S. Gygax, S. Soerensen, E. Altshuler, A. Hnatiw, J. Jang, M.A.-K. Mohamed, M.K. Yu, G.I. Sproule, J. Chrzanowski, and J.C. Irwin, Physica C185–189, 1379 (1991). 8. G.M. Zhao, M.B. Hunt, H. Keller, and K.A. M�ller, Nature 385, 236 (1997). 9. J. Hofer, K. Conder, T. Sasagawa, G.M. Zhao, M. Willemin, H. Keller, and K. Kishio, Phys. Rev. Lett. 84, 4192 (2000). 10. R. Khasanov, D.G. Eshchenko, H. Luetkens, E. Morenzoni, T. Prokscha, A. Suter, N. Garifianov, M. Mali, J. Roos, K. Conder, and H. Keller, Phys. Rev. Lett. 92, 057602 (2004). 11. C. Grimaldi, E. Cappelluti, and L. Pietronero, Europhys. Lett. 42, 667 (1998). 12. A.S. Alexandrov, Europhys. Lett. 56, 92 (2001). 13. T. Schneider and H. Keller, Phys. Rev. Lett. 86, 4899 (2001). 14. A. Deppeler and A.J. Millis, Phys. Rev. B65, 224301 (2002). 15. P. Paci, M. Capone, E. Cappelluti, S. Ciuchi, C. Grimaldi, and L. Pietronero, Phys. Rev. Lett. 94, 036406 (2005). 16. A. Lanzara, P.V. Bogdanov, X.J. Zhou, S.A. Kellar, D.L. Feng, E.D. Lu, T. Yoshida, H. Eisaki, A. Fuji- mori, J.-I. Shimoyama, T. Noda, S. Uchida, Z. Hus- sain, and Z.-X. Shen, Nature 412, 510 (2001). 17. G.-H. Gweon, T. Sasagawa, S.Y. Zhou, J. Graf, H. Takagi, D.-H. Lee, and A. Lanzara, Nature 430, 187 (2004). 18. A.F. Hebard, M.J. Rosseinsky, R.C. Haddon, D.W. Murphy, S.H. Glarum, T.T.M. Palstra, A.P. Ramirez, and A.R. Kortan, Nature 350, 600 (1991). 19. K. Tanigaki, T.W. Ebbesen, S. Saito, J. Mizuki, J.S. Tsai, Y. Kubo, and S. Kuroshima, Nature 352, 222 (1991). 20. R.M. Fleming, A.P. Ramirez, M.J. Rosseinsky, D.W. Murphy, R.C. Haddon, S.M. Zahurak, and A.V. Mak- hija, Nature 352, 787 (1991). 21. K. Holczer, O. Klein, S.-M. Huang, R.B. Kaner, K.-J. Fu, R.L. Whetten, and F. Diederich, Science 252, 1154 (1991). 22. J. Nagamatsu, N. Nakagawa, T. Muranaka, Y. Zeni- tani, and J. Akimitsu, Nature 410, 549 (2001). 23. Y.J. Uemura, L.P. Le, G.M. Luke, B.J. Sternlieb, W.D. Wu, J.H. Brewer, T.M. Riseman, C.L. Seaman, M.B. Maple, M. Ishitawa, D.G. Hinks, J.D. Jorgen- sen, G. Saito, and H. Yamochi, Phys. Rev. Lett. 66, 2665 (1991). 24. Y.J. Uemura, A. Keren, L.P. Le, G.M. Luke, B.J. Sternlieb, W.D. Wu, J.H. Brewer, R.L. Whetten, S.M. Huang, S. Lin, R.B. Kaner, F. Diederich, S. Donovan, G. Gr�ner, and K. Holczer, Nature 352, 605 (1991). 25. A.B. Migdal, Zh. Eksp. Teor. Fiz. 34, 1438 (1958) [Sov. Phys. JETP 7, 996 (1958)]. 26. L. Pietronero and S. Str�ssler, Europhys. Lett. 18, 627 (1992). 27. G. Grimvall, The Electron-Phonon Interaction in Me- tals, North-Holland, Amsterdam (1981). 28. A.A. Abrikosov, L.P. Gorkov, and I.E. Dzyaloshinski, Methods of Quantum Field Theory in Statistical Phy- sics, Dover, New York (1975). 29. G. Rickayzen, Green’s Functions and Condensed Mat- ter, Academic Press, London (1980). 476 Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 L. Pietronero and E. Cappelluti 30. G.M. Eliashberg, Zh. Eksp. Teor. Fiz. 38, 966 (1960) [Sov. Phys. JETP 11, 696 (1960)]. 31. L. Pietronero, Europhys. Lett. 17, 365 (1992). 32. I.G. Lang and Yu.A. Firsov, Zh. Eksp. Teor. Fiz. 42, 1843 (1962) [Sov. Phys. JETP 16, 1301 (1963)]. 33. A.S. Alexandrov, Phys. Rev. B38, (1988). 34. A.S. Alexandrov and N.F. Mott, Polarons and Bipo- larons, World Scientific, Singapore (1995). 35. A.J. Millis, R. Mueller, and B.I. Shraiman, Phys. Rev. B54, 5389 (1996). 36. S. Ciuchi, F. de Pasquale, S. Fratini, and D. Fein- berg, Phys. Rev. B56, 4494 (1997). 37. L. Pietronero, S. Str�ssler, and C. Grimaldi, Phys. Rev. B52, 10516 (1995). 38. C. Grimaldi, L. Pietronero, and S. Str�ssler, Phys. Rev. B52, 10530 (1995). 39. C. Grimaldi, L. Pietronero, and S. Str�ssler, Phys. Rev. Lett. 75, 1158 (1995). 40. A.S. Alexandrov and P.P. Edwards, Physica C331, 97 (2000). 41. V.N. Kostur and P.B. Allen, Phys. Rev. B56, 3105 (1997). 42. A.S. Alexandrov, Europhys. Lett. 56, 92 (2001). 43. M. Grabowski and L.J. Sham, Phys. Rev. B29, 6132 (1984). 44. V.N. Kostur and B. Mitrovic, Phys. Rev. B48, 16388 (1993). 45. V.N. Kostur and B. Mitrovic, Phys. Rev. B50, 12774 (1994). 46. E.J. Nicol and J.K. Freericks, Physica C235, 2379 (1994). 47. J.K. Freericks, V. Zlatic, W.K. Chung, and M. Jar- rell, Phys. Rev. B58, 11613 (1998). 48. P. Miller, J.K. Freericks, and E.J. Nicol, Phys. Rev. B58, 14498 (1998). 49. H.R. Krishnamurthy, D.M. Newns, P.C. Pattnaik, C.C. Tsuei, and C.C. Chi, Phys Rev. B49, 3520 (1994). 50. P. Paci, E. Cappelluti, C. Grimaldi, and L. Pietro- nero, Phys. Rev. B65, 012512 (2002). 51. C. Grimaldi, L. Pietronero, and M. Scattoni, Eur. Phys. J. B10, 247 (1999). 52. M. Grilli and C. Castellani, Phys. Rev. B50, 16880 (1994). 53. M.L. Kulic and R. Zeyher, Phys. Rev. B49, 4395 (1994). 54. R. Zeyher and M.L. Kulic, Phys. Rev. B53, 2850 (1996). 55. M.L. Kulic, Phys. Rep. 338, 1 (2000). 56. Z.B. Huang, W. Hanke, E. Arrigoni, and D. J. Scalapino, Phys. Rev. B68, 220507 (2003). 57. E. Cappelluti, B. Cerruti, and L. Pietronero, Phys. Rev. B69, 161101 (2004). 58. G. Baym and L.P. Kadanoff, Phys. Rev. 124, 287 (1961). 59. L.P. Kadanoff and G. Baym, Quantum Statistical Mechanics, Benjamin, New York (1962). 60. D.J. Scalapino, in: Superconductivity, D.R. Parks (ed.), Dekker, New York (1969). 61. J.P. Carbotte, Rev. Mod. Phys. 62, 1027 (1990). 62. M. Scattoni, C. Grimaldi, and L. Pietronero, Euro- phys. Lett. 47, 588 (1999). 63. E. Cappelluti, C. Grimaldi, and L. Pietronero, Phys. Rev. B64, 125104 (2001). 64. M. Botti, E. Cappelluti, C. Grimaldi, and L. Pietro- nero, Phys. Rev. B66, 054532 (2002). 65. P. Benedetti, C. Grimaldi, L. Pietronero, and G. Varelogiannis, Europhys. Lett. 28, 351 (1994). 66. P.W. Anderson, J. Phys. Chem. Solids 11, 26 (1959). 67. D. Fay and J. Appel, Phys. Rev. B20, 3705 (1979). 68. D. Fay and J. Appel, Phys. Rev. B22, 1461 (1980). 69. C. Grimaldi and L. Pietronero, Europhys. Lett. 47, 681 (1999). 70. S. Engelsberg and J.R. Schrieffer, Phys. Rev. 131, 993 (1963). 71. F. Marsiglio, M. Schossmann, and J.P. Carbotte, Phys. Rev. B37, 4965 (1988). 72. E. Cappelluti and L. Pietronero, Phys. Rev. B68, 224511 (2003). 73. X.J. Zhou, T. Yoshida, A. Lanzara, P.V. Bogdanov, S.A. Kellar, K.M. Shen, W.L. Yang, F. Ronning, T. Sasagawa, T. Kakeshita, T. Noda, H. Eisaki, S. Uchi- da, C.T. Lin, F. Zhou, J.W. Xiong, W.X. Ti, Z.X. Zhao, A. Fujimori, Z. Hussain, and Z.-X. Shen, Na- ture 423, 398 (2003). 74. O. Gunnarsson, Rev. Mod. Phys. 69, 575 (1997). 75. W.E. Pickett, in: Solid State Physics, v. 48, H. Eh- renreich and F. Spaepen (eds.), Academic Press, New York (1994). 76. E. Cappelluti and L. Pietronero, Phys. Status Solidi B242, 133 (2005). 77. For a review see: R.S. Markiewicz, J. Phys. Chem. Solids 58, 1179 (1997). 78. E. Dagotto, Rev. Mod. Phys. 66, 763 (1994). 79. P. Fulde, Electron Correlations in Molecules and Solids, Springer Verlag, Heidelberg (1995). 80. Correlated Electron Systems, V.J. Emery (ed.), World Scientific, Singapore (1993). 81. A. Georges, G. Kotliar, W. Krauth, and M. Rozen- berg, Rev. Mod. Phys. 88, 13 (1996). 82. M.C. Gutzwiller, Phys. Rev. Lett. 10, 159 (1963). 83. M. Lavagna, Phys. Rev. B41, 142 (1990) 84. J.H. Kim and Z. Tesanovic, Phys. Rev. Lett. 71, 4218 (1993). 85. J.D. Lee, K. Kang, and B.I. Min, Phys. Rev. B51, 3830 (1995). 86. M. Mierzejewski, J. Zielinski, and P. Entel, Phys. Rev. B57, 590 (1998). 87. T. Yildirim, L. Barbedette, J.E. Fischer, C.L. Lin, J. Robert, P. Petit, and T.T.M. Palstra, Phys. Rev. Lett. 77, 167 (1996). 88. J. Bernholc, Phys. Today 52, 30 (1999). 89. C.M. Varma, J. Zaanen, and K. Raghavachari, Sci- ence 254, 989 (1991). 90. M. Schluter, M. Lannoo, M. Needels, G.A. Baraff, and D. Tom�nek, Phys. Rev. Lett. 68, 526 (1992). 91. M. Schluter, M. Lannoo, M. Needels, G.A. Baraff, and D. Tom�nek, J. Phys. Chem. Solids 53, 1473 (1992). Nonadiabatic breakdown and pairing in high-Tc compounds Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 477 92. O. Gunnarsson and G. Zwicknagl, Phys. Rev. Lett. 69, 957 (1992). 93. E. Koch, O. Gunnarsson, and R.M. Martin, Phys. Rev. Lett. 83, 620 (1999). 94. J.E. Han, O. Gunnarsson, and V.H. Crespi, Phys. Rev. Lett. 90, 167006 (2003). 95. E. Cappelluti, C. Grimaldi, L. Pietronero, and S. Str�ssler, Phys. Rev. Lett. 85, 4771 (2000). 96. E. Cappelluti, C. Grimaldi, L. Pietronero, S. Str�s- sler, and G.A. Ummarino, Eur. Phys. J. B21, 383 (2001). 97. J.C. R. Faulhaber, D.Y.K. Ko, and O.R. Briddon, Phys. Rev. B48, 661 (1993). 98. V.P. Antropov, O. Gunnarsson, and A.I. Liechten- stein, Phys. Rev. B48, 7651 (1993). 99. S.K. Watson, K. Allen, D.W. Denlinger, and F. Hellman, Phys. Rev. B55, 3866 (1997). 100. M.S. Fuhrer, K. Cherrey, A. Zettl, M.L. Cohen, and V.H. Crespi, Phys. Rev. Lett. 83, 404 (1999). 101. For a review see: Y. Iwasa and T. Takenobu, J. Phys.: Condens. Matter 15, 495 (2003) and references therein. 102. O. Zhou, R.M. Fleming, D.W. Murphy, M.J. Ros- seinsky, A.P. Ramirez, R.B. van Dover, and R.C. Haddon, Nature 362, 433 (1993). 103. M.R. Norman, D. Pines, and C. Kallin, cond-mat/0507031 (2005). 104. J. Rossat-Mignod, L.P. Regnault, C. Vettier, P. Bourges, P. Burlet, J.Y. Henry, and G. Lapertot, Phy- sica C185–189, 86 (1991). 105. V. Barzykin and D. Pines, Phys. Rev. B52, 13585 (1995). 106. P. Monthoux and D. Pines, Phys. Rev. B50, 16015 (1994). 107. Ch. Renner, B. Revaz, J.-Y. Genoud, K. Kadowaki, and O. Fischer, Phys. Rev. Lett. 80, 149 (1998). 108. V.M. Loktev, R.M. Quick, and S.G. Sharapov, Phys. Rep. 349, 1 (2001). 109. M. Grilli, R. Raimondi, C. Castellani, C. Di Castro, and G. Kotliar, Phys. Rev. Lett. 67, 259 (1991). 110. J.M. Tranquada, J.D. Axe, N. Ichiwara, A.R. Moo- denbaugh, Y. Nakamura, and S. Uchida, Phys. Rev. Lett. 78, 338 (1997). 111. V.J. Emery and S.A. Kivelson, Physica C209, 597 (1993). 112. V.J. Emery and S.A. Kivelson, Physica C235, 189 (1994). 113. C. Castellani, C. Di Castro, and M. Grilli, Z. Phys. B103, 137 (1997). 114. R.S. Markiewicz, J. Phys. Chem. Solids 59, 1737 (1998). 115. J.W. Loram, K.A. MIrza, J.R. Cooper, and J.L. Tallon, J. Phys. Chem. Solids 10–12, 2091 (1998). 116. E. Cappelluti and R. Zeyher, Phys. Rev. B59, 6475 (1999). 117. S. Onoda and M. Imada, J. Phys. Soc. Jpn. Suppl. B69, 32 (2000). 118. J. Stajic, A. Iyengar, K. Levin, B.R. Boyce, and T. Lemberger, Phys. Rev. B68, 024520 (2003). 119. L. Boeri, E. Cappelluti, C. Grimaldi, and L. Pietro- nero, Phys. Rev. B68, 214514 (2003). 120. C.C. Tsuei, J.R. Kirtley, C.C. Chi, L.S. Yu-Jahnes, A. Gupta, T. Shaw, J.Z. Sun, and M.B. Ketchen, Phys. Rev. Lett. 73, 593 (1994). 121. G. Blumberg, A. Koitzsch, A. Gozar, B.S. Dennis, C.A. Kendziora, P. Fournier, and R.L. Greene, Phys. Rev. Lett. 88, 107002 (2002). 122. J.A. Skinta, M.-S. Kim, T.R. Lemberger, T. Greibe, and M. Naito, Phys. Rev. Lett. 88, 207005 (2002). 123. A. Biswas, P. Fournier, M.M. Qazilbash, V.N. Smo- lyaninova, H. Balci, and R.L. Greene, Phys. Rev. Lett. 88, 207004 (2002). 124. N.-C. Yeh, C.-T. Chen, G. Hammerl, J. Mannhart, A. Schmehl, C.W. Schneider, R.R. Schulz, S. Tajima, K. Yoshida, D. Garrigus, and M. Strasik, Phys. Rev. Lett. 87, 087003 (2001). 125. Y. Dagan and G. Deutscher, Phys. Rev. Lett. 87, 177004 (2001). 126. A. Sharoni, O. Millo, A. Kohen, Y. Dagan, R. Beck, G. Deutscher, and G. Koren, Phys. Rev. B65, 134526 (2002). 127. A.I. Lichtenstein and M.L. Kulic, Physica C245, 186 (1995). 128. P. Paci, C. Grimaldi, and L. Pietronero, Eur. Phys. J. B17, 235 (2000). 129. See for instance: P.B. Allen, Z. Fisk, and A. Migli- ori, in: Physical Properties of High-Temperature Su- perconductors, D.M. Ginsberg (ed.), World Scientific, Singapore (1989). 130. E. Cappelluti and L. Pietronero, Europhys. Lett. 36, 619 (1996). 131. E. Cappelluti and L. Pietronero, Phys. Rev. B53, 932 (1996). 132. J. Kortus, I.I. Mazin, K.D. Belashchenko, V.P. Ant- ropov, and L.L. Boyer, Phys. Rev. Lett. 86, 4656 (2001). 133. J.M. An and W.E. Pickett, Phys. Rev. Lett. 86, 4366 (2001). 134. T. Yildirim, O. G�lseren, J.W. Lynn, C.M. Brown, T.J. Udovic, Q. Huang, N. Rogado, K.A. Regan, M.A. Hayward, J.S. Slusky, T. He, M.K. Haas, P. Khali- fah, K. Inumaru, and R.J. Cava, Phys. Rev. Lett. 87, 037001 (2001). 135. Y. Kong, O.V. Dolgov, O. Jepsen, and O.K. An- dersen, Phys. Rev. B64, 020501 (2001). 136. A.Y. Liu, I.I. Mazin, and J. Kortus, Phys. Rev. Lett. 87, 087005 (2001). 137. K.-P. Bohnen, R. Heid, and B. Renker, Phys. Rev. Lett. 86, 5771 (2001). 138. E. Cappelluti, S. Ciuchi, C. Grimaldi, L. Pietronero, and S. Str�ssler, Phys. Rev. Lett. 88, 117003 (2002). 139. E. Cappelluti, S. Ciuchi, C. Grimaldi, and L. Pietro- nero, Phys. Rev. B68, 174509 (2003). 140. L. Boeri, G.B. Bachelet, E. Cappelluti, and L. Piet- ronero, Phys. Rev. B65, 214501 (2002). 141. L. Boeri, E. Cappelluti, and L. Pietronero, Phys. Rev. B71, 012501 (2005). 478 Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 L. Pietronero and E. Cappelluti