You are on page 1of 14

Interplay of Charge Density Waves and Superconductivity

Jason Sadowski April 22, 2010

Abstract The nature of the relationship between CDW and SC is not fully understood, and the literature is full of conicting experimental evidence. This paper will provide a review of the Greens function formalism for both charge-density waves and superconductors, as well as discuss the nature of the long ranged order inherent to each state. Experimental evidence will be given illustrating the non-intuitive nature of the interplay between charge-density waves and superconductivity. Contents 1 Introduction 2 Theory of CDWs 2.1 2.2 2.3 2.4 Greens Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Random Phase Approximation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Dielectric Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Diagonal Long Range Order . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 3 3 4 5 6 7 7 8 9 10 10 10 11

3 Superconductivity 3.1 3.2 3.3 Results from the BCS Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Anomalous Greens Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . O-Diagonal Long Range Order . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

4 Evidence of Density Waves in Superconductors 4.1 4.2 Competition of CDW and SC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Enhancement of SC by CDW . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

5 Conclusion

A Appendix A.1 Summary of Diagram Rules for the Single Particle Propagator . . . . . . . . . . . . . . . . . . . . . . A.2 Dielectric Response in 3D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

12 12 13

Introduction

be saved by the gap opening at the Fermi surface? To his complete surprise [2] this distortion of the lattice was always energetically favorable and such a metal would always undergo what is now known as a Peierls Transition. Since the distortion of the lattice causes the electron density to distort in a periodic fashion, it is also known as a charge-density wave transition (CDW). The second problem faced by solid state theory was a theoretical explanation for superconductivity. Superconductivity was discovered in 1911 by Keike Kamerlingh Onnes while investigating the low temperature properties of metals. It was thought at the time that perhaps the electron phonon interaction of the Peierls transition could be the mechanism responsible for superconductivity. Bardeen and Frhlich were able to show that in a o pure 1D metal the sliding of electrons could generate a so called ideal conductor, as well as the required energy gap at the Fermi surface. However they quickly realized that this theory could not explain the multitude of 3D superconductors, or the Meissner eect. It wasnt until 1957 when Bardeen, Cooper, and Schrieer (BCS) were able to successfully use the quantum theory to describe the superconducting state in a true many body fashion. Although it has been realized that the Peierls transition can not be reconciled with superconductivity, there is a large body of evidence which in fact shows that these two states can coexist with each other. It has been the consensus that the CDW state (which exhibits insulating properties) should compete with superconductivity. However there is also experimental evidence to show that the superconducting state can be enhanced by the electronhole pairing at the Fermi surface by the CDW state. At this time the issue is not about the coexistence of CDW and SC, but whether or not the CDW is favorable or

The quantum theory of the solid state was developed in the mid 1930s and had seen many triumphs in the prediction of many material properties, however there were two obstacles standing in the way of the solid state theories success. The rst was the absolute failure of the solid state theory in predicting certain metal-insulator transitions (now referred to as Mott-Insulator metals). According to the theory compounds such as NiO should be a metal since the electrons form a d band which is not completely occupied. The experimental fact however, is that NiO is an insulator. The explanation rst given by Mott in 1949 says that this discrepancy comes from approximating the electrons as only weakly interacting. It turns out that the Coulomb repulsion (which is not accounted for in this approximation) is enough to cause them to become localized at a particular lattice site, making them insulators. These metals provided some of the rst examples where the many-body eects from the electron interactions simply could not be ignored. In the mid 1930s Rudolph Peierls was writing an introductory textbook on the quantum theory of the solid state [1], and was in the process of designing a manybody exercise when he made an accidental discovery. He was considering what would happen if the crystal lattice could be distorted, and eectively double the size of the unit cell. It was well established that the periodic potential of the lattice will cause a gap to form at the edge of the Brillouin zone (In a one dimensional lattice of length a the gap is at k = /a). Doubling the size of the unit cell, he thought, should eectively halve the size of the Brillouin zone and the gap would open at the Fermi surface. The question he was asking was how much energy would

destructive to superconductivity. The remainder of this paper will discuss the many-body theories of both superconductivity,charge density waves, as well as the interplay between the two sates. Section 2 and Section 3 discusses the theoretical description of CDW and SC ground states and examines intrinsic dierences between them. Finally, Section 4 provides examples of some recent experiments which illustrate conicting evidence about the interplay of CDW and superconductivity. 2 Theory of CDWs

of momentum k2 at time t2 , if we were to inject a particle of momentum k1 into the ground state of the full interacting system at time t1 . Thus, knowledge of the Greens function provides information about the nature of the many body interactions, as it represents the propagation of a particle in the full many body system. Since it is usually impossible to determine the true many body Greens function, it is expanded in a perturbation series to try to incorporate the most dominant contributions. A summary of the diagram rules for the single particle propagator perturbation expansion is given in Appendix A.1. The expansion of the propagator using these rules is given by [3]:

As already mentioned the charge density wave ground state was serendipitously discovered by Peierls while preparing materials for his book on solid state physics. At the time he did not consider the result anything more than an academic curiosity, since in practice there is no such thing as a strictly one dimensional metal. It was later discovered that some materials do in fact exhibit CDW properties at low temperatures, in agreement with his earlier results. In his book More Surprises in Theoretical Physics he says, In this case the rst surprise was in the mathematical result for the one-dimensional case; a second surprise was that this had some connection with reality. In this section the many-body formulation of the CDW state, particularly the properties of the Greens function, will be discussed.

However the third and fourth terms in the sum are said 2.1 Greens Functions to be composed of repeated simple parts since they are composed of elements similar to the rst and second. Recall that the single particle Greens function for the many-body system is given by: E D G (k2 , t2 ; k1 , t1 ) = i 0 T {ck2 (t2 )c 1 (t1 )} 0 k These terms which can not be further broken down are referred to as the irreducible self energy contribution to (1) the Greens function. It is possible to sum over diagrams which are irreducible using partial summation. For example the usual Hartree-Fock approximation to the Greens function is given by the lowest order irreducible self energies due to exchange and Coulomb interactions. Summing

where T {} is the usual time ordering operator. Physically the Greens function represents the amplitude for nding the system in its ground state with an added particle

over all of the repeated self-energies gives: 1


1

are considered in the contributions to the irreducible self energy. This approximation was rst introduced by GellMann and Brueckner who attempted to model the screening eects in the electron gas. In order to see why these

graphs are the dominant contribution we consider the expression for the diagram given by,

where the term represents the sum over irreducible self energies.

DRP A = XZ (4e2 )2 d d (1) iG0 (k q, )iG0 (q + p, + )iG0 (p, ) q4 (2)2 q,p It is clear that the integral diverges at q = 0 due to the q 4 in the denominator, which is due to the long range nature of the Coulomb potential. Similar diagrams of higher order will have the same behavior, and will also diverge. Therefore the most divergent diagrams are ones which have the same momentum q transferred, and we pick up

(2)

Using the rules in Appendix A.1 Equation 2 becomes: 1 G(k, ) = (3) k (k, ) + ik Equation 3 is referred to as Dysons Equation and is the starting point in most many-body calculations. Although the partial summation which was performed is exact, it is in general not possible to determine (k, ) and some approximations have to be made. For a given system, one must determine which contributions to will be the most dominant. For example in the low density limit the only important contributions to the sum are the so called ladder graphs, which are graphs containing only one hole line. In the next section we will look at the contribution to the irreducible self energy in the high density limit, also known as the random phase approximation. 2.2 Random Phase Approximation

a factor of 1/q 2 in the integral. Evidently the most dominant contribution to the self energy are those due to the ring diagrams.

After performing a partial summation over the repeated graphs this sum over innite divergences manages to give a nite result! It is found that the eect of the random phase approximation is to renormalize the Coulomb potential lines leaving an eective interaction. It can be shown that this eective interaction is given by: iVef f (RP A) =
RP A

= 1

(4)

In the random phase approximation only diagrams of the q form:

Looking at the diagram for the eective interaction we can see that the Coulomb interaction is renormalized as a DRP A = kq p p+q consequence of the sum over the fermion pair loops. If a horizontal line were drawn across the diagram representing a particular instant in time, the primary contribution q

is due to an electron-hole interaction. In a sense the system has become virtually polarized, and the sum is occasionally referred to as a sum over polarization diagrams. Using the diagram rules Equation 4 can be written as: Vef f (RP A) = Vq 1 + Vq 0 (q, ) Vq RP A (q, ) (5)

The reason the dielectric response leads to a Charge Density Wave state is because the screened interaction of the RPA causes the electrons to form an induced charge density, and modulation of the surrounding lattice ions. It can be shown that when the dielectric response diverges, the charge density for large distances takes the form: cos(2kf r + ) r3 (7)

and 0 (q, ) is the interaction of an electron and hole pair. The generalized dielectric constant,
RP A (q, ), is a result

of the polarization of the medium due to an applied eld. Is is the properties of this dielectric response that are of importance in understanding the Charge Density Wave state. 2.3 Dielectric Response

where is an associated phase [4]. This is the origin of the term charge density wave. Consequently, the modulation of the lattice has the same period of oscillation. It is interesting to note that the screened interaction is the origin of the so called Kohn Anomaly. The physical reason for this is given by the eective interaction in Equation 4. All of the interactions are replaced by the screened interactions, which depend on the dielectric response
RP A (q, ).

Equation 5 provides the denition for the generalized dielectric constant for a system of interacting electrons. This result is important because the dielectric response
RP A (q, )

As a result electrons and phonons

characterizes how the electron gas will respond

interact via this screened eld. The singularity which develops in the dielectric response manifests itself as dramatic dip in the phonon frequency spectrum at q = 2kf , resulting in the Kohn Anomaly.

to an applied potential. It has been shown by other means that the response is given by: Z fo (k + q) fo (k) d3 k e2 RP A (q, ) = 1 2 Ek+q Ek (2)3 0q

(6) Nesting In one dimension the Fermi Surface is simply 2 parallel lines at k = kf . An electron at k = kf and a hole at k = +kf have the same energy and are separated by q = 2kf . The wavevectors q which connect the divergent points are referred to as nesting vectors. Since there are an innite number of these points in 1D the response is strongly divergent. In higher dimensions however the number of these points is reduced as shown in Figure 2.3

Where fo (Ek ) is the Fermi-Dirac distribution function for energy Ek , and and q are the Fourier components of the perturbation potential. The interesting characteristic of the dielectric response is that in 1D it diverges. By setting = 0 for a time independent potential the integral can be performed explicitly to give:
1D RP A (q, )

= 1 + e g(Ef ) ln

q + 2kf q 2kf

Where g(Ef ) is the density of states at the Fermi level. This result is divergent at q = 2kf which implies that the system of electrons is unstable. Where does this divergence come from? Taking a closer look at Equation 6 shows that the most dominant contributions to the integral occur when Ek+q Ek = 0. That is, when two states spanned by the same wave-vector q have the same energy, the integral diverges, and the system becomes unstable.

Figure 1: Nesting in 1,2 and 3 Dimensions. The number of points which lead to divergent terms in the 5

integral is reduced in higher dimensions


It is apparent that the strength of the divergence is strongly Dependant on the particular topology of the Fermi Surface. Likewise the integral in Equation 6 can be carried out in 3D to give the result rst derived by Lindhard
1

which is evidently the Fourier transform of the momentum distribution nk . Now recall that a macroscopically occupied mode is a coherent quantum many particle state. In the coherent state only one momentum states is macroscopically occupied and thus, nk |k |2

3D RP A (q, 0)

|1 + | e2 1 2 =1+ ln g(Ef ) 1 + 2 2 |1 | 0q (8)

for the coherent state |k1 , k2 , k3 , Hence the one particle density matrix is: (r r ) = 1 X ik(rr ) e |k |2 V
k,k

This expression is no longer divergent at q = 2kf , and as a result CDWs in three dimensional materials are quite rare. The calculation is performed for zero temperature, but since Equation 6 contains the Fermi-Dirac distribution function the divergence of the dielectric response, and thus the CDW state, is also a strong function of temperature. 2.4 Diagonal Long Range Order

If only one of these states has a nite occupation, say No at ko then the momentum distribution has the form: nk No k,ko + f (k) (11)

where f (k) is assumed to be some suciently smooth function. The corresponding density matrix (which is proportional to the single particle Greens function) by Fourier transform has the form: Z

The CDW state is characterized by the interaction between the electrons and phonons. At temperatures below a certain threshold the system transforms into one with modulated lattice ions and charge density. As it was stated elsewhere [5], the CDW exhibits a macroscopically occupied phonon mode that characterizes an ordered state with an associated transition temperature. In an ordered state the value of some property of the system at a given point is correlated to the value at some other point, potentially a long distance away. The one particle density matrix is given by: D E (r r ) = (r)(r ) (9)

(r r ) =

No 2 + V (2)3

eik(rr ) f (k)d3 k

(12)

where V is the volume of the system. In a normal (disordered) system the one particle density matrix should decrease with the separation distance r r , usually exponentially. However, since the density decays to a constant value when r r in Equation 12 the system is said to exhibit long range order. Additionally, since the density matrix goes to a constant value for large r, it means that the two points in Equation 9 are statistically independent. This means that the thermal average of the operators can be computed independently. Hence, D E D E (r r ) = (r)(r ) (r) (r )

and is clearly proportional to the one particle Greens function. Expanded in its Fourier coecients: 1 X ik(rr ) D E e ak ak (r r ) = V
k,k 1 According

(10)

to many textbooks, the Lindhard expression is the result of a trivial integral. As in most cases however, the

result is far from trivial, and is derived in Appendix A.2.

Systems which exhibit this property are said to have diagonal long range order because the density matrix corresponds to the normal Greens Functions dened previously. In the following sections, we will discuss the

pair. Bardeen, Cooper and Schrieer then tried to solve the reduced Hamiltonian, Hred = 2 X
k k bk bk

X
k,k

Vk,k b bk k

(13)

anomalous Greens functions and the o-diagonal long ranged order in superconductors. 3 Superconductivity where the bk s are the Cooper pair creation and anhilation operators bk = ck ck . This Hamiltonian describes a gas of Cooper pairs interacting above a lled Fermi sea. Although the Hamiltonian is reduced, it is still not possible The CDW state has many features in common with the superconducting state, namely they both are due to interaction of electrons with the lattice, and both have an energy gap opening at the Fermi surface. These two states have many features in common, but have physical properties which are worlds apart. The CDW state tends to have very high eective dielectric constant, and thus exhibits insulating behavior, while the SC state exhibits zero resistivity, as well as expelling a magnetic eld from the interior of the sample. As was mentioned in the Introduction, the CDW and SC state were quickly realized to be incompatible with each other and were assumed to compete. Strangely, there is experimental evidence to show that these two states can in fact coexist and do not always compete with each other. In this section we will briey review the results from the usual BCS theory as well as discuss the Greens function approach to the superconducting ground state. In this way the primary dierences between the CDW and SC state can be illustrated. 3.1 Results from the BCS Theory
2 vk

to determine the exact form of the wavefunction. The real breakthrough came when Schrieer suggested a form of coherent state wavefunction using a variational approach. He suggested that the ground state wavefunction could be written as: |0 = Y uk + vk b |0 k
k

(14)

where uk and vk are the hole and particle amplitudes for a Cooper pair state respectively. The amplitudes uk and vk were treated as variational parameters, and are chosen such that it makes the ground state energy of the reduced Hamiltonian minimum. After performing such a procedure one nds: 1 = 2 1
k

Ek

(15)

2 and u2 +vk = 1. In addition we nd the energy dispersion k

law: Ek =

2 k

+ 2 k

(16)

where k represents a gap at the Fermi surface, given by the self consistent equation: X
k

Before continuing into the Greens function approach to SC it will be benecial to briey review some of the results from the BCS theory. The major discovery made by Cooper was that there existed certain scenarios where the interaction between two electrons could be attractive via the electron phonon interaction. He later came to realize that if two electrons of opposite momentum were interacting above a lled Fermi sea, then these electrons would always form a bound state now known as a Cooper k =

Vkk

k 2Ek

(17)

In contrast to the Bose-Einstein condensates, the BCS wavefunction represents a coherent state of Cooper pairs, which are fermions. It is this fact, in addition to the Pauli exclusion principle, which lead to the interesting characteristics of the superconductor. Excited states of the SC are determined using the reduced Hamiltonian, and

can be obtained through the Bogoliubov-Valatin canonical transformations. These transformations are given by [6]

The expansion of the propagator will be the same as usual, however we should replace the Coulomb interaction with the BCS interaction given by Cooper. The graphs which make the dominant contribution to the irreducible self energy should be those in which two particles interact with no net momentum transfer q = 0. These graphs are the so called ladder graphs and the sum of them dene the tmatrix, which also obeys a Dyson like integral equation.

= =

up c vp cp p up cp + vp c p

as well as their Hermitian conjugates. When these operators act on the superconducting ground state |0 one nds the relations, t
p |0 p |0

= = = =

|p |p 0 0 t(q = 0, )

, For the attractive interaction in the BCS problem the t-matrix has the form: 1 1 + + i i

p |0 p |0

Evidently the operators p and p create and de-

Taking the Fourier transform of this equation one can see that the pole in the right half plane will cause the result to diverge, and the solution is unstable when time t . Although the usual perturbation theory can not be performed for the SC state, it is possible to dene a new type of Greens function where the usual techniques will apply. The technique was rst introduced by Nambu and Gorkov in an attempt to save the nice Dyson equations for the SC state. The major obstacle in dening a perturbative expansion of the Greens function comes from the form of the reduced Hamiltonian, where one has to deal with terms of the form: Hint X
k

stroy excitations in the superconducting state, where |0 is the vacuum state for the superconductor. These excitations are sometimes referred to as Bogoliubov quasi particles or Bogolons. 3.2 Anomalous Greens Function

Although the Greens function approach has been tremendously successful for describing many metals in the normal state, the method breaks down when applied to superconductivity, and a new approach must be found. A quick calculations shows why the usual perturbation expansion is no longer valid. It is well established that the poles of the Greens function correspond to the energies of the system, and from Equation 16 we would expect something of the form: G(k, ) p 1 2 2 k + k + i

Vk

c c + H.c k k

(18)

In this case one does not have the usual type of one particle interaction terms. Nambu discovered that it is possible to eliminate this problem if we dene the two component eld operator:

k =

ck c k

= k

c k ck

3.3 (19)

O-Diagonal Long Range Order

It was stated in the last section that the BCS wavefunction is a coherent state of Cooper pairs. This means that below the critical temperature the density of Cooper pairs picks up a nite occupation number and becomes macroscopically occupied state. The density matrix of Cooper pairs is D E (r r ) = (r)(r ) and is now the wavefunction for a Cooper pair. Similarly to Section 2.4 the Fourier transform is E 1 X ik(rr ) D e ck c ck ck k V
k,k

n Then o the commutation relations are given by k , = kk 1 and {k , k } = 0, with these defk initions the complication of Equation 18 are removed [7], and one can perform the perturbation expansion in the usual way. The Greens function becomes, G (p, t) = i 0|T {p (t)p (0)} |0 (20)

and the propagator written explicitly in matrix form is ! G(p , t) F (p, t) G= (21) F (p, t) G(p , t) From the denition of the two component spinors (19) and the Greens function it is clear that the anomalous Greens function F is given by: F (p, t) F (p, t)

(r r ) =

(27)

= =

i 0|T {ck ck } |0 D n o E i 0|T c c |0 k k

(22) (23)

which is clearly proportional to the anomalous Greens functions. All of the arguments from Section 2.4 still hold, however now it is the anomalous Greens function which become correlated as r r . In this case the SC state is said to exhibit o-diagonal long range order since F (p, t) comprises the o-diagonal components of the Greens function G.

and now the perturbation can be written down by using the matrix propagator G and the usual diagrams for G. G therefore obeys the Dyson equation with all the interactions now evaluated in the Nambu formalism: 1
1

(24)

The o-diagonal long ranged order in the anomalous Greens functions is the primary dierence from the charge density wave state. It is generally believed that ODLRO is a sucient condition (although not necessary) for the appearance of super-phenomena, and was proved to be capable of maintaining both the Meissner eect and ux quantization. The CDW state exhibits DLRO, and is responsible for the coherence properties such as sliding and the locking of phase. Although they are dierent, it is possible for the two types of order to exist in the same material. The next section will give examples of materials demonstrating coexistence of superconductivity and charge-density waves, as well as illustrate the non-intuitive nature of the interplay between these two states.

and the Greens function is: 1 (25) 1 k 3 (k, ) where t3 is the Pauli matrice. When the reduced HamilG(k, ) = tonian is solved using G the BCS results are easily reproduced, and give for G11 of G: G11 = p u2 k 2 + k 2 + i k + + p
2 vk 2 k

+ 2 i k

(26)

which is precisely what was expected based on physical grounds. For a detailed account of how to arrive at this result see Schrieer (pg 178).

Evidence of Density Waves in Superconductors

The goal of this section is to illustrate that the CDW state and SC can in fact coexist with each other, and that often the results are very non intuitive. One would expect that due to the insulating properties of the charge density wave state that CDW should compete with superconductivity. In many experiments this is in fact the case, however there are examples where CDW can enhance SC. This section will describe a pair of examples which show conicting evidence regarding the nature of the interplay of CDW and SC.

Figure 2: The specic heat measurements for the 3 and


40 day samples. The three day samples show the SC phase transition at TC = 4K, and demonstrates the Meissner

4.1

Competition of CDW and SC

Eect. The 40 sample has TCDW = 7K and is not effected by a magnetic eld [8]

Recently (2009) H. Fu et al at the Lawrence Berkeley National Laboratory were studying the eects of charge density wave in samples of N a0.3 CoO2 1.3H2 O. These samples exhibit superconductivity at 4.5K, and the strength of the superconducting state depends on the age of the sample. As the sample ages oxygen vacancies begin forming causing pair breaking to occur, and eventually superconductivity is killed. The group performed specic heat measurements for samples aged at three, ve, and forty days. Superconductivity is present in the three day sample and shows measurable pair breaking by 5 days. By 40 days the superconductivity in the sample is completely destroyed. In the 40 day sample however the specic heat measurements still show a second order phase transition, but it is moved from 4K to 7K. It was concluded that the phase transition in this sample is due to a CDW state since the transition is unaected by a magnetic eld. Figure 2 show the specic heat measurements for the 3 and 40 day samples.
4.2 Enhancement of SC by CDW What is interesting about this experiment is that the charge density wave transition was anticipated to occur at 7K by a theoretical calculation. Intriguingly however the CDW state only occurs in the 40 day sample, and since TCDW > TC it should occur in all of the samples. Only once superconductivity has been destroyed does the CDW state appear. The conclusion is that the superconducting state competes with CDW, and pair breaking frees portions of the Fermi surface allowing the CDW state to form.

In 2007 Kiss et al performed angle resolved photo emission (ARPES) measurements of N bSe2 and were able to map the Fermi surface in both the CDW and SC state. It was shown that regions of the Fermi surface where CDW occurs correspond to regions where the SC gap is maximum. The Fermi surface mapping is shown in Figure 3.

10

being made both experimentally and theoretically. 5 Conclusion

In this review it has been shown that in many cases the CDW and SC state are capable of coexisting with each other. At this time there is no consensus as to whether charge density waves compete with, or enhance superconductivity. There seems to be numerous experimental evidence to support both arguments, and it would not be possible to list all of them here. For a review of the current state of experimental studies regarding the CDW and SC one should look at [10]. Although CDW and SC both

Figure 3: Mapping of k-points corresponding to CDW


and k-dependant SC gap [9]. The upper portion corresponds to measurements in the CDW state while the lower portion is in the SC state. Red and blue circles identify primary and secondary CDW nesting vectors, respectively, and colored dots give the magnitude of the superconducting gap across the Fermi surface. The upper portion corresponds to the CDW measurements and show the location of the CDW nesting vectors. Primary vectors (red circles) form a triangle on the K2 sheet and blue circles lying on the kogome lattice labeled 2 correspond to secondary CDW vectors. The sample was cooled into the SC state, and the magnitude of the gap was measured at dierent regions of the Fermi surfaces as shown in the lower portion of Figure 3. It can be seen that the points where the SC gap is maximum correspond precisely to the location of the CDW nesting vectors. These results indicate that a charge density wave may enhance superconductivity in certain regions of momentum space. These results are completely contrary to those given in the previous example. This problem is not restricted to these particular results, and in general it possible to nd many examples of such conict in the literature. It is clear that there is still much to learn in the eld of many-body physics, and everyday new developments are

stem from the electron phonon interaction and have very similar characteristics, their physical properties are very dierent. It was shown that the two states show dierent types of long ranged order ,diagonal long range order in the charge density waves, and o-diagonal for superconductivity. It is believed that the fundamental dierence between the two types of long ranged order are what leads to the interesting properties of these two states. Despite the large volume of experimental evidence and theoretical might of the Greens function approach, the true nature of the interplay between charge density waves and superconductivity is still elusive, and remains a scientic challenge waiting to be solved.

11

A A.1

Appendix Summary of Diagram Rules for the Single Particle Propagator Diagram Element Factor

or

iG(k, )

or

iG0 (k, ) =

i k +i

q m n

iVklmn or iVq (Vklmn ( ) for time dependent interactions)

Fermion loops:

(1)

Intermediate Energy

d 2

Intermediate momentum k

or

d3 k (2)3

(including sum over spins for unit volume)

12

A.2

Dielectric Response in 3D

To nd the dielectric response in three dimensions we have to evaluate an integral of the form: Z f0 (k + q) f0 (k) e2 d3 k (q, ) = 1 RP A 2 E(k + q) E(k) (2)3 0q (A.1) Where f0 (k) is the Fermi-Dirac distribution function for energy E(k). In three dimensions the volume element in momentum space is d k k sin dddk. For the free
3 2

factor of 2, and after a little manipulation: 2 3 Z Z kf 2 1 k2 k2 5 4 I = d(cos) + dk q 2 1 0 1 + 2kq 1 2kq 2 2 = = 2 q2 2 q Z


0 kf

Z
0

kf

1 q 2k cos q 2k cos ln(1 + ) ln(1 ) 2k q 2k q 1 ! 2k 1+ q k ln dk 1 2k q


2k q

Now perform the change of variable x = =


q 2kf

and

electron gas the energy-momentum dispersion relation is: k Ek = 2m For simplicity we will only consider the zero frequency response at zero temperature. We are considering the scattering between states of momentum k and k and are connected by wavevector q = k k.
2 2

then the integral becomes: I= (A.4)

Z q 1/ 1+x x ln dx 2 0 1x Performing the integration by parts we get: q I= 2 x2 ln 2

! 1/ Z 1+x 1 1/ x2 x2 + dx 1 x 0 2 0 1+x 1x

Now we use the result: Z b2 ax2 2bx x2 dx = 3 ln (|ax + b|) + + C (A.5) ax + b a 2a2 to arrive at: |1 + | q 1 2 1+ ln (A.6) 2 2 |1 | The zero frequency dielectric response in three dimenI=

Figure 4: Scattering of electrons from state k to k . sions is then: At zero temperature all states less k < kf are occupied and states k > kf are unoccupied
Then at zero temperature for all states k below the Fermi-level we have:
3D RP A (q, 0)

=1+

|1 + | 1 2 2me2 kf 1+ ln (2)3 2 0 q 2 2 |1 | (A.7)

Recall that for a free electron gas the density of states at the Fermi level is given by: 0 (k < kf )

f0 (k q) =

And the integral can be split into two parts for q


3D RP A (q, 0)

3/2 1 2m kf (A.8) 2 2 2 2m Substituting this into the expression for (q, 0) and g(Ef ) = accounting for both spin directions gives the Lindhard result for the response in 3D |1 + | 1 2 e2 g(Ef ) 1 + ln 2 2 |1 | 0q (A.9)

=1

e2 2m 2 3 0 q (2)

(A.2)

I=

1 1 + d3 k (A.3) (k + q)2 k2 (k q)2 k 2 |kf kf |

3D RP A (q, 0)

=1+

Since the integral is independent of we pick up a

13

References [1] Rudolph Peierls. Quantum Theory of Solids. Clarendon Press, Oxford, 1955. [2] Rudolph Peierls. More Surprises in Theoretical Physics. Princeton University Press, 1991. [3] R.D. Mattuck. A Guide to Feynman Diagrams in the Many Body Problem, page 145. McGraw-Hill, 1967. [4] T.D. Schultz. Quantum Field Theory and the Many-Body Problem, page 99. Gordan and Breach, Science Publishers, 1964. [5] Jason Sadowski. Interplay of charge density waves and superconductivity. Presentation for PHYS 894, March 2010. [6] J. R. Schrieer. Theory of Superconductivity, pages 4748. W.A. Benjamin, Inc., Publishers, 1964. [7] J. R. Schrieer. Theory of Superconductivity, pages 173176. W.A. Benjamin, Inc., Publishers, 1964. [8] H.Fu and N. Oeschler et al. Competition between superconductivity and charge-density wave in Na0.3 CoO2 1.3H2 O. Journal of Superconductivity and Novel Magnetism, 22(3):295 298, 2009. [9] T. Kiss, T. Yokoya, A. Chainani, S. Shin, T. Hanaguri, M. Nohara, and H. Takagi. Charge-order-maximized momentum-dependent superconductivity. Nature Physics, 3:720725, October 2007. [10] A. M. Gabovich, A. I. Voitenko, and M. Ausloos. Charge- and spin-density waves in existing superconductors: competition between cooper pairing and peierls or excitonic instabilities. Physics Reports, 367(6):583 709, 2002. [11] Goerge Grner. Density Waves in Solids. Perseus Publishing, Massachusetts, 1996. u

14

You might also like