Friedel Oscillations as a Probe of Fermionic Quasiparticles
Emanuele G. Dalla Torre,1, 2 David Benjamin,2 Yang He,2 David Dentelski,1 and Eugene Demler2
2
1
Department of Physics, Bar Ilan University, Ramat Gan 5290002, Israel
Department of Physics, Harvard University, Cambridge, MA 02138, U.S.A.
arXiv:1507.04345v1 [cond-mat.str-el] 15 Jul 2015
When immersed in a see of cold electrons, local impurities give rise to density modulations known
as Friedel oscillations. In spite of the generality of this phenomenon, the exact shape of these
modulations is usually computed only for non-interacting electrons with a quadratic dispersion
relation. In actual materials, one needs to take into account several additional factors, such as (i)
the details of the band structure, (ii) the lifetime of quasiparticles, (iii) in superconductors, the
presence of a pairing gap. Studying how these effects influence Friedel oscillations is a viable way
to access the properties of fermionic excitations in strongly-correlated materials. In this work we
analyze the signatures of Friedel oscillations in STM and X-ray scattering experiments, focusing on
the concrete example of cuprates superconductors. A detailed comparison with recent experiments
reveals a rich interplay between local modulation of the chemical potential and of the pairing gap.
Contents
I. Introduction
1
II. Theoretical framework
A. Lindhard formula
B. Local density of states
C. Wannier functions and Bragg peaks
D. Scattering from local impurities
E. Impurities in a paired state
2
2
2
3
3
4
III. Experiments
A. X-ray: effects of the band structure
B. STM: dispersive vs non-dispersive peaks
C. STM and REXS: identifying the impurities
D. STM: Masking procedure
E. REXS: Temperature dependence
4
4
6
7
7
10
IV. Summary and outlook
11
A. Technical details
1. Average over impurities
2. REXS: Green’s function approach
3. REXS: Spin-orbit effects
4. REXS: Polarization dependence
References
I.
12
12
13
14
15
16
INTRODUCTION
Friedel oscillations are density modulations generated
by local impurities acting on mobile charges, such as electrons in a metal. At the lowest order of perturbation
theory, these modulations are given by the static densitydensity response function of the unperturbed charges.
For free electrons in three dimensions, this function is
peaked at twice the Fermi momentum, making Friedel oscillations a direct tool to measure the electrons’ density.
In more complex materials these oscillations can deliver
important information about the band structure of the
fermionic quasiparticles, their lifetime, and the presence
of a pairing gap.
The experimental measurement of Friedel oscillations
is however challenging: the most direct procedure is given
by hard X-ray scattering experiments. This approach
was successfully implemented in vanadium-doped blue
bronze1 , a charge density wave (CDW) material. For this
material, accurate X-ray scattering experiments revealed
two distinct incommensurate diffraction peaks. These
peaks where respectively identified with the CDW wavevector, and with Friedel oscillations at twice the Fermi
wavevector. In other materials, the experimental observation of Friedel oscillations is made difficult by the presence of stronger Bragg peaks induced by the underlying
lattice, which can conceal weak Friedel oscillations.
A possible path to overcome this difficulty and enhance
the susceptibility of the conducting electrons is offered by
resonant elastic X-ray scattering (REXS). In this experiment, X-ray photons resonantly excite electrons from a
core level to the conduction band. After a short time (of
the order of pico-seconds), the core level is refilled by a
conduction-band electron with the emission of a photon.
The momentum difference between the incoming and outgoing electron gives precise information about scattering
events occurring in the conduction band. As pointed out
by Abbamonte et al.2 , this process is analogous to the
tunneling of an electron from the tip of a scanning tunneling microscope (STM) to the conduction band. Both
REXS and STM involve the transition of an electron from
a localized state to the conduction band and probe modulations of the local density of states. STM was indeed
employed to directly probe Friedel oscillations around individual Zn and Ni impurities3,4 . The precise relation
between hard X-ray, REXS, and STM experiments is the
subject of the present work.
For concreteness, in this paper we employ phenomenological parameters relevant to superconducting cuprates,
and compare our findings with static modulations recently observed in experiments5–15 . In Sec. II we review
the framework necessary to relate charged quasiparticles
to X-ray, STM, and REXS experiments. In Sec. III we
compare these theoretical predictions with the experi-
2
mental findings. As summarized in Sec. IV, our results
provide a new insight into the nature of fermionic quasiparticles in cuprates, as well as the dominant source of
disorder. Specifically, we outline a procedure to distinguish between different types of impurities. Actual experiments suggest that local modulations of the pairing
gap are enhanced in the pseudogap phase. The resulting
short-ranged Friedel oscillations can be interpreted as a
precursor of the long-range-ordered phase, observed at
large magnetic fields16–18 .
II.
0
where [·, ·] is the commutation relation and ρ(x, t) =
ψ † (x, t)ψ(x, t) is the charge density. For quasiparticles
with a dispersion relation εk and a finite lifetime Γ,
Eq. (1) becomes
X
k
nk − nk+q
.
εk − εk+q + 2iΓ
(2)
where nk = [1+exp((εk −µ)/T )]−1 is the Fermi-Dirac distribution function, and T the temperature. By neglecting
interactions between quasiparticles, Eq. (2) disregards
possible collective modes such as spin waves and paramagnons. In the case of free electrons (with εk = k 2 /2m
and Γ → 0+ ) and at T = 0, Eq. (2) can be evaluated
analytically (see Ref. [20] for a review). In two dimensions χ(q) is momentum-independent for q < 2kF , and
decays algebraically for q > 2kF , where kF is the Fermi
momentum80 . In actual materials, the dispersion relation is more complex and an exact analytical evaluation
of Eq. (2) is generically not possible. We therefore resort
to a numerical evaluation of this expression. As we will
see in Sec. III A, this calculation leads to sharp peaks in
χ(q), which are commonly interpreted as signatures of
static charge density waves (CDW).
B.
Incoming
photon
Conduction
band (d)
Core level (c)
Conduction
band (d)
Core level (c)
(b)
k+q
Outgoing
photon
FIG. 1: Schematic diagram of a typical REXS experiment, in
which X-ray photons scatter electrons from a core level to the
conductions band (a), and viceversa (b).
Lindhard formula
We open our theoretical description of Friedel oscillations by considering (hard) X-ray scattering experiments.
In these experiments the intensity of the scattered light
is proportional to the zero-frequency density-density response function19
Z ∞
Z
h
i
χ(q) =
dt
dx eiq·x h ρ(x, t), ρ(x, 0) i, (1)
χ(q) =
k
conductivity dI(r)/dV is proportional to the local density of states g(r, ω = V ), given by the imaginary part of
the retarded Green’s function:
THEORETICAL FRAMEWORK
A.
(a)
Local density of states
In contrast to hard X-ray measurements, STM and
REXS temporarily change the number of electrons in
the conduction band and couple to the density of states,
rather than to the density-density response function.
Specifically, at zero temperature the STM differential
g(r, ω) = Im[G(r, ω)]
(3)
Z ∞
0
dt e−iω(t−t ) h[ψ † (r, t), ψ(r, t0 )]i .
= Im
0
(4)
For disordered materials, g(r, ω) varies in space and is in
general unpredictable. It is therefore common to compute the two-dimensional Fourier transform of the signal
(at fixed voltage)
Z
g(q, ω) = dd r eiq·r g(r, ω) .
(5)
As we will explain in detail below, the absolute value
of g(q, ω) depends on the types of scatterers present in
the material, but not on their position (assuming that
the sample is large enough to enable self-averaging of
the scatterers’ position). Another common method to
analyze STM experiments is the function
Z
g(r, ω)
Z(q, ω) = dd r eiq·r
.
(6)
g(r, −ω)
This function cures the “STM normalization problem”
which affects g-maps (see for example SI-10 of Ref. 21),
although its theoretical calculations is more challenging.
If we assume g(r, ω) to be given by the sum of a homogeneous contribution g0 (ω) and a weak spatially-dependent
part δg(r, ω), we can approximate the Z function as
Z
h g(r, ω) g(r, −ω) i
Z(q, ω) ≈ dd r eiq·r
−
. (7)
g0 (ω)
g0 (−ω)
g(q, ω) g(q, −ω)
=
−
.
(8)
g0 (ω)
g0 (−ω)
Resonant elastic X-ray scattering (REXS) offers an
alternative way to measure the local density of states,
g(q, ω). As mentioned in the introduction, STM and
REXS describe analogous processes: in STM electrons
tunnel to the sample’s conduction band from an atomicsize tip, while in REXS they are coherently pumped from
a local core level (see Figure 1). Based on this analogy,
Abbamonte et al.2 modeled the intensity of the REXS
3
dressed Green’s function
signal (at zero temperature) by
Z
IREXS (q, ω) = A
∞
dω
0
0
GR
c (ω
𝐺 𝑘, 𝑘 + 𝑞, 𝜔
2
− ω )g(q, ω)
0
𝑔 𝑞, 𝜔
(9)
𝑘
bare Green’s function
(in the absence of disorder)
−1
GR
c (ω)
Here
= [(ω + iΓc )] is the retarded Green’s function of the core level and Γd ≈ 300meV is its lifetime. In
Appendix A 2 we provide a rigorous derivation of Eq.(9)
based on the Keldysh Green’s function formalism, which
allows to extend this expression to finite temperatures.
Notably, Eq. (9) neglects the effects of the core-hole potential on the evolution of the conduction band. This effect is fundamental to understand resonant inelastic scattering (RIXS)22 processes, but can probably be neglected
in the case of REXS.
The prefactor A in Eq. (9) describes the transition amplitude for the excitation of a single core hole. As shown
in Appendix A 3, this quantity does not depend on the
details of the core orbital and, in particular, is unaffected
by the spin-orbit coupling. In the absence of magnetic
impurities, the dipole approximation results into
A ∝ hd| (η̂i · r) (η̂o∗ · r) |di; ,
We now consider the effects of non trivial Wannier functions on the Fourier-transformed local density of states g(q, ω). To achieve this goal, we first
express Eq. (5) in terms of the
R Fourier-transformed
fermionic operators ψk (t) =
dd r eik·r ψ(r, t) and
their
retarded Greens function G(k, k + q, t) =
R∞
dt
e−iωt h[ψ(k, t), ψ † (k + q, t)]i as
0
X
G(k, k + q, ω) ,
𝐺(𝑘, 𝜔) 𝛿𝑞
≈
outgoing
quasiparticle
𝐺(𝑘 + 𝑞, 𝜔)
X
𝑇𝑘 + 𝑇𝑘+𝑞
𝐺(𝑘, 𝜔)
+
𝑘
local impurity
FIG. 2: Schematic representation of the diagrams considered
in the present analysis. Our approach is based on the firstorder perturbation theory in the strength of the disorder and
does not include the effects of interactions among quasiparticles. The function g(q, ω) is the Fourier-transformed local
density of states.
single site, ci , through the Wannier function W (r − ri ),
X
ψ(r, t) =
W (r − ri )ci .
(12)
i
Wannier functions and Bragg peaks
g(q, ω) = Im
incoming
quasiparticle
(10)
where |di denotes the orbital wavefunction of the electrons forming the conduction band, r is displacement vector in this state, and η̂i/o is the polarization of the incoming/outgoing photon. Eq. (10) is used in Appendix A 4 to
compare the theoretical predictions of the present singleband model with the experimental results of Ref. 23.
C.
=
(11)
k
To derive Eq. (11) we assumed the system to be symmetric under r → −r.R This symmetry allowed us to
invert the order of the dd k and Im operators (see SI-2
of Ref. 21). This assumption is valid for example in the
presence of a single scatterer at the origin of the axis. In
the presence of several scatterers at random locations the
present analysis applies to the absolute value of the measured quantity (see App. A 1). We will further comment
about the distinction between the phase and the amplitude of the experimentally measured g(q, ω) in Sec. III D,
when describing the STM experiments of Ref. 24.
For a single-band model, the operator ψ(r, t) is related
to the annihilation of an electron (quasiparticle) on a
Combining Eq.s (11) and (12) we arrive to the
expression21,25–27
X
g(q, ω) = Im
Wq∗ Glattice (k, k + q, ω) Wk+q , (13)
k
R
where
W (k) = dd reik·r W (r), Glattice (k, k + q, t) =
R∞
P
dt e−iωt h[c(k, t), c† (k + qt)]i, and ck = i eik·xi ci .
0
Note that by definition, Glattice (k, k + q, ω) is a periodic
function of k and q with a period given by the Bravais
lattice vectors, G.
In thePabsence of impurities Glattice (k, k + q, ω) =
G(k, ω) G δ(q − G), where G(k, ω) = G(k, k, ω). This
expression gives rise to well-defined Bragg peaks in
g(q, ω). The approximate shape of Wk can then be inferred by comparing the intensity of the central Bragg
peak with those of higher-order peaks (usually only the
first order peaks are visible in the experiment due to the
discrete steps at which STM spectra are measured). In
this paper we approximate W (k) as a Gaussian wavefunction with width σk = 1.8(2π/a), where a is the lattice constant.
D.
Scattering from local impurities
In actual materials, as a consequence of disorder, the
Fourier-transformed density of states g(q, ω) is non zero
even for wave-vectors that do not correspond to a lattice
vector. Performing a first-order perturbation theory in
the scattering potential (Born approximation) one finds28
X
G(k, k + q, ω) = G(k, ω)
δ(q − G)
G
+ G(k, ω)T (k, q)G(k + q, ω) ,
(14)
4
where G(k, ω) = G(k, k, ω), and T (k, q) describes the
scattering of quasiparticles from momentum k to momentum k + q.
One of the main goals of this paper is to consider the
effects of different types of impurities, defined through
their scattering matrices Tk . We consider here only perturbations that are static and quadratic in the quasiparticles’ creation and anihilation operators. Any such
perturbation can be described by the Hamiltonian
X
Hpert = Vi,j c†i cj =
T (k, q)c†k ck+q
(15)
k,q
iq·xj ik·(xi −xj )
where T (k, q)
=
e
+
i,j Vi,j (e
iq·xi ik·(xj −xi )
e
e
).
If the scatterer is centered around xi = 0, the scattering amplitude is given by the sum of two terms, which
depend respectively on the momentum of the incoming
and outgoing quasiparticles only: T (k, q) = Tk + Tk+q .
Combining this expression with Eq. (11) we find
X
X
g(q, ω) =
Im [Wk∗ G(k, ω)Wk+q ]
δ(q − G)
P
k
+
X
G
h
i
Im Wk∗ G(k, ω)(Tk + Tk+q )G(k + q, ω)Wk+q .
k
(16)
Eq. (16) is at the basis of the present analysis: the first
line corresponds to the density of states in an ideal lattice,
while the second line describes the effects of the impurities (see also Fig. 2). In what follows we will mainly
consider this latter contribution.
E.
Impurities in a paired state
The above-mentioned formalism can be easily extended
to include the effects of a spectral gap. For concreteness, we describe underdoped cuprates in terms of single
spectral gap, the paring gap ∆. Following Ref. 21, we
propose that the second energy scale observed in many
experiments corresponds to the quasiparticle’s lifetime Γ,
rather than to a distinct (competing) gap. This would explain the uncertainty in determining the precise value of
the gap in underdoped samples, varying roughly between
∆ − Γ and ∆ + Γ, depending on the type of experiment29 .
The relatively-large value of Γ30 in underdoped cuparates
might be related to enhanced phase fluctuations, which
lead to a loss of global phase coherence at the critical temperature Tc (see for example Ref.s 31–39). The density of
states is a gauge-invariant object and, as such, depends
only on the amplitude of ∆ but not on its phase. For
simplicity, we assume a pairing gap of pure d-wave form,
∆k = ∆0 /2(cos kx − cos ky ). The temperature dependence of ∆0 and Γ is further studied in Sec. III E.
In the presence of a pairing gap, quasiparticles are conveniently represented as 2 × 2 matrices in Nambu space
(whose two entries are respectively particles and holes).
In this notation the retarded Green’s function of a quasiparticle with momentum k and energy ω is given by
ω − εk + µ + iΓ
∆k
−1
G (k, ω) =
.
∆−k
ω + ε−k − µ + iΓ
(17)
The dispersion relation εk , the pairing gap ∆0 , and
the quasiparticles lifetime Γ relevant to superconducting
cuprates are provided in Sec. III and in Table I.
Static and quadratic perturbations can be divided
into two main categories, CDW and PW, depending on
whether they conserve the total number of quasiparticles
(∼ c†i cj , see Eq. (15))), or not (∼ ci cj + c†i c†j ). In this
paper we restrict our analysis to three specific types of
impurities: two of them have a simple physical interpretation and correspond to local modulations of the chemical potential (sCDW) and of the pairing gap (dPW). The
third type (dCDW) corresponds to a local modulation of
the intra-unit-cell nematic order40 and has d-wave symmetry. These three types of impurities correspond to the
real-space Hamiltonians H sCDW = c†0,0 c0,0 , H dCDW =
(c†0,0 c1,0 +c†0,0 c−1,0 )−(c†0,0 c0,1 +c†0,0 c0,−1 )+H.c., H dP W =
(c†0,0 c†1,0 + c†0,0 c†−1,0 ) − (c†0,0 c†0,1 + c†0,0 c†0,−1 ) + H.c.. The
associated scattering matrices to be used in Eq. (16) are
1 0
sCDW
Tk
=
,
0 −1
dk 0
dCDW
Tk
=
,
0 −dk
0 dk
TkdP W =
(18)
dk 0
where dk = cos kx − cos ky . Note that sCDW and dCDW
are diagonal in Nambu space, while dPW is off-diagonal.
The Fourier transformed density of states g(q, ω) is obtained by numerically integrating Eq. (16), with G(k, ω)
and Tk defined by (17) and (18) respectively. As we will
see, a comparison between the resulting plots and the experimental findings suggests a coexistence of sCDW and
dPW, but rules out the presence of dCDW local modulations.
III.
A.
EXPERIMENTS
X-ray: effects of the band structure
As mentioned in Sec. II A, X-ray experiments couple
to the density-density response function and can directly
measure Friedel oscillations. To employ the Lindhard
formula (2) it is necessary to know to dispersion relation
εk , the chemical potential µ, and the quasiparticles’ lifetime Γ. In the case of superconducting cuprates, these
parameters can be directly read from the high-precision
angle-resolved photoemission spectroscopy (ARPES) experiments. Following the common approach, we assume
electrons to move within isolated CuO planes and map
5
the conduction band in terms of the two-dimensional dispersion relation
t0
(cos kx + cos ky ) + t1 cos kx cos ky
2
t3
t2
+ (cos kx + cos ky ) + (cos 2kx cos ky + cos kx cos 2ky )
2
2
t5
+ t4 cos 2kx cos 2ky + (cos 2kx cos kx + cos 2ky cos ky ) .
2
(19)
εk =
In this work we specifically refer to three distinct compounds:
Bi2 Sr2x Lax CuO6+δ (La2201),
Bi2 Sr2 CaCu2 O8+δ (Bi2212), YBa2 Cu3 O7−x (Y123),
whose band structure were experimentally determined
by King et al.41 , Norman et al.42 , and Schabel et
al.43 , respectively. The relevant parameters t0 − t6 are
reproduced in Table I. Note that Y123 material has
inequivalent bonding (B) and antibonding (A) bands:
Table I refers to the former only.
The chemical potential µ is uniquely determined by the
charge doping through the Luttinger count. Following
the common convention, we denote by p the density of
additional holes with respect to half filling:
P
k nk
p = 2x − 1 , where x = P
,
(20)
k
and k runs over the Brillouin zone. In the case of Y123,
we identify the nominal doping with the algebraic average
of the doping in the bonding and antiboding bands, p =
(pA + pB )/2. The resulting Fermi surfaces are plotted in
Fig. 3(a). As shown in the inset, the Fermi surfaces are
not circular, and display a significant amount of nesting
at the antinodes81 .
Using the phenomenological parameters listed in Table I, we can directly evaluate the Lindhard susceptibility (2). Fig. 3(b) presents χ(q) along the direction
(q, 0), for the three different materials. In all three cases,
we observe a pronounced peak at a wave-vector ranging between 0.2 and 0.3. The exact position of the peak
depends on the choice of the chemical potential, and is
roughly equal to the distance between two adjacent antinodes. The width of the peak is of order 0.03 − 0.1, leading to a correlation length of about 10 − 30 unit cells, or
40 − 120A. Its value is mainly determined by the amount
of nesting at the antinodes82 : Among the three materials
considered here, the sharpest peak is predicted in Y123,
where the amount of nesting is maximal. In contrast, the
Fermi surfaces of Bi2212 and La2201 involve a lower level
of nesting, resulting in broader peaks. This could explain
why, so far, X-ray experiments have revealed Friedel oscillations in Y123 only.
The specific choice of the band structure determines
the details of the predicted signal. In the case of Y123,
Fig. 3(b) compares the signal resulting from the band
structure of Pasani and Atkinson44 (continuous blue
curve) and of Shabel et al.43 (dashed blue curve). As
shown in the inset of Fig. 3(a), the latter band structure
Band Structure
µ
t0
t1
t2
t3
t4
t5
Doping, p
Lifetime, Γ
Gap, ∆0
Bi2212
[42]
0.0234
-0.5951
0.1636
-0.0519
-0.1117
0.0510
0
0.04
La2201
Y123(B)
[41]
[44]
[43]
-0.148
-0.03 -0.1256
-0.5280 - 0.42 -1.1259
0.2438 0.1163 0.5540
-0.0429 - 0.0983 -0.1774
-0.0281 -0.353 -0.0701
-0.0140
0 0.1286
0
0
-0.1
0.11
pB =-0.04
(pA +pB )/2=0.12
0.004 0.020
0.002
0.001
0.040 0.080
0.030
0.030
TABLE I: Phenomenological band structures used in this paper. The parameters t0 − t5 correspond to hopping terms in
a tight-binding model and are defined in Eq. 19. With respect to the originally published band structures, the chemical potential has been shifted to achieve the required doping
p (through the Luttinger count). Additionally, the parameter t5 has been added to the band structure of Y123(B) in
order to cure a spurious back-banding of the band structure
(See SI-7 of Ref. [21]). Γ is the quasiparticle lifetime, used in
Eq.s (2) and (17), and ∆0 the zero-temperature pairing gap
used in (17). Their value can be read from the voltage dependence of the Fourier-transformed STM signal12,21,23,24,30 ,
or from ARPES experiments45,46 .
predicts a larger amount of nesting, in agreement with
the experiment by Okawa et al.47 . When used to predict
the intensity of the REXS signal, the band structure by
Ref. [44] predicts a peak at wavevector q = 0.28 with
width δq ≈ 0.05, while the band structure by Ref. [43]
predicts a peak at q = 0.3 with δq ≈ 0.03. For comparison, the experiment of Chang et al.11 shows a peak with
maximal intensity at q = 0.31 and width δq ≈ 0.03.
For completeness we mention that the sharp peak observed in X-ray scattering experiments could be additionally enhanced by effected that are not included in
the present analysis. In particular, the Linhard formula
(2) disregards the effects of the electron-phonon coupling.
This coupling was instead found to be relatively strong in
Y123 at this wavevector, leading to a significant phonon
softening48 . Electron-phonon coupling will generically
lead to a sharpening of the X-ray response function, as
well as to a renormalization of the position of maximal
intensity. It seems plausible that the combination of the
band structure of Ref. [44] with of electron-phonon effects could deliver a quantitative agreement with the experiments as good as the one obtained from the band
structure of Ref. [43]. As a side remark, we also note
that the two predicted Lindhard response for Y123 differ
by an overall multiplicative factor (see continuous and
dashed blue curves in Fig. 3(b)). This difference can be
traced back to the different bandwidth predicted by the
two models (∼ 0.3eV in Ref. [44] and ∼ 1eV in Ref. [43)].
Current experiments involve an unknown normalization
6
(a)
1
1
0.5
ky/π
0.5
0
0.2
0
0.3
0.4
−0.5
−1
−1
−0.5
0
k /π
0.5
x
(b)
1
0.8
0.5
0.6
0.4
0.4
0.3
0.2
0.2
0
0
0.1
0.2
0.3
qx/(2π)
FIG. 4: Lindhard response, Eq.(2), as a function of q =
(2π/a) × (qx , qy ) for Bi2212. The response of the other materials is qualitatively similar (although the peak position is
shifted away from q = 0.25).
0.8
0.6
1
χ(q)
1
0.4
0.2
0.2
0.3
0.3
0.5
Bi2212
La2201
Y123(B) [44]
Y123(B) [43]
FIG. 3: (a) Fermi surfaces resulting from the band structures
listed in Table I. (b) Lindhard response, Eq.(2), along the line
q = (2π/a) × (q, 0) for the same materials.
factor and are therefore not sufficient to measure the actual value of χ(q) and distinguish between these two scenarios. Different experiments, and in particular resonant
inelastic scattering (RIXS), might be able to fill in this
information (see for example Ref. [49]).
We now consider the full Lindhard susceptibility as a
function of the two dimensional wavevector q = (2π/a)×
(qx , qy ). Fig. 4 represents the results for Bi2212 (whose
band structure is known to the highest degree of precision) and displays three inequivalent local maxima. The
global maximum occurs around the wavector qπ,π =
(2π/a) × (±0.5, ±0.5). This peak occurs in the other two
materials as well (not shown) and its position is found to
be independent on the doping level. Interestingly, qπ,π
corresponds to the wavevector of the anti-ferromagnetic
order observed in the parent compound. Because the
Lindhard formula describes both spin and charge sus-
ceptibility, the predicted scattering enhancement around
qπ,π is a precursor of the long-ranged spin order achieved
in the absence of doping (Mott insulator).
A second broad peak appears at q = (2π/a) ×
(±0.25, ±0.25). The exact position of this peak is
material-dependent and ranges between |qx | = |qy | = 0.2
and |qx | = |qy | = 0.3 depending on the details of the
band structure and the doping level, in analogy to the
qy = 0 cut shown in Fig. 3(b). Notably, this peak might
easily escape experimental probes: due to its broadness,
it seems to merge with the stronger peak at qπ,π , especially if only the cut along the line qx = qy is available.
We will come back to this point in Sec. III C. Finally, the
third local maximum occurs at q = (±0.25, ±0.1): the
wavevector q = (2π/a) × (±0.25, 0) is predicted to be
a saddle point sitting between these local maxima. We
note that a similar behavior was observed in a recent experiments by Thampy et al.50 , who found sharp peaks at
q = (2π/a) × (0.25, ±0.015), separated by a saddle point
at q = (2π/a) × (0.25, 0)83 .
B.
STM: dispersive vs non-dispersive peaks
We now move to STM experiments, by first offering a
brief summary of the main results of Ref. [21]. Specifically, in this paper we related the emergence of nondispersive peaks in underdoped cuprates to their relatively large inverse quasiparticle lifetime Γ. In materials where Γ is small (such as overdoped cuprates), the
STM probe excites quasiparticles with an energy that
precisely corresponds to the tip-sample voltage. In this
case, energy and momentum conservation leads to the
well-known “octet model”51 . This model predicts the
emergence of seven inequivalent dispersive peaks, which
can be found by connecting points on the Fermi surface
where the pairing gap is equivalent to the tip-sample voltage. As shown by Nowadnick et al.28 , these peaks are
7
V [meV]
(a) Bi2212 (Γ=1meV)
(b) Bi2212 (Γ=20meV)
40
40
30
30
20
20
10
10
0
0
0.25
qx
0.5
0
max
0
0.25
qx
0.5
min
FIG. 5: STM spectra g(q, ω) along the line q = (qx , 0)×2π for
Bi2212 (see Table I) and T = T dP W . A long quasiparticle’s
lifetime ((a) Γ = 1meV) leads to dispersive peaks, while a
short lifetime ((b) Γ = 20meV) leads to non dispersive peaks.
indeed reproduced by Eq. (16) in the limit of Γ → 0.
In contrast, for a finite Γ, the argument leading to the
octet model does not apply because the quasiparticles’
energy is not conserved. In this case, a numerical evaluation of Eq. (16) is necessary. As shown in Ref. [21]
these calculations lead to non-dispersive peaks around
the wavevectors connecting the antinodes. These scattering wavevectors are enhanced at all voltages for two
reasons: (i) Any scattering is enhanced at the antinodes
due to the Fermi surface nesting (in analogy to the analysis of Sec. III A); (ii) The modulations of the pairing gap,
TkdP W in Eq. (18), are proportional to the pairing gap
∆k and are therefore enhanced at the antinodes, where
the latter is maximal. The effect of Γ on the calculated
STM maps is highlighted in Fig. 5, where Γ varies from
1meV (subplot (a)) to 20meV (subplot (b)), while all
other parameters are kept fixed. The former plot displays
dispersive peaks, while the latter mainly non-dispersive
ones. Importantly, the temperature dependence of Γ can
explain the transition between dispersive peaks (at low
temperatures) and non-dispersive peaks (at higher temperatures) reported in Ref. [52].
C.
STM and REXS: identifying the impurities
To further clarify the nature of the main source of
disorder (sCDW,dCDW, or dPW), we now consider the
STM and REXS experiments of Comin et al.12 . Their
two-dimensional Fourier-transformed STM signal is reproduced in Fig. 6(e). The intensity of the signal is maximal in a χ-shaped region, oriented in the (±q, 0) and
(0, ±q) directions. Fig.s 6(a-c) represent our theoretical
calculations for the three types of impurities defined in
Eq. (18). The correct shape of the signal is reproduced
only by local modulations of the pairing gap (dPW – subplot (b)), suggesting that this is the dominant sources of
disorder. The experimental REXS measurement of the
same material is reproduced in Fig. 7(e). It shows a pronounced peak in the (q, 0) direction, and a monotonous
behavior in the (q, q) direction. A comparison with the
theoretical curves, Fig. 7(a-c), reveals that this effect is
reproduced only by local modulations of the chemical potential (sCDW – subplot (a)).
This analysis leads to an apparent inconsistency: STM
reveals local modulations of the d-wave pairing gap
(dPW), while REXS reveals local modulations of the
chemical potential (sCDW). The solution of this apparent paradox is hidden in the intrinsic properties of the
two probes: STM measurements refer to low voltages and
probe the scattering of quasiparticles with small energy
E ≈ V < ∆0 . In contrast, REXS probes the scattering
of quasiparticles with energy E ∼ Γd ∆0 . Due to
the coherence factors appearing in Eq. (16), quasiparticles at different energies are mainly affected by different
sources of disorder: low-energy quasiparticles are mainly
affected by modulations of the pairing gap, while highenergy quasiparticles are mainly affected by modulations
of the chemical potential (see also SI-3 of Ref. [21]). This
effect becomes evident in the present calculation: the
intensity of the STM signal is significantly stronger for
dPW (Fig. 6(c)) than for sCDW (Fig. 6(a)), while the intensity of REXS is stronger for sCDW (Fig. 1(a)) than for
dPW (Fig. 1(c)). The experimental results are then best
reproduced by a coherent superposition of both types of
modulations (Fig.s 6(d) and 7(d)). This result is in line
with Jeljkovic et al.53 , who found a strong correlation
between local perturbations of the pairing gap and of
the chemical potential (identified there as atypical oxygen vacancies). Notably, local modulations of the intra
unit-cell nematic order (dCDW) are inconsistent with the
q dependence of both STM and REXS signals. We will
come back to this last point in more detail in Sec. III D.
Let us now discuss a theoretical prediction made in
Ref. [21], which appears to be in contradiction by the
experiment of Cominet al.23 . Specifically, Ref. [21] predicted the existence of a peak in the REXS signal at
wavevector (0.25, 0.25). In contrast, the experimental
measurements of Ref. [23] (orange curve in Fig. 1(e)) does
not show any significant peak at (0.25, 0.25). We believe
that the absence of the peak at (0.25, 0.25) is due to its
blending with the larger and broader peak at (0.5, 0.5)
along the same direction.84 This phenomenon is clearly
demonstrated in Fig. 8, showing the predicted REXS intensity as a function of the two-dimensional wavevector
q. Note the close analogy with the results of the Lindhard susceptibility shown in Fig. 4. We hope that future
experiments will be able to confirm the hereby prediction
of an increased scattering in the (q,q) direction.
D.
STM: Masking procedure
In an independent experiment, Fujita et al.24 presented
a new method to analyze the symmetry of static modulations observed in STM experiments. Their analysis consists of isolating the local spectra associated with oxygen
Ox and Oy sites, and performing their Fourier transform
independently. The resulting Z spectra for Ox + Oy and
Ox − Oy are reproduced in Fig. 9(e). The latter quan-
8
(a) sCDW
(b) dCDW
g(q,25meV)
0
0.17
(c) dPW
g(q,25meV)
0.33
0
0.13
(d) sCDW+dPW
g(q,25meV)
0.25
0
0.31
g(q,25meV)
0.63
0
0.5
1
FIG. 6: (a-d) Theoretical calculations of Fourier-transformed STM spectra for different types of impurities: local modulation
of the chemical potential (sCDW), local modulations of the intra-unit-cell nematic order (dCDW), local modulations of the
pairing gap (dPW). Model details: see column La2201 of Table I. (e) Experimental measurement reproduced from Ref. [23].
(a) sCDW
I
REXS
(a.u.)
20
(b) dCDW
(c) dPW
15
10
Q=(0,q)
Q=(q,q)
3
10
5
1
0.2
q
0.4
0
0
0.2
q
0.4
0
0
15
10
2
5
5
0
0
20
4
15
(d) sCDW+dPW
5
0.2
q
0.4
0
0
0.2
q
0.4
FIG. 7: (a-d) Theoretical calculations of Fourier-transformed REXS spectra for different types of impurities (see caption of
Fig. 6). In order to easy the comparison with the experimental plot, the theoretical predictions in the (q, q) direction (orange
diamonds) have been reduced by a factor of 0.25 with respect to the (q, 0) direction (green squares). Model details: see column
La2201 of Table I. (e) Experimental measurement reproduced from Ref. [23].
FIG. 8: Two dimensional plot of the predicted REXS signal. Pronounced peaks are observed at q = (2π/a) × (0.25, 0)
and q = (2π/a) × (0.5, 0.5). Model details: Bi2201 with
sCDW+dPW impurities.
tity displays pronounced peaks in correspondence to the
presumed CDW wavevector q = (2π/a) × (0.25, 0). This
finding is presented as evidence of CDW with d-wave
form factor, or equivalently, modulations of the intraunit-cell nematic order24 .
Here we show that the experimental findings are fully
consistent with the present model of Friedel oscillations.
To theoretically model the experiment of Fujita et al.24
one needs to follow the same procedure as the one employed in the experiment, which consists of multiplying the experimental measurements by an appropriate
mask in real space and subsequently applying the Fourier
transform. Specifically, we introduce the mask functions
Mx (r) and My (r), associated respectively with the Ox
and Oy orbitals. The Fourier-transformed masked signal
is then given by
Z
Oα (q, ω) = dd r eiq·r Mα (r)g(r, ω) ,
(21)
where α = x, y.
Let us now temporarily neglect the finite resolution of
the STM probe and assume that the Ox and Oy orbitals
can be modeled as delta functions shifted by half a lattice constant. The resulting mask functions are shifted
9
(a) sCDW
(b) dCDW
(c) dPW
(d)=(a)+(c)
(e) Exp.
FIG. 9: Orbital-selective STM: (a-d) Theoretical predictions for the Z maps obtained for different types of local impurities T .
The maps were obtained by numerically integrating Eq. (6) over ωε(0, 2∆0 ) with the simplifying assumption g0 (ω) = const.
Model parameters: see column Bi2212 of Table I. Red and blue colors refer to positive and negative amplitudes respectively.
(d) Experimental data reproduced from Ref. [24].
P
combs: Mx (x, y) = n,m δ(x − (n + 1/2)a)δ(y − mb) and
P
My (x, y) = n,m δ(x − na)δ(y − (m + 1/2)a). Applying
the convolution theorem, we obtain
X
Ox (q, ω) = cos (πqx )
g(q + G, ω)
(22)
G
Oy (q, ω) = cos (πqy )
X
g(q + G, ω)
(23)
G
where q = (2π/a)×(qx , qy ), and G are the Bravais lattice
vectors. This approximation is sufficient to describe the
main properties of Fig. 9(e). In particular, we observe
that Ox − Oy ∼ cos(qx π/2) − cos(qy π/2) changes sign
along the direction (q, q), giving rise to pronounced peaks
in the directions (q, 0) and (0, q). In contrast Ox − Oy ∼
cos(qx π/2) + cos(qy π/2) is approximately isotropic. This
distinction is simply due to the masking procedure and
is not associated with the symmetry of the underlying
modulations of the local density of states. In addition,
Eq. (23) displays the following symmetry
qx →qx +2π
Ox − Oy −−−−−−−→ Ox + Oy ,
(24)
which was indeed experimentally observed in Ref. [24].
In reality, the mask function Mα (r) is not given
by delta functions, but rather by windows with a finite width. In this case, we can rely on the formalism presented in Sec. II C and describe the masks in
terms of modified Wannier functions. Specifically, we
2
2
choose85 W (k) = e−k /2σk cos(kx π) for Ox and W (k) =
2
2
e−k /2σk cos(ky π) for Oy . To compare the results of our
calculation with the experiment of Ref. [24], it is necessary to recall that (i) these experiments refer to Z maps,
which are defined in Eq. (6) and approximated by Eq. (8);
(ii) the function Z(q, ω) is integrated over the variable ω
from zero to an upper cutoff (which we arbitrarily chose
to be equal to 2∆0 ).
The end product of these calculations is presented in
Fig. 9(a-d), for different types of impurities, where the
color coding refers to the sign of the signal. We find that
both modulations of the chemical potential (Fig. 9(a))
and of the pairing gap (Fig. 9(c)) lead to pronounced
peaks at q ≈ (2π/a) × (±0.25, 0) in the Ox − Oy combination (lower panel), but not in the Ox +Oy combination
(upper panel). As before, an equal superposition of both
contributions provides the best agreement with the experiment. Note that our theoretical calculations refer to
the effects of a single impurity centered at the axis origin. As explain in Appendix A 1, the random position of
the impurities determines an unpredictable phase of the
experimental signal. This explains why the phase of the
central peak in Ox + Oy is predicted to be constant phase
in the theoretical calculations, while it severely fluctuates
in the experiment. Fig. 9(b) shows that local modulation
of an intra-unit-cell nematic order (dCDW) lead to pronounced peaks in the Ox + Oy signal, which were not
observed in the experiment. We deduce that this type of
disorder is generically absent (or very weak) in cuprates.
Our conclusions are in apparent contradiction with Fujita et al.24 , who interpreted this experiment as evidence
for the existence of a d-wave static order. However, a
more careful analysis reveals that the present work and
Ref. [24] follow different conventions in defining the symmetry of the modulation. Indeed, for modulations with
a finite wavevector, the rotation symmetry is explicitly
broken and the distinction between s and d-wave modulations can be drawn in different ways. In the present
paper we consider only local perturbations, modeled by
(15), and classify them according to their global symmetry. In contrast, Ref. [24] considers long-ranged waves
10
(c) qx=0.36
(a)
0
20
Ox+Oy
Ox+Oy
Ox+Oy
Ox−Oy
Ox−Oy
Ox−Oy
40
60
ω [meV]
(d) qx=0.12
80 0
χ( q,ω)
1
40
60
ω [meV]
(e) qx=0.24
80 0
1
0.5
0
0
20
20
40
60
40
60
ω [meV]
(f) qx=0.36
80
1.5
1.5
1
0.5
0
0
50
∆0 [meV]
100
1
0.5
0
0
100
Γ [meV]
200
FIG. 11: Dependence of the REXS signal (a) on the pairing
gap ∆0 for Γ=20meV, and (b) on the quasiparticle lifetime Γ
for ∆0 =80meV. It is found that the REXS signal is weakly
dependent on ∆0 , while it is strongly affected by modifications of Γ. Numerical parameters: band structure of La2201,
impurity model T CDW + T dP W , and q = (2π/a) × (0.25, 0).
0.5
χ+
χ+
χ−
χ−
20
40
60
χ+
χ−
0
80 0
20
40
60
FIG. 10: Voltage dependence of the Ox + Oy and Ox − Oy
signals for fixed values of q = (2π/a) × (qx , 0). Subplots
(a-c) refer to the Z function of Eq. (8) with the simplifying
assumption g0 (ω) =const. Subplots (d-f) refer to the ratio
between the two signals defined in Eq. (25).
and classifies them with respect to the symmetry under
rotation of k only (while keeping q fixed). The absence
of dCDW impurities claimed in this paper is therefore
not necessarily orthogonal to the identification of d-wave
order in Ref. [24]. These two conventions reflect a different approach towards the role of disorder: in our view
the incommensurate modulations are intrinsically shortranged and generated by the disorder. In the approach
by Fujita et al. the modulations are instead intrinsically
long-ranged, and their correlation length is made finite
by the disorder (see also Ref.54 ).
We now present the theoretical predictions for the voltage dependence of the masked signals. Fig. 10 presents
the results of our calculations for three distinct wavevectors. At small wave-vectors (qx = 0.12), we find a singual
behavior around V = 50meV, where Ox − Oy is strongly
peaked and Ox + Oy changes sign. This energy scale
corresponds to a well-known Van Hove singularity and
is not related to the presence of a gap86 . In contrast,
at qx = 0.24 we find Ox − Oy to be peaked at ω ≈ ∆0 .
Finally at qx = 0.36, we find both functions to grow till
ω ≈ ∆0 and develop into a plateau. To highlight the
different energy dependence of Ox + Oy and Ox − Oy , we
introduce the following normalization
χ± (q, ω) =
2
1
0.5
0
80 0
20
(b)
2
I REXS (q)
I REXS (q)
(b) qx=0.24
Z′( q,ω)
(a) qx=0.12
|Ox ± Oy |
|Ox + Oy | + |Ox + Oy |
(25)
Notably, this function can be computed without the actual knowledge of the homogeneous conductance g0 (ω)
appearing in Eq. (8), as long as g0 (ω) = g0 (−ω). This
simplification gives us the hope to obtain a quantitative
80
agreement between the predicted and observed χ maps.
Starting from the curves presented in Fig. 10(a-c), it is
straightforward to compute χ+ and χ− . The results of
these calculations are shown in Fig. (9)(d-f) and demonstrate sharp peaks in correspondence to the local maxima
of Ox + Oy and Ox − Oy . Of particular interest is subplot (e), which corresponds to the wavevector where pronounced peaks were observed in both g maps and REXS
experiments. In this case, we observe that χ+ is peaked
at low voltages, while χ− is peaked at around ∆0 .
E.
REXS: Temperature dependence
In this section we consider the temperature dependence
of X-ray scattering experiments. Ghiringhelli et al.10 and
Chang et al.11 found a sharp cusp at the critical temperature Tc , which was interpreted as direct evidence for
the competition between superconductivity and charge
ordering. In particular, Hayward et al.55 were able to reproduce this cusp using a field-theoretical model in which
superconducting and charge ordering fluctuations are introduced as orthogonal components of an emergent order
parameter.
The present model does not describe a direct competition between charge modulations and superconductivity.
The predicted signal is nevertheless non-trivially dependent on the temperature T for two main reasons: (i) T
directly appears in the Lindhard formula (2) (through
the Fermi-Dirac distribution function) and in the expression for the REXS signal (see Appendix A 2); (ii) the
pairing gap ∆0 and the inverse quasiparticle lifetime Γ
can directly depend on the temperature. Explicit calculations demonstrate that the first effect is negligible. The
corrections due to a finite temperature are suppressed by
1/Ef in the Lindhard formula and 1/Γh in the REXS intensity (see Eq. (A7). Both quantities are of the order of
300meV and much larger than the typical experimental
temperature T ∼ 100K ∼ 10meV. As a consequence, the
11
explicit dependence on the temperature.
Fig. 11(a) shows that the pairing gap ∆0 has very little
effect on the REXS intensity. This observation is in line
with the findings of Sec. III C, where it was shown that
REXS couples mainly to charge fluctuations, rather than
to fluctuations of the pairing gap. Fig. 11(b) shows that
the REXS signal is roughly inversely proportional to the
quasiparticle lifetime. A similar behavior is observed in
the Lindhard response (2 (not shown). We deduce that
the cusp observed in Y12310,11 at the critical temperature Tc could be related to a non-analytic behavior of Γ
at the this point. The temperature dependence of Γ is
a largely-debated subject. Its behavior was for example
assessed by pump-probe experiments56–58 , as well as by
several theoretical approaches, including self-consistent
fluctuation-exchange (FLEX) approximations59,60 , dynamical mean-field theories (DMFT)58,61,62 , and functional RG methods63 . According to our approach, the
sharp maximum in the intensity of the X-ray scattering
in Y123 could imply that Γ displays a sharp minimum at
Tc . However, we also remark that the generality of this
peak is questioned by recent measurements of Hg120115 ,
where no significant cusp was observed: in this material
the REXS signal is approximately constant for T < Tc
and monotonously decrease for T > Tc . In our language,
this measurement could imply that the exact behavior of
Γ at Tc may be material dependent and not universal.
IV.
SUMMARY AND OUTLOOK
In this paper we presented a theoretical modeling of
recent X-ray, REXS, and STM measurements of underdoped cuprates, with specific attention to Ghiringhelli
et al.10 , Chang et al.11 , Comin et al.23 , and Fujita et
al.24 . These authors assume the existence of a competing order, distinct from superconductivity, and associated with the spontaneous breaking of translational symmetry. This scenario was also adopted in interpreting
earlier STM5–8 , nuclear magnetic resonance (NMR)17,67 ,
and quantum oscillations16,68 experiments revealing enhanced static spatial modulation. The pseudogap energy scale would then correspond to the excitation gap
required to restore the translational invariance. The association between the charge ordering and the pseudogap phase is however undermined by recent X-ray experiments revealing the same type of charge ordering in
electron-doped cuprates69 , where a pseudogap phase is
not expected to subsist.
Here we follow a different and complementary approach, and associate these findings to Friedel oscillations around local sources of disorder21 .
Superconducting cuprates include a large density of chemical dopants and impurities, which can generically seed
Friedel oscillations53 . Periodic modulations were indeed
observed in early STM experiments performed around
Zn and Ni atomic impurities3,70 . Employing the Born approximation (first-order perturbation theory in the impu-
rity strength), we consider each scatterer independently
and predice the correlation length of the modulations to
be independent on the concentration of impurities. This
prediction has been now confirmed by two experimental observations: (i) Achkar et al.71 modified the amount
of disorder in Y123 through a thermal quench and observed that the correlation length of the observed oscillations was unchanged. (ii) The analysis of materials
with similar band structure and different amount of intrinsic disorder (such as La220112 , Bi220113 , Bi221214 ,
and Hg120115 ) revealed an approximately constant correlation length. These findings are not consistent with
theories of competing orders, in which the predicted correlation length should be directly related to the amount
of external disorder54 .
By considering the scattering of short-lived quasipartice from local impurities, we can quantitatively reproduce all the experimental findings: Our model correctly
predicts the wavevector and correlation length of the spatial modulations that were observed in X-ray (Fig.s 3(b)
and 4), STM (Fig. 5), and REXS (Fig. 7) experiments.
The wavevector is approximately equal to the distance
between adjacent antinodes, where the Fermi surface is
often quite nested. In addition, our approach reproduces
experimental observations that where interpreted as evidence for the d-wave symmetry of the oscillations (Fig.s 7
and 9).
To reproduce the experimental results, we introduced
different models of local impurities and found that the
most dominant type corresponds to local modulations of
the chemical potential and of the pairing gap. In STM
maps, the former contribution is generically dominant
along the (q,q) direction, while the latter is dominant
along the (0,q) direction (see Fig.6). The interplay between these two sources of disorder connects to the earlier
analysis of STM data in the presence of a magnetic field
performed by Hanaguri et al.64 and He et al.66 . These
authors found that the ratio between the (q,q) and (0,q)
components generically increases with magnetic field87 .
This effect can be understood by noting that in typeII superconductors external magnetic fields generate isolated vortices, in whose core the pairing gap is locally
suppressed.
In this sense, vortices are similar to other types of local impurities, and generate Friedel oscillations around
their center. When their density reaches a critical value,
vortices can depin from local defects and give rise to a
long-range-ordered phase. Following the proposal of Wu
et al.17 , we believe this effect to be responsible for the
formation of a long-range-ordered phase at large magnetic fields that was observed by quantum oscillations16 ,
NMR17 , and sound velocity18 experiments. Indeed, the
measured critical field ∼ 20T corresponds
to an average
p
distance between vortices of d = φ0 /B ∼ 100A, which
is comparable with the correlation length of Friedel oscillations. The CDW phase observed in cuprates would
then be analogous to the field-induced spin density waves
(FISDW) observed for example in Bechgaard salts (see
12
T
T
PG
PG
SC
SC
CDW
H
p
CDW
H
FIG. 12: Proposed phase diagram of BSCCO and YBCO
compounds, following Ref. [67]: the long-range-ordered CDW
phase observed at high magnetic field is distinct from the
pseudogap (PG) phase, characterized by an incoherent pairing gap of preformed pairs. The inset refers to doping levels
p ≈ 0.1, where a direct transition between the superconducting (SC) and WC phases is observed67 .
Ref.[72] for a review). Because the correlation length
of Friedel osciilations depends on the amount of nesting
at the antinodes, it is natural to expect the magnetic
phase to be enhanced around p = 0.1, where the antinodes are maximally nested. The resulting phase diagram
is plotted in Fig. 12, and highlights our claim that the
long-range ordered CDW phase is distinct from the pseudogap (PG) phase observed at zero magnetic field.
We now discuss how to exploit STM maps to further
compare the effects of magnetic fields, temperature, and
doping. Fig. 13 shows that approaching the pseduogap phase (by increasing the temperature, or decreasing
the doping) generically leads to an increase of scattering in the (0, q) direction, which is associated with local
modulations of the pairing gap. This observation is in
agreement with recent muon spin rotation73 (µSR) and
NMR67 experiments, which detected enhanced static inhomogeneities in the pseudogap phase. A similar conclusion was reached in Ref. [74], where the effects of disorder were found to be similar to the effects of temperature
and magnetic fields (see also Ref. [75], where a pseudogap phase was found in disordered thin films). Inhomogeneities of the pairing gap are naturally accompanied by
a reduction of the long-range coherence: the transition
to the pseudogap phase may be due to a loss of coherence
of the pairing gap34 , rather than to its disappearance.
From the prospective of fermionic quasiparticles, the
transition to the pseudogap phase is generically associated with an increase of the inverse lifetime Γ. The role
of this quantity on ARPES measurement is well known
and offers a simple explanation for the “Fermi arcs” observed in underdoped cuprates21,76,77 . STM21,30 and
transport78 measurements show that the inverse quasiparticle lifetime Γ is strongly reduced in underdoped
cuprates, probably going to zero at the transition to
the Mott insulator. This observation suggests a possible relation between Γ and the critical temperature of
cuprates21 . Measuring the temperature dependence of Γ
would allow to distinguish between the effect of disorder
(elastic scattering, which exists down to zero temperature) from the effects of interactions (inelastic scattering, which is supposed to diasppear at zero temperature).
Surprisingly, although the inverse quasiparticle lifetime
Γ is commonly used in fitting virtually any experimental
spectroscopic data, a systematic study of this quantity
as a function of doping, temperature, and magnetic field
has not been performed yet. We hope that the present
work will motivate a new analysis of existing data along
these lines.
We acknowledge useful discussions with Marc-Henri
Julien, Riccardo Comin, Max Metlistki, Nigel Hussey,
Steve Kivelson, Giacomo Ghiringhelli. EGDT is thankful to the Aspen Center for Physics, where part of this
work was performed during the winter-2014 workshop on
“Unconventional Order in Strongly Correlated Electron
Systems”.
Appendix A: Technical details
1.
Average over impurities
In this section we consider the effects of several identical impurities, located at random positions. We find
that the absolute value of g(q, ω) is independent on their
position and therefore is an intrinsic property of the system. In the text, Eq. (11), we assumed g(r, ω) to be
symmetric under r → −r. This approximation is valid
in the presence of a single impurity located at the origin
of the axis88 . To extend this treatment to systems with
several impurities, we first notice that within the present
framework (first order perturbation theory in the impurity strength), g(q 6= G, ω) is given by a sum of terms,
each referring to the scattering
of quasiparticles from a
P
single impurity: g(q, ω) = i gi (q, ω), where i runs over
all the impurities. To compute gi (q, ω) we first consider
a coordinate system whose origin is located at the center
of the ith impurity, where Eq. (16) applies. We then shift
back gi (q, ω) to the common lab frame by the multiplication with eiq·ri , and sum all the terms. In the case of
N identical impurities we obtain
g(q, ω) =
N
X
eiq·ri g0 (q, ω) .
(A1)
i=1
Here g0 (q, ω) is the scattering amplitude from an isolated impurity located at the axis origin, computed from
13
(a) small H
(c) small T
(e) large p
(g) dominant direction
(1,1)
(−1,1)
(b) large H
(d) large T
(f) small p
(h) dominant direction
(0,1)
(1,0)
FIG. 13: Fourier-transformed STM measurements of different materials: (a-b) Ca2x Nax CuO2 Cl2 (Tc =28K) at low and high
magnetic field. Reproduced from Ref.64 ; (c-d) Bi2212 (Tc = 37K) at low and high temperature. Reproduced from Ref. [65];
(e-f) Pb-Bi2201 at large and small hole doping. Reproduced from Ref.66 . (d) Deep in the superconducting phase the main
source of scattering is along the (1,1) and (1,-1) directions. (h) When approaching the pseudogap phase the scattering is mainly
along the (0,1) and (1,0) directions, signaling the presence of phase inhomogeneities.
Eq. (16). In Eq. (A1) the phase of g(q, ω) is determined by the random positions ri and is therefore not
predictable. In contrast, its absolute value averages to
h|g(q, ω)|i = N g0 (q, ω) .
(A2)
Here we used the observation that by definition, g0 (q, ω)
is a real function.
2.
REXS: Green’s function approach
In this section we derive Eq. (9) within the Keldysh
path-integral formalism. This approach allows us to extend the results of Abbamonte et al.2 to finite temperatures. In REXS experiments X-rays are scattered upon
the material to be examined at a frequency that allows
the creation of a core hole, i.e. the excitation of an inner
orbital of the atom to the conduction band (see Fig. 1).
The action of the incoming field can be described as
Vin = Ein eiωt δ(t)d† c + H.c., where Ein > 0 describes the
amplitude of the incoming x-rays, ω its frequency, d and
c are fermionic operators describing electrons (quasiparticles) respectively in the conduction band (in cuprates
formed by d orbitals) and in the core level. Shortly after,
an electron from the conduction band fills in the core level
and emits an X-ray photon, which is observed by the experimental setup. This decay process can be described by
the operator Vout (t) = a†out e−iωt c† d, where a†out creates
an outgoing photon and λ is the light-matter coupling.
In perturbation theory, the outgoing field is given by
Eout (t) = haout i ≈ hc† de−iωt + H.c.i
(A3)
where we assumed the initial state to be empty of outgoing photons. Applying perturbation theory (and neglecting oscillating terms), we obtain the Kubo formula
Eout (t) = iEin Θ(t)h[d† (0)c(0), c† (t)d(t)]i
(A4)
where Θ(t) is the Heaviside theta function and [..., ...]
is the commutation relation. In Keldysh notation, Eq.
(A4) becomes the sum of eight terms with an odd number
ˆ Four
of “classical” fields c, d and of “quantum” fields ĉ, d.
of these terms contain three quantum fields and their
expectation values are identically equal to zero. We are
then left only with terms containing three classical fields
and one quantum field:
Eout (t) =
i
Ein Θ(t)hdˆ∗ (0)c(0)c∗ (t)d(t) + d∗ (0)ĉ(0)c∗ (t)d(t)
2
ˆ
+d∗ (0)c(0)ĉ∗ (t)d(t) + d∗ (0)c(0)c∗ (t)d(t)i
(A5)
14
For t > 0 only the first two terms are non zero (and for
t < 0 the last two are non zero). Eq. (A5) can be further simplified by introducing the retarded and Keldysh
∗ˆ
K
∗
Green’s functions GR
d = hd di and Gd = hd di:
R
K
R
Eout = iEin GK
(A6)
d (t)Gc (t) + Gc (t)Gd (t)
Each of the two terms of Eq. (A6) corresponds to the
product of two Greens functions, evaluated at the same
time, or equivalently their convolution in the frequency
domain:
Z ∞
i
0
R
0
Eout (ω) = Ein
dω 0 GK
d (ω − ω )Gc (ω )
2
−∞
0
R
0
+GK
(A7)
c (ω )Gd (ω − ω ) .
At thermal equilibrium the Keldysh components satisfy the fluctuation-dissipation theorem GK
=
d (ω)
R
K
2Im[GR
=
d ] tanh(ω/2T ) ≈ 2Im[Gd ]sign(ω) and Gc
R
2Im[GR
c ] tanh((ω − Eh )/2T ) ≈ −2Im[Gc ]. In the limit of
T → 0 we find
Z ∞
0
R
0
Eout (ω) = iEin
dω 0 Im[GR
c ](ω − ω )Gd (ω )
−∞
0
R
0
+ sign(ω 0 )Im[GR
d ](ω )Gc (ω − ω ) . (A8)
The real component of Eout (the component that is in
phase with Ein ) has a particularly simple form
Z ∞
0
R
0
Re[Eout ](ω) = 2Ein
dω 0 Im[GR
c ](ω − ω )Im[Gd (ω )]
j = 3/2 are excited to the valence band. Thus the total polarization-dependent intensity is the sum over spinorbit eigenstates mj = −3/2, −1/2, 1/2, 3/2:
X
I(η̂i , η̂o ) ∝
η̂o∗ · h2p3/2
mj |r|3dx2 −y 2 , σi
mj
× η̂i · h3dx2 −y2 , σ|r|2p3/2
i
,
mj
(A12)
To compute the dipole matrix elements, we introduce a
unit operator in the basis of separate spin and orbital
angular-momentum eigenstates |m` , ms i
X
2 −y 2 , σi
η̂o∗ · h2p3/2
|m
,
m
ihm
,
m
|r|3d
I∝
`
s
`
s
x
mj
mj ,m` ,ms ,m0` ,m0s
× η̂i · h3dx2 −y2 , σ|r|m0` , m0s ihm0` , m0s |2p3/2
mj i
(A13)
=
X
η̂o∗
·
h2p3/2
mj |m` , σihm` |r|3dx2 −y 2 i
mj ,m` ,m0`
× η̂i · h3dx2 −y2 |r|m0` ihm0` , σ|2p3/2
i
mj
(A14)
The Clebsch-Gordan matrix elements vanish unless mj =
m` + σ = m0` + σ, and hence we require m0` = m` . We
then have the further simplification:
2
X I∝
η̂o∗ · hm` |r|3dx2 −y2 i
h2p3/2
mj |m` , σi
mj ,m`
× η̂i · h3dx2 −y2 |r|m` i
(A15)
0
(A9)
Applying Karmers-Kronig relation we then obtain
Z ∞
0
R
0
Eout (ω) = 2Ein
dω 0 GR
c (ω − ω )Im[Gd (ω )] (A10)
0
For a featureless core level with response function
−1
GR
, we recover exactly the same
c (q, ω) = [(ω + iΓc )]
expression as in Ref. [2] and Eq. (9) with A = 2Ein .
3.
REXS: Spin-orbit effects
In this section we study the dependence of REXS scattering on the polarization of the incoming (i) and outgoing (o) photons. As an important result, we will show
that in the absence of magnetic impurities, the intensity
of the REXS signal is not affected by spin-orbit effects.
For an isolated atom, the REXS intensity I is given by
the product of dipole matrix elements for the absorption
and the emission:
I(η̂i , η̂o ) ∝ (η̂o∗ · hψi |r|ψn i) (η̂i · hψn |r|ψi i) ,
(A11)
where ψi is the initial (and final) core electron state and
ψn is a valence 3dx2 −y2 orbital with spin σ at the same
site. In typical experiments one selects a resonance so
that only 2p core levels with total angular momentum
If the Hamiltonian is spin-independent, i.e. if there is no
spin-density wave, the amplitude is independent of the
spin σ of the photoelectron in the intermediate state and
thus the two spins contribute equally to the coherent sum
over histories, and we have


2
X X  ∗

η̂o · hm` |r|3dx2 −y2 i
I∝
h2p3/2
mj |m` , σi
m`
mj ,σ
× η̂i · h3dx2 −y2 |r|m` i
(A16)
2
P
3/2
Now mj ,σ h2pmj |m` , σi is the probability that a core
electron with orbital angular momentum m` and unknown spin is in a total spin-j = 3/2 state. By spherical
symmetry this is obviously independent of m` , since m`
is coordinate-dependent but j is not. Since this is an
m` -independent quantity, we obtain
X
I∝
η̂o∗ · hm` |r|3dx2 −y2 i η̂i · h3dx2 −y2 |r|m` i
m`
(A17)
=h3dx2 −y2 | (η̂i · r) (η̂o∗ · r) |3dx2 −y2 i
(A18)
We now have a tensorial matrix element that is not modulated by the spin-orbit effect except for the aforementioned constant prefactor that represents the contribution to resonant scattering only from j = 3/2 core states.
15
𝒌′
z
𝝅′
o
𝝈′
𝜃
𝜃
IREXS ∝
q
x
y x
Our starting point is Eq. (10). Because the outgoing
beam is not filtered according to its polarized, the measured signal is proportional to the sum of the intensities
of the two outgoing polarizations:
sample
𝒌
𝛼
x
𝝈𝝅
3.5
Exp. BSCCO
Exp. YBCO
Th. F =0
Iσ / Iπ (a.u.)
z
Th. Fz=0.1
2.5
2
1.5
1
0.5
0
50
100
α [deg.]
150
200
FIG. 15: Polarization dependence – comparison between theory and experiment. The continuous curves corresponds to
the ratio between Eq. A20 and Eq. A21 (multiplied by a fixed
number), for different values of Fz /Fx = 0 (blue), 0.15 (red),
-0.15 (green). The best agreement between theory and experiment is obtained for Fz = −0.15Fx . The errorbars indicate
the experimental values, reproduced from Ref. 23.
4.
(A19)
o=σ 0 ,π 0
FIG. 14: Experimental setup of Ref. [23].
3
2
X η̂o · M · η̂i REXS: Polarization dependence
In this section we study the dependence of the REXS
signal on the wavevector of the incoming photon k. This
dependence was experimentally measured by Comin et
al.23 , and used to identify the dominant type of charge
modulations. These authors concluded that the best
agreement between theory and experiments is achieved
for charge modulations with a d-wave symmetry. Comin
et al. did not consider the effect of the pairing gap, nor of
any other type of pseudogap, although the experiments
were performed at T ≈ Tc << T ∗ . From their analysis it
is therefore unclear whether the observed signal is due to
charge modulations, or rather modulations of a gap. In
addition, their analysis of is based on a multi-band approach, in which both Oxygen and Copper orbitals are
considered. In what follows we show that a simpler analysis, based on a single-band model, correctly reproduces
the experimental observation (within the experimental
uncertainty).
where the tensor M is defined by Mα,β = hd|rα rβ |di,
i = σ, π is the incoming polarization, and o = σ 0 , π 0 is
the outgoing polarization. We denote by F the diagonal matrix corresponding to M in the principal axis of
the lattice. Its three non-zero entries are Fx = hd|x2 |di,
Fy = hd|y 2 |di, and Fz = hd|z 2 |di. The ratio between
these quantities was measured in Ref. [23] and found to be
Fz /Fx ≈ Fz /Fy ≈ 0.1. The smallness of Fz indicates that
the conduction band has a small extension in the z direction, in agreement with the theoretical calculations79 ,
which predict a dominant x2 − y 2 character.
In the experiment the direction and polarization of the
incoming photons are kept fixed, while the wavevector k
is modified by rotating the sample around the vector q
(Fig. 14): the dipole matrix M is then given by M =
RT (α) F R(α), where the matrix R represents a rotation
of α degrees around the q axis. Without loss of generality
we choose to work in Cartesian coordinates in which the
incoming photon moves in the direction k = (0, 0, 1) with
polarizations σ = (0, 1, 0) and π = (1, 0, 0). As shown in
Fig. 14, the polarization of the outgoing photon are σ 0 =
(0, −1, 0) and π 0 = (cos 2θ, 0, sin 2θ), and q ≡ k0 − k ∼
(cos θ, 0, sin θ). For Fx = Fy = 1, Fz = 0 we then find :
2
α
α
Iσ (α) = 4 cos3 sin cos2 θ sin θ
2
2
2
2
2
+ cos α + sin α sin2 θ
(A20)
2
α
α
Iπ (α) = 4 cos sin3 cos2 θ sin θ
2
2
2
+ cos4 θ − sin2 θ(sin2 α + cos2 α sin2 θ)2 (A21)
For Fz 6= 0 one obtains more complex expressions,
numerically depicted in Fig. 15 for different values of
Fz /Fx = −0.15, 0, 0.15. Following the procedure described in Ref. [23], we first compute Iσ /Iπ and then
normalize this quantity by its minimal value (occurring
at α = 90 deg.). We find that Iσ /Iπ (α = 0) ≈ 1.8.
The experimental measurements of YBCO shows that
Iσ /Iπ (α = 0) ≈ 3 ± 1. The agreement between theory and experiment therefore sits at the boundaries of
the statistical significance. The experimental measurement of BSCCO are in better agreement with the theory. However, due to the different q vector, this material
requires a different value of θ, leading to a lower predictions. Surprisingly, this difference angle is neglected in
the theoretical analysis of Ref. [23], who applied the same
model to both materials. Further experiments with lower
signal-to-noise will enable to identify a possible systematic discrepancy between experiments and theory.
16
1
2
3
4
5
6
7
8
9
10
11
12
13
S. Rouzière, S. Ravy, J.-P. Pouget, and S. Brazovskii.
Friedel oscillations and charge-density wave pinning in
quasi-one-dimensional conductors: An x-ray diffraction
study. Phys. Rev. B, 62:R16231–R16234, Dec 2000.
Peter Abbamonte, Eugene Demler, J.C. Séamus Davis, and
Juan-Carlos Campuzano. Resonant soft x-ray scattering,
stripe order, and the electron spectral function in cuprates.
Physica C: Superconductivity, 481(0):15 – 22, 2012.
S. H. Pan, E. W. Hudson, K. M. Lang, H. Eisaki, S. Uchida,
and J. C. Davis. Imaging the effects of individual zinc
impurity atoms on superconductivity in Bi2 Sr2 CaCu2 O8+ .
Nature (London) , 403:746–750, February 2000.
M. J. Lawler, K. Fujita, J. Lee, A. R. Schmidt, Y. Kohsaka,
C. K. Kim, H. Eisaki, S. Uchida, J. C. Davis, J. P. Sethna,
and E.-A. Kim. Intra-unit-cell electronic nematicity of the
high-Tc copper-oxide pseudogap states. Nature, 466:347–
351, July 2010.
J. E. Hoffman, E. W. Hudson, K. M. Lang, V. Madhavan,
H. Eisaki, S. Uchida, and J. C. Davis. A four unit cell
periodic pattern of quasi-particle states surrounding vortex
cores in Bi2 Sr2 CaCu2 O8+δ . Science, 295(5554):466–469,
2002.
C. Howald, H. Eisaki, N. Kaneko, M. Greven, and A. Kapitulnik. Periodic density-of-states modulations in superconducting Bi2 Sr2 CaCu2 O8+δ . Phys. Rev. B, 67:014533,
Jan 2003.
T. Hanaguri, C. Lupien, Y. Kohsaka, D.-H. Lee,
M. Azuma, M. Takano, H. Takagi, and J. C. Davis. A
‘checkerboard’ electronic crystal state in lightly hole-doped
Ca2−x Nax CuO2 Cl2 . Nature, 430:1001–1005, August 2004.
Michael Vershinin, Shashank Misra, S. Ono, Y. Abe, Yoichi
Ando, and Ali Yazdani. Local ordering in the pseudogap state of the high-tc superconductor Bi2 Sr2 CaCu2 O8+δ .
Science, 303(5666):1995–1998, 2004.
P Abbamonte, A Rusydi, S Smadici, GD Gu,
GA Sawatzky, and DL Feng.
Spatially moduNature Physics,
lated’mottness’ in La2−x Bax CuO4 .
1(3):155–158, 2005.
G. Ghiringhelli, M. Le Tacon, M. Minola, S. BlancoCanosa, C. Mazzoli, N. B. Brookes, G. M. De Luca,
A. Frano, D. G. Hawthorn, F. He, T. Loew, M. Moretti
Sala, D. C. Peets, M. Salluzzo, E. Schierle, R. Sutarto, G. A. Sawatzky, E. Weschke, B. Keimer, and
L. Braicovich. Long-range incommensurate charge fluctuations in (y,nd)ba2cu3o6+x. Science, 337(6096):821–825,
2012.
J. Chang, E. Blackburn, A. T. Holmes, N. B. Christensen,
J. Larsen, J. Mesot, R. Liang, D. A. Bonn, W. N. Hardy,
A. Watenphul, M. V. Zimmermann, E. M. Forgan, and
S. M. Hayden. Direct observation of competition between superconductivity and charge density wave order
in YBa2 Cu3 O6.67 . Nature Physics, 8:871–876, December
2012.
R. Comin, A. Frano, M. M. Yee, Y. Yoshida, H. Eisaki,
E. Schierle, E. Weschke, R. Sutarto, F. He, A. Soumyanarayanan, Yang He, M. Le Tacon, I. S. Elfimov, Jennifer E. Hoffman, G. A. Sawatzky, B. Keimer, and A. Damascelli. Charge order driven by fermi-arc instability in
bi2sr2xlaxcuo6+. Science, 343(6169):390–392, 2014.
Eduardo H. da Silva Neto, Pegor Aynajian, Alex Frano,
Riccardo Comin, Enrico Schierle, Eugen Weschke, An-
14
15
16
17
18
19
20
21
22
23
24
25
26
drs Gyenis, Jinsheng Wen, John Schneeloch, Zhijun Xu,
Shimpei Ono, Genda Gu, Mathieu Le Tacon, and Ali Yazdani. Ubiquitous interplay between charge ordering and
high-temperature superconductivity in cuprates. Science,
343(6169):393–396, 2014.
M. Hashimoto, G. Ghiringhelli, W.-S. Lee, G. Dellea,
A. Amorese, C. Mazzoli, K. Kummer, N. B. Brookes,
B. Moritz, Y. Yoshida, H. Eisaki, Z. Hussain, T. P.
Devereaux, Z.-X. Shen, and L. Braicovich. Direct observation of bulk charge modulations in optimally-doped
Bi1.5 Pb0.6 Sr1.54 CaCu2 O8+δ . ArXiv e-prints, March 2014.
W. Tabis, Y. Li, M. L. Tacon, L. Braicovich, A. Kreyssig,
M. Minola, G. Dellea, E. Weschke, M. J. Veit, M. Ramazanoglu, A. I. Goldman, T. Schmitt, G. Ghiringhelli,
N. Barišić, M. K. Chan, C. J. Dorow, G. Yu, X. Zhao,
B. Keimer, and M. Greven. Charge order and its connection with Fermi-liquid charge transport in a pristine highTc cuprate. Nature Communications, 5:5875, December
2014.
Nicolas Doiron-Leyraud, Cyril Proust, David LeBoeuf,
Julien Levallois, Jean-Baptiste Bonnemaison, Ruixing
Liang, DA Bonn, WN Hardy, and Louis Taillefer. Quantum oscillations and the fermi surface in an underdoped
high-tc superconductor. Nature, 447(7144):565–568, 2007.
T. Wu, H. Mayaffre, S. Krämer, M. Horvatić, C. Berthier,
W. N. Hardy, R. Liang, D. A. Bonn, and M.H. Julien.
Magnetic-field-induced charge-stripe order
in the high-temperature superconductor YBa2 Cu3 Oy .
Nature (London) , 477:191–194, September 2011.
David LeBoeuf, S Krämer, WN Hardy, Ruixing Liang,
DA Bonn, and Cyril Proust. Thermodynamic phase diagram of static charge order in underdoped yba2cu3oy.
Nature Physics, 9(2):79–83, 2013.
N. D. Mermin N. W. Ashcroft. Solid State Physics. Saunders College Publishing, 1976.
B. Mihaila. Lindhard function of a d-dimensional Fermi
gas. ArXiv e-prints, November 2011.
Emanuele G Dalla Torre, Yang He, David Benjamin,
and Eugene Demler. Exploring quasiparticles in high-tc
cuprates through photoemission, tunneling, and x-ray scattering experiments. arXiv preprint arXiv:1312.0616, 2013.
Luuk JP Ament, Michel van Veenendaal, Thomas P Devereaux, John P Hill, and Jeroen van den Brink. Resonant
inelastic x-ray scattering studies of elementary excitations.
Reviews of Modern Physics, 83(2):705, 2011.
R Comin, R Sutarto, F He, EH da Silva Neto, L Chauviere,
A Frano, R Liang, WN Hardy, DA Bonn, Y Yoshida, et al.
Symmetry of charge order in cuprates. Nature materials,
2015.
K. Fujita, M. H. Hamidian, S. D. Edkins, C. K. Kim,
Y. Kohsaka, M. Azuma, M. Takano, H. Takagi, H. Eisaki,
S. Uchida, A. Allais, M. J. Lawler, E.-A. Kim, S. Sachdev,
and J. C. Séamus Davis. Intra-unit-cell Nematic Density
Wave: Unified Broken-Symmetry of the Cuprate Pseudogap State. ArXiv e-prints, April 2014.
Daniel Podolsky, Eugene Demler, Kedar Damle, and B. I.
Halperin. Translational symmetry breaking in the superconducting state of the cuprates: Analysis of the quasiparticle density of states. Phys. Rev. B, 67:094514, Mar
2003.
Peayush Choubey, T. Berlijn, A. Kreisel, C. Cao, and
17
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
P. J. Hirschfeld. Visualization of atomic-scale phenomena in superconductors: Application to fese. Phys. Rev.
B, 90:134520, Oct 2014.
A. Kreisel, Peayush Choubey, T. Berlijn, W. Ku, B. M.
Andersen, and P. J. Hirschfeld. Interpretation of scanning
tunneling quasiparticle interference and impurity states in
cuprates. Phys. Rev. Lett., 114:217002, May 2015.
E. A. Nowadnick, B. Moritz, and T. P. Devereaux. Quasiparticle interference and the interplay between superconductivity and density wave order in the cuprates. Physical
Review B, 86(13):134509, October 2012.
S Hüfner, MA Hossain, A Damascelli, and GA Sawatzky.
Two gaps make a high-temperature superconductor?
Reports on Progress in Physics, 71(6):062501, 2008.
J. W. Alldredge, J. Lee, K. McElroy, M. Wang, K. Fujita, Y. Kohsaka, C. Taylor, H. Eisaki, S. Uchida, P. J.
Hirschfeld, and J. C. Davis. Evolution of the electronic excitation spectrum with strongly diminishing hole density
in superconducting Bi2 Sr2 CaCu2 O8+δ . Nature Physics,
4(4):319–326, 2008.
Y. J. Uemura, G. M. Luke, B. J. Sternlieb, J. H. Brewer,
J. F. Carolan, W. N. Hardy, R. Kadono, J. R. Kempton,
R. F. Kiefl, S. R. Kreitzman, P. Mulhern, T. M. Riseman, D. Ll. Williams, B. X. Yang, S. Uchida, H. Takagi,
J. Gopalakrishnan, A. W. Sleight, M. A. Subramanian,
C. L. Chien, M. Z. Cieplak, Gang Xiao, V. Y. Lee, B. W.
Statt, C. E. Stronach, W. J. Kossler, and X. H. Yu. Unins
versal correlations between Tc and m
∗ (carrier density over
effective mass) in high-Tc cuprate superconductors. Phys.
Rev. Lett., 62:2317–2320, May 1989.
S. Doniach and M. Inui. Long-range coulomb interactions
and the onset of superconductivity in the high-tc materials.
Phys. Rev. B, 41:6668–6678, Apr 1990.
Mohit Randeria, Nandini Trivedi, Adriana Moreo, and
Richard T. Scalettar. Pairing and spin gap in the normal state of short coherence length superconductors. Phys.
Rev. Lett., 69:2001–2004, Sep 1992.
VJ Emery and SA Kivelson. Importance of phase fluctuations in superconductors with small superfluid density.
Nature, 374(6521):434–437, 1995.
Qijin Chen, Ioan Kosztin, Boldizsár Jankó, and K Levin.
Pairing fluctuation theory of superconducting properties in
underdoped to overdoped cuprates. Physical review letters,
81(21):4708, 1998.
M Franz and AJ Millis. Phase fluctuations and spectral
properties of underdoped cuprates. Physical Review B,
58(21):14572, 1998.
Guy Deutscher. Coherence and single-particle excitations in the high-temperature superconductors. Nature,
397(6718):410–412, 1999.
Hyok-Jon Kwon and Alan T Dorsey. Effect of phase fluctuations on the single-particle properties of underdoped
cuprates. Physical Review B, 59(9):6438, 1999.
E. W. Carlson, S. A. Kivelson, V. J. Emery, and
E. Manousakis. Classical phase fluctuations in high temperature superconductors. Phys. Rev. Lett., 83:612–615,
Jul 1999.
MJ Lawler, K Fujita, Jhinhwan Lee, AR Schmidt,
Y Kohsaka, Chung Koo Kim, H Eisaki, S Uchida,
JC Davis, JP Sethna, et al. Intra-unit-cell electronic
nematicity of the high-tc copper-oxide pseudogap states.
Nature, 466(7304):347–351, 2010.
P. D. C. King, J. A. Rosen, W. Meevasana, A. Tamai,
E. Rozbicki, R. Comin, G. Levy, D. Fournier, Y. Yoshida,
42
43
44
45
46
47
48
49
50
51
52
53
H. Eisaki, K. M. Shen, N. J. C. Ingle, A. Damascelli, and F. Baumberger. Structural origin of apparent fermi surface pockets in angle-resolved photoemission
of bi2 sr2−x lax cuo6+δ . Phys. Rev. Lett., 106:127005, Mar
2011.
M. R. Norman, M. Randeria, H. Ding, and J. C. Campuzano. Phenomenological models for the gap anisotropy
of Bi2 Sr2 CaCu2 O8 as measured by angle-resolved photoemission spectroscopy. Phys. Rev. B, 52:615–622, Jul 1995.
Matthias C. Schabel, C.-H. Park, A. Matsuura, Z.-X. Shen,
D. A. Bonn, Ruixing Liang, and W. N. Hardy. Angleresolved photoemission on untwinned YBa2 Cu3 O6.95 . i.
electronic structure and dispersion relations of surface and
bulk bands. Phys. Rev. B, 57:6090–6106, Mar 1998.
K. Pasanai and W. A. Atkinson. Theory of (001) surface
and bulk states in Y1−y Cay Ba2 Cu3 O7−δ . Phys. Rev. B,
81:134501, Apr 2010.
Takeshi Kondo, Tsunehiro Takeuchi, Syunsuke Tsuda, and
Shik Shin. Electrical resistivity and scattering processes in
(Bi, Pb)2 (Sr, La)2 CuO6+δ studied by angle-resolved photoemission spectroscopy. Phys. Rev. B, 74:224511, Dec
2006.
I. M. Vishik, M. Hashimoto, R.-H. He, W.-S. Lee,
F. Schmitt, D. Lu, R. G. Moore, C. Zhang, W. Meevasana,
T. Sasagawa, S. Uchida, K. Fujita, S. Ishida, M. Ishikado,
Y. Yoshida, H. Eisaki, Z. Hussain, T. P. Devereaux, and Z.X. Shen. Phase competition in trisected superconducting
dome. Proceedings of the National Academy of Science,
109:18332–18337, November 2012.
M. Okawa, K. Ishizaka, H. Uchiyama, H. Tadatomo,
T. Masui, S. Tajima, X.-Y. Wang, C.-T. Chen, S. Watanabe, A. Chainani, T. Saitoh, and S. Shin. Superconducting electronic state in optimally doped yba2 cu3 O7δ
observed with laser-excited angle-resolved photoemission
spectroscopy. Phys. Rev. B, 79:144528, Apr 2009.
M Le Tacon, A Bosak, SM Souliou, G Dellea, T Loew,
R Heid, KP Bohnen, G Ghiringhelli, M Krisch, and
B Keimer. Inelastic x-ray scattering in yba2cu3o6. 6 reveals
giant phonon anomalies and elastic central peak due to
charge-density-wave formation. Nature Physics, 10(1):52–
58, 2014.
David Benjamin, Dmitry Abanin, Peter Abbamonte, and
Eugene Demler. Microscopic theory of resonant soft-x-ray
scattering in materials with charge order: The example of
charge stripes in high-temperature cuprate superconductors. Phys. Rev. Lett., 110:137002, Mar 2013.
V. Thampy, M. P. M. Dean, N. B. Christensen, L. Steinke,
Z. Islam, M. Oda, M. Ido, N. Momono, S. B. Wilkins,
and J. P. Hill. Rotated stripe order and its competition
with superconductivity in la1.88 sr0.12 cuo4 . Phys. Rev. B,
90:100510, Sep 2014.
Qiang-Hua Wang and Dung-Hai Lee. Quasiparticle scattering interference in high-temperature superconductors.
Phys. Rev. B, 67:020511, Jan 2003.
C. V. Parker, P. Aynajian, E. H. da Silva Neto, A. Pushp,
S. Ono, J. Wen, Z. Xu, G. Gu, and A. Yazdani. Fluctuating stripes at the onset of the pseudogap in the high-Tc
superconductor Bi2 Sr2 CaCu2 O8+x . Nature, 468:677–680,
December 2010.
Ilija Zeljkovic, Zhijun Xu, Jinsheng Wen, Genda Gu,
Robert S Markiewicz, and Jennifer E Hoffman. Imaging the impact of single oxygen atoms on superconducting bi2+ ysr2–ycacu2o8+ x. Science, 337(6092):320–323,
2012.
18
54
55
56
57
58
59
60
61
62
63
64
65
66
L. Nie, G. Tarjus, and S. A. Kivelson. Quenched disorder
and vestigial nematicity in the pseudo-gap regime of the
cuprates. ArXiv e-prints, November 2013.
L. E. Hayward, D. G. Hawthorn, R. G. Melko, and
S. Sachdev. Angular fluctuations of a multi-component
order describe the pseudogap regime of the cuprate superconductors. ArXiv e-prints, September 2013.
V. V. Kabanov, J. Demsar, B. Podobnik, and D. Mihailovic. Quasiparticle relaxation dynamics in superconductors with different gap structures: Theory and experiments on yba2 cu3 o7−δ . Phys. Rev. B, 59:1497–1506, Jan
1999.
Y. H. Liu, Y. Toda, K. Shimatake, N. Momono, M. Oda,
and M. Ido.
Direct observation of the coexistence
of the pseudogap and superconducting quasiparticles in
bi2 sr2 cacu2 o8+y by time-resolved optical spectroscopy.
Phys. Rev. Lett., 101:137003, Sep 2008.
F. Cilento, S. Dal Conte, G. Coslovich, S. Peli, N. Nembrini, S. Mor, F. Banfi, G. Ferrini, H. Eisaki, M. K. Chan,
C. J. Dorow, M. J. Veit, M. Greven, D. van der Marel,
R. Comin, A. Damascelli, L. Rettig, U. Bovensiepen,
M. Capone, C. Giannetti, and F. Parmigiani. Photoenhanced antinodal conductivity in the pseudogap state of
high-Tc cuprates. Nature Communications, 5:4353, July
2014.
T. Dahm and L. Tewordt. Physical quantities in nearly
antiferromagnetic and superconducting states of the twodimensional hubbard model and comparison with cuprate
superconductors. Phys. Rev. B, 52:1297–1308, Jul 1995.
Chien-Hua Pao and N. E. Bickers. Superconductivity in
the two-dimensional hubbard model: One-particle correlation functions. Phys. Rev. B, 51:16310–16326, Jun 1995.
Michel Ferrero, Pablo S. Cornaglia, Lorenzo De Leo,
Olivier Parcollet, Gabriel Kotliar, and Antoine Georges.
Pseudogap opening and formation of fermi arcs as an
orbital-selective mott transition in momentum space.
Phys. Rev. B, 80:064501, Aug 2009.
E. Gull, M. Ferrero, O. Parcollet, A. Georges, and A. J.
Millis.
Momentum-space anisotropy and pseudogaps:
A comparative cluster dynamical mean-field analysis of
the doping-driven metal-insulator transition in the twodimensional hubbard model. Phys. Rev. B, 82:155101, Oct
2010.
M. Ossadnik, C. Honerkamp, T. M. Rice, and M. Sigrist.
Breakdown of landau theory in overdoped cuprates
near the onset of superconductivity. Phys. Rev. Lett.,
101:256405, Dec 2008.
T. Hanaguri, Y. Kohsaka, M. Ono, M. Maltseva, P. Coleman, I. Yamada, M. Azuma, M. Takano, K. Ohishi,
and H. Takagi. Coherence factors in a high-tc cuprate
probed by quasi-particle scattering off vortices. Science,
323(5916):923–926, 2009.
K. Fujita, A. R. Schmidt, E.-A. Kim, M. J. Lawler, D. H.
Lee, J. C. Davis, H. Eisaki, and S.-i. Uchida. Spectroscopic Imaging Scanning Tunneling Microscopy Studies of
Electronic Structure in the Superconducting and Pseudogap Phases of Cuprate High-Tc Superconductors. Journal
of the Physical Society of Japan, 81(1):011005, January
2012.
Yang He, Yi Yin, M. Zech, Anjan Soumyanarayanan,
Michael M. Yee, Tess Williams, M. C. Boyer, Kamalesh Chatterjee, W. D. Wise, I. Zeljkovic, Takeshi
Kondo, T. Takeuchi, H. Ikuta, Peter Mistark, Robert S.
Markiewicz, Arun Bansil, Subir Sachdev, E. W. Hudson,
67
68
69
70
71
72
73
74
75
76
77
78
79
and J. E. Hoffman. Fermi surface and pseudogap evolution
in a cuprate superconductor. Science, 344(6184):608–611,
2014.
Tao Wu, Hadrien Mayaffre, Steffen Krmer, Mladen Horvati, Claude Berthier, W.N. Hardy, Ruixing Liang, D.A.
Bonn, and Marc-Henri Julien. Central role of disorder in the pseudogap state of high-tc superconductors.
Unpulished.
Neven Barišić, Sven Badoux, Mun K Chan, Chelsey
Dorow, Wojciech Tabis, Baptiste Vignolle, Guichuan Yu,
Jérôme Béard, Xudong Zhao, Cyril Proust, et al. Universal
quantum oscillations in the underdoped cuprate superconductors. Nature Physics, 9(12):761–764, 2013.
Eduardo H. da Silva Neto, Riccardo Comin, Feizhou
He, Ronny Sutarto, Yeping Jiang, Richard L. Greene,
George A. Sawatzky, and Andrea Damascelli. Charge ordering in the electron-doped superconductor nd2xcexcuo4.
Science, 347(6219):282–285, 2015.
E. W. Hudson, K. M. Lang, V. Madhavan, S. H. Pan,
H. Eisaki, S. Uchida, and J. C. Davis. Interplay of magnetism and high-Tc superconductivity at individual Ni
impurity atoms in Bi2 Sr2 CaCu2 O8+ . Nature (London) ,
411:920–924, June 2001.
AJ Achkar, X Mao, Christopher McMahon, R Sutarto, F He, Ruixing Liang, DA Bonn, WN Hardy, and
DG Hawthorn. Impact of quenched oxygen disorder on
charge density wave order in yba 2 cu 3 o {6 + x}. arXiv
preprint arXiv:1312.6630, 2013.
PM Chaikin. Field induced spin density waves. Journal de
Physique I, 6(12):1875–1898, 1996.
Z. Lotfi Mahyari, A. Cannell, E. V. L. de Mello,
M. Ishikado, H. Eisaki, Ruixing Liang, D. A. Bonn, and
J. E. Sonier. Universal inhomogeneous magnetic-field response in the normal state of cuprate high-Tc superconductors. Phys. Rev. B, 88:144504, Oct 2013.
T. Cren, D. Roditchev, W. Sacks, J. Klein, J.-B. Moussy,
C. Deville-Cavellin, and M. Laguës. Influence of disorder
on the local density of states in high- ¡span class=”apsinline-formula”¿¡math¿¡mrow¿¡msub¿¡mrow¿¡mi
mathvariant=”italic”¿t¡/mi¿¡/mrow¿¡mrow¿¡mi
mathvariant=”italic”¿c¡/mi¿¡/mrow¿¡/msub¿¡/mrow¿¡/math¿¡/span¿
superconducting thin films. Phys. Rev. Lett., 84:147–150,
Jan 2000.
Benjamin Sacépé, Thomas Dubouchet, Claude Chapelier,
Marc Sanquer, Maoz Ovadia, Dan Shahar, Mikhail Feigelman, and Lev Ioffe. Localization of preformed cooper pairs
in disordered superconductors. Nature Physics, 7(3):239–
244, 2011.
M. R. Norman, A. Kanigel, M. Randeria, U. Chatterjee,
and J. C. Campuzano. Modeling the fermi arc in underdoped cuprates. Phys. Rev. B, 76:174501, Nov 2007.
T. J. Reber, N. C. Plumb, Z. Sun, Y. Cao, Q. Wang,
K. McElroy, H. Iwasawa, M. Arita, J. S. Wen, Z. J. Xu,
G. Gu, Y. Yoshida, H. Eisaki, Y. Aiura, and D. S. Dessau.
The origin and non-quasiparticle nature of Fermi arcs in
Bi2 Sr2 CaCu2 O8+δ . Nature Physics, 8:606–610, August
2012.
N.E. Hussey. What drives pseudogap physics in high-tc
cuprates? a view from the (resistance) bridge. Journal
of Physics and Chemistry of Solids, 72(5):529 – 532,
2011.
Spectroscopies in Novel Superconductors 2010
{SNS} 2010.
FC Zhang and TM Rice. Effective hamiltonian for the
superconducting cu oxides. Physical Review B, 37(7):3759,
19
80
81
82
83
84
1988.
Note that current experiments would detect this situation
as a pronounced peak at q = 2kF due to the common
practice of “background subtraction”.
The antinodes are here defined as the points at which the
Fermi surface reaches the boundaries of the first Brillouin
Zone
Here we use the term “nesting” to indicate segments of the
Fermi surface that are parallel to each-other, leading to
an enhanced scattering at the corresponding wave-length
difference
The quantitative discrepancy between our predictions and
their finding may be attributed to either the different material ( La1.88 Sr0.12 CuO4 ), or the different technique (resonant elastic X-ray scattering).
In the experimental curves reproduced in Fig. 1(e), the
signal in (0, q) and (q, q) directions were normalized by
a different multiplicative factor Indeed, if the same normalization had been used, the two curves should had converged in the q → 0 limit. As a consequence, the plot of
Fig. 1(e) does not convey any information about the ra-
85
86
87
88
tio between the (0.25, 0) and (0.25, 0.25) wavevectors, but
only describes the behavior of the REXS intensity in the
(0, q) and (q, q) direction, independently.
By comparing the intensity of the Bragg peaks we deduce
that σx ≈ 0.25a, or σk = 4(2π/a).
We verified that the position of this peak is uniquely determined by the chemical potential. Note that the wavevector
qx = 0.12 approximately corresponds to the distance between the antinodes and the Van Hove singularity, located
in the middle between two adjacent antinodes.
Because the wavevector (q, 0) connects quasiparticles with
opposite signs of the pairing gap, while the (q, q) wavevector connects quasiparticles with the same sign, these two
contributions are often referred to as “sign-reversing” and
“sign-preserving”.
Here we assume the impurity itself to be symmetric with
respect to r → −r. This assumption applies for example
to impurities with s and d symmetry, but does not apply
to impurities with p symmetry