High Pressure and Road to Room Temperature

Superconductivity
Lev P. Gor’kov1 and Vladimir Z. Kresin2
1
NHMFL, Florida State University, 1800 E. Paul Dirac Drive,
Tallahassee, Florida, 32310, USA

2
Lawrence Berkeley Laboratory, University of California,
1 Cyclotron Road, Berkeley, CA 94720, USA

PACS numbers:74.20.-z, 74.25.Kc, 63.20.kd,64.60.F

Abstract

High pressure serves as a path finding tool towards novel structures,

including those with very high Tc. The superconductivity in sulfur
hydrides with record value of Tc =203 K (!) is caused by the phonon
mechanism. However, the picture differs from the conventional one in
important ways. The phonon spectrum in sulfur hydride is both broad
and has a complex structure. High value of Tc is mainly due to strong
coupling to the high-frequency optical modes, although the acoustic
phonons also make a noticeable contribution. New approach is
described; it generalizes the standard treatment of the phonon
mechanism and makes it possible to obtain an analytical expression
for Tc . It turns out that, unlike in the conventional case, the value of
the isotope coefficient varies with the pressure and reflects the impact
of the optical modes. The phase diagram, that is the pressure
dependence of Tc, is rather peculiar. A crucial feature is that
increasing pressure results in a series of structural transitions,
including the one, which yields the superconducting phase with the
record Tc . In a narrow region near P≈150 GPa the critical
temperature rises sharply from Tc ≈120K to Tc ≈200K. The sharp
structural transition, which produces the high Tc phase, is a first-order
phase transition caused by interaction between the order parameter
and lattice deformations. A remarkable feature of the electronic
spectrum in the high Tc phase is the appearance of small pockets at
the Fermi level. Their presence leads to a two-gap spectrum, which

1
can, in principle, be observed, with the future use of tunneling
spectroscopy. This feature leads to non-monotonic and strongly
asymmetric pressure dependence of Tc. Other hydrides can be
expected to display even higher values of Tc, up to room temperature.
The fundamental challenge lays in creation a structure capable of
displaying high Tc at ambient pressure.

CONTENTS
I. Introduction
II. Hydrides
A. Metallic hydrogen
B. Superconductivity in hydrides: main properties

III. Electron-phonon interaction. Critical temperature

A. Main equations. The coupling constant and Tc.
B. Function α2(Ω)F(Ω). Tunneling spectroscopy
IV. Sulfur hydrides
A. Phonon spectra and electron-phonon interaction
B. Generalized equation and two coupling constants.
C. Critical temperatures for different phases
V. Isotope effect

VI. Energy spectrum in the high Tc phase: two-gap structure

and non-monotonic behavior of Tc
A. Structural transitions. High Tc phase.
B. The ”Migdal” criterion and small pockets

2
 C. Broad bands and small pockets
D. Two-gap spectrum. Slow decrease in Tc at P>Pcr.
VII. Other hydrides
A. Calcium hydride; MgH6
B. Palladium hydride
C. Organic hydrides
D. Transition of ice under high pressure and by doping
VIII. Main challenges.
IX. Concluding remarks.
Acknowledgements
References

I. Introduction.
In a recent dramatic development, superconductivity with a
critical temperature Tc=203 K (!) has been observed in sulfur hydride
under high pressure (Drozdov, Eremets et al., 2015). This
development is the most significant breakthrough since the discovery

of the high TC oxides (Bednorz and Mueller, 1986). There is every

reason to anticipate even higher values of Tc for other hydrides,

which means that achieving superconductivity at room temperature
now appears perfectly realistic.
We focus below on theoretical aspects, which can provide a clue
to understanding the specifics of the new superconducting state. It
should be emphasized that in many respects the system is quite
unusual. As will be discussed below, the observed phenomenon can
be explained by the phonon mechanism. Nevertheless, a number of

3
key features reveal that the picture differs from the conventional one.
In this Colloquium we devote particular attention to these features.
One of them is the peculiar pressure dependence of the critical
temperature. Not only does external pressure result in metallization
and the appearance of superconductivity, but also the value of Tc
grows dramatically with a further rise in pressure. Moreover, this
variation turns out to be non-monotonic. It also should be noted that
while Cooper pairs formation is mediated by phonons, the complex
structure of the phonon spectrum and its broad range (all the way up
to Ω≈2000 K, where Ω is the phonon frequency) makes it necessary
to modify the conventional treatment. While the strong isotope effect
affirms the action of the phonon mechanism, the value of the isotope
coefficient turns out to vary with pressure. Below we discuss all these
interesting aspects of the new development in superconductivity.
Despite its rich and interesting prehistory, the discovery came as
a surprise, especially for those who accept the notion that electron-
phonon interaction cannot give rise to such a high Tc. The
background of this notion will be discussed below (Sec. IIIA). One
also should not lose track of the fact that this impressive discovery
came about thanks to the remarkable progress in high-pressure
technique and to experimental innovations by the M.Eremets group
(Max Planck Institute, Mainz, Germany ). One should stress also that
the study of the hydrides has attracted an attention of many
theoretical groups (see below,Ch.3). One should give a special credit
to the group of Y.Ma and also to D.Duan, T.Cui and their
collaborators (Julin University, Changchun, China) (see
below,Chs.III,VII). Their remarkable studies brought a special

4
attention to the sulfur hydrides and motivated the key experimental
studies.
A few remarks on the historical perspective are in order. The
phenomenon of superconductivity was discovered more than one
hundred years ago by H. Kamerlingh Onnes (1911). While
measuring the temperature dependence of the electrical resistance
R(T) of mercury he observed that at the temperature of 4.2 K the
resistivity suddenly vanishes. The dissipationless (R=0) state, which
emerged, was named the superconducting state. Subsequently
superconductivity was discovered in many other materials. Moreover,
it was soon realized that the loss of resistance was only one facet of
the superconducting state, hence the latter corresponded to a
qualitatively new state of matter. Its most fundamental feature, the so-
called Meissner effect ( Meissner and Ochsenfeld, 1933), is
manifested in the expulsion of the magnetic field from the bulk of the
sample (anomalous diamagnetism).
The microscopic theory of superconductivity was created by
Bardeen, Cooper and Schrieffer (BCS) in 1957, almost 50 years after
the experimental discovery. According to the BCS theory, the key
microscopic factor behind the phenomenon is the attraction between
electrons mediated by the exchange of phonons, such that below Tc
within the electronic system there forms a macroscopic manifold of
bound electron pairs (known as the Cooper pairs). Thus the attraction
has its origin in the ionic system; qualitatively the effect may be
related to the dielectric function changing its sign at low frequencies.

The well-known expression for the critical temperature is

5
 ⎛ 1 ⎞
!
Tc ≈ Ωexp ⎜− ⎟ (1.1)
⎝ λ − µ *⎠

! is the characteristic phonon frequency (typically on

Here Ω
the order of the Debye temperature), λ is the electron-phonon
coupling constant, and µ* is the so-called Coulomb pseudopotential
which characterizes the direct electron-electron repulsion (usually
µ*≈0.1-0.15). Eq.(1.1) is valid in the weak coupling approximation
(λ<<1). Note that the specific value of the pre-exponential factor is
determined by the renormalization effect (see below,Sec.IIIA).
The search for superconducting materials with higher critical
temperatures has been ongoing. Fig.1 shows how the maximum
transition temperatures within different superconductor families has
grown with time. In the first 75 years progress was rather modest
(from Tc=4.2 K for mercury up to Tc≈23 K in Nb3Ge). A breakthrough
came in 1986 when Bednorz and Müller discovered a new family of
superconducting materials, the copper oxides (cuprates) and
observed a Tc close to 40 K in the La-Sr-Cu-O compound.
Subsequent research on cuprates raised their Tc all the way up to
130 K in the HgBaCaCuO compound (Schilling et al., 1993). For
what follows, it is noteworthy that under pressure Tc was raised up to
≈160 K (Gao et al., 1994). Until very recently, this remained the
highest critical temperature ever observed. (Fig.1)

6
II. Hydrides
A. Metallic hydrogen.

As noted in the Introduction, recent work on sulfur hydrides under

high pressure led to the observation of a superconducting state with
the record value of Tc=203K . This achievement was preceded by
developments, which started almost fifty years ago (Ashcroft, 1968).
According to the BCS theory, Tc is proportional to the characteristic
! ∝1 / M [see Eq.(1.1)]. One may expect,
phonon frequency Ω
therefore, that metallic hydrogen should have a high value of Tc : for
such a light metal the characteristic phonon frequency in the prefactor
of Eq.(1.1) is high and (ignoring for the moment the magnitude of the
exponential factor) Tc may also turn out to be rather high. However,
this prediction can be verified only under very high pressure. Indeed,
hydrogen first must become metallic, but the transition from the
molecular phase into the metallic state is known to require high
pressure (Wigner and Huntington,1935). This is the dissociative
transition to an atomic lattice, which occurs through compression of
solid molecular hydrogen.
A conductive (probably, semimetalic ) state of hydrogen was
observed by Eremets and Troyan (2011) at room temperature under
the pressure of 260-270 GPa (recall that 100 GPa corresponds to a
million atmospheres).
The phase diagram for hydrogen is rather complicated (Fig. 2),
and the determination of this diagram was a non-trivial task. The main
complication derives from the fact that the usual technique for

7
Fig.1. Increase in Tc;max with time. During the next 75 years since the
discovery the increase was ΔTc≈18K ; the main focus was on the Nb-
based materials. The highest Tc for the cuprates at ambient pressure
(Tc ≈130K) was raised under the pressure up to Tc ≈160K. At
present, the curve is extended up to Tc;max = 203K observed for sulfur
hydride under high pressure.

the structure determination, which underlies the phase diagrams of

materials, namely X-ray diffraction, is not conclusive because
scattering by hydrogen is very weak.

8
The diagram in Fig. 2 is based on Raman and infrared measurements
along with resistivity data. The low temperature phases I-III wer
observed at relatively low pressures (≤150 GPa, see the review by
Mao et al.,1994).

Fig. 2. The phase diagram of hydrogen. Letters I-VI indicate domains

for different phases. Phases I-III correspond to molecular state of
pure hydrogen. Their boundaries can be determined from Raman and
infrared measurements. Phase IV is mixed atomic-molecular phase,
phase V-atomic liquid. Recently observed phase VI corresponds to
the metallic state. From Eremets et al., 2016.

9
 Recently Eremets et al. (2016) observed a new phase (phase VI
in Fig.2) at pressures >! 360 GPa and temperatures <200 K. This
phase displays a drastic drop in resistivity, characteristic featureless
Raman spectra, and the absence of a photoconductive response.
These properties are characteristic of the metallic state.
As concerns the superconducting state of metallic hydrogen,
theoretical calculations (McManon et al.,2012) show that a high-Tc
state of pure metallic hydrogen requires pressures on the order of
≈500 GPa. Hopefully, this phenomenon will be observed in the near
future.

B. Hydrides. High Tc superconductivity.

The reasoning for an elevated value of Tc outlined above
applies not only to pure hydrogen but also to hydrides, that is, to
materials containing hydrogen as one of their components (Ashcroft,
2004; Wang and Ma, 2014). The presence of hydrogen results in the
appearance of high-frequency optical phonon modes, whereas heavy
ions provide additional acoustic modes, which also contribute to the
pairing. At the same time, the metallization of such compounds does
not require extraordinarily high pressures. In fact, hydrides are even
more promising materials than pure metallic hydrogen. Indeed, the
presence of more than one ion in the unit cell, in this case an
additional hydrogen ion, leads to an appearance of high frequency
optical modes which, in addition to high frequency , are characterized
by a high density of states. The latter is beneficial for
superconductivity.

10
 A number of density-functional theory studies supported the high
promise of hydride compounds. For example, calculations (Gao et al.,
2008) suggested that GaH3 at P≈160 GPa would display Tc≈140 K
and Si2H6 at ≈275 GPa would display Tc≈73 K – 86 K (Jin et al.,2010)
.The most thermodynamically stable structures were established by
calculating the enthalpy- difference curves. Among others, the
minima hopping method was employed (Goedeker, 2004).
Initial experiments on SiH4 ( Eremets et al.,2008) demonstrated
that hydrides can indeed support a superconducting state, although
the critical temperature was a relatively modest ≈17K. Later,
following the discovery of the record high Tc in sulfur hydrides, a high
Tc state (≈100 K) in covalent hydride phosphine (P-H) was also
observed (Drozdov et al., 2015). Theoretical analysis (Flores-Divas et
al, 2016) suggests that the PH1,2,3 systems indeed have a rather high
Tc , but the material is probably, in a metastable state.
In 2014 Li et al. made the remarkable prediction that metallic
sulfur hydride would become superconducting with Tc ≈80 K under
the relatively low pressure P≈100 GPa . Following the first
experimental observation of such a superconducting state, Eremets
and his collaborators continued increasing the pressure and
discovered that Tc goes up significantly all the way to Tc≈203 K
(Drozdov, Eremets et al., 2014, 2015a; see also the review by
Eremets and Drozdov, 2016). Such a remarkable observation was
explained by mixed valence of S and formation of sulfur hydride with
higher hydrogen content. This assumption was in agreement with

11
very interesting independent theoretical study by Duan et al (2014)
More specifically, increase in pressure is accompanied by the
formation of H3S structural units (Fig.3a) via the transformation:

3 H2S ! 2H3S+S ( 2.1 )

The transformation H2S! H3S also has been confirmed by

detailed calculations by Bernstein et al.,2015 and Errea et al.,2015.

Sulfur hydride at the pressure P≈90GPa has Tc≈90 K. As

mentioned above, the subsequent increase in pressure leads to a
large increase in the value of the critical temperature up to Tc ≈200
K. This pressure dependence implies that the presence of hydrogen
is necessary but not sufficient by itself for reaching the highest Tc
values. Indeed, the sample at 90 GPa contains hydrogen and
consequently high frequency modes. The Tc of 90 K is very high and
would be sensational 30 years ago, but it is much below the 203 K
that is achieved under higher pressure. Therefore the pressure
increase brings in some additional factors. As mentioned above,
Drozdov, Eremets et al. (2014,2015a) suggested that the rise in Tc is
due to the formation of new compounds with higher valence states of
sulfur.
As will be described below, the pressure increase changes the
crystal structure of sulfur hydride, thereby signifying a structural
phase transition. The structural transformation plays the

12
Fig. 3. Structures of the “low Tc” (R3m; Tc ≈120K) and
“high Tc” ( Im-3m; Tc≈200K) phases: (a) structure of the R3m phase
(top view); one can see the H3S entities. Large (small) spheres
denote sulfur (hydrogen) ions. After Duan et al.,2014 (b) structure of
the R3m phase (side view); one can see that d1≠ d2, di (i=1.2) are the
distances between the H ion and neighboring S-ions .From Errea et
al.,2016. (c) structure of the Im-3m phase . Unlike the “low
Tc “ phase, the distances d1 and d2 are equal in the cubic high Tc
phase Im-3m. From Duan et al.,2014 .

the crucial role in the observed behavior.(Fig.3).

13
 Below we focus on the two phases with the highest values of Tc
(Fig.3) . One of them (Figs.3a,b) has the crystal structure ,which
corresponds to R3m symmetry (see, e.g., Massa, 2004). Increasing
the pressure further leads, at P≈150 GPa, to the appearance of a
different phase (Fig.3c). It is this new structure that displays the
record- high Tc .The symmetry group of the lattice is cubic Oh (Im-
3m). The theoretically predicted structure is in agreement with X-rays
data (Einaga et al.,2016; Goncharov et al.,2016;2017).
The usual X-rays spectroscopy does not allow to determine the
structure of the hydrides with a high accuracy, because the light H
ions do not provide strong scattering. In addition, the multiphase
nature of the studied samples (Goncharov et al., 2016) also
represents a serious complication. Recently, advance spectroscopy
has been employed to monitor how the structure evolves with
pressure (Goncharov et al., 2017). More specifically, the synthesis
performed out of S and molecular hydrogen along with cyclotron XRD
technique and Raman spectroscopy did allow the researchers to
study all structures formed with an increase in pressure. It turns out
that the transformation (2.1) occurs at pressure P> 40 GPa . It has
been also demonstrated that the phase at P> 110GPa (Tc≈120 K)
has R3m symmetry (Figs.3 a,b). As for the most interesting high Tc
phase, its structure (Fig. 3c) , indeed, is characterized by the Im-3m
symmetry.
The transformation (2.1) supports the suggestion by Drozdov et
al. (2015, see above) about the mixed valence state of sulfur. The
latter factor turns out to be essential (see Sec. IVC below).
C. Superconductivity in hydrides: Main properties.

14
 Experimentally, the onset of the high Tc superconducting state
in sulfur hydrides ( Drozdov et al.,2015) is detected by the drastic
resistance drop near Tc . A sharp transition was observed in
annealed samples. The measured resistance was at least two
orders of magnitude below that of pure copper.
The critical temperature shifts downward in the presence of an
external magnetic field. Magnetic susceptibility measurements reveal
an abrupt transition into the diamagnetic state (the Meissner effect).
This key result was also confirmed by a direct observation (Troyan et
al.,2016) of magnetic field expulsion as detected by the response of a
thin Sn film placed inside the bulk sample. The Meissner effect was
also observed later by Huang et al. (2016) by means of AC magnetic
susceptibility measurements.
Below we concentrate on the two phases with the highest Tc. One
of them (R3m; Fig.3a,b) has Tc ≈120 K , and the other (Im-3m;
Fig.3c) has the highest Tc of ≈200 K. Below, we will refer to the
former structure as the “low Tc phase” (although this name sounds
ironic for Tc ≈120 K), and to the second structure as the “high Tc
phase.”
The question of the mechanism of superconductivity with such a
record Tc is of fundamental interest. The strong isotope effect
indicates that pairing is provided by phonon exchange. The main
contribution comes from the high-frequency optical modes.
Nevertheless, as stressed in the Introduction, the picture is far from
conventional.
The phase diagram, that is, the pressure dependence of Tc is
very peculiar. Indeed, Fig. 4 demonstrates that Tc is strongly

15
dependent on the applied pressure (Einaga et al, 2016). It increases
from ≈100 – 120 K up to the record ≈200 K over the relatively narrow
pressure interval 125-150 GPa. We argue below that such a rapid Tc
change is a fingerprint of a first-order structural transition.

Fig. 4. Pressure dependence of Tc. The data for annealed samples

are presented. One can see a large increase in the value of Tc in the
region near P=140GPa. Adapted from Einaga et al.,2016.

16
Note that once Tc reaches its maximum value, it decreases upon
further rise in applied pressure. The decline is rather slow, so that the
Tc dependence is strongly asymmetric relative to Tc ;max. Such a
dependence is unusual and requires an explanation. We interpret this
feature (see Sec. VIID) as deriving from the impact of the
superconducting order parameters on the main Fermi surfaces and
on small pockets, with the latter appearing in the electronic spectrum
of the high Tc phase in the course of a first-order phase transition.
Note also that because of the complex structure and the width of
the phonon spectrum evaluation of Tc and of the isotope coefficient
(whose value turns out to be pressure-dependent) must be carried
out with considerable care.
The next chapter contains a general description of the phonon
mechanism with emphasis on the strong coupling case. Subsequent
chapters focus on the challenges described above.

III. Electron-phonon interaction. Critical temperature.

A. Main equations. Coupling constant and Tc.
Our main goal is to evaluate the value of the critical
temperature for the hydrides. The usual BCS model is not applicable
here ,because it has been developed in a weak coupling
! ). The
approximation ( λ<<1 and correspondingly, 2π Tc << Ω
equations describing strong-coupled superconductors (Migdal,1958;
Eliashberg,1960) contain the so-called phonon propagator (see
below, Eq.(3.1), and therefore, the phonon frequency Ω.
The equation for the pairing order parameter Δ(ωn) has the form:

17
 α 2 (Ω) F (Ω) Δ (ω m )
Δ (ω n ) Z = π Tc ∑ ∫ dΩ D (Ω,! ω n − ω m )
m Ω | ωm | (3.1)
Here
Ω2
D= 2
Ω2 + (ω n − ω m )

(3.2)
is the phonon propagator, Ω is the phonon frequency, ωn=(2n+1)πTc
(the method of thermodynamic Green’s functions is employed; see,
e.g., Abrikosov et al.,1975 ); one should add also the Coulomb
pseudopotential µ*. The factor Z is the renormalization function
determined by the relation:
α 2 (Ω) F (Ω) ω
Z = 1+ (π Tc / ω n )∑ ∫ dΩ D (Ω,! ω n − ω m ) m (3.3)
m Ω | ωm |
The renormalization function describes the “dressing” of electrons
moving through the ionic lattice.
Eqs.(3.1),(3.3) contain the important quantity, the function
α 2 (Ω) F (Ω) .Here F(Ω) is the phonon density of states, α2(Ω)

describes the electron-phonon interaction and contains the

corresponding matrix element (See e.g., Scalapino,1969;
Grimvall,1981; Wolf,2012).
In addition, one can introduce an important parameter, so-called
coupling constant λ, defined by the relation:

α 2 (Ω) F (Ω)
λ = ∫ dΩ
Ω (3.4)

18
 Note that Eqs. (3.1), (3.3) do not explicitly contain the
coupling constant λ. Indeed, generally speaking, this constant cannot
be factored out, because the phonon frequency enters not only in
the factor α 2 (Ω) F (Ω) but also in the phonon propagator D(ωn-ωm) ,

which, in addition, depends on ωn-ωm (see Eq.(3.2) ). It is apparent

from Eqs. (3.1), or (3.3 ) that the coupling constant can be factored
out, if Eq. (3.1 ) does not contain a phonon propagator function (e.g.,
D≈1 for the weak coupling case) or if the phonon propagator D
slowly depends on the frequency Ω , so that Ω in D can be replaced
by its average value. When the function α 2 (Ω) F (Ω) is known, the

value of the critical temperature can be evaluated from the non-linear

equation which looks like (3.1) with the replacement:

1/2
| ω n |→ ⎡⎣ω n2 + Δ 2 (ω n )⎤⎦ in the denominator of the integrand. Such an

equation describes the order parameter at any temperature ( then

ωn=(2n+1)πT).The calculation can be performed numerically ,without
invoking the coupling constant concept. We will discuss the
corresponding method below (Sec. IVA) while focusing on the sulfur
hydrides.

At the same time the possibility to introduce the coupling constant, λ

,is very beneficial for the analysis. The concept of coupling constant
is commonly used to study usual superconductors. It allows one to
deduce the analytical expressions for Tc and interpret its
dependences on other parameters in the problem. Such an approach
is justified, because usually the function α 2 (Ω) F (Ω) is characterized

19
by the peak structure in phonon density of states F(Ω) (see, e.g.,
Wolf ,2012, and also Fig.5b ). The latter structure corresponds to the
short-wavelength part of the spectrum where the mode dispersion is
weak. The phonon propagator changes slowly on the scale
corresponding to the peak structure, and this permits the replacement
! ; the latter
of Ω in the phonon propagator by its average value Ω
! =<Ω2>1/2 (see, e.g., Louie and Cohen ,1977,
can be taken either as Ω
!
and the reviews by Grimvall ,1981; Kresin and Wolf (2009)) , or as Ω
! =<Ω2>1/2 (Allen and Dynes ,1975;
=< Ωlog>which is close to Ω
Carbotte,1990) . The average < f > is defined by the relation:

! =<Ωlog>,then f= log
< f (Ω) >= ( 2 / λ ) ∫ dΩ f (Ω) α 2 (Ω) F (Ω) Ω−1 . If Ω

! =<Ω2>1/2, so that
. Below we have chosen Ω

< Ω2 >= ( 2 / λ ) ∫ dΩΩα 2 (Ω) F (Ω) ; λ is defined by Eq.(3.4). As a result,

Eqs.( 3.1 ),(3.3) can be written in the form

Δ (ω m )
Δ (ω n ) Z = π Tc λ ∑ D (Ω,! ω n − ω m ) (3.5)
m | ωm |

ωm
Z = 1+ (π Tc / ω n )λ ∑ D (Ω,! ω n − ω m ) (3.6)
m | ωm |

D is defined by Eq.(3.2) and

! =<Ω2>1/2
Ω (3.7)

20
 (a) (b)

(c)

Fig.5.a) Phonon spectrum of Pb ; Ω and q are the phonon frequency

and momentum, from Brockhouse et al. ,1962; b) the function
α 2 (Ω) F (Ω) for Pb. Adapted from McMillan and Rowell,1965; c)

phonon spectrum of sulfur hydride (R3m phase; from Duan et

al.,2014)

21
As is known, the solution of the equation (3.5) can be written in
analytical form and the explicit expression depends on the strength of
the electron-phonon interaction (see the review by Kresin et al.,2014).
! ), D≈1, Z≈1, and
For the weak coupling case (λ<<1; then 2π Tc << Ω
we obtain a well-known BCS expression, Eq.(1.1). For the
intermediate coupling one should take into account the
renormalization of the coupling constant ,since Z≈1+λ, and Tc is
described by the corresponding equation (Nakajima and Watabe
,1963, see Grimvall, 1977). Note that even in the weak coupling case
the renormalization function affects the value of the pre-exponential
factor (Karakosov et al,1976; Wang and Chubukov,2013),which
! ,see below Eqs.(4.5),(7.1). A well-
appears to be equal to a ≈ 0.25Ω
known McMillan-Dynes expression (McMillan ,1968;Dynes ,1972) is
widely used in literature:

⎡ ⎤
(! /1.2 exp ⎢− 1.04 (1+ λ ) ⎥
Tc = Ω ) ⎢⎣ λ − µ * (1+ 0.62 λ ) ⎥⎦ (3.8)

This expression is valid for λ <! 1.5 . Note that for small λ Eq. (3.8) can
! , which is close
be written in the form similar to Eq. (1.1) with a ≈ 0.3Ω
to the value mentioned above. For larger values of λ one should use
the different expression, which was obtained by numerical
modification of Eq. (3.8) (Allen and Dynes, 1975), see below,
Sec.IVA. One can use also the expression valid for any λ (Kresin,
1987), see below, Eq. (7.1). Note that for the case of very strong

22
 ! ) the dependence of Tc on λ is entirely
coupling ( λ >! 5 ; then π Tc >! Ω

different from Eqs. (1.1), (3.8) (we omitted µ* for simplicity) and has a
form:

!
Tc = 0.18λ 1/2Ω
(3.9)

(see Allen and Dynes, 1975; Kresin et al.,1984) .

In addition, one can introduce the following important relation

(McMillan,1968):

!2
λ =< I >2 ν / M Ω (3.10)

(I describes the electron-phonon scattering, ν is the electronic density

of states)

The important question is whether there exists an upper limit of the

value of Tc . A “myth”, which persists even today, claims that the
phonon mechanism is unable to provide the values of the critical
temperature higher that ∼30K. For some scientists the argumentation

is based on Eqs. ( 3.8),(3.10 ). Indeed, neglecting µ* for simplicity

! , one can easily find the maximum value of Tc
and calculating ∂Tc / ∂Ω
; this value corresponds to λ 2. However the McMillan equation is
valid only for λ <! 1.5 . This limitation was mentioned above, below
Eq.(3.8). In connection with this, let us note that the MacMillan-Dynes
equation (3.8) was obtained by taking into consideration the
renormalization function Z≈1+λ ( see Eqs.(3.1) and (3.3)) and by

23
fitting the coefficients to describe the data on Nb. The same limitation
was obtained by Geilikman et al (1975), who derived analytically the
equation similar to (3.8) ,see below, Eq.(4.6);the derivation is valid for
λ< ! )2<<1. Therefore, the value
! 1.5 (then (πTc/ Ω λ 2 is outside the
range of the applicability of the MacMillan-Dynes equation. One can
see from Eq. ( 3.9) that the mentioned upper limit for Tc does not
exist.

Another erroneous restriction was imposed not on the

dependence of Tc on λ (see above), but on the limiting value of
the coupling constant λ itself. In the framework of the so-called
Froelich Hamiltonian : Ĥ = Ĥ el + Ĥ ph + Ĥ int ,where Ĥ ph contains the

phonon frequency Ω0 ( q) and Ĥ int describes the electron-phonon

! ,renormalized by the
interaction, one obtains for the frequency Ω
! = Ω (1− 2 λ )1/2 .One
electron-phonon interaction, the expression: Ω 0

would conclude that the lattice becomes unstable at value of the

coupling constant ! (see
=0.5, and therefore the value Tc ≤ 0.1Ω
! ≈ Ω ) provides the upper limit of the critical temperature.
Eq.(1.1), Ω D

At the same time, as we know, there exist many superconductors

with λ > 0.5 (e.g., Sn, Pb, Hg ). To clarify this point, an analysis,
based on rigorous adiabatic theory was carried out ( Brovman and
Kagan, 1967, Geilikman ,1971). It has been shown that the use of the
experimentally observable acoustic law for Ω0 ( q) is not self-

consistent. The thing is that the electron-ion interaction participates in

24
formation of the phonon spectrum in the system, so that one is
dealing with a double counting.
Starting from the adiabatic theory with the Hamiltonian
! !
Ĥ = T̂r! + T̂R! + V (r, R) ( T̂r! and T̂R! are kinetic energy operators for
! !
electrons and ions, and V (r, R) is the sum of the Coulomb
interactions) one can evaluate the electron-phonon interaction and
the phonon spectrum rigorously. The electron-phonon interaction,
indeed, leads to the formation of the experimentally observed phonon
spectrum, and the aforementioned limitation on value of the coupling
constant is absent.
There are superconductors with large values of the coupling
constant (e.g., λ≈2.6 for Am-Pb0.45Bi0.55 , see Wolf, 2012). The
combination of high characteristic phonon frequency and large
coupling constant can provide the high temperature superconducting
state. This is the case for the sulfur hydrides, where such
combination leads to high value of Tc (see below, Ch. IV)
Note also, that the complex structure of the phonon spectra
requires a modification of the usual methods, as it will be discussed
below (Sec.IVB).

B. Function α 2 (Ω) F (Ω) .Tunneling spectroscopy.

Tunneling spectroscopy of ordinary metals is the uniquely powerful

tool allowing us to obtain an important information about the energy
spectrum of a superconductor. To be more concrete, with use of this
technique, one can measure the value of the energy gap ,which

25
includes the case of the multi-gap structure of the spectrum (the
case is relevant to the hydrides, see Sec. VID ). Moreover, it allows
one to evaluate such important quantity as the function α 2 (Ω) F (Ω) ,

see above, Eqs.(3.1)-(3.4)

The tunneling contact contains two electrodes, separated by a
barrier. For the most interesting case of the S-I-N system (S stands
for a superconductor, N for a normal metal, and I for an
insulator),one can obtain the following relation(Schrieffer et al.,
1963;see also review by
Scalapino,1969)
−1/2
σ s / σ N =| ω | ⎡⎣| ω |2 −Δ 2 (ω )⎤⎦ (3.11)

Here σs is the tunneling conductivity; σs=∂j/∂V, j is the

tunneling current, and V is the applied voltage;Δ(ω)
corresponds to the analytical continuation of Δ(ωn) to re
al axis ; σ N is the conductivity for the N-I-N junction .The value of the

energy gap ε 0 is determined by the relation ω = Δ (iω ) .

The quantity σs/ σn can be measured experimentally (see

Eq.(3.11). Since this quantity has a sharp peak at ω=ε0, the tunneling
can be used for measuring the value of the energy gap that
corresponds to the peak in the density of states. Note that the
observation, say, of two peaks would manifest the presence of the
two energy gaps ( see below, Sec.VI D ).
The special inversion procedure, allowing to reconstruct the
function α 2 (Ω) F (Ω) and the value of µ* was developed by McMillan

and Rowell (see, e.g., Rowell,1969; Wolf, 2012). Usually the function

26
α 2 (Ω) F (Ω) contains peaks ; they correspond to peaks in phonon

density of states . Note that, in turn, the function F (Ω) can be

measured independently by neutron scattering technique. The
coincidence of the peaks obtained by these two methods (i.e.,
tunneling spectroscopy and neutron scattering) is a crucial evidence
of the fact that the pairing, indeed, is caused by the phonon
mechanism.
For the sulfur hydrides the tunneling spectroscopy has not been
performed yet. It would be interesting to develop tunneling
spectroscopy and determine the important function for
these new materials.

Note that the tunneling measurements under pressure were

performed by Zavaritskii et al.,1971, to study properties of Pb . As
mentioned above, this method has not been used so far for sulfur
hydride and the function α 2 (Ω) F (Ω) and the energy gap have not

been determined experimentally. In what follows we are using the

results of several theoretical papers describing the calculations of
α 2 (Ω) F (Ω) performed with use of the density functional formalism

(see below, Sec.IVA).

27
V. Sulfur hydrides.

A. Phonon spectrum and the electron-phonon interaction

Let us now turn our attention to the material of interest, sulfur
hydride. As was noted above, the Cooper pairing in the
superconducting state is provided by the electron-phonon interaction
and the main role is played by high frequency optical modes; this
mechanism is manifested in the large value of the isotope coefficient
for substitution of deuterium for hydrogen (see below, Ch. )

In principle, the value of the critical temperature can be evaluated

with help of Eqs. (3.1),(3.3),which contain the function α 2 (Ω) F (Ω) .

As mentioned above, the tunneling measurements allowing to

reconstruct this function have not been performed yet. Instead ,we
use results of the calculations carried out by several groups. For
example, Fig. 6 shows the function α 2 (Ω) F (Ω) calculated by Duan

et al.(2014) for both, the high Tc phase (Tc=203K) and for the
structure with the lower value of the critical temperature (Tc≈120K).
For comparison, one can see the pictures for Pb, the conventional
superconductor (Fig.5). The lattice dynamics and superconducting
properties were treated with use of the density functional theory (see,
e.g.,Baroni et al.,2001) and the quantum expresso code
(Paolo,2009). Results for the function α 2 (Ω) F (Ω) are also given in

Errea et al.,2015;Sono et al.,2016;Durajsky et al.,2015. As was

mentioned above, all these calculations are based

28
Fig. 6. The spectral function α 2 (Ω) F (Ω) and the electron-phonon
coupling constant λ(Ω) (dash lines) for the “low” Tc and high Tc
phases of the sulfur hydride. The function λ(Ω) is defined by
Eq.(3.4) with the phonon frequency Ω as the upper limit : (a) the
functions α 2 (Ω) F (Ω) and λ(Ω) for the “low” Tc phase; (b) the
functions α 2 (Ω) F (Ω) and λ(Ω) for the high Tc phase. Units:
dimensionless quantities α 2 (Ω) F (Ω) and λ(Ω) (horizontal axis) and
the phonon frequency ( vertical axis). The acoustic and optical
branches are separated at ∼15THZ .From Duan et al.,2014.

29
on the density functional formalism, but ,because of using different
codes, they differ in details, which lead to some spread in the
numerical results. Nevertheless, the calculated density of states ,
F(Ω) ,and the functions α 2 (Ω) F (Ω) are rather similar.

One can see from Fig. 6 that for sulfur hydride the function
α 2 (Ω)F(Ω) has a rather complicated structure. Indeed, in addition to

acoustic modes, it contains a number of optical phonons. Moreover,

because of high frequencies of optical modes, the phonon spectrum
is broad and extended up to 200 meV (Fig.5c).

As mentioned above (Sec. IIIA), the value of the critical

temperature can be found without invoking the coupling constant
concept (Errea et al.,2015;Akashi et al.,2015;Flores-Livas et
al.,2016;Sano et al.,2016). Such a program was carried out in the
framework of the superconducting density functional theory (see, e.g.,
Luders et al.,2005). Errea et al. ,2015, used the equation, which
1/2
looks like Eq.(3.1) with the replacement: | ω n |→ ⎡⎣ω n2 + Δ 2 (ω n )⎤⎦ in the

denominator of the integrand. This equation is valid at any

temperature. The value of Tc was calculated from such a non-linear
equation by successive approximations. With each iteration the value

of order parameter decreases, and Tc was identified as the point

30
(after up to 512 iterations) where the order parameter approaches
zero.

According to Errea et al.(2015 ) and Sano et al.(2016), an

important role is played by anharmonicity. For example, taking
anharmonicity into account noticeably shifts the value of T c ( by as
much as up to ∼ 60K : from Tc ≈260K to Tc ≈200K in the high Tc
phase. The role of anharmonicity and quantum effects was analyzed
with the use of the so-called stochastic self-consistent harmonic
approximation (SSCHA), developed by Errea et al. (2014). The
impact of the zero point motion was analyzed by Bianconi and
Jarlborg (2015a) and by Sano et al. (2016). The effects of
anharmonicity and zero point motion could be essential, because of
the small mass of the hydrogen ions. The problems raised in the
aforementioned papers deserve a further study.

As was noticed in the previous section, the concept of the coupling

constant was very fruitful for studying conventional superconductors.
Similar approach was used by Li et al.(2014),by Papaconstanopoulos
et al.(2015), to calculate Tc for sulfur hydrides. Since the value of the
total coupling constant is large λ >! 2 (see Fig. 6), they did not use
Eq.(3.8),but more general expression (Allen and Dynes,1975):

!
f1 f2Ω ⎡ 1.04 (1+ λ ) ⎤
log
Tc = exp ⎢− * *⎥
1.2 ⎣ λ − µ − 0.62 λµ ⎦
(4.1)

31
Eq.(4.1) is similar to Eq.(3.8), but the pre-exponential factor is
multiplied by the product f1f2; the functions f1 and f2 are numerically
fit for the solution valid at larger λ and have a form:

1/3
⎡ 3/2
⎤ 2 ! 2 −1
( )
f1 = ⎣1+ ( λ / Λ1 ) ⎦ ; f2 =1+ λ Ω / Ωlog −1 ( λ + Λ 2 )
2

Λ1 = 2.46 (1+ 3.8µ ∗ ) ; Λ 2 =1.82 (1+ 6.3µ ∗ ) Ω! / Ωlog

( )

The large value of the constant for the coupling to characteristic

phonon frequency leads to a high Tc close to that observed
experimentally.

The phonon spectrum of the sulfur hydrides is broad and rather

complex: it contains a number of optical and acoustic branches. On
the other hand, these branches are well separated, and this
separation allows us to develop different approach, which will be
described in the next section.

B. Generalized equation and two coupling constants.

In the more general approach (Gor’kov and Kresin,2016) the value

of Tc was evaluated analytically. As will be shown, it allows us to
compare the relative contributions of the optical and acoustic modes
for different parts of the phase diagram.

32
 The phonon spectrum contains two well-separated frequency
intervals for the optical and acoustic branches. Let us introduce the
coupling constants λopt. and λac for each of these regions and the
! ,Ω
corresponding average frequencies Ω ! and .Then the equation
opt ac

for the order parameter at T= Tc takes the following form

(cf. Eq. ( 3.5)):

( ) ! , ω − ω ⎤ Δ (ω m )
! ,ω − ω + λ D Ω
Δ (ω n ) = π Tc ∑⎡⎣( λopt − µ * ) D Ω ( )
opt n m ac ac n m ⎦
m | ωm |

(4.2)

Here

2 2
λi = ∫ dΩα (Ω)F (Ω) / Ω;Ω =< Ω
i >1/2
i
i (4.2’)

< Ω2 >i = ( 2 / λ ) ∫ dΩΩα 2 (Ω)F (Ω) ; i ≡ {opt., ac.}.

i

The critical temperature must be calculated with the use of Eq. (4.2).
This equation is the generalization of Eq.(3.5) for the presence of two
phonon groups , the acoustic and optical modes.

B. Critical temperatures for different phases.

Let us assume that in the high-Tc phase (Im-3m) λopt. ≫ λac

.We also suppose that λopt <! 1.5 . As will be shown below, these

33
 conditions are indeed satisfied. Then from Eqs. (4.2) and (4.2’) one
can obtain expression for Tc in an analytical form and, hence,
.evaluate the value of the critical temperature for the sulfur
hydrides.
Let us write Tc as

Tc = Tc0 + ΔTcac ;Tc0 ≡ Tcopt (4.3)

and assume that ΔTcac << Tc0 . Based on Eq.(4.2),with use of these

assumptions, one can obtain the following analytical expression for

the critical temperature ion the high Tc phase:(Gor’kov and
Kresin,2016):

⎡ ⎤
⎢ λac 1 ⎥T 0
Tc = 1+ 2 2 c (4.4)
⎢ λopt − µ * 1+ (π Tc0 / Ωac ) ⎥
⎣ ⎦

For T 0 , one can use the McMillan-Dynes expression (Eq.(3.8)), which

C

is valid for λopt <! 1.5 :

⎡ 1.04 (1+ λopt ) ⎤

! /1.2)exp ⎢−
Tc0 = (Ω ⎥
opt *
λ
⎢⎣ opt − µ (1+ 0.62 λ opt ) ⎥
⎦ (4.5)

34
One can use the close expression, obtained analytically by
Geilikman et al.(1975), valid also for λopt <! 1.5 :

⎡ 1+1.5λopt ⎤
0 !
T ≈ Ωopt exp ⎢− ⎥ (4.6)
c *
⎢⎣ λopt − µ (1+ 0.5λopt ) ⎥⎦

As is known that the coefficients in the Eq.(4.5) were selected to fit

the data for Nb. As for Eq.(4.6),it was obtained by analytical solution
of the Eq. (3.5). Note that these expressions are rather similar.
Indeed, by neglecting µ* for simplicity, one can write Eqs.(4.5),(4.6) in
the form: Tc≈Aexp(-1/λopt), with close values of the pre-exponential
factor.

For ordinary superconductors the values of the coupling

constants and µ* (usually µ*≈ 0.1 ÷ 0.15) can be determined from
tunneling spectroscopy measurements (see, e.g., Wolf, 2012). For
sulfur hydride, we deduce the coupling constants λopt and λac from
several theoretical calculations of α 2 (Ω) F (Ω) . Although the

corresponding theoretical results differ somewhat, they are relatively

close. The values of λopt and λac can be directly determined from
λ(Ω) (Fig.6).We estimate λopt ≈ 1.5 and λac ≈0.5; see Fig. 6b .We
assumed (see above) that λopt <! 1.5 and λopt >> λac . One can see

that the obtained values are consistent with the above

approximations. Using these coupling constants and taking the
! = 1700K and Ω
values Ω ! = 450K ( µ*≈ 0.14, which is close to
opt ac

35
that for usual superconductors and was also calculated by Flores-
Livas et al.,2016), we obtain Tc0 ≡ Tcopt =170K and ΔTcac = 45K , so that
in total Tc ≈ 215 K, in quite good agreement with the value of Tc
≈ 203 K observed by Drozdov, Eremets et al. ( 2015). The main
contribution comes from the optical phonons; this confirms the self-
consistency of our approach.

The fact that the coupling constant λopt in the cubic phase is so
large is the key factor underlying the observed high Tc ≈ 203 K.
Qualitatively, this comes about due to the ability of sulfur to retain
several hydrogen atoms in its proximity, that is, to the presence of
many light ligands near the S atoms. There are six (!) ligands in the
high Tc phase (see Fig. 3c ).

To demonstrate the importance of this point, let us evaluate the

value of Tc for the “low Tc ” phase (R3m structure). One can see from
Fig. 6a (Duan et al.,2014) that the coupling constants for this phase
are: λopt≈λac≈1. It is interesting that in this case the value of the total
coupling constant λT=λopt+λac ≈2 and is close to that in the high Tc
phase (see above).However, the relative contributions are shifted
towards to low frequencies; the value of λac is larger for the “low Tc ”
! << Ω
phase. In this case Tc < Ω ! , and one can estimate Tc within
ac opt

the usual BCS logarithmic approximation while adding the

renormalization function Z≈1+ λT into the exponent (Grimvall,1981).:

λopt λac ⎡ 1+ λT ⎤
(!
Tc ≈ Ω ) ( )
λT !
Ω λT
exp ⎢− ⎥ ( 4.7 )
opt ac
⎣ λT − µ * ⎦

36
 ! ≈ 105meV , Ω
With Ω ! ≈ 26meV (Duan et al.,2014),we obtain
opt ac
Tc ≈120K.

The transition into the high Tc phase is accompanied by re-

distribution of interaction of electrons with optical phonons and
their interaction with acoustic branches. This re-distribution is
manifested in an increase in a number of hydrogen ligands, caused
by the structural transition. This is a key factor determining record
high value of Tc .

Dividing the phonon spectrum and, correspondingly, the

electron-phonon interaction in two parts, turns out to be rather
fruitful. First of all, the value of λopt is within the range of
applicability of Eq. (4.4). Moreover, one can evaluate the relative
contribution of the optical and acoustic branches of the phonon
spectrum to Tc. For the high Tc phase the contribution from the
optical phonons comprises ~80% and only ~20% is due to the
acoustic part. The impact of acoustic phonons is noticeably smaller
than that of optical branches (45K vs. 170K), but still is essential.

The method proposed above can be of relevance for other

materials as well. A promising example is calcium hydride, CaH6
(see below, Sec. VIIA ).

It is known that for a number of superconductors the value of the

electron-phonon coupling constant is large. Among them Pb

37
(λ≈1.55), Hg (λ≈1.6), AmPb0.45Bi0.55 (λ≈2.6), see, e.g., Wolf , 2012.
However, because of low values of characteristic phonon
frequencies, the values of the critical temperature for them are not
large . An uniqueness of the hydrides is that they combine the strong
coupling, especially to optical modes, with high values of the
characteristic frequencies.

V. Isotope effect

According to Drozdov et al.(2015) ,the substitution of deuterium for

hydrogen noticeably affect the value of the critical temperature.
Observation of this isotope effect is of fundamental importance, since
it proves (a) that the high Tc state is caused by the electron-phonon
interaction, and (b) that namely the high frequencies hydrogen modes
determine the value of Tc. Indeed, the optical modes are mainly due
to motion of hydrogen, whereas for the acoustic modes the
participation of sulfur ions prevails. Therefore the magnitude of the
isotope shift for the deuterium for hydrogen substitution reflects
indirectly the relative contributions of the each group (the optical vs.
acoustic ones) into the observed Tc.
Defining the isotope coefficient via the relation Tc ∝ M −α one can

obtain the following expression for α (see,e.g.,Gor’kov and

Kresion,2016):

( ! ∂Ω
)(
α = − ( M / Tc ) ∂Tc / ∂Ω ! / ∂M ) ( 5.1 )

38
Since the deuterium substitution affects the optical modes, one can
write the expression (5.1 ) in the form ( in the harmonic
! ∝ M −1/2 ):
approximation; then Ω
! / T ∂T / Ω
(
α = 0.5 Ω )( ! ) (5.2)
opt c opt

The value of the isotope coefficient in the high Tc phase can be

calculated with use of Eqs. (4.4) and (5.2). As a result, we obtain:
⎡ ⎤
1⎢ λac ρ2 ⎥
α ≈ 1− 4 (5.3)
2⎢ λopt ( ρ 2 +1)2 ⎥
⎣ ⎦

! / π T 0 . With λopt ≈ 1.5,λac ≈ 0.5 , Ω

Here ρ = Ω ! ≈450K ,we obtain
ac c opt

α≈0.35 , in a good agreement with the experimental data in Fig.4 .

One should use the different expression, Eq.( 4.7), for Tc in the
“low Tc ” phase. Then, with use of Eq.( 5.2 ), one can find α≈0.25,
which is noticeably smaller than that for the high Tc phase. Note that
the agreement between the value obtained from Eq.(5.3) and the
experimental data is rather good for the high Tc phase. As for the
“low” Tc phase, the data are not so well determined. It would be
interesting to perform the measurements at lower pressures, at the
values more distant from the region of the transition into the high Tc
phase.
As was mentioned above (Sec. IIB) , at low pressures a formation of
the H3S phase occurs (see Eq. (2.1)). According to Akashi et al.

39
(2016) , the transformation occurs through intermediate structures. In
other words, we are dealing with the coexistence of phases and the
percolation scenario, so the percolation threshold corresponds to
formation of the so-called infinite cluster, that is, to the metallic state.
As a result and the R3m phase contains some inclusions. Then one
can expect the pressure dependence of the isotope coefficient inside
of the R3m phase.

Note that the usual analytical derivation of the value of the isotope
coefficient is carried out under the assumption that the ionic mass is
the continues variable. It is essential that the obtained value of α
appears to be independent of M. It allows to use it for different values
of the ionic mass, even the latest has discrete values. Qualitatively
this means that the shift in the value of Tc stays the same within the
phase, so that the dependences Tc (P) for different isotopes are
parallel to each other. For example, in our case the shift is described

by the relation : TcH / TcD = ( M D /M H ) .In the high Tc phase α≈0.35 (see

above),and ,therefore , TcH / TcD = 1.25 . Since TcH = 203K , we obtain

TcD ≈165 K. This value is in a rather good agreement with the

measurements by Drozdov et al.(2015)

The value of the isotope coefficient in the high Tc phase is

relatively large and it reflects the fact that the pairing in this phase is
dominated by the optical H-modes, whereas in the “low Tc” phase
the contributions of the optical and acoustic modes are comparable.
The impact of the isotopic substitution in the region of smaller Tc is

40
weaker than in the high Tc phase. A smaller α is in agreement with
the larger role played by the optical phonons in the cubic high Tc
phase

Notice that the value of α can be affected by the anharmonicity

! ,although
(Errea et al.,2015) and by the dependence of µ* on Ω opt

the last contribution is of the order of (µ*/λopt)2 and is small. However,

the main conclusion that the value of the isotope coefficient depends
on pressure and is different in different phases, remains valid and
reflects the relative contributions of the optical and acoustic modes.

VI. Energy spectrum of the high Tc phase: two-gap structure and

non-monotonic dependence of Tc.

A. Structural transition. High Tc phase.

As was mentioned above, the pressure dependence of is highly

asymmetric relative to its maximum value Tc;max =203K. The

value of Tc ≈120K at P≈125GPa sharply increases to Tc ≈200K at
P≈150GPa (Drozdov et al.,2015; Einaga et al.,2016;) .A rapid
increase in Tc is followed by a slow decrease at P>Pcr.≈150GPa
(Fig.4). Remarkably, the structural transition into the high Tc phase
takes place somewhere in the same pressure interval where it is
accompanied by a sharp increase in the value of the critical
temperature. Currently, it is generally accepted that the Bravais
lattices of the high- Tc and the low- Tc phases belong to different

41
symmetries (the Im-3m group for the high Tc cubic phase and the
trigonal R3m symmetry group for the low-Tc phase) and that the
structural phase transition between them occurs at a pressure
somewhere in between P≈150GPa and P≈125GPa . The
sharpness of the increase prompts the question whether transition
into the high- Tc phase could be the first order transition. This
scenario was discussed by us (Gor’kov and Kresin,2017) with use of
the group theory and taking into consideration the impact of lattice
deformations. The picture is similar to those considered by Larkin and
Pikin (1969) and later by Borzykin and Gor’kov (2009): the coupling to
lattice can transform the second order transition into the transition of
the first order. The idea of the first order transition allows us to
explain self-consistently the slow decrease in Tc with an increase in
pressure above the pressure Pcr corresponding to the maximum of Tc

≈203K . The appearance of the two-gap spectrum is an important

ingredient of the picture (see below, Sec VIC)

According to the band structure calculations (Duan et al.,2014;

Flores-Levas et al.,2016; Errea et al.,2015,2016; Akashi et
al.,2015;Neil and Boeri,2015), the high Tc compound is
characterized by broad energy bands ( large Fermi surface) and
strong interaction between electrons and high frequency optical
phonons. The calculated values of Tc and the isotope coefficient are
in a good agreement with the experimental data (see above,
Sec.IVC, Ch.V ).

42
 Meanwhile, the calculations also revealed the presence in the

high- TC phase of small Fermi-pockets. The importance of their

existence was emphasized by Bianconi and Jarlborg (2015a,b,c),who

suggested that the pockets play an especial role in increasing Tc .
Note, however, that the analysis of the electron-phonon mediated
pairing on pockets should be carried out with a considerable care.
Let us discuss this point in more details.

B. Migdal adiabaticity criterion and small pockets.

The complex structure of the Fermi surface with small pockets

emerging in addition to several large Fermi sheets is not uncommon
for many novel superconductors such, for instance, as the high- Tc
oxides , low-dimensional organic conductors, the so-called heavy
fermions (see, e.g., Gor’kov,2012). Here we focus on the possible
impact of small pockets on the superconductivity in hydrides.

The main equation, Eq. (3.1) is valid if the so-called adiabatic

! /E
parameter Ω( ) ( ! / E <<1 ;Migdal,1958, see also
is small ( Ω )
F F

review by Scalapino , 1969). Then one can neglect all higher order
corrections (so called “vertex corrections” ),containing the products
of the matrix elements of the electron-phonon interaction. Then the
right side of equation for the order parameter is linear in the
coupling constant λ (see Eq.(3.1) and also Eq.(3.3)).The value of
the coupling constant is expressed by Eq. (3.10). According to

43
 Migdal (1958), the correction to Eq.(3.1) contains an additional term
! /E .
∝λ3 Ω
( )
F

! << E
Therefore, the inequality Ω allows us neglect the higher
F

order corrections . If Ω >! EF one should include the contribution of all

higher terms . But the rigorous calculations of even the second term
(Grimaldi et al.,1995) appear to be the non-trivial task. At this point
we meet with the problem, which at the present time remains
unresolved.

The case of the weak electron-phonon coupling (λ<<1) is the

exception (see Gor’kov, 2016). Because of the smallness of λ, the
vortex corrections can be neglected .

! / E << 1 is satisfied for most conventional

The condition Ω ( )
F

superconductors, since in usual metals the Fermi energy is large

! ≈ Ω . However, this is not
compared with the Debay energy EF >> Ω D

the case for pockets in the sulfur hydrides, since for these materials
! ≈ (1.5 − 2)100meV ,
the characteristic frequency of optical modes Ω opt

whereas the Fermi energy of a pocket is of order ≈ (40-50) meV. The

electron-phonon interaction on the pockets can be rigorously treated
only in the case of weak coupling, that is, if the corresponding
coupling constant λP<<1.

44
 C. Broad bands and pockets.

The calculated spectrum of electrons display small pockets only

inside of the high- Tc phase and it may be tempting to relate the high
value of Tc to the appearance of the pockets (Bianconi and Jarlborg,
2015a,b,c; Quan and Pickett, 2016). In this scenario, the major pairing
interaction occurs on the pockets. As for the electron-phonon
interaction on the larger bands, it is weak and is playing only a
secondary role.

At this point it is worth noting that the calculations of Tc performed

assuming the prevailing role of the large bands and sufficiently strong
coupling are in a good agreement with the experimental data (see
Ch. IV), so that there is no special need for modifying the picture.
Besides of that, if one is trying to assign the leading role to pockets,
then it is clear that the on- pocket interactions should be rather strong
in order to provide high Tc . However, in this case the rigorous
treatment is not known, because of the violation of the Migdal
theorem (see above). On the other hand, if the main contributions
into the interaction were coming from the large bands, then the
contribution of the pockets could be assumed weak, and the case can
be analyzed self-consistently.

Note also-and this is a strong argument- that if the leading role of

pockets were due to a peak in their density of states ,this would
produce a pre-factor in the expression for Tc of an electronic origin.
However, such a pre-factor cannot depend on the ionic mass, in the

45
strong contradiction with the observed isotope effect (Drozdov et al.,
2015).

D. Two-gap spectrum. Slow decrease in Tc at P>Pcr

In the superconducting state the pockets are characterized by the

energy gaps in their electronic spectrum. Below we consider such
two-gap model with one gap corresponding to the broad band and the
second gap describing the excitations on the pocket.

The two-gap model was introduced shortly after the creation of

the BCS theory (Suhl et al.,1959; Moskalenko,1959). From that
follows, we stipulate that under the notion of the two gap spectrum we
mean the presence of two peaks in the density of states.

Each band has its own set of the Cooper pairs. Since a single
pair is formed by two electrons with equal and opposite momenta,
one can neglect pairing between electrons belonging to different
bands. Indeed, in general, the electrons on the Fermi level, which
belong to different bands, have different values of the momenta.
However, in the two-band model, the absence of the interband
pairing does not mean that the pairing within each band is insensitive
to the presence of the other band. Indeed, the presence of the
second band gives rise to an additional pairing channel. Namely, the
electron originally located on the first band can radiate phonon and
make the virtual transition into the second band. The second
electron can absorb the phonon and also make transition into

46
second band forming the pair the first electron. Therefore, owing to
the interband electron-phonon scattering the on-the-pocket electrons
can form the Cooper pairs on the large Fermi surface and vice versa.

Let us stress one important point. As noted above, the two

gap model was introduced shortly after the creation of the BCS
theory. Nevertheless, the two gap phenomenon has not been
essential for the conventional superconductors. This is due to the
large coherence length; more specifically, the inequality l<< ξ ( l is
the mean free path) ,which hold for usual superconductors , leads to
the averaging caused by the interband impurity scattering. As a
result ,the two gap picture is washed out and the usual one gap
picture is applicable.

The two-gap spectrum was observed for the first time in the
Nb-doped SrTiO3 system (Binnig et al,1980) with use of the scanning
microscope (STM) technique. The second gap appears as a result of
doping and filling the second gap. At present, the two-gap picture is
important feature of the novel superconducting systems, and this is
due to their short coherence length. It has been observed in the
cuprates (Greek et al.,1988),in MgB2 (Uchiyama et al., 2002 ;
Tsuda et al.,2003), see review by Kresin et al.(2014).
At the formulation of the two-gap model for the high- Tc phase of
sulfur hydrides one can introduce three coupling constants: λL ,
responsible for strong electron-phonon interactions on the large
band , λP <<1 ( weak coupling on the pockets), and λLP <<1,
describing the transitions from large band electrons to the pairing

47
states on the pockets. The coupling constants λL , λP and λLP are
described by Eq.(3.10);the constant λLP contains the matrix
element describing the interband transitions caused by the electron-
phonon interaction. Note also that, because of the interband
transitions, the system has the common temperature of the
superconducting transition Tc. In addition, their presence is beneficial
for superconductivity.
Performing the calculations (Gor’kov and Kresin,2016),one can
show that the shift in Tc caused by the presence of pockets is
proportional to the density of the states on the pockets:ΔTc ∝νF

Return to the problem of the strong asymmetry in the pressure

dependence of Tc relative to the position of its maximum value (at
Tc;max =203K ( at Pcr.≈150GPa) posed in Sec.IIC. Assume, as
above, that the sharp increase in Tc (from Tc ≈120K to Tc;max≈

200K ) is the result of the first order structural transition into high TC

cubic phase. This phase is characterized by the coexistence of

broad band (responsible for the large part of the Fermi surface) and
small pockets.
As mentioned above, the interaction between the large band and
the pocket leads to the shift in the temperature of transition
ΔTc = Tc - Tc;0 ,which is proportional to the density of states on the
pocket νP(EF)∝mPPF;poc where mP and PF;poc are the effective
mass and momentum for the pocket’s states, Tc;0 is the value of
the critical temperature in the absence of the pockets. It is essential
whether the pockets appear instantly, as a result of the

48
 discontinuous first order transition or the transition into the high- Tc
phase is either of the second order or is of a topological nature at
which the pocket’s size would grow continuously with the further
increase in pressure.

Above we have given arguments in favor of the first order transition,

which is accompanied by the emerging singularity in the density of
states in the form of pockets. The further increase in pressure leads
to shrinking of the pocket with an effective decrease in their Fermi
momentum PF;poc and the corresponding depression of the two-gap
picture. Since the two-gap scenario is beneficial for superconductivity,
such a depression leads to a decrease in Tc .This explains the

observed slow decrease in TC after the transition; the small scale of

the decrease in TC at P>Pcr is related to small values of λP and λPL

The two-gap spectrum and its evolution with pressure, including the
decrease in the amplitude of the second gap at P>Pcr must be
confirmed by future tunneling experiments. The presence of the
second energy gap will be manifested as the second peak in the
density of states.

VII. Other hydrides

S-H system appears to be the first hydride, which displays record

high value of Tc. We described above (Ch.II) the development of the

49
field. Let us discuss here several other studied hydrides, which
display interesting and promising properties.

As was emphasized in the Introduction and Ch.IV, the theoretical

studies containing predictions of specific hydrides along with values
of Tc and the pressures, deserve a special credit. In this Chapter we
will be talking about promising compounds. As was mentioned above,
the group of Y.Ma made a very important prediction related to sulfur
hydrides (Li et al,2014). This group made several other interesting
predictions, which are waiting their future confirmations (Peng et
al.,2017;see the review by Zhang et al.,2017). They predicted the
high Tc superconducting state for calcium hydride (Tc≈240K; Wang et
al,2014) and for YH6 (Tc≈264K; Li et al.,2015).Note that the
compound MgH6 ,similar to CaH6 was also studied (Feng et al.,2015)
and the values of Tc ≈ 260K at P >! 300GPa were predicted.
Even higher values of Tc were obtained by Szczesniak and Durajski
(2016). The next section contains the description of the calcium
hydride and MgH6; these are examples of the future high Tc
compounds.

A. Calcium hydride; MgH6

Calcium hydride, CaH6, has been analyzed by Wang et al.
(2012) and looks very promising. Wang et al. evaluated its structure
and the function α 2 (Ω) F (Ω) (Fig.7 ).Based on numerical solution of

Eqs.( 3.1),(3.3), they predicted that at pressures P≈150GPa the

value of Tc will be higher than that for H3S.

50
 The value of Tc for CaH6 can be evaluated with use of Eq.(4.2
).Indeed, in accordance with an approach described in Sec. IVB , the
electron-phonon interaction can be separated in two parts. From Fig.
! =820 sm-1 , Ω
7 one can determine that λopt ≈2.1, λac≈0.6, Ω !
opt ac

=350sm-1. Correspondingly, one can write that Tc = Tc0 + ΔTcac ;Tc0 ≡ Tcopt
is determined by contribution of the optical modes. However, because
of large value of λopt, the MacMillan-Dynes expression for Tc0 ,
(Eq.(3.7),is not applicable (it is valid for λopt <! 1.5 ). One can use the

modified MacMillan equation, Eq. (4.1) , valid for larger λ . Another

option is to use the analytical expression, valid for any value of the
coupling constant (Kresin,1984;see also review by Kresin and
Wolf,2009):

!
0.25Ω
Tc0 = opt
1/2
⎡e 2/λeff −1⎤ (7.1)
⎣ ⎦
λeff = ( λopt − µ *)⎡⎣1+ 2µ * +λopt µ * t ( λopt )⎤⎦;t(x) = 1.5exp (−0.28x )

With use of Eq. (7.1) and the parameters for CaH6, we obtain Tc
≈230K. This value contains the contributions of the optical ( Tc0 ) and

acoustic (ΔTc) modes: Tc0 ≈180K, and ΔTc≈50K.Therefore, the optical

and acoustic modes contribute 78% and 22% to the total value of the
critical temperature, correspondingly. As for the isotope coefficient,
one can obtain from Eq. ( 5.2 ) the value α≈0.36; it is close to that for
the high phase of the sulfur hydrides.

51
Fig. 7. Superconducting calcium hydride:(a) structure of CaH6 ; (b)
the spectral function α 2 (Ω) F (Ω) for CaH6. From Wang et al., 2012.

The hydride MgH6 studied by Feng et al.,2015, and by Szczesniak

and Durajski,2016 has properties similar to those for the calcium
hydride. The structure contains the sodalite-like hydrogen cage with
interstitial Ca (Mg) atoms

52
(Fig.7). This is not occasional, since the Mg and Ca atoms have
similar chemical properties. As mentioned above, the predicted
values of Tc are even higher than for sulfur hydrides.

B. Palladium hydride.

In principle, palladium hydride ( Pd-H), is not a new

superconductor. It has been discovered in 1972 ( Stritzer and
Buckel,1972 ); its Tc≈8-10K. This hydride was known by the value of
the isotope coefficient, which appears to be negative. This
phenomenon was explained by strong anharmonicity
(Ganzuly,1973; Klein and Cohen,1992), namely, by the peculiar
dependence of the phonon frequency on ionic mass.
According to the study by Syed et al.,2016, one can
observe the superconducting state of Pd-H at higher temperatures
(Tc ≈54K for Pd-H and Tc ≈60K for Pd-D; the isotope coefficient is
negative, as for the usual compound).It is important that such an
increase in Tc has been observed at the ambient pressure. It
occurs, thanks to special sample preparation, namely, because of
fast cooling of the hydride.
The result looks interesting, because the superconducting
state persists up to high temperature at ambient pressure. The
authors observed a drastic drop in resistance. However, the
Meissner effect has not been demonstrated yet, as well as the
impact of the magnetic field on Tc. Of course, it would be interesting
if future experiments confirm the presence of superconductivity at
elevated temperatures.

53
 C. Transition of ice under high pressure and by doping

In general, the transition of an insulator into metallic state can

proceed through two channels: either via doping (e.g., high Tc
cuprates, fullerides), or by increase in pressure (e.g., hydrides).
The recent paper (Flores-Livas et al., 2016) is interesting, because
the authors consider theoretically the transition, caused by
combination of these channels. They studied the properties of H2O,
which can be transferred into solid phase (ice) by applied pressure
( up to P≈150GPa). The crystal continues to be an insulator at this
pressure. It has been proposed that the sample in the solid phase
can be doped by nitrogen. The calculation demonstrates that, as a
result of such a doping under the pressure, the material becomes
metallic and even superconducting with rather high value of Tc
≈60K.
Note that at ambient pressure and also at low pressure (up to
P≈110GPa) the oxygen ion has four hydrogen neighbors (so called
phase I). Two of these neighbors are covalently bonded with oxygen
and form the H2O molecule ,and other two ions formed additional
hydrogen bonds. The lengths of the bonds are different and the
structure is asymmetric. However at higher pressures (of the order
of P≈300GPa) the so-called ice -X phase is formed and it is
characterized by symmetric O-H bonds ( see Goncharov et al.,
1999)
As was noted above, at P≈150GPa the ice crystal is still in the
insulating state. To prompt a transition into metallic state one needs
to use doping and it turns out that the nitrogen is the best dopant.

54
The nitrogen for the oxygen substitution leads to the hole
conductivity. Moreover, the transition into a metallic state is
accompanied by changes in the phonon spectrum. All these
changes provide the transition into superconducting state. The
calculations show that best value of the superconducting
parameters correspond to relatively low doping (4-6%). The
calculated α 2 (Ω) F (Ω) function leads to the value of Tc ≈60K. Of

course, this value is below Tc ≈203K observed for H3S phase of

sulfur hydride, but still is rather high.
The idea combining high pressure and doping is elegant and
looks promising. One has to wait whether the future experiments will
confirm this interesting prediction.

D. Organic hydrides

The presence of hydrogen and corresponding high vibrational

frequency is very beneficial for the formation of the
superconducting state. As was noted in the previous section, it can
be manifested even at ambient pressure. In connection with this,
the recent discovery of superconductivity in the organic
compound, consisting of C and H elements, so-called poly(p-
phenylene) and doped with potassium (Wang et al.,2017) looks
very promising. The value of Tc is rather high (Tc≈120K); this
value is the highest among organic superconductors. Organics is a
relatively young family of superconducting materials. The first
organic superconductor was discovered in 1980 by D. Jerome:

55
 the complex material (TMTSF)2PF6 displayed the property at
Tc≈0.9K under the pressure of around 9 kbar. The recent
discovery was made by Wang et al. in the high pressure
laboratory, but the effect was observed at ambient pressure. The
mechanism of superconductivity in this new material and an
impact of the hydrogen bonding and high frequency modes should
be studied in full details, As a whole, this new class of organic
hydrides looks very important and deserves a further study.

VIII. Main challenges.

The recent discovery of the record-breaking high Tc compound,
sulfur hydride, signifies the arrival of a novel family of high
temperature superconductors: the hydrides. Even higher values of Tc
can be expected. That is, it finally becomes perfectly realistic to
envision the detection of superconductivity at room temperatures.
The search for novel hydrides with still higher values of Tc (including
CaH6, YH6) is a very important direction of future research.
More detailed X-ray diffraction of sulfur hydrides will establish
the position of the sulfur atoms with high accuracy and clarify the
nature of the phase transition between the “low Tc” and “high Tc”
phases.
Development of tunneling spectroscopy of the high Tc hydrides
is another important forthcoming task. Because of the large values of
Tc and the energy gap, the tunneling I(V) characteristic needs to be
measured for a wider energy interval as compared to that for
conventional superconductors. But similar measurements have been

56
performed for the high Tc cuprates by Aminov et al. (1994) and
Ponomarev et al. (1999) with the use of the break junction technique,
and by Lee et al. (2006) by scanning tunneling spectroscopy, see
review by Kresin and Wolf (2009) . Therefore tunneling spectroscopy
should be successfully applicable for sulfur hydrides as well. As a
result, it will be possible to reconstruct the α 2 (Ω) F (Ω) function as

well as to determine the Coulomb pseudopotential µ*. Tunneling

spectroscopy also can be employed for measuring other important
parameters of the system, including the energy gap, and for
observing the multi-gap structure and its evolution with pressure (see
Ch. VI).
And, of course, following detailed studies of the structure of the
high Tc phase under pressure, there remains the most intriguing
question: is it possible to create analogous structures stable at
ambient pressure?

IX. Concluding remarks.

The discovery of pressure-induced superconductivity in the

hydride family opens new prospects in research on high temperature
superconductivity. In this Colloquium we have focused mainly on the
theoretical aspects of this new development.

From the fundamental point of view, it is remarkable that the

high Tc superconducting state manifests itself under such extreme
experimental conditions. Sulfur hydrides offer a remarkable
combination of strong electron-phonon coupling and high optical-

57
phonon frequencies. As long as the Migdal adiabaticity criterion is not
violated, observations of superconductivity at even higher
temperatures now can be anticipated.

There are, however, novel features, which necessitate

significant deviations from the conventional Migdal-Eliashberg
approach. First of all, the hydride phonon spectra are quite broad (up
to 200 meV) and contain both optical and acoustic modes. We
propose that the electron-phonon interaction can be treated with the
use of the general equation (4.1) by employing two coupling
constants λopt and λac together with two corresponding average
frequencies. This leads to an analytical expression for Tc applicable
to a number of cases, and permits analysis of other relevant factors
for different phases.
The experimentally observed isotope effect (deuterium-hydrogen
substitution) turns out to be not universal in the sense that the isotope
coefficient depends on the pressure and has distinct values for
different phases. We show that this reflects the relative contributions
of optical and acoustic modes.
We point out that the sharp increase in Tc (from ≈120K up
to≈200K over a narrow pressure interval near ≈150GPa) is a
signature of a first-order structural transition into the high Tc phase.
This picture also explains the curiously slow decrease in Tc at P>Pcr.
It has been predicted in a number of theoretical papers that the
transition into the high Tc phase is accompanied by the appearance
of additional small Fermi pockets. As a consequence, a two-gap

58
structure appears . It should be observable by tunneling
spectroscopy, which also will be useful for determination of the
characteristic function α 2 (Ω) F (Ω) .

It may be expected that there should exist other hydrides

capable of displaying a high Tc superconducting state under
pressure, possibly with even higher values of Tc all the way up to
room temperature. But the most challenging question relates to the
possibility of creating superconducting structures stable at ambient
pressure. In this regard, we are encouraged by the recent
observation by Syed et al.,2016, of the superconducting state of Pd-H
at temperatures ≈54 K and ≈60 K for Pd-D.

Acknowledgements

The authors thank M. Eremets, A. Drozdov, M. Calandra, M.Einaga,

and Y.Ma for interesting and stimulating discussions. The work of
LPG is supported by the National High Magnetic Field Laboratory
through NSF Grant No. DMR-1157490, the State of Florida and the
U.S. Department of Energy. The work of VZK is supported by the
Lawrence Berkeley National Laboratory, University of California at
Berkeley, and the U.S. Department of Energy.

59
References

Abrikosov, A., Gor’kov, L., Dzyaloshinski, I., 1963, Methods of

Quantum Field Theory in Statistical Physics, Dover, NY

Akashi, R., Sano, W., Arita, R.. Tsuneyuki., R., 2016, Phys. Rev. Lett.
117, 075503

Akashi, R., Kawamura, M., Tsuneyuki, S. Nomura, Y. Arita. R., 2015,

Phys. Rev. B 91, 224513

Allen, P., Dynes, R., 1975, Phys. Rev. B 12, 995

Aminov B., Hein M., Miller G., Piel H., Ponomarev Y., Wehler
D.,Boockholt M., Buschmann L., Gunherodt G., 1994, Physica C 235,
240,1863

Ashcroft, N.,1968, Phys. Rev. Lett. 21, 1748

Ashcroft, N., 2004, Phys. Rev. Lett. 92, 187002

Bardeen, J., Cooper, L., Schrieffer R., 1957, Phys.Rev. 108, 1175

Baroni, S., de Gironcoli, S., Dal Corso, A., Gianozzi, P., 2001, Rev.
Mоd. Phys. 73, 515

Barzykin,V., Gor’kov,L., 2009, Phys.Rev.B 79,134510

Bednorz,G., Mueller,K.,1986, Z. Physik B 64,189

Bernstein,N., Hellberg, C., Johannes, M., Mazin, I., Mehl, M., 2015,
Phys. Rev. B 91, 060511 (R) .

Bianconi, A., Jarlborg, T., 2015, Novel Superconducting Materials 1,

15

60
Bianconi, A., Jarlborg, T., 2015, Europhys. Lett. 112, 37001

Binnig, G., Baratoff, A., Hoenig, H., Bednorz, G., 1980, Phys. Rev.
Let.. 45, 1352

Born, M., Huang, K., 1954, Dynamic Theory of Crystal Lattices,

Oxford, NY

Born, M., Oppenheimer, R., 1927, Ann.Phys. 84, 457

Brockhouse,B.,Arase,T.,Caglioti,K., Rao, K., Woods, A., 1962,

Phys.Rev. 128,1099

Browman, Y., Kagan, Y., 1967, JETP 25, 365

Carbotte, J., 1990, Rev. Mod. Phys. 62, 1027

Drozdov, A., Eremets, M., Troyan,2014,arXiv:1412.0460

Drozdov, A., Eremets, M., Troyan, I., Ksenofontov, V., Shylin, S.,
2015a, Nature 525, 73

Drozdov, A., Eremets, M.,Troyan,I.,arXiv : 2015b, 1508.06224

Duan, D., Liu, Y., Tian, F., Li, D., Huang, X., Zhao, Z., Yu, H., Liu, B.,
Tian, W., Cui, Ti., 2014, Scientific Rep. 4, 6968

Durajski, A., Szczesniak, R., Li, Y., 2015, Phys. C 515, 1

Dynes R., 1972, Solid State Comm. 10, 615

Einaga,M.,Sakata,M.,Ishikawa,T.,Shimizi,K.,Eremets,M.,
Drozdov,A.,Tpoyan,I.,Hirao,N.,Ohishi,Y.,2016,Nature Physics 12, 835

Eisaki, H., Uchida, S., Davis, J., 2006, Nature 442, 546

M.Eremets, A.Drozdov, Uspechi 186,1257 (2016)

Eremets, M., Troyan, I., 2011, Nature Mater. 10, 927

61
Eremets, M., Troyan, I., Drozdov, A., 2016, ArXiv: 1601. 04479 v1

Eremets, M., Troyan, I., Medvedev, S., Tse, J., Yao, Y., 2008,
Science 319, 1506

Errea, I., Calandra, M., Pickard, C., Nelson, J., Needs, R., Li, Y., Liu,
H., Zhang, Y., Ma, Y., Mauri, F., 2015, Phys Rev. Lett. 114, 157004

Errea, I., Calandra, M., Pickard, C., Nelson, J., Needs, R., Li, Y., Liu,
H., Zhang, Y., Ma, Y., Mauri, F., 2016, Nature 532, 81

Errea, L., Calandra, M., Mauri, F., 2014, Phys. Rev. B 89, 064302

Feng,X.,Zhang,J.,Gao,G.,Liu,H.,Wang,H., 2015, RSV Adv. 5, 59292

Flores-Livas, J., Sanna, A., Davydov, A., Godecker, S., Marques, M.,
1916, ArXiv: 1610.04110

Flores-Livas, J., Amsler, M., Heil, C., Sanna, A., Boeri, L., Profeta, G.,
Wolverton, C., Goedecker, S., Gross, E., 2016, Phys. Rev. B 93,
020508

Flores-Livas,J., Sanna, A., Gross,E., 2016, Euro. Phys. J. B 89, 63

Ganzuly, B., 1973, Z. Phys. 265, 433

Gao, G., Oganov, A,m Bergara, A., Martinez-Canales, M., Cui, T.,
Iitaka, T., Ma, Y., Zou, G., 2008, Phys. Rev. Lett. 101, 107002

Gao, Y., Xue, Y., Chen, F., Xiang, Q., Meng, R., Ramirez, D., Chu, C.,
Eggert, J, Mao, H., 1994, Phys Rev. B 50, 4260

Geerk, J., Xi, X., Linker, G., 1988, Z. Phys. B 73, 329

Geilikman, B., 1971, J. Low Temp. Phys. 4, 189

Geilikman, B., 1975, Uspekhi 18,190

Geilikman, B., Kresin, V., Masharov, N., 1975, J. Low Temp. Phys. 18,
241

62
Goedecker, 2004, J. of Chem. Phys. 120, 9911

Goncharov,A., Struzhkin,V., Mao,H., Hemley,R., 1999,

Phys.Rev.lett.83,1998

Goncharov, A., Lobanov,S.,Kruglov,I., Zhao,X., Chen,X., Oganov,

A.,Konopkova, Z., Prakapenka, V.,2016, Phys.Rev. B 93,174105

Goncharov, A., Lobanov,S., Prakapenka,V., Greenberg,E., 2017,

Phys.Rev.B 95,14011

Gor’kov, L., 1958, JETP 7, 505

Gor’kov, L., 2012, Phys. Rev. B 86, 060501

Gor’kov, L., 2016, Phys. Rev. B 93, 054517

Gor’kov, L., G. Gruner, G.,1989, In Charge Density Waves Ch. 1, 1–

14, North-Holland

Gor’kov, L., Kresin V., 2016, Nature, Sci. Rep. 6, 25608

Gor’kov, L., Kresin V., 2017,arXiv:1705.03165

Grimaldi, G., Pietronero, L., Strassler, S., 1995, Phys. Rev. Lett. 75,
1158

Grimvall, G., 1981, The Electron-Phonon Interaction in Metals, North-

Holland, Amsterdam

Huang.X.,Wang,X.,Duan,D.,Sundqvist,B.,Li,X.,Huang,Y., Li,F.,
Zhou,Q.,Liu,B.,Cui,T.,2016, arXiv:1610.02630

Jarlborg, T., Bianconi, A., 2016, Sci. Rep. 6, 24816

Jerome, D.,Mazaud,A., Ribault,M.,Bechgaard,K., 1980, J.Phys. Lett.

(Paris) 41,L95

63
Jin, X., Meng, X., He, Z., Ma, Y., Liu, B., Cui, T., Zou, G., Mao, H.,
2010, Proc. NAS 107, 9969

Kamerlingh-Onnes, H., 1911, Leiden Commun. 124C

Karakosov,A., Maksimov,E., Mashkov,S., 1976,JETP,41,971

Klein, B., Cohen, R., 1992, Phys. Rev. B 45, 12405

Kresin V., 1987, Phys. Lett. A 122, 434

Kresin, V., Gutfreund, H., Little, W., 1984, Solid State Commun. 51,
339

Kresin,V., Morawitz,H., Wolf,S., Superconducting State;Mechanisms

and Properties,2014, Oxford Univ.Press

Kresin, V., Wolf, S., 2009, Rev. of Mod. Phys. 81, 481

Larkin, A.,Pikin, S.,1969,JETP 29,891

Lee J., Fujita K., McElroy K., Slezak J. A., Wang M., Aiura
Y., Bando, H., Ishikado, M., Mazui, T., Zhu, J., Balatsky, A.,
Eisaki, H.,Uchida, S., Davis, J., 2006, Nature 442, 546
Li, Y., Hao, J., Liu, H., Li, Y., Ma., Y., 2014, J. Chem. Phys. 140,
174712

Li, Y., Hao, J., Liu, H., Tse,J., Wang,Y., Ma., Y.,2015, Sci.Rep.
5,9948

Louie, S., Cohen, M., 1977, Solid State Commun. 10, 615

Luders,M.,Margues,M.,Lathiotakis, N., Floris, A.,

Profeta,G.,Fast,L.,Continenza,A.,Massidda, S., Gross, E.,
2005, Phys. Rev. B72,024545

Mao, H., Hemley, R., 1994, Rev. of Mod. Phys. 66, 671

Massa, W., 2004, Crystal Structure Determination, Springer, Berlin

64
McMahon, J., Morales, M., Pierleoni, C., Ceperley, D., 2012, Rev. of
Mod. Phys. 84, 1607

McMillan, W., Powell J., 1965,Phys.Rev.Lett. 14,108

McMillan, W., 1968, Phys. Rev. 167, 331.

McMillan, W., Powell J.1969. in Superconductivity, Parks R., Ed. v.1,

Marcel Dekker, NY

Migdal A., 1960, JETP 37,176

Moskalenlo, V., 1959, Fizika Metal Metalloved 8, 503

Neil, C., Boeri. L. , 2015, Phys. Rev. B 92, 06058 .

Nicol, E., Carbotte, J., 2015, Phys. Rev. B 91, 220507 .

Paolo, G., 2009, J. Phys.: Con. Matter 21, 395502

Papaconstantopoulos,D., Klein, B. Mehl, M., Pickett, W., 2015, Phys.

Rev. B 91, 184511

Peng,F.,Sun,Y.,Pickard, C.,Needs,R.,Wu,Q.,Ma,Y., 2017, Phys..Rev.

Lett.119,107001

Ponomarev, Y., Tsokur, E., Sudakova, M., Tchesnokov, S., Habalin,

S., Lorenz, S., Hein S., Muller, S., Piel, S., Aminov, S., 1999, Solid
State Comm. 111, 513

Quan, Y., Pickett, W., 2016, Phys. Rev. B 93, 104526

Rowell, J., 1969, in Tunneling Phenomena in Solids, Burnstein, E.,

Lindqvist, S., Eds., Plenum, NY

Sano, W., Korensune, T., Tadano, T., Akashi, R., Arita, R., 2016,
Phys. Rev. B 93, 094525 .

Scalapino D., 1969, in Superconductivity, p.449, Parks R.,Ed.,Marcel

Dekker, NY

65
Schilling, A., Cantoni, M., Guo, J., Ott, H., 1993, Nature, 363, 56

Schrieffer, J., Scalapino, D., Wilkins, J., 1963, Phys. Rev. Lett. 10,
336

Stritzer, B., Buckel, W., 1972, Z.Phys.257,1

Suhl, H., Mattias, B., Walker, L., 1958, Phys. Rev. Lett, 3, 552
Syed, H., Gould, T., Webb, C., Cray, E., 1910, ArXiv: 1608.01774

Szczesniak,R.,Durajski.A.,2016,Front. of Phys.11,117406

Troyan, I., Gavriliuk, A., Ruffer, R., Chumakov, A., Mironovich, A.,
Lyubutin, I., Perekalin, D., Drozdov, A., Eremets, M., 2016, Science
351, 1303

Tsuda, S., Yokoya, T., Takano, Y., Kito, H., Matsushita, A., Yin, F.,
Itoh, J., Harima, H., Shin, S., 2003, Phys. Rev. Lett. 91, 127001

Uchiyama, H., Shen, K., Lee, S., Damacelli, A., Lu, D., Feng, D.,
Shen Z.-X., Tajima, S., 2002, Phys. Rev. Lett. 88, 157002

Wang, H., Tse, J., Tanaka, K., Litaka, T., Ma, Y., 2012, PNAS 109,
6463

Wang, R., Gao,Y.,Huang,Z.,Chen,X.,2017, arXiv: 1703.06641

Wang,Y., Chubukov, A., 2013,Phys.Rev.B 88, 024516

Wang, Y., Ma, Y., 2014, J. Chem. Phys. 140, 040901

Wigner, E., Huntington, H., 1995, J. Chem. Phys. 3, 764

Wolf E., 2012, Principles of Electron Tunneling Spectroscopy, Oxford,

NY

Zavaritskii, N., Itskevich, E., Voronovskii, A., 1971, JETP 33,

762

Zhang,L.,Wang,Y.,Lv,J., Ma,Y., 2017,Nature Rev.,2,17005

66
67