Spectroscopic Evidence of Pairing Gaps to 60 Kelvin or Above in Surface-Doped pTerphenyl Crystals
HaoxiangLi1†,XiaoqingZhou1†,StephenParham1,ThomasNummy1,JustinGriffith1,KyleGordon1,
EricL.Chronister2andDaniel.S.Dessau1,3
1DepartmentofPhysics,UniversityofColoradoatBoulder,Boulder,CO80309,USA
2DepartmentofChemistry,UniversityofCalifornia,Riverside,CA,92521,USA
3CenterforExperimentsonQuantumMaterials,UniversityofColoradoatBoulder,Boulder,CO80309,USA
Dated:Apr.13,2017
†Co-firstauthor.
[email protected]
[email protected]
[email protected]
[email protected]
1
The possibility of high temperature superconductivity in organic compounds has been
discussed since the pioneering work of Little in 1964 [1], with minimal progress until the recent
reports of a weak Meissner shielding effect to 43 Kelvin [2] and then 120 Kelvin [3] in potassiumdoped
para-terphenyl
samples.
To date however, no other signals of the superconductivity have
been shown, including the zero resistance state or evidence for the formation of the Cooper pairs
that are inherent to the superconducting state. Here, using high resolution photoemission
spectroscopy on potassium surface-doped p-terphenyl crystals, we show spectroscopic evidence for
pairing gaps at the surfaces of these materials, with the gaps persisting to 60 K or above. Together
with the Meissner signal reported in [3], this greatly increases the likelihood of high temperature
superconductivity in organics, opening the field up for future study of the mechanism of this
pairing as well as for the possible application of earth-abundant “plastic” superconducting devices.
Para-terphenyl is a simple organic molecule composed of three benzene rings arranged end-to-end,
as illustrated in Fig. 1a and 1b, and it is available commercially at a modest price. These molecules can
be packed together in single crystalline form (Fig. 1c), in which case the molecules arrange themselves
in a unidirectional stacking as shown in Fig. 1a and 1b. The experiments that reported the Meissner
shielding effect were prepared from non-crystalline p-terphenyl powders, annealed with potassium in an
evacuated tube at temperatures between 443 and 533 K. Meissner signals from these experiments
initially formed at 7 K [4], with subsequent experiments yielding signals beginning at 43 K [2] and
finally 120 K [3]. The evolution with time in the onset temperature presumably has to do with the doping
level of the K in the samples and/or the cleanliness/quality of the samples.
The Meissner effect is just one signature of superconductivity, so it is in need of confirmation from
other techniques such as transport or spectroscopy, with the latter also able to give critical information
needed to understand the origin of the possible superconductive pairing. The Meissner effect signal in
these papers was also extremely weak, implying that only a tiny fraction of the end products became
superconducting. This raises the possibility that the end products might consist of multiple components
or material phases, since the organic molecules are chemically active at annealing temperature, are
sensitive to impurities such as oxygen and water vapor, and have polymorphism in general. In fact, an
earlier study5 from the same group suggested that the end products also contained C60, graphite, and
possibly other compounds as well. Thus a critical question is whether there is indeed superconductivity,
and if yes, which component or phase is responsible and what is the strength of the superconductive
pairing. Motivated by these questions, we performed a photoemission study on pristine p-terphenyl
2
single crystals (Fig. 1c) with controlled in-situ potassium metal (K) evaporation in ultra-high vacuum,
aiming to directly detect the presence of the Cooper pairs that are at the heart of all known
superconductors.
Figure 1d shows a schematic of the experiment. The crystals were initially annealed in ultrahigh
vacuum at 100C, which due to the high vapor of p-terphenyl will sublimate off any dirty exterior layers.
Sub monolayer coverages of K were then consecutively dosed onto the clean surface at T=300 K, with
the surfaces monitored by x-ray core level spectroscopy (XPS) as well as by high-resolution
photoemission of the near-Fermi level features. Fig. 1e shows spectra for a variety of consecutive doses.
There are at least 4 peaks in the spectra at binding energies (energy below EF) near 4 eV, 7 eV, 9 eV and
13.5 eV respectively, corresponding to various peaks in the valence band/occupied molecular orbitals.
For the pristine compound there is vanishingly small spectral weight for the first 2 eV below the
chemical potential, consistent with the optical gaps that are of order 3-4 eV [6]. With consecutive K
surface dosings, a potassium 3p core level develops at the binding energy around 18 eV, indicating that
potassium is incorporated onto/into the surface. Even in the presence of K-dosing, the original valence
peak features remain robust. This indicates that the potassium doping is perturbative in nature, only
minimally modifying the large-scale electronic structure of p-terphenyl. On the other hand, the peak
positions were monotonically shifted away from EF, indicating a change in chemical potential and the
spectral weight in the vicinity of the chemical potential grows (not visible in the wide scale scan of Fig.
1e). This is consistent with the idea that potassium donates extra electrons to the lowest energy
conduction bands.
With sufficient K-dosing (e.g. the 34 doses shown in Fig. 1e), weak metallic spectral weight
appeared near the chemical potential and the material became much more conductive, as evidenced by
the lack of sample charging at low temperatures (see Extended Data Fig. 1) that plagued our
experiments on pristine and lightly doped samples. To date, minimal angle-dependent changes have
been observed, which is presumably due to a) the very weak dispersion expected in organic crystals in
which the constituent components are far separated, and/or b) disorder of the underlying crystal lattice or
K overlayers, which were not annealed after the K deposition. For this reason, the present spectra are
not labeled by momentum-space positions, and should initially be viewed as representing the average
effect across the Brillouin zone.
Figure 2a shows the very low energy regime for sample #2 as a function of temperature between
10 K and 120 K (see Extended Data Fig. 2 for some data on sample #1). The leading edge of the 10 K
3
spectrum is pulled away from the chemical potential, as also evidenced by an overlay of this spectrum
with that from a metallic gold film measured under identical conditions right after the measurement of
the doped p-terphenyl (Extended Data Fig. 3). An alternative view of the same spectra is presented in
Fig. 2b, which shows the data of Fig. 2a symmetrized about EF, which has been developed as a powerful
way to remove the effect of the Fermi function and better visualize the presence of any low energy gaps
[7]. Here we see that there is a strong suppression of low energy spectral weight (a gap) that gradually
disappears as the temperature is warmed towards 120 K. The most natural explanation for this gap is
that it is a signature of Cooper pairs of electrons, with the gap energy equal to the energy cost to remove
an electron from a pair, i.e. pairing strength. A measurement of the gap such as this is the most direct
spectroscopic signature of Cooper pairing in a superconductor.
In a conventional Bardeen-Cooper-Schrieffer (BCS) superconductor there is a direct
correspondence between the presence of the pairing gap and of superconductivity, that is, the pairing
gap is finite for all temperatures that are superconducting, being reduced to zero at the transition to the
normal state. In other, more exotic superconductors such as the cuprates, there is much evidence for
pairing at temperatures higher than the superconducting temperature (especially in the underdoped
“pseudogap” regime [8,9], though the understanding of this situation is still under serious debate. Also
important and outside the realm of simple BCS theory, are the electron scattering rates or “self-energy”
effects, which typically act as pair-breakers. If these effects are strong as in the cuprate superconductors
the pairing gap may show a “filling” behavior with increasing temperature [10,11,12,13], rather than the
“gap closing” of the simple BCS result.
The gap data of Fig. 2a clearly shows the gap filling behavior with temperature, similar to the
cuprate high temperature superconductors. Such a “filling-in” behavior is most commonly and simply
modeled with the Dynes model for the superconducting density of states NSC(E,T) [14]:
𝑁"# 𝐸, 𝑇 = 𝑁( (𝐸)𝑅𝑒
./01(2)
(./01(2))3 /4(2)3
,
(1)
where NN(E) is the normal state density of states and Γ(T) represents a scattering rate or pair-breaking
effect that competes with the superconducting gap Δ(T). The dotted lines in Fig. 2b (shown non-overlaid
in Extended Data Fig. 4a) show fits to the experimental data using this equation convolved with the
measured experimental resolution function, and with NN(E)=a+bE, i.e. a linearly varying density of
states. The parameters extracted from these fits are shown in Fig. 2c. The gap has a low temperature
magnitude of Δ(0)~12 meV and slowly decreases with rising temperature. Our data also shows that the
4
pair-breaking parameter G rises rapidly with temperature, which is unexpected for a conventional BCS
superconductor but is a well-known effect for cuprate high temperature superconductors [10,11,12,13]. Our
fits show that G becomes larger than D at approximately 60 K. Above this temperature the rate at which
pairs are broken will be faster than the rate at which they are created, so we expect that 60 K would be
the nominal Tc for our samples. This expectation is based upon evidence in conventional
superconductors in which an effective G term was varied due to static disorder [15], and from cuprate
superconductors which also show a G that grows rapidly with temperature [13]. So even though our fits
indicate the likely presence of pairing gaps at temperatures above 60 K, these gaps are unlikely to create
an actual superconducting state, and we also note that the fits become less and less reliable when G is
larger than D, due both to the intrinsic broadening effects of the G term as well as the smaller D term.
Other doping levels or different momentum space cuts that we have not yet explored may have larger
gap values and/or smaller G values, both of which would raise the superconducting Tc.
Our results therefore strongly enhance the likelihood of high temperature superconductivity in this
class of materials, also indicating that the host of the superconductivity is the K-doped or K-intercalated
p-terphenyl itself. In contrast to the original report [2,5] in which the end product is likely a mixture of
different components, in our experiment it is highly unlikely that chemical reactions substantially
modified the material phase. Our estimated temperature scale of 60 K is already significantly higher than
all other organic superconductors known to-date [ 16 ], with many potential avenues for significantly
increasing the Tc of this broad class of materials, including varying the doping level of the K, utilizing
other methods to dope the samples, and other related compounds such as those with different numbers of
benzene rings.
The mechanism of the pairing in these materials is potentially quite different from other known
superconductors, not just because of the high Tc’s but also because of the unusual structure and
chemistry of organic molecular solids. Little’s original proposal suggested that a fully-electronic (nonphononic) mechanism may be possible [1] and other proposals for organic superconductors including
Resonating Valence Bond (RVB) physics [17] as well as bipolaronic pairing mechanisms [18] have been
discussed in the context of organic superconductivity. The new findings of high temperature
superconductivity in these compounds therefore potentially opens new and exciting venues into the most
fundamental aspects of superconductivity as well.
5
Methods
Single crystal samples of p-terphenyl were grown by the zone-refining method [19]. Its crystal
structure was identified using X-ray diffraction. 2 crystals of similar size (around 0.2*0.5*0.5 mm) were
used. Their surfaces were prepared through sublimation at <373K for one hour in 1E-9 Torr vacuum.
Experimental PES measurements were carried out at the Stanford Synchrotron Radiation Lightsource
(SSRL) beamline 5.4 with 32 eV linearly polarized light and 2E-11 Torr scale ultra-high vacuum. The
experimental energy resolution was 14 meV. Fermi energy references were repeatedly obtained from the
in-situ Au Fermi edge installed on the same sample manipulator. In-situ potassium dosing was
performed using a commercial SAES getter source. It took the form of consecutive doses, with one dose
corresponding to a heating current of 5.5A that lasts 60 seconds. To avoid sample charging, significant
potassium dosing was performed at 300 K before cooling to low temperature. The possibility of sample
charging was ruled out through comparison of spectral weight at different photon fluxes, and the
possibility of sample aging was ruled out by comparison of spectral weight through various thermal and
time cycles.
Acknowledgements
This work was funded by DOE project DE-FG02-03ER46066 to the University of Colorado, Boulder.
We thank Drs. D. H. Lu and M. Hashimoto for technical assistance on the ARPES measurements, and
Sean Shaheen, Gang Cao, Gerald Arnold, Bruce Normand, Jennifer M. Reed, Justin Waugh and
Miranda Thompson for help and valuable discussions. SSRL is supported by the Director, Office of
Science, Office of Basic Energy Sciences, of the U.S. Department of Energy under Contract No. DEAC02-05CH11231.
6
a
b
c
e
1.0
K (3p)
d
Intensity (arb. unit)
0.8
0 dose
8 doses
20 doses
34 doses
0.6
0.4
0.2
0.0
-20
-15
-10
-5
0
E-EF (eV)
Figure 1. Materials, preparation, and broad overview. a, b, two views of p-terphenyl molecules
arranged in a single crystal. c, Pictures of our bulk crystals of p-terphenyl. d, The experimental
schematic. K was repeatedly dosed onto the surfaces of the crystals, with photoemission spectra taken at
these different doses. e, A wide overview of photoemission intensity vs. doping level. The growing
intensity of the humps and peak around 18 eV (K 3p) in the photoemission spectra indicate the
increasing doping level that follows the number of doses. The shift of the four spectral peaks of the
valence band around 14, 10, 7, and 5 eV shows the consistent change of chemical potential.
7
a
120 K 1.0
100 K
60 K
30 K
10 K 0.8
Intensity (arb. unit)
0.8
0.6
b
120 K
100 K
60 K
30 K
10 K
Fit
0.6
0.4
0.4
c
0.025
Δ
Γ
0.020
Energy (eV)
1.0
0.015
0.010
0.005
0.2
0.2
0.000
0.0
-0.08
-0.04
0.00
E-EF (eV)
0.04
-0.08
-0.04
0.00
E-EF (eV)
0.04
0.08
0
40
80
120
Temperature (K)
Fig. 2, Spectral gap and its temperature evolution. a, Temperature dependence of
the photoemission spectra with effective doping of potassium. b, The correspondent
Figure 2. Spectral
gaps from a sample with 34 K doses. a, Temperature dependence of the very
symmetrized , and the correspondent fit to Dynes formula. c, The superconducting gap
(Δ) and the scattering
(𝝘) extracted
the afitting.
The extrapolated
low energy photoemission
spectra. rate
b, The
spectra offrom
panel
symmetrized
about Etemperature
F so as to remove the
scale from the superconducting gap is above 100 K.
effect of the Fermi function. This data clearly shows the presence of a gap at low temperatures, with the
gap “filling in” as the temperature is increased. The dashed lines are fits to the data using equation 1,
which has two key parameters – a gap D and a scattering rate G. c, The gap Δ and scattering rate G as a
function of temperature. We expect that the superconducting Tc of our sample would be near the
crossing point of these two parameters, i.e. around 60 K.
8
1.0
Check Charging
Intensity (arb. unit)
0.8
0.6
0.4
Low Flux
High Flux
0.2
0.0
-0.10
-0.05
0.00
0.05
E-EF (eV)
Data
Figure 3, Photoemission
spectra photon
with different
flux.of the
Extended Extended
Data Figure
1, Photoemission
spectra with different
flux. Thephoton
step edges
two spectraThe
withstep
different
photon
consistent.
indicates
that there
is noconsistent.
surface charging
edges
of theflux
twoarespectra
withThis
different
photon
flux are
This that
could shifteliminate
the spectral
edge.
the possibility of surface charging that shifted the spectral edge.
9
1.2
a
b
Sample 2
Sample 2
1.0
Intensity (arb. unit)
1.0
0.8
0.8
0.6
0.6
Sample 1
0.4
Sample 1
0.4
0.2
0.2
0.0
0.0
-0.08
-0.04
0.00
E-EF (eV)
0.04
-0.05
0.00
0.05
E-EF (eV)
Extended Data Figure 1, Spectral gap at 10 K. a,
Extended Data Figure
2, Spectral spectra
gap at 10
Photoemission
of two different samples
Photoemission
of K.
twoa,different
samplespectra
with effective
doping
of potassium.
Both show
spectra
showspectral
a clearedge
stepbelow the Fermi
with effective surface surface
doping of
potassium.
Both spectra
a clear
edge below the Fermi level. b, The correspondent symmetrized
level. b, The corresponding
symmetrized spectra.
spectra
10
1.0
Au Fermi edge
K doped p-Terphenyl at 10K
Intensity (arb. unit)
0.8
0.6
0.4
0.2
0.0
-0.06 -0.04 -0.02
0.00
0.02
0.04
E-EF (eV)
Extended Data Figure 2, Compare sample spectrum
Extended Data Figure
Comparing
sample spectrum
to Fermi level
reference.
to 3,
Fermi
level reference.
The spectrum
of the
sampleThe spectrum of the
sample shows a leading
edge a
well
pushededge
away at
from
that ofbelow
the Au the
reference
shows
leading
-7meV
Fermispectrum.
level
judging from the Au.
11
a
b
120 K
100 K
60 K
30 K
10 K
Fit
-0.06
E-EF (eV)
0.06
120 K
100 K
60 K
30 K
10 K
Fit
-0.06
E-EF (eV)
0.06
Data4,Figure
4, Fitting
with without
and without
scattering
term.a, Fitting the
Extended Extended
Data Figure
Fitting
with and
the the
scattering
raterate
term.
a, Fitting
symmetrized
spectra
term
(𝝘),asthe
symmetrized
spectrathe
with
the scattering
rate with
term the
(𝝘),scattering
the same rate
set of
data
in same
Fig. 3b. b, Fitting
without thesetscattering
ratein term
(𝝘),b,i.e.
simple
BCS.the
The
fit quality
unacceptable
of data as
Fig. 3b.
Fitting
without
scattering
rateisterm
(𝝘). The without the
scattering rate
term. peak is much sharper when the scattering rate term is missing.
coherence
12
a
Intensity (arb. unit)
0.00
k (1/Å)
-0.05
-0.10
-0.15
-0.20
-0.25
-0.20
-0.10
0.00
E-EF (eV)
1.0
c
20
EDC
10
0
-0.12
0.8
-0.08
-0.04
0.00
0.04
E-EF (eV)
d
0.012
Δ
Γ
0.010
0.7
0.6
60 K
70 K
80 K
85 K
Fit
0.5
0.4
Energy (eV)
Intensity (arb. unit)
30
0.014
0.9
0.3
-0.08
b Spectral weight
0.008
0.006
0.004
0.002
0.000
-0.04
0.00
0.04
0.08
50
60
70
80
E-EF (eV)
Temperature (K)
Extended Data Figure 5, Comparing to another high Tc cuprate superconductor. a,
ARPES5,spectrum
of BSCCO
taken at 60K (Tc=85K).
b, The
ExtendedA typical
Data Figure
Comparing
to highsuperconductor
Tc cuprate superconductor.
a, A typical
ARPES
weight
and energytaken
distribution
as near
the green
line in b,
a. The
c, The
spectrumspectral
of BSCCO
superconductor
at 60 K curve
(Tc=85(EDC)
K) in the
nodal region.
spectral
spectral weight
of different
temperature
with Dynes Curve
formula.
d, at
weight (integrated
over the
momentum
range of and
this the
cut)correspondent
and an EnergyfitDistribution
(EDC)
The
extracted
superconducting
gap
(Δ)
and
scattering
rate
(𝝘)s.
k=kF (green line in panel a). c, The symmetrized spectral weight at different temperatures and the
corresponding fits to formula 1 of the main text. d, The extracted superconducting gaps (Δ) and
scattering rates (𝝘), which have a temperature dependence similar to that of Fig 2c.
13
References:
1. Little, W. A., Possibility of synthesizing an organic superconductor. Phys. Rev. 134, A1416
(1964).
2. Wang, R. S., Gao, Y., Huang, Z. B. and Chen X. J., Superconductivity at 43 K in a single CC
bond linked terphenyl. arXiv: 1703.05804 (2017)
3. Wang, R. S., Gao, Y., Huang, Z. B. and Chen X. J., Superconductivity above 120 Kelvin in a
chain link molecule. arXiv: 1703.06641 (2017)
4. R. S. Wang, Y. Gao, Z. B. Huang and X. J. Chen, Superconductivity in p-terphenyl. arXiv:
1703.05803 (2017)
5. Gao, Y., Wang, R. S., Wu, X. L., Chen, J., Deng, T. G., Yan X. W. and Huang Z. B., Search
superconductivity in potassium doped p-terphenyl. acta Physica Sinica 65, 077402 (2016)
6. Nijegorodov N. I., Downey, W. S. and Danailov, M. B., Systematic investigation of
absorption, fluorescence and laser properties of some p- and m-oligophenylenes. Spectr.
Acta. Part A 56, 783 (2000)
7. Norman, M. R., Destruction of the Fermi surface in underdoped high-Tc superconductors,
Nature 392, 157 (1998)
8. Wang, Y. Y., Lu, L. and Ong, N. P., Nernst effect in high-Tc superconductors. Phys. Rev. B
73, 024510 (2006)
9. Kondo, T. et al. Disentangling Cooper-pair formation above the transition temperature from
the pseudogap state in cuprates. Nat. Phys. 7, 21 (2011)
10. Reber, T. J. et al. The origin and non-quasiparticle nature of Fermi arcs in Bi2Sr2CaCu2O8+δ.
Nat. Phys. 8, 606 (2012)
11. Pasupathy, A. N. et al. Electronic origin of the inhomogeneous pairing interaction in the
high Tc superconductor Bi2Sr2CaCu2O8+δ. Science 320, 196 (2008)
12. Hwang J., Timsuk, T., Gu, G. D., Doping dependent optical properties of Bi2Sr2CaCu2O8+δ.
J. Phys. Condens. Matter 19, 125208 (2007)
13. Reber, T. J. et al. Pairing, pair-breaking, and their roles in setting the Tc of cuprate high
temperature superconductors. arXiv: 1508. 06252 (2015)
14
14. Dynes, R. C, Narayanamurti, R. and Garno, J. P., Direct measurement of quasiparticlelifetime broadening in a strong-coupled superconductor. Phys. Rev. Lett. 41, 1509 (1978)
15. Dynes, R. C., Garno, J. P., Hertel, G. B., and Orlando, T. P., Tunneling study of
superconductivity near the metal-insulator transition, Phys. Rev. Lett. 53, 2437 (1984)
16. Lang, M. and Mueller, J., Organic superconductors. Chapter in “The Physics of
Superconductors – Vol II”, Springer-Verlag Berlin Heidelberg (2004)
17. Powell, B. J., Mackenzie, R. H., Half-filled layered organic superconductors and the
resonating- valence-bond theory of the Hubbard-Heisenberg model, Phys. Rev. Lett. 94,
047004 (2005)
18. L. D. Kispert, J. Joseph, G. G. Miller and R. H. Baughman “EPR study of polarons in a
conducting polymer with nondegenerate ground states: Alkali metal complexes of poly (p‐
phenylene) and phenylene oligomers” The Journal of Chemical Physics 81, 2119 (1984)
19. Rice, A. P., Tham, F. S. and Chronister, E. I., A temperature dependent X-ray study of the
order-disorder enantiotropic phase transition of p-terphenyl, J. Chem. Crystallogr. 43, 14
(2013)
15