Page 1
1
Extremely Strong Coupling Superconductivity in
Artificial Two-dimensional Kondo Lattices
Y. Mizukami1, H. Shishido1+
D. Watanabe1, M.Yamashita1, H. Ikeda1, T. Terashima2, H. Kontani3, and Y. Matsuda1*
1Department of Physics, Kyoto University, Kyoto 606-8502, Japan
2Research Center for Low Temperature and Materials Sciences, Kyoto University,
Kyoto 606-8501, Japan
3Department of Physics, Nagoya University, Furo-cho, Nagoya 464-8602, Japan
+Present address: Department of Physics and Electronics, Osaka Prefecture University, Sakai, Osaka 599-8531,
Japan
*matsuda@scphys.kyoto-u.ac.jp
When interacting electrons are confined to
low-dimensions, the electron-electron correlation effect is
enhanced dramatically, which often drives the system
into exhibiting behaviours that are otherwise highly
improbable. Superconductivity with the strongest
electron correlations is achieved in heavy-fermion
compounds, which contain a dense lattice of localized
magnetic moments interacting with a sea of conduction
electrons to form a three-dimensional (3D) Kondo
lattice1. It had remained an unanswered question
whether superconductivity would persist upon
effectively reducing the dimensionality of these materials
from three to two. Here we report on the observation of
superconductivity in such an ultimately
strongly-correlated system of heavy electrons confined
within a 2D square-lattice of Ce-atoms (2D Kondo
lattice), which was realized by fabricating epitaxial
superlattices2,3 built of alternating layers of
heavy-fermion CeCoIn5
YbCoIn5. The field-temperature phase diagram of the
superlattices exhibits highly unusual behaviours,
including a striking enhancement of the upper critical
field relative to the transition temperature. This implies
that the force holding together the superconducting
electron-pairs takes on an extremely strong coupled
nature as a result of two-dimensionalisation.
The layered heavy-fermion compound CeCoIn5 has the
highest superconducting transition temperature (Tc=2.3 K)
among rare-earth-based heavy-fermion materials4. Its
electronic properties are characterized by anomalously large
value of the linear contribution to the specific heat
(Sommerfeld coefficient ~1 J/mol K2) indicating heavy
effective masses of the 4f electrons which contribute greatly
to the Fermi surface. The tetragonal CeCoIn5 crystal
structure is built from alternating layers of CeIn3 and CoIn2
stacked along the [001] direction. This compound possesses
several key features for understanding the unconventional
superconductivity in strongly correlated systems5-7. The
superconductivity with dx2-y2 pairing symmetry8-11 which
occurs in the proximity of a magnetic instability is a
manifestation of magnetic fluctuations mediated
superconductivity5-7,12. A very strong coupling
superconductivity, where electron-pairs are bound together
, T. Shibauchi1, M. Shimozawa1, S. Yasumoto1,
4 and conventional metal
by strong forces, is revealed by a large specific heat jump4 at
Tc representing a steep drop of the entropy below Tc, and a
large superconducting energy gap needed to break the
electron-pair9. Despite its layered structure, the largely
corrugated Fermi surface13, 3D-like antiferromagnetic
fluctuations in the normal state14, and small anisotropy of
upper critical field15, all indicate that the electronic, magnetic
and superconducting properties are essentially 3D rather
than 2D. Therefore it is still unclear to which extent the 3D
nature is essential for the superconductivity of CeCoIn5.
Recently the state-of-the-art technique has been developed
to reduce the dimensionality of the heavy electrons in a
controllable fashion by the layer-by-layer epitaxial growth of
Ce-based materials. Previously a series of antiferromagnetic
superlattices CeIn3/LaIn3 have been successfully grown2, but
it remains open whether heavy electrons in a single Ce-layer
forming a 2D Kondo lattice can be superconducting. Here
we fabricate multilayers of CeCoIn5 sandwiched by
nonmagnetic and nonsuperconducting metal YbCoIn5
(Yb-ion is divalent in closed-shell 4f(14) configuration)
forming (n:m) c-axis oriented superlattice structure, where n
and m are the number of layers of CeCoIn5 and YbCoIn5 in a
unit cell, respectively. Small lattice mismatch between
CeCoIn5 and YbCoIn5 offers a possibility of providing an
ideal heterostructure. The high resolution cross-sectional
transmission-electron-microscope (TEM) results (Figs. 1a-c),
and distinct lateral satellite peaks in X-ray diffraction pattern
(Fig. 1d) demonstrate the continuous and evenly spaced
CeCoIn5 layers with no discernible interdiffusion even for
n=1 cases (see Supplementary Fig. S1 for quantitative
analysis of interdiffusion by X-ray). The epitaxial growth
of each layer with atomic flatness is shown by the streak
patterns of the reflection-high-energy-electron-diffraction
(RHEED) (Fig.1e) and atomic-force-microscopy (AFM)
image (Fig. 1f). We investigate the transport properties for
the (n:5) superlattices by varying n. The resistivity of
CeCoIn5 thin film (Fig. 2a) reproduces well that of bulk
single crystals4; Below ~100 K (T) increases upon cooling
due to the Kondo scattering, decreases after showing a peak
at around the coherence temperature Tcoh~30 K, and drops to
zero at the superconducting transition. The hump structure of
(T) at ~Tcoh is also observed in the superlattices but
becomes less pronounced with decreasing n. The
Page 2
2
Figure 1 | Epitaxial superlattices (n:5) of CeCoIn5(n)/YbCoIn5(5). a, High-resolution cross sectional TEM image of n=1
superlattice. The bright dot arrays are identified as the Ce layers and the less bright dots are Yb atoms, which is consistent with
the designed superlattice structure in the left panel. The intensity integrated over the horizontal width of the image plotted
against vertical position (b) indicates clear difference between the Ce and Yb layers, showing no discernible atomic
interdiffusion between the neighboring Ce and Yb layers. Upper right inset is the fast Fourier transform (FFT) of the TEM
image, which shows clear superspots along the [001] direction (c). d, Cu-K1 x-ray diffraction patterns for n=1, 2, 3, 4, 5, and 7
superlattices with typical total thickness of 300 nm show first (red arrows) and second (blue arrow) satellite peaks. The position
of the satellite peaks and their asymmetric heights can be reproduced by the step-model simulations (green lines) neglecting
interface and layer-thickness fluctuations29 (see also Supplementary Information for the detailed analysis of the satellite peak
intensity). e, Streak patterns of the RHEED image during the growth. f, Typical AFM image for n=1 superlattice. Typical
surface roughness is within 0.8 nm, comparable to one unit-cell-thickness along the c axis of CeCoIn5.
superconducting transition to zero resistance is observed in
the superlattices for n?3 (Fig. 2b). For n=2 and 1, (T)
decreases below ~1 K, but it does not reach zero. However,
when the magnetic field is applied perpendicular to the
layers for n=1, (T) increases and recovers to the value
extrapolated above 1 K at 5 T, while the reduction of (T)
below 1 K remains in the parallel field of 6 T (Fig. 2c). The
observed large and anisotropic field response of (T) is
typical for layered superconductors, demonstrating the
superconductivity even in n=1 superlattice with 2D square
lattice of Ce-atoms. The critical temperature Tc determined
by the resistive transition gradually decreases with
decreasing n (Fig. 2d). The residual resistivity 0 of the
superlattices is in the same order as 0 of single-crystalline
film (Fig. 2d) and is much lower than 0 of Yb-substituted
CeCoIn5 single crystals16. An important question is whether
the superconducting electrons in the superlattices are heavy
and if so what is their dimensionality. As displayed in Figs.
2c, 3a and 4a, the parallel and perpendicular (to the layers)
upper critical field, ???|| and ???, of the superlattices at
low temperature are significantly larger than those in
conventional superconductors with similar Tc. The magnetic
field destroys the superconductivity in two distinct ways, the
orbital pair-breaking effect (vortex formation) and Pauli
paramagnetic effect, a breaking up of pair by
spin-polarization. The zero-temperature value of the orbital
upper critical field in perpendicular field ???
the effective electron mass in the plane ???
???
superlattices from the initial slope of ??? at Tc by the
relation, ???
are comparable or in the same order of ???
bulk single crystal, providing strong evidence for the
superconducting ``heavy'' electrons in the superlattices. We
stress that even a slight deviation of the f-electron number
from unity leads to a serious reduction of the heavy electron
????0? reflects
∗, ???
????0? ∝
∗?, and is estimated to be 6, 11, 12 T for n=3, 5, and 7
????0?=0.69Tc (-d???/dT)Tc. These magnitudes
????0? (=14 T) in
Page 3
3
Figure 2 | Superconductivity in superlattices (n:5)
of CeCoIn5(n)/YbCoIn5(5).
dependence of electrical resistivity (T) for n=1, 2, 3, 5,
7, and 9, compared with those of 300-nm-thick
CeCoIn5 and YbCoIn5 single-crystalline thin films. b,
Low-temperature part of the same data as in a. c, (T)
for n=1 at low temperatures in magnetic field parallel
(dotted line) and perpendicular (solid lines) to the ab
plane. d, Superconducting transition temperature as a
function of n (left axis). The circles are the mid points
of the resistive transition and the bars indicate the
onset and zero-resistivity temperatures. The residual
resistivity 0 as a function of n is also shown (right
axis).
mass17. Moreover the band structure calculation for n=1
superlattice shows the number of f-electrons in each
CeCoIn5 layer is very close to unity (Supplementary
Information). These indicate that the f-electron wave
functions are essentially confined to Ce-layers. The magnetic
two-dimensionality is shown by estimating the strength of
Ruderman-Kittel-Kasuya-Yosida (RKKY) interaction, an
intersite magnetic exchange interaction between the
localized f-moments, which decays with the distance as 1/r3.
This interaction between the Ce-ions in different layers of
(n:5) superlattices reduces to less than 1/200 of that between
neighboring Ce-ions within the same layer.
The superconducting order parameters in the CeCoIn5
layers of the superlattices are expected to be coupled
weakly by the proximity effect through the normal metal
YbCoIn5 layers. The proximity induced superconductivity in
YbCoIn5 layers is expected to be very fragile and destroyed
when a weak field is applied18. If the thickness of CeCoIn5
layer is comparable to the perpendicular coherence length
(~2.1 nm for CeCoIn5) and the separation of
superconducting layers (~3.7 nm for (n:5) superlattices)
a, Temperature
Figure 3 | Superconducting anisotropy in superlattices
(n:5) of CeCoIn5(n)/YbCoIn5(5). a, Magnetic-field
dependence of the resistivity for n=3 superlattice at several
field angles from ab) to 90 deg (H//c) (10 deg step)
at T=0.1 K. b, Anisotropy of Hc2, Hc2||/Hc2, as a function of
reduced temperature T/Tc for n=3, 5, and 7 superlattices and
for the bulk CeCoIn5. c, Upper critical field Hc2() at
several temperatures as a function of field angle . Hc2 is
determined by the mid-point of the transition except for 1.0
K, where a 80% resistivity criterion has been used. The solid
blue and red lines are the fits to the 3D anisotropic mass
model represented as Hc2() = Hc2||/(sin2+2cos2)1/2 with =
Hc2||/Hc2and Tinkham's formula | Hc2() cos/ Hc2|+{Hc2()
sin/ Hc2||}2=1 for a 2D superconductor, respectively19.
exceeds , each CeCoIn5 layer acts as a 2D
superconductor19. This 2D feature is revealed by diverging
Hc2||/Hc2 of n=3, 5 and 7 superlattices with approaching Tc
(Fig. 3b) in sharp contrast to bulk CeCoIn5
cusp-like angular dependence Hc2() near parallel field for
n=3 superlattice (Fig. 3c), which is qualitatively different
from that expected in the 3D anisotropic-mass-model but is
well fitted by the model in 2D limit19. Based on the above
2D features observed in all electronic, magnetic and
superconducting properties, we conclude that the observed
heavy-electron superconductivity is mediated most likely by
2D electron correlation effects. A fascinating issue is how the
two-dimensionalisation changes the pairing nature. The fact
that ???
well exceeds the actual ??? at low temperatures indicates
the predominant Pauli paramagnetic pair-breaking effect
even in perpendicular field. Therefore Hc2 at low
temperatures is dominated by the Pauli effect in any field
directions. This is reinforced by the result that the cusplike
behaviour of Hc2 becomes less pronounced well below Tc
(Fig. 3c), which is the opposite trend to Hc2 behaviour of
the conventional multilayer systems21. In fact
Pauli-limited upper critical field ???
???
20, and by a
???(0) estimated from the initial slope of Hc2(T) at Tc
??????given by
?????√2g
where g is the gyromagnetic ratio determined by the Ce
crystalline electric field levels, varies smoothly with field
direction, consistent with Hc2 of the present superlattices
at low temperatures. Figure 4a displays the H-T phase
diagram of the superlattices. What is remarkable is that with
Page 4
4
Figure 4 | Superconducting phase diagrams of superlattices (n:5) of CeCoIn5(n)/YbCoIn5(5). a, Magnetic field vs.
temperature phase diagram of n=3, 5, and 7 superlattices in magnetic field parallel (open symbols) and perpendicular (closed
symbols) to the ab plane compared with the bulk CeCoIn5 data. The mid-point of the transition in the (T) (circles) and (H)
(squares) has been used to evaluate Hc2(T). b, Superconducting transition temperature, Tc (open triangles), the reduced critical
fields Hc2/Tc in parallel (filled blue circles) and perpendicular (filled red squares) fields as a function of dimensionality
parameter 1/n (right panel). The pressure dependence of these quantities28 is also shown for comparison (left panel).
decreasing n, Tc decreases rapidly from the bulk value, while
Hc2 does not exhibit such a reduction for both field directions.
In fact, at low temperatures, Hc2|| of n=5 and 7 is even larger
than that of the bulk. This robustness of Hc2 (and hence of
against n-reduction indicates that the superconducting
pairing-interaction is hardly affected by
two-dimensionalisation. This provides strong evidence that
the superconductivity in bulk CeCoIn5 is mainly mediated
by 2D spin-fluctuations, although neutron spin resonance
mode is observed at 3D () position below Tc
In sharp contrast to Hc2, the thickness reduction
dramatically enhances Hc2/Tc from the bulk value (Fig. 4b).
A comparison with the pressure dependence results22, which
represent the increased three dimensionality, reveals a Tc
dome and a general trend of enhanced Hc2/Tc with reduced
dimensionality. Through the relation of Eq. (1), this trend
immediately implies a remarkable enhancement of /Tc by
two-dimensionalisation. We note that the enhanced impurity
scattering cannot be primary origins of the Tc reduction, as
these effects do not significantly enhance the /Tc ratio in
d-wave superconductors23. This is supported by no
discernible interdiffusion by TEM results and 0 of
superlattices in the same order as 0 of the bulk CeCoIn5.
The reduction of Tc may be caused by the reduction of
density-of-states (DOS) in the superlattices, but this scenario
is also unlikely because the DOS reduction usually reduces
the pairing interaction, which results in the reduction of
Using the reported value of 2/kBTc=6 in the bulk single
crystal9, 2/kBTc for the n=5 superlattice is estimated to
exceed 10, which is significantly enhanced from the
weak-coupling BCS value of 2/kBTc=3.54. It has been
suggested theoretically that d-wave pairing mediated by
10.
antiferromagnetic fluctuations in two-dimension can be
much stronger than that in three-dimension24-26. The striking
enhancement of 2/kBTc associated with the reduction of Tc,
a situation resemblant to underdoped high-Tc cuprates,
implies that there appear to be additional mechanisms, such
as 2D phase-fluctuations27 and strong pair-breaking effect
due to inelastic scattering28. Further investigation,
particularly probing electronic and magnetic excitations in
the normal and superconducting states, is likely to bridge the
physics of highly unusual correlated electrons in the 2D
Kondo lattice and in the 2D CuO2 planes of cuprates. The
fabrication in a wide variety of nanometric superlattices also
opens up a possibility of nanomanipulation of
heavy-electrons, providing a unique opportunity to produce
a novel superconducting system and its interface.
Method. CeCoIn5/YbCoIn5 superlattices are grown by
the molecular-beam-epitaxy (MBE) technique. The pressure
of the MBE chamber was kept at 10-7 Pa during the
deposition. (001) surface of MgF2 with rutile structure
(a=0.462 nm, c=0.305 nm) was used as a substrate. The
substrate temperature was kept at 550˚C during the
deposition. Atomic-layer-by-layer MBE provides for
digital control of layer thickness, which we measure by
counting the number of unit cells2. Each metal element was
evaporated from individually controlled Knudsen-cells.
15-unit-cell-thick (uct) YbCoIn5 was grown after CeIn3 (28
nm) was grown on the (001) surface of substrate MgF2 as a
buffer layer. Then n-uct CeCoIn5 layers and m-uct YbCoIn5
(typically m=5) were grown alternately, typically repeated
for 30-60 times. The deposition rate was monitored by a
Page 5
5
quartz oscillating monitor and the typical deposition rate was
0.01-0.02 nm/s.
Acknowledgments. We acknowledge discussions with R.
Arita, A.V. Chubukov, M. J. Graf, P.A. Lee, N.P. Ong, S.A.
Kivelson, T. Takimoto, and I. Vekhter. This work was
supported by KAKENHI from JSPS and MEXT and by
Grant-in-Aid for the Global COE program ``The Next
Generation of Physics, Spun from Universality and
Emergence" from MEXT.
References
1. Hewson A. C. The Kondo problem to heavy fermions
(Cambridge Univ. Press, Cambridge, 2003).
2. Shishido, H. et al. Tuning the dimensionality of the heavy
fermion compound CeIn3. Science 327, 980-983 (2010).
3. Logvenov, G. & Bozovic, I. Artificial superlattices grown
by MBE: Could we design novel superconductors?. Physica
C 468, 100-104 (2008).
4. Petrovic, C. et al. Heavy-fermion superconductivity in
CeCoIn5 at 2.3 K. J. Phys.: Condens. Matter 13, L337-L342
(2001).
5. Monthoux, P., Pines, D. & Lonzarich, G. G.
Superconductivity without phonons. Nature 450, 1177-1183
(2007).
6. Sarrao, J. L. & Thompson, J. D. Superconductivity in
cerium- and plutonium-based ‘115’ materials. J. Phys. Soc.
Jpn. 76, 051013 (2007).
7. Pfleiderer, C. Superconducting phases of f-electron
compounds. Rev. Mod. Phys. 81, 1551-1624 (2009).
8. Izawa, K., Yamaguchi, H., Matsuda, Y., Shishido, H.,
Settai, R. & Onuki, Y. Angular position of nodes in the
superconducting gap of
superconductor CeCoIn5. Phys. Rev. Lett. 87, 057002
(2001).
9. Park, W. K., Sarrao, J. L. Thompson, J. D. & Greene, L. H.
Andreev reflection in heavy-fermion superconductors and
order parameter symmetry in CeCoIn5. Phys. Rev. Lett. 100,
177001 (2008).
10. Stock, C., Broholm, C., Hudis, J., Kang, H.J., & Petrovic,
C. Spin resonance in the d-wave superconductor CeCoIn5.
Phys. Rev. Lett. 100, 087001 (2008).
11. An K. et al. Sign reversal of field-angle resolved heat
capacity oscillations in a heavy fermion superconductor
CeCoIn5 and dx2-y2 pairing symmetry. Phys. Rev. Lett. 104,
037002 (2010).
12. Mathur, N. D. et al. Magnetically mediated
superconductivity in heavy fermion compounds. Nature 394,
39-43 (1998).
13. Settai, R., Takeuchi, T. & Onuki, Y. Recent advances in
Ce-based heavy-fermion superconductivity and Fermi
surface properties. J. Phys. Soc. Jpn. 76, 051003 (2007).
quasi-2D heavy-fermion
14. Kawasaki, Y. et al. Anisotropic spin fluctuations in
heavy-fermion superconductor CeCoIn5: In-NQR and
Co-NMR studies. J. Phys. Soc. Jpn. 72, 2308-2311 (2003).
15. Bianchi, A., Movshovich, R., Capan, C., Pagliuso, P. G.
&Sarrao, J. L. Possible Fulde-Ferrell-Larkin-Ovchinnikov
superconducting state in CeCoIn5. Phys. Rev. Lett. 91,
187004 (2003).
16. Shu, L. et al. Correlated electron state in Ce1-xYbxCoIn5
stabilized by cooperative valence fluctuations. Phys. Rev.
Lett. 106, 156003 (2011).
17. Rice, T. M. & Ueda, K. Gutwiller variational
approximation to the heavy-fermion ground state of the
periodic Anderson model. Phys. Rev. Lett. 55, 995-998
(1985).
18. Serafin, A., Fletcher, J. D., Adachi, S., Hussey, N. E. &
Carrington, A. Destruction of chain superconductivity in
YBa2Cu4O8 in a weak magnetic field. Phys. Rev. B 82,
140506 (2010).
19. Schneider, T. & Schmidt, A. Dimensional crossover in
the upper critical field of layered superconductors. Phys. Rev.
B 47, 5915–5921 (1993).
20. Tayama, T. et al. Unconventional heavy-fermion
superconductor CeCoIn5: dc
temperatures down to 50 mK. Phys. Rev. B 65, 180504(R)
(2002).
21. Broussard, P. R. & Geballe T. H. Critical fields of Nb-Ta
multilayers. Phys. Rev. B 35, 1664–1668 (1987).
22. Knebel, G. et al. Competition and/or coexistence of
antiferromagnetism and superconductivity in CeRhIn5 and
CeCoIn5. Phys. Status Solidi B 247, 557–562 (2010).
23. Sun, Y. & Maki, K. Impurity effects in d-wave
superconductors. Phys. Rev. B 51, 6059-6063 (1995).
24. Takimoto, T. & Moriya, T. Superconductivity and
antiferromagnetism in the three-dimensional Hubbard model.
Phys. Rev. B. 66, 134516 (2002).
25. Arita, R., Kuroki, K. & Aoki, H. Spin-fluctuation
exchange study of superconductivity in two- and
three-dimensional single-band Hubbard models. Phys. Rev.
B 60, 14585-14588 (1999).
26. Monthoux, P. & Lonzarich, G.G., Magnetically
mediated superconductivity in quasi-two- and three
dimensions. Phys. Rev. B 63, 054529 (2001).
27. Emery, V. J. & Kivelson, S. A. Importance of phase
fluctuations in superconductors with small superfluid density,
Nature 374, 434-437 (1995).
28. Lee, P. A. & Read, N. Why is Tc of the oxide
superconductors so low?. Phys. Rev. Lett. 58, 2691-2694
(1987).
29. Fullerton, E. E., Schuller, I. K., Vanderstraeten, H. &
Bruynseraede, Y. Structural refinement of superlattices from
x-ray diffraction. Phys. Rev. B 45, 9292-9310 (1992).
magnetization study at
Page 6
6
SUPPLEMENTARY INFORMATION
Extremely Strong Coupling Superconductivity in Artificial
Two-dimensional Kondo Lattices
X-ray diffraction (XRD) analysis
Figure S1 shows the XRD satellite peak near the (004) peak for (1:5)
superlattice and the results of our simulation with various interdiffusion ratio. The
intensity I is calculated from the Laue function G and crystal structure factor F:
I = |F|2G.
The crystal structure factor of the superlattice can be described as
FF
where FCe and FYb are the crystal structure factors of CeCoIn5 and YbCoIn5,
respectively. To estimate the Ce/Yb interdiffusion quantitatively, we assume that the
interdiffuion occurs only at the CeCoIn5/YbCoIn5 interface layers (diffusion ratio p),
ignoring the interdiffusion in the next layers. (The step model used in Fig. 1c
corresponds to p=0.) Then FCe and FYb of the superlattice (1:5) having l unit blocks
perpendicular to the film plane are given by
) 1 ( exp)2 exp(2
Yb YbCeCe In In
z pczcp iqff
,
YbCe
F
1 (
) 1 () 1 () exp
) 1 (exp) 1 (
YbYbCe Ce
YbCe Co YbCe Ce
z pczcp
pc
cp iqf pffpF
1 (
1
1
1
1
1
) 1 (
2
) 1 (
2
exp
22
exp)2exp(2
22
exp
22
2
4
2
2 exp) exp
Yb YbCe CeYb YbCeCe InIn
YbCe CoYbCe
YbCeYb CeYb
zc
p
zc
p
zc
p
zc
p
iqff
c
p
c
p
iqff
p
f
p
c
p
c
p
iqpffpF
,1 exp) exp()exp(2)exp(
22
12exp
3
1
Yb Yb YbInIn YbCo Yb
j
YbCe
zz iqc
ffiqc
ff
c
p
jc
p
iq
where
sin2
q
.
Here is the wave length of the X-ray, is the diffraction angle, fCe, fYb, fCo and fIn are
the atomic form factors for Ce, Yb, Co and In, respectively, and cCe, cYb are the c-axis
lattice constants of CeCoIn5 and YbCoIn5, respectively. In the equation of FYb, 1st , 2nd
and 3rd lines correspond to the YbCoIn5 layers at the interfaces (2nd and 6th layers of the
(1:5) superlattice) involving Ce/La interdiffusion with a ratio p/2, 4th and 5th lines
correspond to 3 pure YbCoIn5 layers. The Laue function G for the superlattice (n:m)
having l unit blocks is written as
Page 7
7
2
sin
sin
YbCe
YbCe
mcncq
mcnc ql
G
.
To compare with the XRD data, we further use the Gaussian distribution function for
calculating I(2) to account for the instrumental resolution of 0.18 deg. The height of
the satellite peaks normalized by the main (004) peak can be reproduced well by
assuming no interdiffusion between Ce and Yb atoms near the interfaces (p=0). A 10 %
interdiffusion makes the peak height discernibly lower.
Combining the TEM, and XRD analyses, we conclude that the superlattice
structures are realized in an epitaxial form without noticeable interdiffusions.
Supplementary Figure S1. XRD satellite peaks (black dots) around (004) main peak in
the (1:5) superlattice with 60 unit blocks. The satellite peaks are compared with our
simulations assuming that 0 (red solid line), 10 (orange), 20 (green) and 30 % (blue) of
Ce atoms at the interfaces are replaced with Yb. We used c-axis lattice constants of
0.75513 nm for CeCoIn5 and 0.7433 nm for YbCoIn5.
Electronic band structure of the superlattice
We investigate the electronic band structure in the (1:5) superlattice based on the ab
initio density functional theory. The band structure calculations for LnCoIn5 (Ln=Ce, La,
and Yb) and for the superlattice have been performed by using the relativistic
full-potential (linearized) augmented plane-wave (FLAPW) + local orbitals method as
implemented in the WIEN2k packageS1. The crystallographical parameters used in the
calculations are summarized in Table S1.
Figures S2(a)-(h) display the energy dispersion, Ek, along the high-symmetry line. In
the Yb case, the almost dispersion-less bands between 0 and -2 eV come from the 4f
electrons, which do not cross the Fermi level (Fig. S2(c)). This indicates that the 4f
electrons do not participate in the conduction, which is consistent with the absence of
Kondo behavior in the transport properties in YbCoIn5. In contrast, the 4f bands extend
to the Fermi level in the Ce case (Fig. S2(a)), in which the Kondo effect is clearly
observed. This difference between Yb and Ce is reinforced by the calculations of partial
Page 8
8
charges in the atomic spheres (Table S2). In the case of Ce, the number of 4f electrons
inside the Muffin-tin sphere is estimated as 0.95 in the FLAPW calculations. This
number is very close to exact unity, as we expect some protrusion (~5%) outside the
sphere. In YbCoIn5, we should also consider that the obtained number of 13.45 for f
electrons is actually close to 14, which is in the closed-shell configuration. (In fact we
verified this fact by an additional calculation of the 4f open-core treatment.) In the
superlattice, we found that these occupation numbers of f orbitals in each layer are
essentially unchanged from the parent materials. This indicates that mobile 4f electrons
are confined only in the Ce layers.
Figures S3(a)-(d) depict the partial density of states (DOS), whose values at the
Fermi level are reported in Table S2. The DOS at the Fermi level in the superlattice is
nearly equal to the total of the DOS in the parent materials, 26.83 ? 8.70 ? 5 ? 70.33
(mJ/mol K2). In addition, the partial DOS of Ce component is also comparable to that in
CeCoIn5.
In Figs. S4(a)-(d), we show the Fermi surface colored by the Fermi velocity,
vkF=|(vx
smaller Fermi velocity, and then larger weight of Ce(4f) component. The Fermi surface
of the superlattice (Fig. S4(d)) is much more two-dimensional than the 3D structure in
the parent materials, which is mostly due to the folding along the c-axis. We note that
the cylindrical bands near the zone center ( point) are hole-like Fermi surface whilst
the bands near the zone corner (M point) are electron-like (see the band dispersions in
Fig. S2(h)). This shape of these sheets suggests a good nesting between these bands,
which would enhance the susceptibility at a Q-vector of (). It is tempting to
suggest that such 2D antiferromagnetic fluctuations are responsible for the observed
strong-coupling 2D superconductivity in the superlattices, which deserves further
studies.
Supplementary Table S1 | All materials studied here have the space group No.123
P4/mmm. Lattice constants in the superlattice CeCoIn5(1)/YbCoIn5(5) are determined
by a=(aCeCoIn5+5aYbCoIn5)/6 and c=cCeCoIn5+5cYbCoIn5, and we keep the relative atomic
positions (Wyckoff positions) in each layer the same as those in the parent materials.
a (Å) c (Å)
CeCoIn5 (S2) 4.612 7.549
LaCoIn5 (S3) 4.634 7.615
YbCoIn5 (S4) 4.559 7.433
(1:5) superlattice 4.568 44.714
Supplementary Table S2 | Occupation number of d and f orbitals inside the Muffin-tin
spheres of Ln and Co atoms and the DOS at the Fermi level. Occupation number in the
superlattice is described as Ce/Yb. Five Yb atoms have almost the same occupation
number. In the last column, values in the parentheses represent the partial DOS of Ce
components.
Ln(f) Ln(d)
CeCoIn5 0.950 0.718
LaCoIn5 0.081 0.646
YbCoIn5 13.45 0.456
(1:5)
superlattice
kF ,vy
kF ,vz
kF)| , where vi
kF = 1/ħ(dEkF/dki
F). The blue parts of Fermi surface have
Wyckoff position z of In(2)
0.305
0.311
0.305
---
Co(d)
7.54
7.55
7.58
7.59
DOS(mJ/mol K2)
26.83 (14.5)
6.81
8.70
69.36 (13.4) 0.986 / 13.48 0.712 / 0.456
Page 9
9
Supplementary Figure S2 | Energy band dispersion along the high-symmetry line in
CeCoIn5 (a), LaCoIn5 (b), YbCoIn5 (c), and the (1:5) superlattice (d). The Fermi level is
indicated by the green line. Lower panels (e-h) are the enlarged figures near the Fermi
level.
Supplementary Figure S3 | DOS in the unit of eV. The partial DOS of Ln ions is
indicated by green dashed lines.
Page 10
10
Supplementary Figure S4 | The Fermi surfaces colored by magnitude of the Fermi
velocity. The unit is 106 (m/sec).
SUPPLEMENTARY REFERENCES
S1. Blaha, P., Schwarz, K., Madsen, G. K. H., Kvasnicka, D. & Luitz, J., WIEN2k, An
Augmented Phane Wave + Local Orbitals Program for Calculating Crystal Properties
(Karlheinz Schwarz, Techn. Universitat Wien, Austria, 2001).
S2. Settai, R., J. Phys.: Condens. Matter 13, L627 (2001).
S3. Macaluso, R. T., J. Solid State Chem. 166, 245 (2002).
S4. Zaremba, V. I., Z Anorg. Allg. Chem. 629, 1157 (2003).
Download full-text