Puzzling pairing in the non-centrosymmetric superconductor LaNiC2
1. ISIS Facility, STFC School of
Rutherford Appleton Laboratory Physical Sciences
Puzzling pairing in the
non-centrosymmetric superconductor
LaNiC2
Jorge Quintanilla
SEPnet, University of Kent
Hubbard Theory Centre, Rutherford Appleton Laboratory
Collaborators: Adrian Hillier (RAL)
Bob Cywinski (Huddersfield)
James F. Annett (Bristol)
Bayan Mazidian (Bristol and RAL)
Funding: STFC, SEPnet
CMMP’10, University of Warwick, 15 December 2010
2. LaNiC2 – a weakly-correlated, paramagnetic
superconductor?
W. H. Lee et al., Physica C 266, 138 (1996)
V. K. Pecharsky, L. L. Miller, and Zy, Physical Review B 58, 497 (1998)
specific heat susceptibility
ΔC/TC=1.26
Tc=2.7 K (BCS: 1.43)
4. Relaxation due to electronic moments _
e
e
(longitudinal) sample
forward
detector
backward
detector
Moment
size Timescale:
~ 0.1G > 10-4s
~
(~ 0.01μB)
Hillier, Quintanilla & Cywinski,
PRL 102 117007 (2009)
5. Relaxation due to electronic moments _
e
e
(longitudinal) sample
forward
detector
backward
detector
Moment
size Timescale:
~ 0.1G > 10-4s
~
(~ 0.01μB)
Hillier, Quintanilla & Cywinski,
PRL 102 117007 (2009)
6. Relaxation due to electronic moments _
e
e
(longitudinal) sample
forward
detector
backward
detector
Moment
size Timescale:
~ 0.1G > 10-4s
~
(~ 0.01μB)
Hillier, Quintanilla & Cywinski,
PRL 102 117007 (2009)
7. Relaxation due to electronic moments _
e
e
(longitudinal) sample
forward
detector
backward
detector
Moment
size Timescale:
~ 0.1G > 10-4s
~
(~ 0.01μB)
Spontaneous, quasi-static fields appearing at Tc
⇒ superconducting state breaks time-reversal symmetry
[ c.f. Sr2RuO4 - Luke et al., Nature (1998) ]
Hillier, Quintanilla & Cywinski,
PRL 102 117007 (2009)
8. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
Symmetry of the gap function
k k
k
ˆ
k k
See J.F. Annett Adv. Phys. 1990.
9. Neutron diffraction
35000
30000
Data from
25000 D1B @ ILL
Intensity (arb units)
20000
15000
10000
5000
Orthorhombic Amm2 C2v
0
30 40 50 60 70 80 a=3.96 Å
b=4.58 Å
o
2
c=6.20 Å
Note no inversion centre.
C.f. CePt3Si (1), Li2Pt3B & Li2Pd3B (2), ...
(1) Bauer et al. PRL’04 (2) Yuan et al. PRL’06
10.
11. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
Singlet, triplet, or both?
0 0 x id y
d dz
k
ˆ
0 0 dz dx id y
singlet triplet
[ 0(k) even ] [ d(k) odd ]
12. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
Singlet, triplet, or both?
Neglect (for now!) spin-orbit coupling:
Singlet and triplet representations of SO(3):
Γns = - (Γns)T , Γnt = + (Γnt)T
Impose Pauli’s exclusion principle: , ' k ', k
k
ˆ either singlet , ' k 0 k i y
ˆ
or triplet , ' k dk .σi y
ˆ ˆ
13. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
Character table
180o
Hillier, Quintanilla & Cywinski,
PRL 102 117007 (2009)
14. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
Character table
C2v Symmetries and Sample basis
their characters functions
Irreducible E C2 v ’v Even Odd
representation
A1 1 1 1 1 1 Z
A2 1 1 -1 -1 XY XYZ
B1 1 -1 1 -1 XZ X
B2 1 -1 -1 1 YZ Y
These must be combined with the singlet and triplet
representations of SO(3).
Hillier, Quintanilla & Cywinski,
PRL 102 117007 (2009)
15. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
Possible order parameters
SO(3)xC2v Gap function Gap function
(unitary) (non-unitary)
1A
1 (k)=1 -
1A
2 (k)=kxkY -
1B
1 (k)=kXkZ -
1B
2 (k)=kYkZ -
3A d(k)=(0,0,1)kZ d(k)=(1,i,0)kZ
1
3A d(k)=(0,0,1)kXkYkZ d(k)=(1,i,0)kXkYkZ
2
3B d(k)=(0,0,1)kX d(k)=(1,i,0)kX
1
3B d(k)=(0,0,1)kY d(k)=(1,i,0)kY
2
Hillier, Quintanilla & Cywinski,
PRL 102 117007 (2009)
16. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
Possible order parameters
SO(3)xC2v Gap function Gap function
(unitary) (non-unitary)
1A
1 (k)=1 -
1A
2 (k)=kxkY -
1B
1 (k)=kXkZ -
1B
2 (k)=kYkZ -
3A d(k)=(0,0,1)kZ d(k)=(1,i,0)kZ
1
3A d(k)=(0,0,1)kXkYkZ d(k)=(1,i,0)kXkYkZ
2
3B d(k)=(0,0,1)kX d(k)=(1,i,0)kX
1
3B d(k)=(0,0,1)kY d(k)=(1,i,0)kY
2
Hillier, Quintanilla & Cywinski,
PRL 102 117007 (2009)
17. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
Possible order parameters
SO(3)xC2v Gap function Gap function
(unitary) (non-unitary)
1A
1 (k)=1 -
1A
2 (k)=kxkY -
1B
1 (k)=kXkZ -
1B
2 (k)=kYkZ -
3A d(k)=(0,0,1)kZ d(k)=(1,i,0)kZ
1
3A d(k)=(0,0,1)kXkYkZ d(k)=(1,i,0)kXkYkZ
2
3B d(k)=(0,0,1)kX d(k)=(1,i,0)kX
1
3B d(k)=(0,0,1)kY d(k)=(1,i,0)kY
2
Hillier, Quintanilla & Cywinski,
PRL 102 117007 (2009)
18. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
Possible order parameters
SO(3)xC2v Gap function Gap function
(unitary) (non-unitary)
1A
1 (k)=1 -
1A
2 (k)=kxkY -
1B
1 (k)=kXkZ -
1B
2 (k)=kYkZ -
3A d(k)=(0,0,1)kZ d(k)=(1,i,0)kZ
1
3A d(k)=(0,0,1)kXkYkZ d(k)=(1,i,0)kXkYkZ
2
3B d(k)=(0,0,1)kX d(k)=(1,i,0)kX
1
3B d(k)=(0,0,1)kY d(k)=(1,i,0)kY
2
Non-unitary
d x d* ≠ 0
Hillier, Quintanilla & Cywinski,
PRL 102 117007 (2009)
19. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
Possible order parameters
SO(3)xC2v Gap function Gap function
(unitary) (non-unitary)
1A
1 (k)=1 -
1A
2 (k)=kxkY -
breaks only SO(3) x U(1) x T *
1B
1 (k)=kXkZ -
1B
2 (k)=kYkZ -
3A d(k)=(0,0,1)kZ d(k)=(1,i,0)kZ
1
3A d(k)=(0,0,1)kXkYkZ d(k)=(1,i,0)kXkYkZ
2
3B d(k)=(0,0,1)kX d(k)=(1,i,0)kX
1
3B d(k)=(0,0,1)kY d(k)=(1,i,0)kY
2
* C.f. Li2Pd3B & Li2Pt3B,
H. Q. Yuan et al. PRL’06
Non-unitary
d x d* ≠ 0
Hillier, Quintanilla & Cywinski,
PRL 102 117007 (2009)
22. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
Non-unitary pairing
Spin-up superfluid
coexisting with spin- ˆ 0 0 0
or
down Fermi liquid. 0 0
0
C.f.
The A1 phase of
liquid 3He.
23. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
Non-unitary pairing
Spin-up superfluid
coexisting with spin- ˆ 0 0 0
or
down Fermi liquid. 0 0
0
C.f.
The A1 phase of
liquid 3He.
Ferromagnetic
F. Hardy et al., Physica B 359-
61, 1111-13 (2005) superconductors.
[ See A. de Visser in Encyclopedia of
Materials: Science and Technology (Eds.
K. H. J. Buschow et al.), Elsevier, 2010 ]
24. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
Ferromagnetic superconductors
But LaNiC2 is a paramagnet !
A. de Visser in Encyclopedia of Materials: Science and Technology
(Eds. K. H. J. Buschow et al.), Elsevier, 2010
25. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
Isn’t there a more simple explanation?
26. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
The role of spin-orbit coupling (SOC)
, ' k spin ' orbit k
,
Gap function may have both singlet and triplet components
0 0 d x id y dz
k
ˆ
d
0 0 z d x id y
• However, if we have a centre of inversion
basis functions either even or odd under inversion
still have either singlet or triplet pairing (at Tc)
• No centre of inversion: may have singlet and triplet (even at Tc)
27. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
The role of spin-orbit coupling (SOC)
G = [SO(3)×Gc]×U(1)×T
28. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
The role of spin-orbit coupling (SOC)
G = [SO(3)×Gc]×U(1)×T
29. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
The role of spin-orbit coupling (SOC)
G = [SO(3)×Gc]×U(1)×T
30. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
The role of spin-orbit coupling (SOC)
G = [SO(3)×Gc]×U(1)×T
31. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
The role of spin-orbit coupling (SOC)
G = [SO(3)×Gc]×U(1)×T
32. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
The role of spin-orbit coupling (SOC)
G = Gc,J×U(1)×T
33. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
The role of spin-orbit coupling (SOC)
G = Gc,J×U(1)×T
34. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
The role of spin-orbit coupling (SOC)
G = Gc,J×U(1)×T
35. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
The role of spin-orbit coupling (SOC)
z
E.g. reflection through a vertical
plane perpendicular to the y axis:
v, J I C2y, J
y x
Quintanilla, Hillier, Annett and Cywinski, PRB 82, 174511 (2010)
36. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
The role of spin-orbit coupling (SOC)
z
E.g. reflection through a vertical
plane perpendicular to the y axis:
v, J I C2y, J
y x
Quintanilla, Hillier, Annett and Cywinski, PRB 82, 174511 (2010)
37. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
The role of spin-orbit coupling (SOC)
z
E.g. reflection through a vertical
plane perpendicular to the y axis:
v, J I C2y, J
y x
Quintanilla, Hillier, Annett and Cywinski, PRB 82, 174511 (2010)
38. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
The role of spin-orbit coupling (SOC)
z
E.g. reflection through a vertical
plane perpendicular to the y axis:
v, J I C2y, J
y x
Quintanilla, Hillier, Annett and Cywinski, PRB 82, 174511 (2010)
39. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
The role of spin-orbit coupling (SOC)
z
E.g. reflection through a vertical
plane perpendicular to the y axis:
v, J I C2y, J
This affects d(k) (a vector under
spin rotations).
y x
Quintanilla, Hillier, Annett and Cywinski, PRB 82, 174511 (2010)
40. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
The role of spin-orbit coupling (SOC)
z
E.g. reflection through a vertical
plane perpendicular to the y axis:
v, J I C2y, J
This affects d(k) (a vector under
spin rotations).
It does not affect 0(k) (a scalar).
y x
Quintanilla, Hillier, Annett and Cywinski, PRB 82, 174511 (2010)
41. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
The role of spin-orbit coupling (SOC)
C2v,Jno t Gap function, Gap function,
singlet component triplet component
A1 (k) = A d(k) = (Bky,Ckx,Dkxkykz)
A2 (k) = AkxkY d(k) = (Bkx,Cky,Dkz)
B1 (k) = AkXkZ d(k) = (Bkxkykz,Ckz,Dky)
B2 (k) = AkYkZ d(k) = (Bkz, Ckxkykz,Dkx)
Quintanilla, Hillier, Annett and Cywinski, PRB 82, 174511 (2010)
42. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
The role of spin-orbit coupling (SOC)
C2v,Jno t Gap function, Gap function,
singlet component triplet component
A1 (k) = A d(k) = (Bky,Ckx,Dkxkykz)
A2 (k) = AkxkY d(k) = (Bkx,Cky,Dkz)
B1 (k) = AkXkZ d(k) = (Bkxkykz,Ckz,Dky)
B2 (k) = AkYkZ d(k) = (Bkz, Ckxkykz,Dkx)
None of these break time-reversal symmetry!
Quintanilla, Hillier, Annett and Cywinski, PRB 82, 174511 (2010)
43. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
How could this happen?
k 0 k i y dk .σi y
ˆ ˆ ˆ ˆ
Gap matrices evolve smoothly as SOC is turned on.
E.g. Ai y
ˆ ( 1A1 )
ˆ ˆ ˆ
A i y Bk y , Ck x , Dk x k y k z .σ i y ( A1 )
for B = C = D = 0
Quintanilla, Hillier, Annett and Cywinski, PRB 82, 174511 (2010)
44. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
How could this happen?
Some instabilities split in two under the influence of SOC:
E.g. 1, i,0k z .σi y
ˆ ˆ ( 3A1(b) )
ˆ
Ak y k z i y Bk z , Ck x k y k z , Dk x .σ i y
ˆ ˆ
( B2 )
with A, B, C, D 0,1,0,0
i
ˆ
Ak x k z i y Bk x k y k z , Ck z , Dk y .σ i y
ˆ ˆ
( B1 )
with A, B, C, D 0,0,1,0
Quintanilla, Hillier, Annett and Cywinski, PRB 82, 174511 (2010)
47. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
The role of spin-orbit coupling (SOC)
The second (lower-Tc) instability can be symmetry-breaking
because it is no longer an instability of the normal state:
B2 The experiments show a
(kz,0,0) transition straight into the
3A (b)
1 broken TRS phase
(k ,ik ,0) B1 ⇒ SOC must be small in
z z
i(0,kz,0)
LaNiC2
N.B. singlet component must
be very small too.
SOC
Quintanilla, Hillier, Annett and Cywinski, PRB 82, 174511 (2010)
49. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
Recap
What have we learned about LaNiC2?
50. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
Recap
What have we learned about LaNiC2?
Experimental observation:
the superconducting state
breaks time-reversal symmetry.
51. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
Recap
What have we learned about LaNiC2?
Experimental observation:
the superconducting state
breaks time-reversal symmetry.
Theoretical implications:
non-unitary triplet pairing ; weak SOC ; split transition.
52. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
Recap
What have we learned about LaNiC2?
Experimental observation:
the superconducting state
breaks time-reversal symmetry.
Theoretical implications:
non-unitary triplet pairing ; weak SOC ; split transition.
What do we not know yet?
53. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
Recap
What have we learned about LaNiC2?
Experimental observation:
the superconducting state
breaks time-reversal symmetry.
Theoretical implications:
non-unitary triplet pairing ; weak SOC ; split transition.
What do we not know yet?
Which of the four pairing symmetries?
54. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
Recap
What have we learned about LaNiC2?
Experimental observation:
the superconducting state
breaks time-reversal symmetry.
Theoretical implications:
non-unitary triplet pairing ; weak SOC ; split transition.
What do we not know yet?
Which of the four pairing symmetries?
Why non-unitary?
55. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
Recap
What have we learned about LaNiC2?
Experimental observation:
the superconducting state
breaks time-reversal symmetry.
Theoretical implications:
non-unitary triplet pairing ; weak SOC ; split transition.
What do we not know yet?
Which of the four pairing symmetries?
Why non-unitary?
Take this home:
56. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
Recap
What have we learned about LaNiC2?
Experimental observation:
the superconducting state
breaks time-reversal symmetry.
Theoretical implications:
non-unitary triplet pairing ; weak SOC ; split transition.
What do we not know yet?
Which of the four pairing symmetries?
Why non-unitary?
Take this home:
•There’s more than Rashba to noncentrosymmetric
superconductors
57. CMMP’10, Warwick, 15 Dec 2010 blogs.kent.ac.uk/strongcorrelations
Recap
What have we learned about LaNiC2?
Experimental observation:
the superconducting state
breaks time-reversal symmetry.
Theoretical implications:
non-unitary triplet pairing ; weak SOC ; split transition.
What do we not know yet?
Which of the four pairing symmetries?
Why non-unitary?
Take this home:
•There’s more than Rashba to noncentrosymmetric
superconductors
•There’s more than strong correlations to unconventional pairing