Anda di halaman 1dari 37

Spin-spiral calculations of the magnetic properties of Fe-based superconductors

Alexander Yaresko Max Planck Institute for Solid State Research, Stuttgart, Germany
FESC 2010 May 11, 2010

FESC 2010

in collaboration with

Guo-Qiang Liu, Viktor Antonov, and Ole Andersen MPI FKF

special thanks to: Lilia Boeri, George Jackeli, Wei Ku for helpful discussions

FESC 2010

overview

band structure, nesting, and susceptibility q-dependence of the total energy from spin-spiral calculations effect of electron and hole doping applicability of the Heisenberg model interlayer coupling conclusions

FESC 2010

new class of unconventional superonductors

150

T
120

90

T(K)

60

SDW
T
C

30

SC
0 0.0 0.2

x in Ba

0.4

1-

x x

Fe As
2

0.6

0.8

1.0

H.Luetkens,et al Nature Mat.8, 305 (2009)

H. Chen, et al, EPL 85, 17006 (2008)

undoped compounds: orthorombic distortions + AFM order when doping or pressure suppress magnetic order, superconductivity
comes into play (TC >50K) LiFeAs: nonmagnetic with (TC 18K) J. Tapp, et al, PRB 78, 060505 (2008) electron-phonon coupling seems to be too weak to explain high TC
L. Boeri, et al, PRL 101, 026403 (2008)

it is important to understand magnetic properties!


FESC 2010 4

crystal structure
MFe2 As2 M=Sr,Ba I4/mmm (122) LiFeAs P4/nmm (111)
La Fe As O
Ba Fe As

LaFeAsO P4/nmm (1111)

Li Fe As

Fe is surrounded by slightly distorted As4 tetrahedra Fe ions form a square lattice FeAs layers are separated by a LaO, M or Li layer(s)
FESC 2010 5

two limits
weak magnetic perturbation strong magnetic perturbation

noninteracting susceptibility

self-consistent calculations for spin spirals

0 (q, )
? ? ? ? ?

do the two limits always lead to the same conclusions how electron and hole doping affect the magnetic properties is stripe AFM order the ground state for doped compounds can the j1 j2 Heisenberg model describe the calculated E(q) similarities and differences between various families compounds

FESC 2010

two limits
weak magnetic perturbation strong magnetic perturbation

noninteracting susceptibility

self-consistent calculations for spin spirals

0 (q, )
? ? ? ? ?

do the two limits always lead to the same conclusions how electron and hole doping affect the magnetic properties is stripe AFM order the ground state for doped compounds can the j1 j2 Heisenberg model describe the calculated E(q) similarities and differences between various families compounds

details: L(S)DA calculations were performed using the LMTO method for experimental high-temperature tetragonal crystal structures orthorhombic distortions were neglected electron ( > 0) or hole ( < 0) doping per Fe ion was simulated by using the virtual crystal approximation
FESC 2010 6

t2 bands for BaFe2 As2

xy

yz,xz
N P X

M T

0
Energy (eV)

-1
T N

-2 T N P T M

three (2dyz,zx ; 1dxy ) hole-like pokets around and two electron-like (2dyz,zx +dxy ) pokets around X

FESC 2010

and for LiFeAs


xy yz,xz
Z A X M

0
Energy (eV)

-1
X

-2 M X Z

similar hole-like and electron-like pockets the size of the dxy FS is larger than dyz,zx ones less effective nesting the width of dxy and dyz,zx bands is larger than in BaFe2 As2
(dFe-Fe =2.67 A vs. 2.80 A)?
FESC 2010 8

noninteracting susceptibility for LaFeAsO

Im ( 0)/ (arb. units)

50 40 30

peak at X due to FS nesting

20 10

M=X

0
Re (0) (arb. units)

90 80 70 60 50
_ _

=0

J. Dong, et al, EPL 83, 27006 (2008), I. Mazin, et al, PRL 101,
X M

057003 (2008). . .
9

FESC 2010

noninteracting susceptibility for LaFeAsO

Im ( 0)/ (arb. units)

50 40 30

peak at X due to FS nesting

20 10

M=X

0
Re (0) (arb. units)

90 80 70 60 50
_ _

=0

J. Dong, et al, EPL 83, 27006 (2008), I. Mazin, et al, PRL 101,
X M

057003 (2008). . .
9

FESC 2010

for LaFeAsO1x Fx
doping sup presses the X peak maximum shifts to an incommensurate q

electron

Im ( 0)/ (arb. units)

50 40 30 20 10 0

=0.2 (x=0.2)

Re (0) (arb. units)

90 80 70 60 50
_ _

=0 =0.1 =0.2

M=X

10

FESC 2010

for LaFeAsO1x Fx
doping sup presses the X peak maximum shifts to an incommensurate q at which FS touch

electron

Im ( 0)/ (arb. units)

50 40 30 20 10 0

=0.2 (x=0.2)

Re (0) (arb. units)

90 80 70 60 50
_ _

=0 =0.1 =0.2

M=X

10

FESC 2010

. . . BaFe2 As2
30
Im ( 0)/ (arb. units)
_

X=X

20
_

10

0
Re (0) (arb. units)

80

=0
_

P=X

70
_

60

50

11

FESC 2010

. . . Ba1x Kx Fe2 As2 (x = 2||)


30
Im ( 0)/ (arb. units)

20

10

hole doping also suppresses

0
Re (0) (arb. units)

the peak at X the maximum shifts


=0 =-0.1 =-0.2

80

70

60

The peak of Re(q) at X is suppressed by either electron or hole doping

50

12

FESC 2010

. . . LiFeAs
30
Im ( 0)/ (arb. units)

20

10

0
Re (0) (arb. units)

80

70

60

nesing is less effective (q) is lower the X peak is split already


in undoped LiFeAs

_ _

50

13

FESC 2010

response to nite AFM perturbation


BaFe2 As2
0.4
xy yz xz 3z2-r2 x2-y2 MFe/5

LiFeAs
0.4
xy yz xz 3z2-1 x2-y2 MFe/5

0.3

0.3

M (B)

0.2

M (B)

0.2

0.1 AFM || x; FM || y 0.0 0.0 0.2 0.4 0.6 0.8 =V -V (eV) 1.0 1.2 1.4

0.1 AFM || x; FM || y 0.0 0.0 0.2 0.4 0.6 0.8 =V -V (eV) 1.0 1.2 1.4

induced moment M vs. = V V

at small the moment is induced faster on obitals forming nested FS


BaFe2 As2 : dMdxy /d = 0.60 dMdx2 y2 /d = 0.18 LiFeAs: dMdxy /d = 0.45 dMdx2 y2 /d = 0.12 the degeneracy of dzx and dyz states is lifted at large nesting is less important: dMdxy /d=dMdx2 y2 /d=0.18 What happens when Fe moments are large?
FESC 2010 14

self-consistent calculations for spin spiral

X ;

q = (q, 0),

q =0.875

X M;

q = (1, q),

q =0.125

EH (j1 + 2j2 ) cos(q) + j1

EH (j1 2j2 ) cos(q) j1

EH is a monotonic function of q
FESC 2010 15

self-consistent calculations for spin spiral

X ;

q = (q, 0),

q =0.

X M;

q = (1, q),

q =1

EH (j1 + 2j2 ) cos(q) + j1

EH (j1 2j2 ) cos(q) j1

EH is a monotonic function of q
FESC 2010 15

spin spirals along M

M;

q = (q, q),

q =0.125

EH 2j1 cos(q) + 2j2 cos2 (q)


a local minimum appears if j2 > j1 /2

FESC 2010

16

LaFeAsO

(0,0)

(,0)

(,)

(0,0)

1.5
MFe (B)

1.0

the E(q) minimum is at X


(stripe AFM order)

0.5

0.0 0
E(q)-E(0) (meV/Fe)

E(q) is reasonably well described


by the j1 j2 Heisenberg model

j1 =81 meV, j2 =57 meV, j2 /j1 =0.71


-50

nonmagnetic solution at small |q|

-100

FESC 2010

17

electron doping in LaFeAsO1x Fx (x = )

(0,0)

(,0)

(,)

(0,0)

deviation from the Heisenberg-like


behaviour increases with doping a new minimum at the XM line

1.5
MFe (B)

1.0 =0 =0.1 =0.2 =0.3

0.5

0.0 0
E(q)-E(0) (meV/Fe)

-50

-100

FESC 2010

18

electron doping in LaFeAsO1x Fx (x = )

(0,0)

(,0)

(,)

(0,0)

deviation from the Heisenberg-like


behaviour increases with doping a new minimum at the XM line

1.5
MFe (B)

1.0 =0 =0.1 =0.2 =0.3

0.5

stripe AFM order becomes unstable already at small

0.0 0
E(q)-E(0) (meV/Fe)

-50

-100

FESC 2010

18

band-structure effect?
LaFeAsO; =0
2
dyz dxz dxy

LaFeAsO1x Fx ; =0.1
20
q=(1,0.0)

DOS (1/eV/atom)

15
0 -1.5 -1.0 -0.5 0.0 0.5 Energy (eV) 1.0 1.5

DOS (states/eV)

10

1.5 1.0 0.5 0.0 -0.5 -1.0 -1.5

dyz dyz

Energy (eV)

0 -1.0
Y X

-0.5

0.0 0.5 Energy (eV)

1.0

1.5

a narrow Fe dyz DOS peak just above EF is lled when LaFeAsO is doped with electrons
FESC 2010 19

band-structure effect?
LaFeAsO; =0
2
dyz dxz dxy

LaFeAsO1x Fx ; =0.1
20
q=(1,0.0) q=(1,0.05) q=(1,0.10)

DOS (1/eV/atom)

15
0 -1.5 -1.0 -0.5 0.0 0.5 Energy (eV) 1.0 1.5

DOS (states/eV)

10

1.5 1.0 0.5 0.0 -0.5 -1.0 -1.5

dyz dyz

Energy (eV)

0 -1.0
Y X

-0.5

0.0 0.5 Energy (eV)

1.0

1.5

a narrow Fe dyz DOS peak just above EF is lled when LaFeAsO is doped with electrons The peak splits at incommensurate q
FESC 2010 19

BaFe2 As2

(0,0)

(,0)

(,)

(0,0)

1.5

MFe (B)

1.0

0.5

0.0 0
E(q)-E(0) (meV/Fe)

the minimum is at X E(q) is Heisenberg-like j1 =95 meV; j2 =73 meV; j2 /j1 =0.77

-50

FESC 2010

20

hole doping in Ba1x Kx Fe2 As2 (x = 2||)

(0,0)

(,0)

(,)

(0,0)

1.5

the E(q) minimum stays at X up

to x =0.6 ( = 0.3) the stabilization energy

MFe (B)

1.0 =0 =-0.1 =-0.2 =-0.3

|E(X) E(0)|
decreases with doping

0.5

0.0 0
E(q)-E(0) (meV/Fe)

-50

FESC 2010

21

hole doping in Ba1x Kx Fe2 As2 (x = 2||)

(0,0)

(,0)

(,)

(0,0)

1.5

the E(q) minimum stays at X up

to x =0.6 ( = 0.3) the stabilization energy

MFe (B)

1.0 =0 =-0.1 =-0.2 =-0.3

|E(X) E(0)|
decreases with doping stripe AFM order is more stable
150

0.5

0.0 0
E(q)-E(0) (meV/Fe)

T
120

90

T(K)

60

SDW
T
C

-50

30

SC
0 0.0 0.2

x in Ba

0.4

1-

x x

Fe As
2

0.6

0.8

1.0

FESC 2010

21

hole vs. electron doping in MFe2 As2 (M=Ba, Sr)


hole doping ( < 0): M1x Kx Fe2 As2 (x = 2|| ) electron doping ( > 0): M(Fe1x Cox )2 As2 (x = )
0
E(q)-E(0) (meV/Fe)

-50
BaFe2As2

qz=
_ _

SrFe2As2

=-0.1 =0 =0.1
_ _

already weak electron doping ( < 0.1) destabilizes stripe AFM order magnetic transition in Ba(Fe1x Cox )2 As2 is suppressed at x 0.06
J. Chu, et al PRB 79, 014506 (2009)

FESC 2010

22

LiFeAs

(0,0)

(,0)

(,)

(0,0)

1.5

MFe (B)

1.0

0.5

Fe moment is much smaller magnetic solution exists in a

0.0 0
E(q)-E(0) (meV/Fe)

narrow range around X stabilization energy is only 12 meV/Fe

The loss of kinetic energy is higher because of compressed lattice?


-50

FESC 2010

23

does the Heisenberg model work for undoped compounds?

q=( ,0):
two decoupled AFM sublattices each Fe has two FM and two AFM nn

EH

= j1 [cos + cos( )] 2j2 = 2j2 = const.

FESC 2010

24

does the Heisenberg model work for undoped compounds?

q=( ,0):
two decoupled AFM sublattices each Fe has two FM and two AFM nn

EH

= j1 [cos + cos( )] 2j2 = 2j2 = const.

FESC 2010

24

does the Heisenberg model work for undoped compounds?

q=( ,0):
two decoupled AFM sublattices each Fe has two FM and two AFM nn

EH

= j1 [cos + cos( )] 2j2 = 2j2 = const.

FESC 2010

24

NO!
50 40
E()-E(0) (meV)

La =0 La =0.2

Ba =0 Ba =-0.2

30 20 10 0 0 45 90 (deg) 135 180

already LSDA prefers collinear stripe AFM order E() sin2 (Si Sj )2 energy variation is suppressed by doping, especially in BaFe2 As2
FESC 2010 25

orbital degrees of freedom?


variation of orbital occupations as a function of
0.8
yz zx xy yz zx xy

0.7
n

0.6

0.5 0 45 90 (deg) 135 180

stripe AFM order lifts the degeneracy of dyz and dzx orbitals (n 0.15)
is this the reason for anisotropic j1 and j1 ? ||

Z. Yin, et al PRL 101, 047001 (2008)


FESC 2010 26

interlayer exchange coupling


stripe AFM order in the ab-plane
E()-E(0) (meV/Fe)

0 -1 -2 -3 -4

Ba Fe As

BaFe2As2 LaFeAsO

90 (deg)

180

E(FMc ) E(AFMc ) 4 meV/Fe


strong AFM interlayer coupling in MFe2 As2

FMc : qz = 0, = 0 AFMc : qz = , = 180


FESC 2010

27

conclusions

In the undoped compounds the minimum of the total energy is found at

q=( ,0) corresponding to stripe AFM order LaFeAsO1x Fx ( > 0): the minimum shifts to incommensurate wave

vectors with doping and stripe AFM order becomes unstable M1x Kx Fe2 As2 ( < 0): stripe AFM order remains stable in a wide range of K concentrations. However, the stabilization energy decreases with doping M(Fe1x Cox )2 As2 ( > 0): stripe AFM order is rapidly destabilized when FeAs layer is doped with electrons LiFeAs: the Fe moment and the stabilization energy are much smaller strong dependence of the energy on shows that the magnetic interactions between Fe spins cannot be described by the simple j1 j2 Heisenberg model the interlayer coupling in MFe2 As2 is much stronger than in LaFeAsO

FESC 2010

28

Anda mungkin juga menyukai