LDA+U calculations using pwscf

Notable failures of LDA/GGA: transition-metal oxides introduction to correlation LDA+U: general formulation Examples and applications Computing U from first-principles

Failures of LDA/GGA: transition metal oxides

Cubic, rock-salt structure TM ion

Oxygen

Antiferromagnetic (AF) ground state possible structural distorsions (FeO)

rhombohedral symmetry and

Conduction properties (exp): insulators (Mott/charge-transfer kind)

FeO: GGA results

Fe2+

Antiferromagnetic ground state: OK crystal structure (cubic): OK but

We obtain a metal !!!

NiO: GGA results

Antiferromagnetic ground state: OK crystal structure (cubic): OK Crystal field produces a band gap, but

The energy gap is too small

Mott insulators: U vs W
Two quantities are to be considered: U W on-site electron-electron repulsion bandwidth (hopping amplitude, related to kinetic energy)

Two different regimes: W/U >> 1: the energy is minimized making the kinetic term as small as possible through delocalization (little price is paid on the occupied atomic sites to overcome repulsion U) W/U << 1: the kinetic energy of electrons is not large enough to overcome the on-site repulsion. Electrons undergo a Mott localization
I. G. Austin and N. F. Mott, Science 168, 71 (1970)

LDA/GGA approximations to DFT always tend to over-delocalize electrons: U is not well accounted for electronic energy functionals are affected by self-interaction

H2 molecule in DFT
Lets consider the spherically and system-averaged pair electronic density:

where

d r12 N ( N 1) 2 f (r12 ) = | (r12 ,R, r3 ,...., rN ) | 4 d Rd r3 ....d rN 2 1 ... N

r12 =| r1 r2 | R = (r1 +r2 ) / 2

P Gori-Giorgi et al. cond-mat/0605174

At very large internuclear distances each electron is still split between the two sites!

From He to 2H: abc of correlation

Two protons a and b (as classical objects, no nuclear forces) and two electrons 1 and 2: electronic ground state at increasing inter-proton distance. We use 1s single electron states (H) or bonding () and antibonding (*) linear combinations (H2+) 1) d = 0, a=b: He atom

(1,2) = N [ a (1) a (2 )]

ionic terms

2) d > 0: H molecule

a ) (1,2) = N [ a (1)b (2 ) + b (1) a (2 )]

b) (1,2) = N [ (1) (2 )] = N {[ a (1)b (2 ) + b (1) a (2 )] + [ a (1) a (2 ) + b (1)b (2 )]}

2) d >> 0: distinct H atoms each electron is on the 1s ground state of each H atom:

(i ) = 1s ( i )

From He to 2H: continuous solution

H molecule: the importance of ionic terms should disappear with distance. The ground state wave function should be something like:

(1,2) = N {[ a (1)b (2) + b (1) a (2)] + (d )[ a (1) a (2) + b (1)b (2)]}

This can be achieved through a linear combination of bonding () and antibonding (*) Slater determinants:

(1,2) = N {a(d )[ (1) (2)] + b(d )[ * (1) * (2 )]} =

a(d) and b(d) can be treated and computed as variational parameters

(multi reference CI methods etc)

Multi-reference ground state and correlation

Ecorr = Eexact - EHF

Multi-reference (multi-determinant) ground state wavefunction

c c

Four-index interactions (instead of two as in HF):

U ijkl 'U ijlk

instead of

U ijij 'U ijji

The LDA+U energy functional

The LDA+U method consists in a correction to the LDA (or GGA) energy functional to give a better description of electronic correlations. It is shaped on a Hubbard-like Hamiltonian including effective on-site interactions. It was introduced and developed by Anisimov and coworkers (1990-1995).
I ELDA+U n ( r ) = ELDA n ( r ) + EHub { nm } Edc { n I }

Fully rotationally invariant formulation (Lichtnestein et al. PRB 1995)

E Hub

[{n }] = 1 {m, m' '| V 2{ }

I mm ' m , I ,
I

I I | m' , m' ' ' nmm ' nm' ''' ee 'm

I I + ( m, m' ' | Vee | m' , m' ' ' m, m' ' | Vee | m' ' ' , m' ) nmm ' nm''m '''

Edc n

[{ }]
k ,v

U I I J I I = n n 1 n n 1 I 2 I , 2

)
n I = n I

Occupations:
I I I nmm ' = f kv kv | m ' m | kv

I n I = nmm m

Electronic interactions
Hartree-Fock formalism (for d states): from the expansion of 1/|r-r| in spherical harmonics we get:

m, m '' | Vee | m ', m ''' = d r d r'

* * m (r) m'' (r') m' (r) m''' (r')

|r-r'|

= ak ( m, m ', m '', m ''' ) F K

k

4 k * ak (m, m' , m' ' , m' ' ') = lm | Ykq | lm' lm' ' | Ykq | lm' ' ' 2k + 1 q = k
The double counting term is evaluated as the Mean Field Approximation of the Hubbard one. So in the expression of EHub we put
I nmm ' = mm '

and get

1 Tr n I 2l + 1

[ ]

1 ' m, m' Vee m, m' = F 0 (2l + 1)2 m,m F2 + F4 1 J= ' m, m'| Vee | m' , m = 14 2l (2l + 1) m ,m U=
k

Keep in mind: we want screened (effective) interactions; F are unscreened

A simplified approach
First order approximation: lets neglect the exchange interaction J:

We get:

J = F2 = F4 = 0
I I EU nmm ' = EHub nmm ' Edc n I =

[{ }]

[{ }]

[{ }]

U = 2

U I I I nmm nmm 'nm 'm = 2 I m , m'

Tr n I ( 1 n I )
I ,

Note: a) U is the only interaction parameter in the functional Note: b) the rotational invariance is preserved. This is the formula implemented in PWscf. We have

ELDA+U

U I {nmm ' } = ELDA [ ] + = ELDA [ ] + EU 2

Tr n I ( 1 n I )
I ,

How does it work?

Because of rotational invariance we can use a diagonal representation:

where

U EU = 2
I n I v m = m v m

I, m

I I m ( 1 m )

I I I m = f kv kv | m m | kv k, v

Potential:

EU U = VU | kv = * kv 2

I I I 1 2m ) | m m | kv ( I , m

I m

I m

1 > VU < 0 2 1 < VU > 0 2

Partial occupations are discouraged Gap opening! Eg U

Input file for LDA+U calculation with PWscf

Only the namelist system is modified: &system . . . lda_plus_u = .true., Hubbard_U(1) = $U1, Hubbard_U(2) = $U2, . . Hubbard_U(ntype) = $Untype, . . /

There is a different U for each distinct type of Hubbard atom U is in eV Typical values: U is rarely larger than 7-8 eV (in most cases 0<U<5 eV)

There might be need of using finer k-points grids for a better evaluation of on-site occupations.

FeO

Does it work?

NiO

M. Cococcioni and S. de Gironcoli PRB 71, 035105 (2005)

Mineral in the Earths mantle: Fe2SiO4

band structure

structural parameters
M. Cococcioni and S. de Gironcoli PRB 71, 035105 (2005)

Cathode material for Li ion batteries: LixFePO4

High voltage (3.5 V) and high theoretical capacity Electrochemical stability Thermal safety Non toxicity Low cost and easy synthesis Issue: low electronic conductivity Properties Crystal structure Magnetic structure Formation Energy LixFePO4 Voltage TM valence states (0< x <1) Exp olivine AF >0 3.5 V Fe2+/Fe3+ GGA olivine AF <0 2.9 V Fe2.5+ GGA+U olivine AF >0 3.5 V Fe2+/Fe3+ PRB 69, 201101 (2004) PRB 70, 235121 (2004)

Similar agreement In other compounds!

Exp GGA+U

V=

E ( Lix2 FePO4 ) E ( Lix1 FePO4 ) ( x2 x1 ) E ( Li ) ( x2 x1 ) F

Molecular chemical reaction: GGA+U studies

Spin states in the Heme group
Fe site

H2 addition-elimination to FeO+

O2

Five coordinated iron

Six coordinated iron (O2, CO, etc)

Magnetic ground state Exp GGA B3LYP HF GGA+U quintuplet (S=2) triplet triplet quintuplet quintuplet singlet (S=0) singlet singlet quintuplet singlet
H. J. Kulik, M. Cococcioni, D. Scherlis and N. Marzari, submitted for publication

D. Scherlis, H. L. Sit, M. Cococcioni and N. Marzari, in preparation

What does U mean?

EU = U 2

I, m

I I m ( 1 m )

The +U correction is the one needed to recover the exact behavior of the energy. What is the physical meaning of U?

Evaluation of U
U is the unphysical curvature of the DFT total energy We want effective interactions: we evaluate U from the DFT ground state A free-electron contribution (due to rehybridization) is to be subtracted in crystals:
2 GGA 2 GGA 0 I 2

d E d E U= I 2 d (n ) d (n )

From self-consistent ground state (screened response)

From fixed-potential diagonalization (Kohn-Sham response)

Second derivatives
Second derivatives of the energy are not directly accessible We apply a potential shift to the d states of each Hubbard atom I and use a Legendre transform:

Legendre transform

I E [{ I }] = min E [n(r )] + I n n(r ) I

I I min

[{n }] = min E[{ }] n = E[{ }] n E

I

d E nJ = I n J d nI

[{ }]

({ })

d 2E nJ
I 2

[{ }] = d ({n }) dn d (n )
J I I

Linear response
Using I as perturbation parameters we can easily evaluate the response matrices:

dn = d J
0 IJ

I 0

dn IJ = d J

the fully interacting (screened) one

is the bare response of the system,

Run a self consistent (unperturbed) calculation. Starting from saved potential and wavefunction add the perturbation The response 0 is evaluated at the first iteration (at fixed potential) The response is evaluated at self consistency

The effective interaction is finally obtained as:

d I d I = 0 1 1 U = I + I dn d n0

II

M. Cococcioni and S. de Gironcoli PRB 71, 035105 (2005)

What screening?
U = 0 1 1
U = 0 1 0 1 0 1

is equivalent to

= 0 + 0U

= 0 + 0U 0

diagrammatic representation:

U is a bare interaction; U is the dressed (effective) one. We use U and let the electrons perform the screening through the relaxation to their self-consistent ground state. But werent we looking for screened effective interactions? What screening are we including?

Linear response: what is inside U?

Lets apply a perturbation to the external potential and study the response of the electronic charge density we can introduce response functions and write:

Vext (r )

n(r )

where:

n(r ) = (r, r ' )Vext (r ' )dr ' = 0 (r, r ' )V (r ' )dr '

(1) (2)

Inverting (1) we get:

1

v xc (r ) n(r' ) V (r ) = Vext + n(r ' )dr ' dr '+ n(r ' ) | r r'|

V (r ) = 0 1 (r, r ' )n(r ' )dr ' Vext (r ) = (r, r ' )n(r ' )dr '

(3)

From (2) and (3) we easily obtain:

vxc (r ) 1 1 1 0 (r, r ' ) (r, r ' ) = + | r r ' | n(r ' )

xc kernel

The Coulomb interaction is screened by the xc kernel

See S. Baroni et al., Rev. Mod. Phys. 73, 515 (2001) and refs quoted therein

unperturbed run

Input file for computing U with PWscf

perturbed run
foreach a (0.d0 0.01 0.01) . &control . restart_mode = from_scratch . / &system . lda_plus_u = .true., Hubbard_U(1) = 1.d-20, . Hubbard_U(ntype) = 1.d-20, Hubbard_alpha(1) = $a, Hubbard_alpha(2) = 0.d0, . / &electrons . startingwfc = file, startingpot = file, diago_thr_init = $ethr_conv . / . end

&system . . . lda_plus_u = .true., Hubbard_U(1) = 1.d-20, Hubbard_U(2) = 1.d-20, . . Hubbard_U(ntype) = 1.d-20, . . / Hubbard_alpha values are symmetrically distributed around 0 (typically from 0.1 to 0.1 eV) the perturbed atom has to be treated as of different kind the series of perturbed runs at different Hubbard_alpha is repeated for every different kind of Hubbard atom diago_thr_init is chosen close to the last ethr of the unperturbed run (ethr_conv)

Constructing the response matrices

The perturbation must be fully accomodated in the considered cell (to avoid interactions with its periodic replicas): we need to run calculations in supercells Supercell: N Hubbard atoms of M distinct types. Response matrices: NxN Perturbed calculations: to be repeated for each of the M types of Hubbard atoms First M columns of the response matrices are calculated straightforwardly. The other N-M columns are obtained attributing the same response to equivalent shells of neighbors when perturbation is applied on equivalent atoms Background term (perturbation must be neutral) and other technical details: please refer to M. C. and S. d. G. PRB 71, 035105 (2005) Converge the computed U values with the size of the supercell

Importance of computing U
Consistency with the expression of the +U functional and with the choice of localized basis set: the computed U is exactly the one needed Sensitiveness to spin states, chemical/physical environments, structural changes (important in studying e.g. chemical reactions, phase transitions) FeO rhombohedral distorsion under pressure (MgxFe1-x)O HS to LS transition under pressure

M. Cococcioni and S. de Gironcoli PRB 71, 035105 (2005)

T. Tsuchiya, R. M. Wentzcovitch, C. R. S. da Silva and S. De Gironcoli, Phys. Rev. Lett. 96, 198501 (2006)

Other useful references

A. Szabo and N. S. Ostlund, Modern quantum chemistry, Dover Publications Inc., 1996 Perdew et al., PRL 49, 1691 (1982) Anisimov, Zaanen, and Andersen, PRB 44, 943 (1991) Anisimov et al., PRB 48, 16929 (1993) Liechtenstein, Anisimov, and Zaanen, PRB 52, R5467 (1995) Anisimov, Aryasetiawan, and Liechtenstein, J. Phys.: Cond Matt 9, 767 (1997) Pickett, Erwin, and Ethridge, PRB 58, 1201 (1998)

Work in progress
Compute J better functionals (for better energetics) auto-consistent calculation of U (from a LDA+U ground state) on-the-fly evaluation of U inter-site interactions phonon+U