Computational Implementations in Electronic Structure … · Computational Implementations in...
Transcript of Computational Implementations in Electronic Structure … · Computational Implementations in...
Computational Implementations in
Electronic Structure Theory
N. N. Lathiotakis
Theoretical and Physical Chemistry Institute
National Hellenic Research Foundation
Athens, Greece
OutlineOutline
1. Electronic structure methods
2. Introduction to density-functional theory (DFT)
• Fundamentals
• Flavors of DFT.
• Successes and failures
3. Computer implementations
• Gaussian basis sets
• Expansion in Plane waves
• APWs, wavelets, SWs, grid methods
4. Introduction to Quantum Espresso
• Basic input, capabilities
5. Some Results
• Graphene force fields
• Graphene allotropes under pressure
• Magnetic impurities in ZnO
BooksBooks
• Fundamentals:• Dreizler R.M., Gross E.K.U. (1990), Density functional theory.
• Parr R. G., Yang W. (1989), Density-Functional Theory of Atoms
and Molecules.
• Applications:• C. Fiolhais, F. Nogueira, M. Marques (Eds) (2003),
A Primer in Density Functional Theory.
• W. Koch, M. C. Holthausen (2000),
A Chemist’s Guide to Density Functional Theory
• Richard M. Martin (2004),
Electronic Structure: Basic theory and practical Methods
• E. Kaxiras (2003), Atomic and electronic structure of solids
• D.S Sholl, J. A. Steckel (2009), Density Functional Theory: a
practical introduction.
1. Electronic structure methods1. Electronic structure methods
The problem in electronic structure theoryThe problem in electronic structure theory
• To solve the quantum mechanical problem of a system of N particles
(e.g. electrons or nuclei) in the presence of external electric and
magnetic fields.
• Born Oppenheimer approximation: electronic and nuclear “motions”
are separated and could be determined independently: Fixed usually
classical nuclear motion.
• Without any applied fields the problem reduces to the quantum
mechanical description of N electrons “living” in the external potential
of the nuclei.
The many-body Schrödinger EquationThe many-body Schrödinger Equation
1 2 1 2ˆ ( , , , ) ( , , , )
N NH r r r E r r rΨ = Ψ
� � � � � �⋯ ⋯
ˆ ˆ ˆ ˆH T V W= + +The Hamiltonian:
We are looking for the solution 1 2( , , , )
Nr r rΨ� � �⋯ of
where
2
2
1 1 , 1
1ˆ ˆ ˆ, ( ), ( , )2 2
N N N
i i i j
i i i j
i j
T V V r W W r rm= = =
≠
= − ∇ = =∑ ∑ ∑ℏ � � �
For an atom:
2 2
( ) , ( , )i i j
i i j
Ze eV r W r r
r r r= − = +
−
� � �� �
Equivalently: Minimization of|
HE
< Ψ Ψ >=
< Ψ Ψ >
How useful is Ψ?How useful is Ψ?
Silicon atom:
~1028 Kg100g/HD
~1029 HD’s1 TB / Hard Disc:
2.5 x 1041 bytes4 bytes / number:
1042 numbers10 entries / coordinate:
42 coordinates14 electrons:
Mass of the earth: 6 x 1026 Kg
• Ψ variation is restricted by constraints imposed by symmetry
but still the problem is impossible to solve.
• The full information contained in Ψ is not necessary for the
expectation values of most observables that correspond to
1- and 2-body operators.
Electronic Structure MethodsElectronic Structure Methods
• Determinantal methods:
• Hartree Fock (Ψ is expressed as a single Slater determinant)
• Configuration Interaction (CI) (Ψ is expressed as a linear
combination of all possible Slater determinants)
• Methods based on functionals of reduced density matrices (RDM)/densities:
• DFT
• Reduced Density Matrix Functional Theory (RDMFT)
Reduced density matrices (RDM)Reduced density matrices (RDM)
• The density matrix:
The expectation value of one-body observable:
•Reduced density matrices (of order p, 1 ≤ p < N ):
•The densities (diagonals of density matrices):
•The density (one electron density):
*
1 2 1 2 1 2 1 2( , , , , , , , ) ( , , , ) ( , , , )N N N N Nr r r r r r r r r r r rγ ′ ′ ′ ′ ′ ′= Ψ Ψ� � � � � � � � � � � �⋯ ⋯ ⋯ ⋯
ˆ ˆ{ }N
A Tr Aγ=
1 2 1 2
3 3
1 1 2 1 1 2 1
( , , , , , , , )
( , , , , , , , , , , , , , )
p p p
p N N p p N p p N
r r r r r r
Nd r d r r r r r r r r r r r
p
γ
γ+ + +
′ ′ ′ =
′ ′ ′
∫ ∫
� � � � � �⋯ ⋯
� � � � � � � � � � � �⋯ ⋯ ⋯ ⋯ ⋯
1 2 1 2 1 2( , , , ) ( , , , , , , , )p p p p p
r r r r r r r r rρ γ=� � � � � � � � �⋯ ⋯ ⋯
3 3 *
2 2 2( ) ( , , , ) ( , , , )N N N
r d r d r r r r r r rρ = Ψ Ψ∫ ∫� � � � � �
⋯ ⋯ ⋯ ⋯
The exact energy functional of γ2The exact energy functional of γ2
The one-body, reduced density matrix (1-RDM):
The one-body density (usual particle density):
3 3 *
1 2 2 2( , ) ( , , , ) ( , , , )N N Nr r d r d r r r r r r rγ ′ ′= Ψ Ψ∫ ∫� � � � � � � � � �
⋯ ⋯ ⋯
1 1( ) ( ) ( , )r r r rρ ρ γ= =� � � �
The exact total energy of an electronic system as a functional
of the 2-RDM, γ2 is:
1
3 3 2
1 2 1 1 1 1 1 1 1
3 3 3 2 1 21 2
1 2
1[ , , ] ( ) [ ( , )]
2
( , )( ) ( )
| |
rE E d r d r r r r r
r rd r V r r d r d r
r r
ρ γ ρ δ γ
ρρ
′ ′ ′= = − − ∇
+ +−
∫ ∫
∫ ∫ ∫
� � � � � �
� �� � � � �
� �
• In principle E can be minimized with respect to γ2.
• Unfortunately, the N-representability constraints for γ2 are complicated
and not completely known.
DFT/RDMFTDFT/RDMFT
• The N-representability constraints for ρ and γ1 are simple:
1. Every ρ integrating to N is N-representable.
2. For γ1 the necessary and sufficient conditions are:
where ni’s are the eigenvalues of γ1.
• Two approximate theories:
1. DFT: The total energy is a functional of ρ.
2. RDMFT: The total energy is a functional of γ1.
, 0 1,i i
i
n N n= ≤ ≤∑
DFT is the workhorse of present day electronic
structure calculations
DFT is the workhorse of present day electronic
structure calculations
2. Introduction to DFT2. Introduction to DFT
Hohenberg-Kohn theoremHohenberg-Kohn theorem
( )V r� Ψ ( )rρ
�
CD
C D
The map CD is invertible:
( )V r�
( )rρ�(1-1)
• Non degenerate ground state:
0 0[ ] [ ] | | [ ] .V
E T W Vρ ρ ρ=< Ψ + + Ψ >Due to D-1 map:
By virtue of Rayleigh-Ritz: E[ρ] is minimized by the density
ρ0 corresponding to V0:
0 00 0[ ] [ ]V V
E E Eρ ρ= ≤
Hohenberg-Kohn theoremHohenberg-Kohn theorem
3VΨ
C D• Degenerate ground state:
C is not a map, C-1 is. D is a map but not D-1: (Ψ[ρ] is not defined).
CD-1 is a map. Subsets in Ψ and ρ are disjoint.
1V
3V
2V
2VΨ
1VΨ
3Vρ
1Vρ
2Vρ
Although D-1 is not a map the universal functional
F[ρ] is still uniquely defined:
3| | ( ) ( )F T W E d r r V rρ=< Ψ + Ψ >= − ∫� � �
Coulomb potential
of the nuclei
Kohn-Sham theoremKohn-Sham theorem
The ground state density of the interacting system can be calculated
as ground state density of a non-interacting system with an effective
potential .( )sv r�
• Kohn-Sham system:
2
2
1
3
0
[ ]( ) ( ) ( )2
( ) ( )
( ')[ ]( ) ( ) ' [ ]( )
'
s j j j
occ
j
j
s xc
v r r r
r r
rv r v r d r v r
r r
ρ ϕ ε ϕ
ρ ϕ
ρρ ρ
=
∇− + =
=
= + +−
∑
∫
� � �
� �
�� � � �
� �
Hartree
potential
Exchange-correlation
potential:
( ) (universal)( )
xcxc
Ev r
r
δ
δρ=�
�
Kohn-Sham systemKohn-Sham system
• Transforms the problem into a problem of self-consistent
1-electron equations.
• The universal functional Exc should be approximated.
Approximations:
• Local Density Approximation (LDA):
• GGA and hybrid functionals
• Orbital functionals (OEP, OPM method)
•Meta-GGA: (use kinetic energy density also)
• Is there any meaning in the 1-electron energies, i.e. the
eigenvalues of the KS hamiltonian?
3 hom
XC xc ( ( ))E d r e rρ= ∫� �
3
XC ( , )E d r g ρ ρ= ∇∫��
XC XC 1 2[ , , , ]NE E ϕ ϕ ϕ= ⋯
3
XC ( , , )E d r g ρ ρ τ= ∇∫��
FunctionalsFunctionals
• LDA
• Slater Exchange ( )
• Corellation: Fit for the Homogeneous Electron Gas
Monte Carlo: Ceperley-Alder, PRL45, 566(1980)
Fit: Vosko, Wilk, Nusair, Can. J. of Phys. 58 1200 (1980)
Perdew-Wang, PRB45 13244 (1992)
• GGA (dependence on the density gradient)
1. Non-empirical
PW91 (PRB33, 8800)
PBE (PRL77, 3865)
2. Hybrid
Becke86 (JCP84, 4524)
LYP/BLYP (PRB37, 785)
B3LYP (JPC98, 11623):
• Orbital Functionals (OEP, OPM)
Exact exchange (EXX)
1/3ρ∼
B3LYP B88 LYP LDA
XC X X0 X C C(1 ) (1 )E a E aE bE cE a E= − + + + + −
Successes of LDASuccesses of LDA
<25%Cohesive energies
<2%Lattice constants of solids
Few %Bands of metals/Fermi
surface
<5%Equilibrium distances
<0.5%Atomic and molecular
ground state energies
Typical deviation from exp.Quantity
Failures of LDAFailures of LDA
• Too small ionization potentials
• Negative ions not bound (no electron affinity)
• Band gaps of semiconductors/insulators too small
• Some transition metal oxides predicted as metals
• Van der Walls interaction not described
LDA and GGALDA and GGA
• Both LDA and GGA underestimate the band gaps
of semiconductors/insulators.
more correctfavors close packingstructure
1%(longer)1% (shorter)Bond length
5%100% (too negative)Ec
0.5%5% (not negative enough)Ex
GGALDAProperty
The fundamental gapThe fundamental gap
The difference of the ionization potential and the electron affinity:
XC
KS
gap
)(2)1()1(
∆+=
−−++=−=
ε
NENENEAIE
where XC XCXC
( ) ( )N N
E E
r rδ δ
δ δ
δρ δρ+ −
∆ = −� �
∆XC is the discontinuity of the xc potential upon adding (removing)
an electron. It is zero for LDA, GGA.
LDA and GGA underestimate gaps by a factor of a 1/2
The fundamental gapThe fundamental gap
EXX results concern the KS gap only. If the discontinuity
is added EXX results are as bad as Hartree-FockStädele et al, PRB59 10031 (1999)
The gap for the HSE hybrid functionalThe gap for the HSE hybrid functional
HSE functional gives gaps close to experiment: Heyd et al JCP 123 174101 (2005)
The Fermi surface of CuThe Fermi surface of Cu
6.30.6700.712
6.50.7480.797
2.70.9730.947
10.50.2560.283
1.51.1661.183
1.41.3581.339
3.20.5920.573
Diff (%)Exp (A-1)LDA (A-1)vector
(1 )
(1 0 0 )Q
( 2 )
(1 0 0 )Q
(1 )
(1 1 0 )Q
( 2 )
(1 1 0 )Q( 3 )
(1 1 0 )Q
( 4 )
(1 1 0 )Q
(1 )
(1 1 1 )Q
LDA: Lathiotakis et al, JMMM 185, 293 (1998); Exp.: M.R. Halse, Phil. Tr. Roy. Soc. London, A 265, 507 (1969).
3. Computer Implementations3. Computer Implementations
Computer codesComputer codes
Codes targeting finite systems:
1. Orbitals expanded in Gaussian type orbitals
• Gaussian (commercial)
• Turbomole (commercial)
• molpro (commercial)
• ORCA (open source)
• GAMESS (free closed source)
• DMoL (commercial)
• Psi3 (open source)
• NWCHEM (open source)
2. Real-Space (grid) with pseudopotentials
• OCTOPUS (open source)
Computer codesComputer codes
Codes targeting periodic systems (solids, surfaces, 2D
Systems, polymers:
1. Orbitals expanded in Gaussian type orbitals
• CRYSTAL (commercial)
2. Orbitals expanded in numerical atom centered basis
• SIESTA (open source)
3. Orbitals expanded in Plane Waves
• Quantum Espresso (open source)
• ABINIT (open source)
• CPMD (open source)
• VASP (commercial)
• CASTEP (commercial)
Computer codesComputer codes
4. Orbitals expanded in Augmented Plane Waves (FP):
• ELK (open source)
• EXCITING (open source)
• Fleur (open source)
• Wien2k (commercial)
5. Orbitals expanded in Wavelets (Daubechies):
• BigDFT (open source)
6. Real space (grid):
• GPAW (open source)
4. Quantum Espresso4. Quantum Espresso
The QE projectThe QE project
• ESPRESSO stands for:
opEn Source Package for Research in Electronic Structure, Simulation, and Optimization.
• Open Source: GNU General Public License
• Developers :
DEMOCRITOS National Simulation Center (Trieste) and SISSA (Trieste)
CINECA National Supercomputing Center in Bologna
Ecole Polytechnique Fédérale de Lausanne
Princeton University
Massachusetts Institute of Technology
University of Oxford
QE capabilitiesQE capabilities
Ground-state calculations:
• Self-consistent calculation of total energies• Kohn-Sham orbitals obtained through minimization• Several k-point sampling, Gamma point calculations • Broadening schemes (Fermi-Dirac, Gaussian, Methfessel-Paxton, and
Marzari-Vanderbilt)• Separable norm-conserving and ultrasoft (Vanderbilt) pseudo-
potentials, PAW (Projector Augmented Waves)• Exchange-correlation functionals: from LDA to generalized-gradient
corrections (PW91, PBE, B88-P86, BLYP) to meta-GGA, exact exchange (HF) and hybrid functionals (PBE0, B3LYP, HSE);
• DFT+U• Polarization• Spin polarized calculations, Spin-orbit coupling and noncollinear
magnetism
QE can also calculateQE can also calculate
1) Geometry optimization and unit cell optimization
2) Ab-initio molecular dynamics: Car-Parrinello, microcanonical, canonical, additional constraints
3) Phonon frequencies and modes, full phonon dispersions, inter-atomic force constants
4) Electron-phonon interactions
5) Infrared and (non-resonant) Raman cross-sections, NMR chemical shifts
6) Quantum transport
QE runs in the large majority of platforms, has been ported to GPUs, parallelization with MPI or OpenMP.
Self-consistent cycle
Initial guess for KS orbitals
2
1
( ) ( )occ
j
j
r rρ ϕ=
=∑� �
3
0
( ')[ ]( ) ( ) ' [ ]( )
's xc
rv r v r d r v r
r r
ρρ ρ= + +
−∫�
� � � �� �
2
{ [ ]( )} ( ) ( )2
s j j jv r r rρ ϕ ε ϕ∇
− + =� � �
convergenceNo
Yes
Exit electronic optimization
mixing
broadening
PseudopotentialsPseudopotentials
• Core electrons do not participate in chemical bond (remain atomic like)
• Core electron orbitals have enormous peaks close to the nuclei • Valence electrons are oscillating rapidly close to the nuclei
Core electrons are not included in the calculation but are
taken into account through pseudopotentials
Example Si-atom: (1s2 2s2 2p6) 3s2 3p2
Core electrons
Large variety of pseudopotentials in QE site: type and
functional in the filename
PseudopotentialsPseudopotentials
In the input file:
ATOMIC_SPECIES
Mg 24.305 Mg.pw91-np-van.UPF
Si 28.085 Si.pw91-n-van.UPF
O 15.999 O.pw91-van_ak.UPF
H 1.008 H.pw91-van_ak.UPF
Same exchange-correlation functional for all pseudopotentials
Typical input file (graphene)Typical input file (graphene)
&control
calculation = 'scf'
restart_mode='from_scratch',
prefix='graphene',
pseudo_dir = '/home/lathiot/espresso-4.3/pseudo/',
outdir='/home/lathiot/tmp/'
/
&system
ibrav= 0, celldm(1) =4.7375, nat= 2, ntyp= 1,
ecutwfc =70.0, ecutrho=560.0
occupations='smearing'
nbnd = 8
degauss=0.001
/
&electrons
diagonalization='david'
electron_maxstep = 100
mixing_mode = 'plain'
mixing_beta = 0.7
conv_thr = 1.0d-8
/
ATOMIC_SPECIES
C 12.000 C.pbe-rrkjus.UPF
CELL_PARAMETERS
1.0 0.0 0.0
0.5 0.8660254 0.0
0.0 0.0 12.0
ATOMIC_POSITIONS alat
C 0.00 0.00 0.00
C 1.0 0.5773503 0.00
K_POINTS automatic
10 10 1 0 0 0
calculation = ‘scf’
‘nscf’
‘relax’
‘vc_relax’
ecutwfc, ecutrho:Kinetic energy
cutoffs for wavefunction and
density
occupations=‘smearing’
‘fixed’
degauss: smearing width (Ryd)
nbnd: number of bands, i.e.
eigenvalues to solve.
Open in browser file:///home/lathiot/espresso-5.0.1/PW/Doc/INPUT_PW.html
Typical input file (graphene)Typical input file (graphene)
&control
calculation = 'scf'
restart_mode='from_scratch',
prefix='graphene',
pseudo_dir = '/home/lathiot/espresso-4.3/pseudo/',
outdir='/home/lathiot/tmp/'
/
&system
ibrav= 0, celldm(1) =4.7375, nat= 2, ntyp= 1,
ecutwfc =70.0, ecutrho=560.0
occupations='smearing'
nbnd = 8
degauss=0.001
/
&electrons
diagonalization='david'
electron_maxstep = 100
mixing_mode = 'plain'
mixing_beta = 0.7
conv_thr = 1.0d-8
/
ATOMIC_SPECIES
C 12.000 C.pbe-rrkjus.UPF
CELL_PARAMETERS
1.0 0.0 0.0
0.5 0.8660254 0.0
0.0 0.0 12.0
ATOMIC_POSITIONS alat
C 0.00 0.00 0.00
C 1.0 0.5773503 0.00
K_POINTS automatic
10 10 1 0 0 0
diagonalization: type of minimization
(david, cg)
mixing_mode, mixing_beta:
mixing parameters for scf convergence
conv_thr: Convergence criterion total energy
CELL_PARAMETERS: Bravais lattice vectors.
, , , ,k G k G i k G i k i k G
G
H C Cε′ ′+ + + +′
=∑
ATOMIC_POSITIONS : atomic positions in
unit cell
K_POINTS: reciprocal lattice mesh
in the Brillouin zone.
(In the example 10x10x1 mesh)
They can also be specified explicitly.
Convergence with k-point sampling must be checked
Finite structures (Gamma point)Finite structures (Gamma point)
Can we calculate non periodic systems, e.g. a molecule?
Yes: Set lattice vectors to large numbers and
K_POINTS gamma, i.e. only 1 k-point at (0,0,0)
5. Results5. Results
Graphene under strain
● Uniaxial:
● Hydrostatic:
Stress-strain plot
● Uniaxial
Young Modulus & Poisson ratio
● Young modulus:
Our 0.96 TPa
Other theoretical (0.91-1.09) TPa
Experiment 1.0 ± 0.1 TPa
Poisson ratio:
Our 0.22
Other theoretical (0.15-0.28)
Stress-strain plot
● Hydrostatic
Fitting of empirical potential:
A) Bond stretching
Fitting of empirical potential:
B) Angle Bending
Fitting of empirical potential:
C) Dihedral Angle Bending
Fitting of empirical potential:
C) Dihedral Angle Bending
Electronic Structure of Pentaheptite
● It occurs through Stone-Wales transformations
● Energetically 0.25 eV/atom above Graphene
● Pentaheptite:
● Pentaheptite is a metal:
Electronic Structure of Pentaheptite
● Fermi Surface
Electronic Structure of Pentaheptite
4) Tetraoctite:3) Pentaheptite B:
2) Pentaheptite A:1) Graphene
Graphene and sp2 allotropes under extreme strain
● Strain vertical to the armchair:
Graphene and sp2 allotropes under extreme strain
Graphene and sp2 allotropes under extreme strain
● Strain vertical to zig-zag:
Magnetism in doped ZnO
• ZnO based materials: Large Technological importance:
• Spintronics
• LED
• Detectors/sensors
• Spintronic Applications:
• Exploiting electron spin in electronic transfer to carry information
• ‘Spin’: an extra degree of freedom in improvement of devices
• Dilute Magnetic Semiconductors: semiconducting materials that have
acquired Ferromagnetic properties.
• Possible path: doping of with Transition Metal atoms
• Increase of FM properties with co-doping
• The Failure of LDA/GGA in the Gap of ZnO:
Experimental Gap: 3.4 eV
LDA/GGA calculated gap: 0.75 - 0.85 eV
• Two additive reasons:
1) LDA and GGA are known to underestimate the Gap by ~50%
2) The d-band of Zn is centered in a false position within LDA/GGA:
enhanced artificial hybridization of Zn-d with O-p
Zn-O Dos
Zn-O Dos
Zn-O Dos
Zn-O Dos
Zn-O Dos
Cobalt doped ZnO
• There are experimental data on the position of Co-d impurity band:
Cobalt doped ZnO
• There are experimental data on the position of Co-d impurity band:
• With LDA+U we can reproduce this result
Calculation Details
• 3x3x3 supercell (108 atoms), 2 k-points.
Calculation Details
• 3x3x3 supercell (108 atoms) with 2 k-points.• 2 Co impurities (4% at)
• Target: To find structures that are Ferromagnetic
• Energy difference of FM from the AFM configuration.
FM
AFM
Structures for Co-doped ZnO
C2:C1:
D: E:
FM vs AFM configuration
0.400.80D
0.010.03E
14.640.72C2
45.3442.50C1
FM-AFM (GGA +U)
(meV)
FM-AFM (GGA)
(meV)
• Two Co impurities in ZnO are AFM coupled.
• Interaction is short-range.
Co-doping with Cu
C1-Cu43 C1-Cu53 C1-Cu59 C1-Cu19
C2-Cu89 C2-Cu67 C2-Cu23
D-C55 D-Cu19 D1-Cu41 E-Cu55
Codoping with Cu - results
-99.2C2-Cu23
-0.8E-Cu55
8.3-35.2C2-Cu67
14.9-2.1C2-Cu89
-22.8C1-Cu19
7.433.9C1-Cu59
-125.3-56.8C1-Cu53
-4.7C1-Cu43
-71.8D-Cu41
87.5-1.9D-Cu19
-44.2-65.9D-Cu55
FM-AFM (GGA+U)
(meV)
FM-AFM (GGA)
(meV)
Structures
Codoping with Cu - results
• In most cases Cu induces a FM coupling
• Symmetric position of Cu (w.r.t Co atoms) induces FM coupling
• The splitting of the Cu-d band in AFM state
Cu-d band splitting
• No splitting for Cu-d in the AFM configuration for D-Cu55
Codoping with Cu - results
• In most cases Cu induces a FM coupling
• Symmetric position of Cu (w.r.t Co atoms) induces FM coupling
• The splitting of the Cu-d band in AFM state
• Cu: Super-exchange mediator
• Cu and Co in different planes (perpendicular to the Wurtzite z axis)
induses FM coupling
• Possibly, growth of Cu and Co alternating impurity planes would lead to
enhanced magnetic properties.
Codoping with Cu - results
• Structures with enhanced FM coupling:
• Structures with enhanced AFM coupling:
Co islands/dots in ZnO
• The 4-atom dot + the vacancy of the common Oxygen
X
60.6(1)
71.5(2)
71.0(1)
77.1(2)
80.1(1)
96.0(2)
0.00(1)
0.00(2)
E (meV)Structure
Co islands/dots in ZnO
• The 4-atom dot + the vacancy of the common Oxygen
(1)Zn Lattice positions, (2)Relaxed Co atoms
Co islands/dots in ZnO
• 3-4 atom dots prefer AFM state
• Far Oxygen vacancy: does not affect magnetic state
• Common Oxygen vacancy: FM state favorable
MgB2
2-D σσσσ-bonding hole pockets
3-D ππππ and ππππ∗∗∗∗ Fermi surfaces
σσσσσσσσ ππππ
ππππ
A
Tc= 39.5 K
ππππ∗∗∗∗σσσσ
Fermi Surface of MgB2
CollaboratorsCollaborators
• G. Kalosakas, K. Papagelis, K. Galiotis,
University of Patras
• Z. G. Fthenakis, Michigan State University
• A. N. Andriotis, IESL FORTH, Heraklio
• M. Menon, University of Kentucky
• E.K.U. Gross group, MPI Halle
Concluding remarksConcluding remarks
• Theoretical electronic structure methods can make real
predictions for material properties
• DFT is the workhorse of present day electronic
structure theory
• QE is a mature software applying DFT approximations
to periodic (and finite) systems able to calculate several
diverse properties like structural, electric, magnetic,
optical, superconducting properties.
Thank you !!!