Excited state properties:

electronic and optical

spectroscopies

Dott.ssa Letizia Chiodo

Nano-bio Spectroscopy Group &
ETSF - European Theoretical Spectroscopy Facility,
Dipartemento de Física de Materiales, Facultad de Químicas,
Universidad del País Vasco UPV/EHU,
San Sebastián-Donostia, Spain
 Outline of the Lectures
• Many Body Problem
• DFT elements; examples
• DFT drawbacks
• excited properties:
electronic and optical spectroscopies. elements of theory
• Many Body Perturbation Theory: GW
• codes, examples of GW calculations
• Many Body Perturbation Theory: BSE
• codes, examples of BSE calculations
• Time Dependent DFT
• codes, examples of TDDFT calculations
• state of the art, open problems
 Outline
• DFT drawbacks
• Photoemission Spectroscopy
• Electron Energy Loss Spectroscopy
• Optical absorption

• Polarizability
• Screening
• Random Phase Approximation
• Local Field Effects
 DFT drawbacks
We have detected two problems with DFT:
1) Kohn-Sham bandgaps are much too small
2) Fermi’s golden rule in the independent particle picture is not reliable
to calculate absorption spectra
How can we understand this?
 DFT drawbacks
We have detected two problems with DFT:
1) Kohn-Sham bandgaps are much too small
2) Fermi’s golden rule in the independent particle picture is not reliable
to calculate absorption spectra
How can we understand this?

L. Chiodo et al., phys. stat. sol. (b) 242, No. 15, 3032–3039 (2005)
 Kohn-Sham bandgaps
DFT is a ground state theory:
KS LUMO-HOMO is not the gap!!!!
KS-gap is strongly underestimated for 2 reasons

Vxc discontinuos
Vxc not exact

adapted from M. van Schilfgaarde et al., PRL 96 226402 (2006)

 Kohn-Sham bandgaps

KS eigenvalues are obtained as Lagrange multipliers, cannot be interpreted as

removal/addition energies: Koopman’s theorem doesn’t hold

to calculate the electronic gap, we can use two calculations of the Kohn-Sham
HOMO energy, one for the N-particle system and another for the N+1-particle
and then apply the ionization potential theorems of KS-DFT
the gap is exact if the exact XC potential is used to calculate
these quantities
this because of the discontinuity of the exchange-correlation potential as the
particle number is varied across an integer
 Optical Absorption
quantitatively and often qualitatively wrong
 other problems…
LDA(GGA) gives:
wrong ground state for the titanium atom
wrong magnetic ground state for bulk iron
wrong adsorption energy of CO on metal surfaces

LDA, due to its local character, cannot describe van der Waals and hydrogen bonding
………..

use of different Vxc can improve e.g. structural results for large organic molecules
(isomers), but cannot correct the Vxc discontinuity, due to the GS nature of DFT

we are describing states that are not GROUND STATES, but EXCITED STATES
therefore not included in DFT

moreover, how to relate macroscopic (experiment) and microscopic (ab initio) world?
 Experimental Techniques

to understand and correct the errors we are doing in DFT, it is usefull to

understand the experimental techniques giving us the excited state properties we
want to study

between Electronic Excitations, there are

•Photoemission Spectroscopy
•Inverse Photoemission Spectroscopy
•Electron Energy Loss Spectroscopy
•Optical Absorption
•Inelastic X-ray Scattering
 Photoemission Spectroscopy
UPS
density of states
AR-UPS
band structure (valence)
AR-IPES
empty states (conduction

UPS,XPS,ARPES, ARIPES
 Photoemission Spectroscopy
UPS
density of states
AR-UPS
band structure (valence)
AR-IPES
empty states (conduction

1 e- is removed (added), the system is charged, and in an excited state

(easy for metals, more difficult for semiconductors and insulators)

UPS,XPS,ARPES, ARIPES
Electron Energy Loss Spectroscopy

energy-loss function, measured in EELS or IXSS, is related to minus

the imaginary part of the inverse macroscopic dielectric function

{
EL(q, ω ) = −ℑ ε M−1 (q, ω )}
dependence on the transferred momentum
Electron Energy Loss Spectroscopy

energy-loss function, measured in EELS or IXSS, is related to minus

LEED the imaginary part of the inverse macroscopic dielectric function

{
EL(q, ω ) = −ℑ ε M−1 (q, ω )}
dependence on the transferred momentum

electron cascade

EELS Auger

HREELS
 Optical Absorption

I ( x ) = I 0 e − αx
ωε 2
α=
νc
ABS(ω ) = ℑ{ε M (q → 0, ω )}

optical transitions:
transferred momentum is 0
 Micro-Macro
from Macro (experiments, Maxwell equations) to micro (ab initio):
averaging procedure

be careful with symmetries of the studied sample (cubic, not cubic, etc…)

differences between the microscopic fields and the averaged (macroscopic) fields:
crystal local fields (due to local fluctuations at micro scale)

microscopic dielectric functions are determined by the elementary medium excitations

interband and intraband transitions, as well as collective excitations

N.B real and reciprocal space are both present in all the following theory
and codes implementation
 Polarizability
ε −1 = 1 + vχ
single particle polarizability
RPA
φ (r )φ *
(r )φ *
( r ' )φ (r ' ) e-
χ 0 = ∑ ( fi − f j ) i j i j
P =
ij ω − ( ε i − ε j ) + i η
h+

hν ωcv

v
 Polarizability

F. Sottile, Ph.D. thesis, 2003

 Screening
from polarizability
microscopic dielectric function ε −1 = 1 + vχ

observable quantities and spectra are related to the macroscopic dielectric function εM
obtained from the microscopic ε by spatially averaging over a distance large enough with
respect to the microscopic structure of the system, e.g. in solids an elementary cell

ε M (r , r ' , ω ) = ε (r , r ' , ω )
1
ε M ( q, ω ) = ≠ ε G =0,G '=0 (q, ω ) = ε MNLF (q, ω )
ε G−1=0,G '=0 (q, ω )
if microscopic dielectric matrix contains off-diagonal terms
 Screening and LFE
crystal local-field effects are neglected ε G =0,G '=0 (q, ω ) = ε MNLF (q, ω )
• absent in the homogeneous electron gas
• marginal in weakly inhomogeneous systems (e.g silicon)
• important in systems presenting strong inhomogeneities in the electronic density

in particular, local-field effects are critical in reduced dimensionality systems

2D surfaces/graphene
1D nanotubes/wires
0D clusters and molecules L. Chiodo et al., phys. stat. sol. (b) 242, No. 15, 3032–3039 (2005)
 Local Field Effects: NWs of Ge

LFE: source of
anisotropy in
response function

left panels: light polarized along the wire axis x

right panels: light polarized perpendicularly to the axis y
M. Bruno, et al., Phys. Rev. B 72, 153310 (2005)
LFE in EELS: zirconia

theoretical dielectric function for c-ZrO2

top panel, real part
center panel, imaginary part
bottom panel, the loss function

solid lines, RPA with local fields.

dashed lines, without local fields.

L. K. Dash, et al., PRB 70, 245116 (2004)

 LFE in EELS: titania

I. G. Gurtubay, et
al., PRB 70,
201201(R) (2004)

Dynamical structure factor (top panel); macroscopic dielectric function (middle and
bottom panels) of TiO2 along the [001] direction. The solid (dotted) line represents the
calculated ALDA spectrum with (without) LFE. The dashed line denotes the calculated
RPA result with LFE.
IP - Random Phase Approximation
RPA non interacting electron-hole pairs, without local field
e-
P=
h+

hν ωcv

v
16π
Im ε M (ω ) = ϕ c v ϕ v δ (ε c − ε v − ω )
2

ω 2 ∑
v ,c
Fermi Golden rule
we want to correct these spectra….

Anatase Titania Absorption Spectrum

DFT is the first step for excited state properties:
electronic properties (DOS, band structure, surface states),
dielectric response at RPA level

.....
local field effects play an important role, to improve properties
connected to screening (absorption, electron energy loss)
because they mix somehow the otherwise independent transitions
but they are not enough
how can we further improve results?

thanks to
Dr. Matteo Gatti
Dr. Silvana Botti