SlideShare a Scribd company logo
1 of 41
Download to read offline
Approximations in DFT
Claudio Attaccalite
CNRS/CINaM, Aix-Marseille Universite (FR)
Outline
Claudio Attaccalite
CNRS/CINaM, Aix-Marseille Universite (FR)
●
Pseudo-potentials
●
Exchange-correlation functions
●
Miscellaneous
Pseudo-potentials
We do not want to simulate core electrons:
1) they do not participate to chemical reactions, atomic structures, etc..
2) including all electron in the simulations increase computational time/memory
An isolated atom in DFT
For a one-electron atom, the Coulomb potential,
is spherically symmetric
The solutions may be split into a radial and an angular part
An isolated atom in DFT
For a one-electron atom, the Coulomb potential,
is spherically symmetric
The solutions may be split into a radial and an angular part
Withint DFT the independent-particle Kohn-Sham states
may be classified by the angular quantum numbers,
subject to the effective potential
An isolated atom in DFT
For a one-electron atom, the Coulomb potential,
is spherically symmetric
The solutions may be split into a radial and an angular part
Withint DFT the independent-particle Kohn-Sham states
may be classified by the angular quantum numbers,
subject to the effective potential
A one particle equation analogous to the Schrödinger equation
for the one-electron atom, can be written as:
Pseudopotential idea 1/2
Replacing the effects of the core electrons with an effective potential.
Replacing the effects of the core electrons with an effective potential.
We rewrite the valence states as a smoother function and
the remaining portion in terms of core states :
Pseudopotential idea 1/2
Replacing the effects of the core electrons with an effective potential.
Replacing the effects of the core electrons with an effective potential.
We rewrite the valence states as a smoother function and
the remaining portion in terms of core states :
Next, we consider the product with one of the core states. Because the valence
state has to be orthogonal to the core states, we have
Pseudopotential idea 1/2
Replacing the effects of the core electrons with an effective potential.
Replacing the effects of the core electrons with an effective potential.
We rewrite the valence states as a smoother function and
the remaining portion in terms of core states :
Next, we consider the product with one of the core states. Because the valence
state has to be orthogonal to the core states, we have
We now rewrite the valence bands in terms of the pseudofunction:
Pseudopotential idea 1/2
Replacing the effects of the core electrons with an effective potential.
Replacing the effects of the core electrons with an effective potential.
We rewrite the valence states as a smoother function and
the remaining portion in terms of core states :
Next, we consider the product with one of the core states. Because the valence
state has to be orthogonal to the core states, we have
We now rewrite the valence bands in terms of the pseudofunction:
Applying the Hamiltonian we get
Pseudopotential idea 2/2
All-electron and pseudo eigenvalues agree for the reference configuration.
Pseudopotential idea 2/2
All-electron and pseudo eigenvalues agree for the reference configuration.
All-electrons and pseudo-wavefunctions agree
beyond a certain cutoff, rc
Pseudopotential idea 2/2
All-electron and pseudo eigenvalues agree for the reference configuration.
All-electrons and pseudo-wavefunctions agree
beyond a certain cutoff, rc
Real and pseudo norm squares integrated from 0 to R
for all R < rc
agree
Pseudopotential idea 2/2
All-electron and pseudo eigenvalues agree for the reference configuration.
All-electrons and pseudo-wavefunctions agree
beyond a certain cutoff, rc
Real and pseudo norm squares integrated from 0 to R
for all R < rc
agree
Additional condition for transferability [see PRL 43, 1494 (1979)]
How a pseudopotential is
constructed
1) Solve the all-electron atomic system
How a pseudopotential is
constructed
1) Solve the all-electron atomic system
2) Decide which of the states are to be taken
as core and which are to be taken as valence
How a pseudopotential is
constructed
1) Solve the all-electron atomic system
2) Decide which of the states are to be taken
as core and which are to be taken as valence
3) Generate the pseudopotentials,
Vnl
(r), for the valence states.
How a pseudopotential is
constructed
1) Solve the all-electron atomic system
2) Decide which of the states are to be taken
as core and which are to be taken as valence
There is no unique way to determine the Vnl
’s. There are two opposing considerations :
1. Good transferability small r
⇒ c
.
2. Large rc
smoother pseudopotentials.
⇒
3) Generate the pseudopotentials,
Vnl
(r), for the valence states.
Exchange-correlation functions
LSDA
GGA
meta GGA
hybrid meta GGA
hybrid GGA
fully non-local
‘Earth’
Hartree Theory
‘Heaven’
Chemical Accuracy
Exchange-correlation functionals
Cited
18298
times
(by
11/2011)
Cited
8578
times
(by
11//2011)
沈呂九
PROBLEM: HK theorem is not a constructive theorem!!
The energy functional
[ ] [ ]
[ ] [ ]
[
= [ ]
]
e
G e
T n V n
F n
E n U n
U n
 


3
[ ] ( ) ( )
U n d r n r U r

 
The F [n] functional is the same for all electronic systems.
“No ones knows the true F [n], and no one will, so it is replaced by varios
uncontrollable approximations.” (Marder, p.247)
We split the problem in various parts….
• Kinetic energy functional
• Hartree energy functional
(exact)
TS [n]=∑i
N v
⟨ϕi|−1
2
∇i
2
|ϕi⟩ where fi
are KS orbitals
• Exchange correlation functional
(unknown)
F[n]=TS[n]+U H [n]+ EXC [n]
€
Exc[n] contains T −Ts and U −UH
EXC [n]
U H [n]=
e2
2
∫∫d
3
r d
3
r '
n(r)n(r' )
|r−r '|
The value of xc at any position r is calculated
using only the value of  at that point.
Ex[ ρ(r)]=∫−
9α
8 (3
p )
1/3
ρ4/3
(r)d r
• Correlation Energy
- no analytic function for the uniform
electron gas. (Cerperley Alder)
- Vosko, Wilk, Nusair functionals
- PZ, PW
• Exchange Energy
((r))
€
Exc
hom ((r))
((r))
€
Exc
hom
Exc
hom
Exc
LSDA
=∫ρ(r )[ ex
hom
( ρ(r))+ec
hom
( ρ(r ))]dr
Local Spin Density Approximation
Local Spin Density Approximation - Performance
• Structural properties are often good
- usually slightly underestimates bulk lattice constants
- bulk moduli are slightly too large
- phonons too stiff
• Binding energies are too negative
- overbinding of molecular and metallic solids, up to several eV
- molecular atomization energies have a large error
• Activation energies for chemical reactions are unreliable
• Relative stabilities of bulk phases can be wrong.
• Density of states and band structure are okay but….
Ge
Local Spin Density Approximation - Performance
• Structural properties are often good
- usually slightly underestimates bulk lattice constants
- bulk moduli are slightly too large
- phonons too stiff
• Binding energies are too negative
- overbinding of molecular and metallic solids, up to several eV
- molecular atomization energies have a large error
• Activation energies for chemical reactions are unreliable
• Relative stabilities of bulk phases can be wrong.
• Density of states and band structure are okay but….
Why does LSDA work?
- Many metallic systems are similar to a homogeneous electron gas.
- One might expect it to fail for molecular systems (it does!)
What are it’s shortcomings?
- self-interaction
Ge
Generalized Gradient Approximation
Based on the idea of doing a Taylor expansion of the density:
€
f (ρ) = f (ρ0) + f '(ρ0)(ρ − ρo) +
f ''(ρ0)(ρ − ρo)2
2!
+L
it turns out that a simple Taylor expansion does not improve upon LSDA
because it violates several of the constraints that the exchange-correlation
functional is known to observe.
Also the gradients in real materials are often large enough that the Taylor
expansion breaks down.
€
Ex/c
GGA
[ρ(r)] = Ex/c
LSDA
[ρ(r)]+ ΔEx/c
∇ρ(r)
ρ4 /3
(r)
⎡
⎣
⎢
⎤
⎦
⎥
lim∇ρ→ 0 Exc = Exc
LSDA
Obey the uniform electron gas limit:
Development of Exchange and Correlation Functionals
Empiricism
•We don’t know the exact form of
the exchange-correlation functional
so write down a form and
parameterize it to fit to a set of
experimental or high-level ab-initio
data.
• B88, mPW, LYP
Nonempiricism
•The exchange and correlation
functional should be developed from
first principles by incorporating
known constraints. Once the right
constraints are met the rest will
come out right.
• Often times the functional forms
used are the same as, or based on
those from empirically based
functionals.
• PW91, PBE
Generalized Gradient Approximation
• Bulk lattice constants increase compared in LSDA
• Cohesive energies decrease compared to LSDA
• Atomic and molecular energies are improved
• Corrects over binding of LSDA
• Improves activation barriers, but they’re still too low.
• Improved relative stability of bulk phases.
Generalized Gradient Approximation
• Bulk lattice constants increase compared in LSDA
• Cohesive energies decrease compared to LSDA
• Atomic and molecular energies are improved
• Corrects over binding of LSDA
• Improves activation barriers, but they’re still too low.
• Improved relative stability of bulk phases.
Problems with LSDA and GGA
●
Self interaction
●
Neglect of strong correlations
●
Exchange-splitting underestimated for narrow d- and f-bands.
●
Many transition-metal compounds are Mott-Hubbard or charge-transfer
(not described by LDA/GGA)
beyond GGA …… 1/2
Meta-GGA Functionals
€
E[ρ(r)] ⇒ ∇2
ρ(r)
If we’re interested in following the Taylor-like expansion of the density the next
logical step is the Laplacian of the density.
B95,B98,KCIS,HCTH,TPSS
LDA+U Method
●
A generalization of the LSDA method that takes into account orbital dependencies of the
exchange and Coulomb interactions.
●
The correction is usually only considered for highly localized atomic-like orbitals on the
same site.
●
The added term serves to shift the energies of the localized orbitals it is applied to relative
to the other orbitals.
SIC -Self-Interaction Correction
Methods that use approximate functionals and add in
a “self-interaction correction” to correct for the unphysical
self-interaction that is present in many functionals.
..it requires U parameter...
beyond GGA …… 2/2
Hybrid Functionals (scaling behaviour N4
)
Exc
hybrid DFT
= (1− a)Ex
LSDA
+ aEx
HF
+ bΔEx
NL
+ Ec
LSDA
+ cΔEc
NL
• Functionals:
B3PW91, B3LYP, O3LYP, X3LYP
Exc
hybrid DFT
= (1− a)(Ex
LSDA
+ ΔEx
NL
) + aEx
HF
+ Ec
LSDA
+ ΔEc
NL
• Functionals:
B1PW91, B1LYP, B1B95, mPW1PW91, PBE1PBE
Hybrid meta Functionals (scaling behaviour N4
)
• Hybrid functionals that contain kinetic energy density, in addition to
Hartree–Fock exchange.
Miscellaneous
V-representability
"a number of 'reasonable' looking densities that
have been shown to be impossible
to be the ground state density for any V."
R. Martin
The central assumption of KS is that such an equivalent system of non-interacting
electrons always exists, i. e., that all interacting density
can be represented by a non-interacting one
Temperature and particles
In a statistical mechanical framework the focus is on the grand canonical potential.
Here, the grand canonical Hamiltonian plays an analogous role as the one played by
the Hamiltonian for the ground-state problem.
Where the entropy operator is defined as
Kohn-Sham equations at non-zero temperature, and the density
Thermal Density Functional Theory in Context, https://arxiv.org/abs/1309.3043
Summary
• As one climbs the Jacob’s ladder of density functionals, the complexity and cost of
the calculation increase, as does the accuracy.
- LSDA  density only
- GGA  density and gradient of the density
- metaGGA  density, gradient of the density, kinetic energy density
- hybrid  density, gradient of the density, Hartree–Fock exchange
- meta hybrid  density, gradient of the density, kinetic energy
density, Hartree–Fock exchange.
• Different functionals within the same class can give very different results.
- its not enough to say that you used GGA. You must specify the use
of PBE, BLYP, etc.
Limitations of DFT
• Some systems cannot be well described by a single Slater determinant.
• Its hard to know how to systematically improve DFT functionals.
• Does not obey the variational principle.
Bibliography
Pseudopotentials
 D.R. Hamann, M. Schlüter, and C. Chiang, Phys. Rev. Lett. 43, 1494 (1979).
 G.P. Kerker, J. Phys. C: Solid St. Phys. 13, L189 (1980).
 L. Kleinman and D.M. Bylander, Phys. Rev. Lett. 48, 1425(1982)
 N. Troullier and J.L. Martins, Phys. Rev. B 43, 1993 (1991)
 D. Vanderbilt, Phys. Rev. B 41, 7892 (1990).
 G. Kresse and D. Joubert, Phys. Rev. B 59, 1758 (1999)
Exchange-correlation funcitions
 Review of approximations for the exchange-correlation energy
in density-functional theory
https://arxiv.org/abs/2103.02645
 Progress in the development of exchange-correlation functionals
https://www.sciencedirect.com/science/article/pii/B9780444517197500676
Miscellaneous
 Thermal Density Functional Theory in Context,
https://arxiv.org/abs/1309.3043
Strength of DFT
From W.Aulbur’s slides
• Neglect of strong correlations
- Exchange-splitting underestimated for narrow d- and f-bands.
- Many transition-metal compounds are Mott-Hubbard or charge-transfer
insulators, but DFT predicts metallic state.
From J. Hafner’s slides
Metal in “LDA”
calculations
LDA DFT Calcs.
(dashed lines)
Rohlfing and Louie PRB 1993
LDA, GGAs, etc. fail in many cases with strong correlations.
• Band-gap problem: HKS theorem not valid for excited states.
Band-gaps in semiconductors and insulators are usually underestimated.
Can be fixed by
GWA…etc
Weakness of DFT
Hybrid Functionals
What’s the rationale for adding Hartree–Fock exchange??
- LSDA and GGA overbind.
- Hartree–Fock theory underbinds.
- LSDA and GGA give bond lengths that are too long.
- Hartree–Fock theory gives bond lengths that are too short.
Perhaps if we mix the two together we’ll get numbers in the middle…
ROH = 0.9572
HOH = 104.52
R(O-H) Θ(H-O-H)
LSDA 0.9701 104.9605
PBE 0.9690 104.2068
BLYP 0.9706 104.5230
PBE1PBE 0.9575 104.9051
B3LYP 0.9604 105.1265
Hartree Fock
– 0.9399 106.2709
Ebind (kcal/mol)
LSDA 7.95
PBE 5.05
BLYP 4.16
PBE1PBE 4.96
B3LYP 4.57
Hartree–Fock 3.49
Accurate 4.99
Hybrid methods also help correct the self-interaction problem, since in Hartree–
Fock theory it cancels out.
How do you Determine the Percent of HF Exchange?
• The most common way to determine the amount of Hartree–Fock exchange to
include it to treat it as an optimizable parameter.
- Take a set of molecules for which the atomization energy, ionization potential,
and proton affinities are well knows, as well as atomic total energies, and vary
the percent of HF exchange until you minimize the unsigned error over this data
set.
- B3PW1, B3LYP,O3LYP, X3LYP
•There are some “parameter free” hybrid density functionals, where the
percentage of Hartree–Fock exchange was determined based on a perturbation-
theory argument
- B1B95, mPW1PW91, PBE1PBE (PBE0), B1LYP
- in reality the percentage of HF exchange used cannot be
determined nonempirically.
Scaling Behavior
Scaling Behavior Method
N3
DFT(LSDA,GGA,meta)
N4
HF, DFT(hybrid,hybrid meta)
N5
MP2
N6
MP3, CISD, MP4DQ, CCSD, QCISD
N7
MP4,CCSD(T),QCISD(T)
N8
MP5,CISDT,CCSDT
N9
MP6
N10
MP7,CISDTQ,CCSDTQ

More Related Content

What's hot

Quantum-Espresso_10_8_14
Quantum-Espresso_10_8_14Quantum-Espresso_10_8_14
Quantum-Espresso_10_8_14
cjfoss
 
BIOS 203 Lecture 4: Ab initio molecular dynamics
BIOS 203 Lecture 4: Ab initio molecular dynamicsBIOS 203 Lecture 4: Ab initio molecular dynamics
BIOS 203 Lecture 4: Ab initio molecular dynamics
bios203
 
Perturbation
PerturbationPerturbation
Perturbation
BHAVANAR12
 

What's hot (20)

Density functional theory
Density functional theoryDensity functional theory
Density functional theory
 
Density Functional Theory
Density Functional TheoryDensity Functional Theory
Density Functional Theory
 
Phonons lecture
Phonons lecturePhonons lecture
Phonons lecture
 
Electronic structure of strongly correlated materials
Electronic structure of strongly correlated materialsElectronic structure of strongly correlated materials
Electronic structure of strongly correlated materials
 
Lecture7
Lecture7Lecture7
Lecture7
 
Introduction to Electron Correlation
Introduction to Electron CorrelationIntroduction to Electron Correlation
Introduction to Electron Correlation
 
Hartree fock theory
Hartree fock theoryHartree fock theory
Hartree fock theory
 
Density Functional Theory.pptx
Density Functional Theory.pptxDensity Functional Theory.pptx
Density Functional Theory.pptx
 
Band structure
Band structureBand structure
Band structure
 
Chapter2 introduction to quantum mechanics
Chapter2 introduction to quantum mechanicsChapter2 introduction to quantum mechanics
Chapter2 introduction to quantum mechanics
 
CHAPTER 6 Quantum Mechanics II
CHAPTER 6 Quantum Mechanics IICHAPTER 6 Quantum Mechanics II
CHAPTER 6 Quantum Mechanics II
 
THE HARTREE FOCK METHOD
THE HARTREE FOCK METHODTHE HARTREE FOCK METHOD
THE HARTREE FOCK METHOD
 
Born–Oppenheimer Approximation.pdf
Born–Oppenheimer Approximation.pdfBorn–Oppenheimer Approximation.pdf
Born–Oppenheimer Approximation.pdf
 
Quantum-Espresso_10_8_14
Quantum-Espresso_10_8_14Quantum-Espresso_10_8_14
Quantum-Espresso_10_8_14
 
Development of highly accurate pseudopotential method and its application to ...
Development of highly accurate pseudopotential method and its application to ...Development of highly accurate pseudopotential method and its application to ...
Development of highly accurate pseudopotential method and its application to ...
 
Quantum Chemistry
Quantum ChemistryQuantum Chemistry
Quantum Chemistry
 
BIOS 203 Lecture 4: Ab initio molecular dynamics
BIOS 203 Lecture 4: Ab initio molecular dynamicsBIOS 203 Lecture 4: Ab initio molecular dynamics
BIOS 203 Lecture 4: Ab initio molecular dynamics
 
Materials Modelling: From theory to solar cells (Lecture 1)
Materials Modelling: From theory to solar cells  (Lecture 1)Materials Modelling: From theory to solar cells  (Lecture 1)
Materials Modelling: From theory to solar cells (Lecture 1)
 
5 introduction to quantum mechanics
5 introduction to quantum mechanics5 introduction to quantum mechanics
5 introduction to quantum mechanics
 
Perturbation
PerturbationPerturbation
Perturbation
 

Similar to Approximations in DFT

Born oppenheimer p1 7
Born oppenheimer p1 7Born oppenheimer p1 7
Born oppenheimer p1 7
Lim Wei
 

Similar to Approximations in DFT (20)

Intro-QM-Chem.ppt
Intro-QM-Chem.pptIntro-QM-Chem.ppt
Intro-QM-Chem.ppt
 
Intro-QM-Chem.ppt
Intro-QM-Chem.pptIntro-QM-Chem.ppt
Intro-QM-Chem.ppt
 
Intro. to quantum chemistry
Intro. to quantum chemistryIntro. to quantum chemistry
Intro. to quantum chemistry
 
NANO266 - Lecture 2 - The Hartree-Fock Approach
NANO266 - Lecture 2 - The Hartree-Fock ApproachNANO266 - Lecture 2 - The Hartree-Fock Approach
NANO266 - Lecture 2 - The Hartree-Fock Approach
 
Hartree-Fock Review
Hartree-Fock Review Hartree-Fock Review
Hartree-Fock Review
 
Electronic spectra of metal complexes-1
Electronic spectra of metal complexes-1Electronic spectra of metal complexes-1
Electronic spectra of metal complexes-1
 
Pot.ppt.pdf
Pot.ppt.pdfPot.ppt.pdf
Pot.ppt.pdf
 
Power point presentation
Power point presentationPower point presentation
Power point presentation
 
UV PES.pdf
UV PES.pdfUV PES.pdf
UV PES.pdf
 
Chm102 14 15_lecture 12
Chm102 14 15_lecture 12Chm102 14 15_lecture 12
Chm102 14 15_lecture 12
 
snak_talk_symmetries_v3
snak_talk_symmetries_v3snak_talk_symmetries_v3
snak_talk_symmetries_v3
 
AI_HF_6.pdf
AI_HF_6.pdfAI_HF_6.pdf
AI_HF_6.pdf
 
Electron Spin Resonance (ESR) Spectroscopy
Electron Spin Resonance (ESR) SpectroscopyElectron Spin Resonance (ESR) Spectroscopy
Electron Spin Resonance (ESR) Spectroscopy
 
1_Unit-1 Electronic Devices.pptx
1_Unit-1 Electronic Devices.pptx1_Unit-1 Electronic Devices.pptx
1_Unit-1 Electronic Devices.pptx
 
SCF methods, basis sets, and integrals part III
SCF methods, basis sets, and integrals part IIISCF methods, basis sets, and integrals part III
SCF methods, basis sets, and integrals part III
 
Quantum Mechanics
Quantum MechanicsQuantum Mechanics
Quantum Mechanics
 
Chapter 4 - Raman Spectroscopy.pdf
Chapter 4 - Raman Spectroscopy.pdfChapter 4 - Raman Spectroscopy.pdf
Chapter 4 - Raman Spectroscopy.pdf
 
Born oppenheimer p1 7
Born oppenheimer p1 7Born oppenheimer p1 7
Born oppenheimer p1 7
 
Molecular orbital theory, 11(3)
Molecular orbital theory, 11(3)Molecular orbital theory, 11(3)
Molecular orbital theory, 11(3)
 
5.Introduction_to_Heisenberg_model.pptx
5.Introduction_to_Heisenberg_model.pptx5.Introduction_to_Heisenberg_model.pptx
5.Introduction_to_Heisenberg_model.pptx
 

More from Claudio Attaccalite

More from Claudio Attaccalite (20)

Non linear electron dynamics in solids
Non linear electron dynamics in solidsNon linear electron dynamics in solids
Non linear electron dynamics in solids
 
Luminescence of hexagonal boron nitride monolayer
Luminescence of hexagonal boron nitride monolayerLuminescence of hexagonal boron nitride monolayer
Luminescence of hexagonal boron nitride monolayer
 
Non-linear response of solids: recent results and new developments
Non-linear response of solids: recent results and new developmentsNon-linear response of solids: recent results and new developments
Non-linear response of solids: recent results and new developments
 
Light induced real-time dynamics for electrons
Light induced real-time dynamics for electronsLight induced real-time dynamics for electrons
Light induced real-time dynamics for electrons
 
CompMatScience.pdf
CompMatScience.pdfCompMatScience.pdf
CompMatScience.pdf
 
A squashed and squeezed h-BN
A squashed and squeezed h-BNA squashed and squeezed h-BN
A squashed and squeezed h-BN
 
Electron-phonon coupling a Yambo overview
Electron-phonon coupling  a Yambo overviewElectron-phonon coupling  a Yambo overview
Electron-phonon coupling a Yambo overview
 
Real Time Spectroscopy
Real Time SpectroscopyReal Time Spectroscopy
Real Time Spectroscopy
 
TSN department
TSN departmentTSN department
TSN department
 
Science ouverte
Science ouverteScience ouverte
Science ouverte
 
Optics properties of hexagonal boron nitride
Optics properties of hexagonal boron nitrideOptics properties of hexagonal boron nitride
Optics properties of hexagonal boron nitride
 
Slides Concours CNRS DR2 2021
Slides Concours CNRS DR2 2021Slides Concours CNRS DR2 2021
Slides Concours CNRS DR2 2021
 
Bethe salpeter equation
Bethe salpeter equationBethe salpeter equation
Bethe salpeter equation
 
Linear response theory
Linear response theoryLinear response theory
Linear response theory
 
Recent theoreritcal developments for 2D heterostructues
Recent theoreritcal developments for 2D heterostructuesRecent theoreritcal developments for 2D heterostructues
Recent theoreritcal developments for 2D heterostructues
 
The story of Academic Publishing: from Galileo to Nature
The story of Academic Publishing: from Galileo to NatureThe story of Academic Publishing: from Galileo to Nature
The story of Academic Publishing: from Galileo to Nature
 
Theory of phonon-assisted luminescence: application to h-BN
Theory of phonon-assisted luminescence: application to h-BNTheory of phonon-assisted luminescence: application to h-BN
Theory of phonon-assisted luminescence: application to h-BN
 
Phonon-assisted luminescence in hexagonal boron nitride
Phonon-assisted luminescence in hexagonal boron nitridePhonon-assisted luminescence in hexagonal boron nitride
Phonon-assisted luminescence in hexagonal boron nitride
 
Non-linear response of two dimensional crystals and layered materials
Non-linear response of two dimensional crystals and layered materialsNon-linear response of two dimensional crystals and layered materials
Non-linear response of two dimensional crystals and layered materials
 
Theoretical Spectroscopy Lectures: real-time approach 2
Theoretical Spectroscopy Lectures: real-time approach 2Theoretical Spectroscopy Lectures: real-time approach 2
Theoretical Spectroscopy Lectures: real-time approach 2
 

Recently uploaded

1029-Danh muc Sach Giao Khoa khoi 6.pdf
1029-Danh muc Sach Giao Khoa khoi  6.pdf1029-Danh muc Sach Giao Khoa khoi  6.pdf
1029-Danh muc Sach Giao Khoa khoi 6.pdf
QucHHunhnh
 
Ecosystem Interactions Class Discussion Presentation in Blue Green Lined Styl...
Ecosystem Interactions Class Discussion Presentation in Blue Green Lined Styl...Ecosystem Interactions Class Discussion Presentation in Blue Green Lined Styl...
Ecosystem Interactions Class Discussion Presentation in Blue Green Lined Styl...
fonyou31
 

Recently uploaded (20)

INDIA QUIZ 2024 RLAC DELHI UNIVERSITY.pptx
INDIA QUIZ 2024 RLAC DELHI UNIVERSITY.pptxINDIA QUIZ 2024 RLAC DELHI UNIVERSITY.pptx
INDIA QUIZ 2024 RLAC DELHI UNIVERSITY.pptx
 
fourth grading exam for kindergarten in writing
fourth grading exam for kindergarten in writingfourth grading exam for kindergarten in writing
fourth grading exam for kindergarten in writing
 
Key note speaker Neum_Admir Softic_ENG.pdf
Key note speaker Neum_Admir Softic_ENG.pdfKey note speaker Neum_Admir Softic_ENG.pdf
Key note speaker Neum_Admir Softic_ENG.pdf
 
9548086042 for call girls in Indira Nagar with room service
9548086042  for call girls in Indira Nagar  with room service9548086042  for call girls in Indira Nagar  with room service
9548086042 for call girls in Indira Nagar with room service
 
Class 11th Physics NEET formula sheet pdf
Class 11th Physics NEET formula sheet pdfClass 11th Physics NEET formula sheet pdf
Class 11th Physics NEET formula sheet pdf
 
1029-Danh muc Sach Giao Khoa khoi 6.pdf
1029-Danh muc Sach Giao Khoa khoi  6.pdf1029-Danh muc Sach Giao Khoa khoi  6.pdf
1029-Danh muc Sach Giao Khoa khoi 6.pdf
 
The Most Excellent Way | 1 Corinthians 13
The Most Excellent Way | 1 Corinthians 13The Most Excellent Way | 1 Corinthians 13
The Most Excellent Way | 1 Corinthians 13
 
microwave assisted reaction. General introduction
microwave assisted reaction. General introductionmicrowave assisted reaction. General introduction
microwave assisted reaction. General introduction
 
Disha NEET Physics Guide for classes 11 and 12.pdf
Disha NEET Physics Guide for classes 11 and 12.pdfDisha NEET Physics Guide for classes 11 and 12.pdf
Disha NEET Physics Guide for classes 11 and 12.pdf
 
Interactive Powerpoint_How to Master effective communication
Interactive Powerpoint_How to Master effective communicationInteractive Powerpoint_How to Master effective communication
Interactive Powerpoint_How to Master effective communication
 
Z Score,T Score, Percential Rank and Box Plot Graph
Z Score,T Score, Percential Rank and Box Plot GraphZ Score,T Score, Percential Rank and Box Plot Graph
Z Score,T Score, Percential Rank and Box Plot Graph
 
Introduction to Nonprofit Accounting: The Basics
Introduction to Nonprofit Accounting: The BasicsIntroduction to Nonprofit Accounting: The Basics
Introduction to Nonprofit Accounting: The Basics
 
Software Engineering Methodologies (overview)
Software Engineering Methodologies (overview)Software Engineering Methodologies (overview)
Software Engineering Methodologies (overview)
 
IGNOU MSCCFT and PGDCFT Exam Question Pattern: MCFT003 Counselling and Family...
IGNOU MSCCFT and PGDCFT Exam Question Pattern: MCFT003 Counselling and Family...IGNOU MSCCFT and PGDCFT Exam Question Pattern: MCFT003 Counselling and Family...
IGNOU MSCCFT and PGDCFT Exam Question Pattern: MCFT003 Counselling and Family...
 
Advance Mobile Application Development class 07
Advance Mobile Application Development class 07Advance Mobile Application Development class 07
Advance Mobile Application Development class 07
 
Advanced Views - Calendar View in Odoo 17
Advanced Views - Calendar View in Odoo 17Advanced Views - Calendar View in Odoo 17
Advanced Views - Calendar View in Odoo 17
 
Ecosystem Interactions Class Discussion Presentation in Blue Green Lined Styl...
Ecosystem Interactions Class Discussion Presentation in Blue Green Lined Styl...Ecosystem Interactions Class Discussion Presentation in Blue Green Lined Styl...
Ecosystem Interactions Class Discussion Presentation in Blue Green Lined Styl...
 
BAG TECHNIQUE Bag technique-a tool making use of public health bag through wh...
BAG TECHNIQUE Bag technique-a tool making use of public health bag through wh...BAG TECHNIQUE Bag technique-a tool making use of public health bag through wh...
BAG TECHNIQUE Bag technique-a tool making use of public health bag through wh...
 
Explore beautiful and ugly buildings. Mathematics helps us create beautiful d...
Explore beautiful and ugly buildings. Mathematics helps us create beautiful d...Explore beautiful and ugly buildings. Mathematics helps us create beautiful d...
Explore beautiful and ugly buildings. Mathematics helps us create beautiful d...
 
Accessible design: Minimum effort, maximum impact
Accessible design: Minimum effort, maximum impactAccessible design: Minimum effort, maximum impact
Accessible design: Minimum effort, maximum impact
 

Approximations in DFT

  • 1. Approximations in DFT Claudio Attaccalite CNRS/CINaM, Aix-Marseille Universite (FR)
  • 2. Outline Claudio Attaccalite CNRS/CINaM, Aix-Marseille Universite (FR) ● Pseudo-potentials ● Exchange-correlation functions ● Miscellaneous
  • 3. Pseudo-potentials We do not want to simulate core electrons: 1) they do not participate to chemical reactions, atomic structures, etc.. 2) including all electron in the simulations increase computational time/memory
  • 4. An isolated atom in DFT For a one-electron atom, the Coulomb potential, is spherically symmetric The solutions may be split into a radial and an angular part
  • 5. An isolated atom in DFT For a one-electron atom, the Coulomb potential, is spherically symmetric The solutions may be split into a radial and an angular part Withint DFT the independent-particle Kohn-Sham states may be classified by the angular quantum numbers, subject to the effective potential
  • 6. An isolated atom in DFT For a one-electron atom, the Coulomb potential, is spherically symmetric The solutions may be split into a radial and an angular part Withint DFT the independent-particle Kohn-Sham states may be classified by the angular quantum numbers, subject to the effective potential A one particle equation analogous to the Schrödinger equation for the one-electron atom, can be written as:
  • 7. Pseudopotential idea 1/2 Replacing the effects of the core electrons with an effective potential. Replacing the effects of the core electrons with an effective potential. We rewrite the valence states as a smoother function and the remaining portion in terms of core states :
  • 8. Pseudopotential idea 1/2 Replacing the effects of the core electrons with an effective potential. Replacing the effects of the core electrons with an effective potential. We rewrite the valence states as a smoother function and the remaining portion in terms of core states : Next, we consider the product with one of the core states. Because the valence state has to be orthogonal to the core states, we have
  • 9. Pseudopotential idea 1/2 Replacing the effects of the core electrons with an effective potential. Replacing the effects of the core electrons with an effective potential. We rewrite the valence states as a smoother function and the remaining portion in terms of core states : Next, we consider the product with one of the core states. Because the valence state has to be orthogonal to the core states, we have We now rewrite the valence bands in terms of the pseudofunction:
  • 10. Pseudopotential idea 1/2 Replacing the effects of the core electrons with an effective potential. Replacing the effects of the core electrons with an effective potential. We rewrite the valence states as a smoother function and the remaining portion in terms of core states : Next, we consider the product with one of the core states. Because the valence state has to be orthogonal to the core states, we have We now rewrite the valence bands in terms of the pseudofunction: Applying the Hamiltonian we get
  • 11. Pseudopotential idea 2/2 All-electron and pseudo eigenvalues agree for the reference configuration.
  • 12. Pseudopotential idea 2/2 All-electron and pseudo eigenvalues agree for the reference configuration. All-electrons and pseudo-wavefunctions agree beyond a certain cutoff, rc
  • 13. Pseudopotential idea 2/2 All-electron and pseudo eigenvalues agree for the reference configuration. All-electrons and pseudo-wavefunctions agree beyond a certain cutoff, rc Real and pseudo norm squares integrated from 0 to R for all R < rc agree
  • 14. Pseudopotential idea 2/2 All-electron and pseudo eigenvalues agree for the reference configuration. All-electrons and pseudo-wavefunctions agree beyond a certain cutoff, rc Real and pseudo norm squares integrated from 0 to R for all R < rc agree Additional condition for transferability [see PRL 43, 1494 (1979)]
  • 15. How a pseudopotential is constructed 1) Solve the all-electron atomic system
  • 16. How a pseudopotential is constructed 1) Solve the all-electron atomic system 2) Decide which of the states are to be taken as core and which are to be taken as valence
  • 17. How a pseudopotential is constructed 1) Solve the all-electron atomic system 2) Decide which of the states are to be taken as core and which are to be taken as valence 3) Generate the pseudopotentials, Vnl (r), for the valence states.
  • 18. How a pseudopotential is constructed 1) Solve the all-electron atomic system 2) Decide which of the states are to be taken as core and which are to be taken as valence There is no unique way to determine the Vnl ’s. There are two opposing considerations : 1. Good transferability small r ⇒ c . 2. Large rc smoother pseudopotentials. ⇒ 3) Generate the pseudopotentials, Vnl (r), for the valence states.
  • 19. Exchange-correlation functions LSDA GGA meta GGA hybrid meta GGA hybrid GGA fully non-local ‘Earth’ Hartree Theory ‘Heaven’ Chemical Accuracy
  • 21. The energy functional [ ] [ ] [ ] [ ] [ = [ ] ] e G e T n V n F n E n U n U n     3 [ ] ( ) ( ) U n d r n r U r    The F [n] functional is the same for all electronic systems. “No ones knows the true F [n], and no one will, so it is replaced by varios uncontrollable approximations.” (Marder, p.247)
  • 22. We split the problem in various parts…. • Kinetic energy functional • Hartree energy functional (exact) TS [n]=∑i N v ⟨ϕi|−1 2 ∇i 2 |ϕi⟩ where fi are KS orbitals • Exchange correlation functional (unknown) F[n]=TS[n]+U H [n]+ EXC [n] € Exc[n] contains T −Ts and U −UH EXC [n] U H [n]= e2 2 ∫∫d 3 r d 3 r ' n(r)n(r' ) |r−r '|
  • 23. The value of xc at any position r is calculated using only the value of  at that point. Ex[ ρ(r)]=∫− 9α 8 (3 p ) 1/3 ρ4/3 (r)d r • Correlation Energy - no analytic function for the uniform electron gas. (Cerperley Alder) - Vosko, Wilk, Nusair functionals - PZ, PW • Exchange Energy ((r)) € Exc hom ((r)) ((r)) € Exc hom Exc hom Exc LSDA =∫ρ(r )[ ex hom ( ρ(r))+ec hom ( ρ(r ))]dr Local Spin Density Approximation
  • 24. Local Spin Density Approximation - Performance • Structural properties are often good - usually slightly underestimates bulk lattice constants - bulk moduli are slightly too large - phonons too stiff • Binding energies are too negative - overbinding of molecular and metallic solids, up to several eV - molecular atomization energies have a large error • Activation energies for chemical reactions are unreliable • Relative stabilities of bulk phases can be wrong. • Density of states and band structure are okay but…. Ge
  • 25. Local Spin Density Approximation - Performance • Structural properties are often good - usually slightly underestimates bulk lattice constants - bulk moduli are slightly too large - phonons too stiff • Binding energies are too negative - overbinding of molecular and metallic solids, up to several eV - molecular atomization energies have a large error • Activation energies for chemical reactions are unreliable • Relative stabilities of bulk phases can be wrong. • Density of states and band structure are okay but…. Why does LSDA work? - Many metallic systems are similar to a homogeneous electron gas. - One might expect it to fail for molecular systems (it does!) What are it’s shortcomings? - self-interaction Ge
  • 26. Generalized Gradient Approximation Based on the idea of doing a Taylor expansion of the density: € f (ρ) = f (ρ0) + f '(ρ0)(ρ − ρo) + f ''(ρ0)(ρ − ρo)2 2! +L it turns out that a simple Taylor expansion does not improve upon LSDA because it violates several of the constraints that the exchange-correlation functional is known to observe. Also the gradients in real materials are often large enough that the Taylor expansion breaks down. € Ex/c GGA [ρ(r)] = Ex/c LSDA [ρ(r)]+ ΔEx/c ∇ρ(r) ρ4 /3 (r) ⎡ ⎣ ⎢ ⎤ ⎦ ⎥ lim∇ρ→ 0 Exc = Exc LSDA Obey the uniform electron gas limit:
  • 27. Development of Exchange and Correlation Functionals Empiricism •We don’t know the exact form of the exchange-correlation functional so write down a form and parameterize it to fit to a set of experimental or high-level ab-initio data. • B88, mPW, LYP Nonempiricism •The exchange and correlation functional should be developed from first principles by incorporating known constraints. Once the right constraints are met the rest will come out right. • Often times the functional forms used are the same as, or based on those from empirically based functionals. • PW91, PBE
  • 28. Generalized Gradient Approximation • Bulk lattice constants increase compared in LSDA • Cohesive energies decrease compared to LSDA • Atomic and molecular energies are improved • Corrects over binding of LSDA • Improves activation barriers, but they’re still too low. • Improved relative stability of bulk phases.
  • 29. Generalized Gradient Approximation • Bulk lattice constants increase compared in LSDA • Cohesive energies decrease compared to LSDA • Atomic and molecular energies are improved • Corrects over binding of LSDA • Improves activation barriers, but they’re still too low. • Improved relative stability of bulk phases. Problems with LSDA and GGA ● Self interaction ● Neglect of strong correlations ● Exchange-splitting underestimated for narrow d- and f-bands. ● Many transition-metal compounds are Mott-Hubbard or charge-transfer (not described by LDA/GGA)
  • 30. beyond GGA …… 1/2 Meta-GGA Functionals € E[ρ(r)] ⇒ ∇2 ρ(r) If we’re interested in following the Taylor-like expansion of the density the next logical step is the Laplacian of the density. B95,B98,KCIS,HCTH,TPSS LDA+U Method ● A generalization of the LSDA method that takes into account orbital dependencies of the exchange and Coulomb interactions. ● The correction is usually only considered for highly localized atomic-like orbitals on the same site. ● The added term serves to shift the energies of the localized orbitals it is applied to relative to the other orbitals. SIC -Self-Interaction Correction Methods that use approximate functionals and add in a “self-interaction correction” to correct for the unphysical self-interaction that is present in many functionals. ..it requires U parameter...
  • 31. beyond GGA …… 2/2 Hybrid Functionals (scaling behaviour N4 ) Exc hybrid DFT = (1− a)Ex LSDA + aEx HF + bΔEx NL + Ec LSDA + cΔEc NL • Functionals: B3PW91, B3LYP, O3LYP, X3LYP Exc hybrid DFT = (1− a)(Ex LSDA + ΔEx NL ) + aEx HF + Ec LSDA + ΔEc NL • Functionals: B1PW91, B1LYP, B1B95, mPW1PW91, PBE1PBE Hybrid meta Functionals (scaling behaviour N4 ) • Hybrid functionals that contain kinetic energy density, in addition to Hartree–Fock exchange.
  • 33. V-representability "a number of 'reasonable' looking densities that have been shown to be impossible to be the ground state density for any V." R. Martin The central assumption of KS is that such an equivalent system of non-interacting electrons always exists, i. e., that all interacting density can be represented by a non-interacting one
  • 34. Temperature and particles In a statistical mechanical framework the focus is on the grand canonical potential. Here, the grand canonical Hamiltonian plays an analogous role as the one played by the Hamiltonian for the ground-state problem. Where the entropy operator is defined as Kohn-Sham equations at non-zero temperature, and the density Thermal Density Functional Theory in Context, https://arxiv.org/abs/1309.3043
  • 35. Summary • As one climbs the Jacob’s ladder of density functionals, the complexity and cost of the calculation increase, as does the accuracy. - LSDA  density only - GGA  density and gradient of the density - metaGGA  density, gradient of the density, kinetic energy density - hybrid  density, gradient of the density, Hartree–Fock exchange - meta hybrid  density, gradient of the density, kinetic energy density, Hartree–Fock exchange. • Different functionals within the same class can give very different results. - its not enough to say that you used GGA. You must specify the use of PBE, BLYP, etc. Limitations of DFT • Some systems cannot be well described by a single Slater determinant. • Its hard to know how to systematically improve DFT functionals. • Does not obey the variational principle.
  • 36. Bibliography Pseudopotentials  D.R. Hamann, M. Schlüter, and C. Chiang, Phys. Rev. Lett. 43, 1494 (1979).  G.P. Kerker, J. Phys. C: Solid St. Phys. 13, L189 (1980).  L. Kleinman and D.M. Bylander, Phys. Rev. Lett. 48, 1425(1982)  N. Troullier and J.L. Martins, Phys. Rev. B 43, 1993 (1991)  D. Vanderbilt, Phys. Rev. B 41, 7892 (1990).  G. Kresse and D. Joubert, Phys. Rev. B 59, 1758 (1999) Exchange-correlation funcitions  Review of approximations for the exchange-correlation energy in density-functional theory https://arxiv.org/abs/2103.02645  Progress in the development of exchange-correlation functionals https://www.sciencedirect.com/science/article/pii/B9780444517197500676 Miscellaneous  Thermal Density Functional Theory in Context, https://arxiv.org/abs/1309.3043
  • 37. Strength of DFT From W.Aulbur’s slides
  • 38. • Neglect of strong correlations - Exchange-splitting underestimated for narrow d- and f-bands. - Many transition-metal compounds are Mott-Hubbard or charge-transfer insulators, but DFT predicts metallic state. From J. Hafner’s slides Metal in “LDA” calculations LDA DFT Calcs. (dashed lines) Rohlfing and Louie PRB 1993 LDA, GGAs, etc. fail in many cases with strong correlations. • Band-gap problem: HKS theorem not valid for excited states. Band-gaps in semiconductors and insulators are usually underestimated. Can be fixed by GWA…etc Weakness of DFT
  • 39. Hybrid Functionals What’s the rationale for adding Hartree–Fock exchange?? - LSDA and GGA overbind. - Hartree–Fock theory underbinds. - LSDA and GGA give bond lengths that are too long. - Hartree–Fock theory gives bond lengths that are too short. Perhaps if we mix the two together we’ll get numbers in the middle… ROH = 0.9572 HOH = 104.52 R(O-H) Θ(H-O-H) LSDA 0.9701 104.9605 PBE 0.9690 104.2068 BLYP 0.9706 104.5230 PBE1PBE 0.9575 104.9051 B3LYP 0.9604 105.1265 Hartree Fock – 0.9399 106.2709 Ebind (kcal/mol) LSDA 7.95 PBE 5.05 BLYP 4.16 PBE1PBE 4.96 B3LYP 4.57 Hartree–Fock 3.49 Accurate 4.99 Hybrid methods also help correct the self-interaction problem, since in Hartree– Fock theory it cancels out.
  • 40. How do you Determine the Percent of HF Exchange? • The most common way to determine the amount of Hartree–Fock exchange to include it to treat it as an optimizable parameter. - Take a set of molecules for which the atomization energy, ionization potential, and proton affinities are well knows, as well as atomic total energies, and vary the percent of HF exchange until you minimize the unsigned error over this data set. - B3PW1, B3LYP,O3LYP, X3LYP •There are some “parameter free” hybrid density functionals, where the percentage of Hartree–Fock exchange was determined based on a perturbation- theory argument - B1B95, mPW1PW91, PBE1PBE (PBE0), B1LYP - in reality the percentage of HF exchange used cannot be determined nonempirically.
  • 41. Scaling Behavior Scaling Behavior Method N3 DFT(LSDA,GGA,meta) N4 HF, DFT(hybrid,hybrid meta) N5 MP2 N6 MP3, CISD, MP4DQ, CCSD, QCISD N7 MP4,CCSD(T),QCISD(T) N8 MP5,CISDT,CCSDT N9 MP6 N10 MP7,CISDTQ,CCSDTQ