time-dependent density-functional theory (td-dft)...p4 haibin su ( 蘇海斌 ) nordita, 4 oct. 2016...

Post on 11-Aug-2021

1 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

TRANSCRIPT

NORDITA, 4 Oct. 2016 1

Mark E. Casida ( 樫田 マーク , 卡西达 马克 ) Laboratoire de Chimie Théorique (LCT)Département de Chimie Moléculaire (DCM)Institut de Chimie Moléculaire de Grenoble (ICMG)Université Grenoble Alpes (UGA)F-38041 GrenobleFrancee-mail: mark.casida@univ-grenoble-alpes.fr

Time-Dependent Density-Functional Theory(TD-DFT)

NorditaStockholm, Sweden

4 October 20161 hour 45 min

NORDITA, 4 Oct. 2016 2

VERI

TAS

LIBE

RANT

“The truth will set you free”(Liberating Truth)

Université de Grenoble

Established: 1339Recreated: 1809 Napolean appointed Joseph Fourier rector of the university.

Dissolved: during the reorganizationof 1970 to create Grenoble I, II, III,and the INPG.Currently: Université Grenoble-Alpes

“Home”

NORDITA, 4 Oct. 2016 3

“A mountain at the end of every street.” –- Henri Beyl STENDHAL*

* Famous French author, originally from Grenoble.

NORDITA, 4 Oct. 2016 4

CHEMISTRY: “Better things for better living through chemistry”Dupont Chemical Motto: 1935-1982

Publish(but don't

mention the10,000 failures)

An idea fora synthesis.Try it out.

Characterizethe product.

Did you get the right molecule?

Why not?New idea for the synthesis.

YesNo

* Physical chemistry and theoretical chemistry start here.

**

******

** *

NORDITA, 4 Oct. 2016 5

NORDITA, 4 Oct. 2016 6

EXPERIMENTAL PHOTOCHEMISTRY

Gomer-Noyes Mechanism [E. Gomer et W.A. Noyes, Jr., J. Am. Chem. Soc. 72, 101 (1950);

T. Ibuki, M. Inasaki et Y. Takesaki, J. Chem. Phys. 59, 2076 (1973).]

NORDITA, 4 Oct. 2016 7

LET'S MODEL IT!

NORDITA, 4 Oct. 2016 8

p4

Haibin Su( 蘇海斌 )

NORDITA, 4 Oct. 2016 9

HOPPING PROBABILITY

Helecr ; R t elec

r , t =i ℏd

dt

elecr ,t

elecr , t =∑m

melecr ; R t C

mt

Expand the time-dependent wave function in terms of the solutions of thetime-independent Schrödinger equation.

Helecr ; R t

melecr ; R t =E

mR t

melecr ; R t

Probability of finding the system on surface m is

Pmt =∣C

mt ∣2

1st order equation.

Cmt =−i E

mt C

mt /ℏ−∑n

⟨m∣d n

dt⟩C

nt

IT IS A

LL GREEK TO M

E!

NORDITA, 4 Oct. 2016 10

p4

Haibin Su( 蘇海斌 )

NORDITA, 4 Oct. 2016 11

NORDITA, 4 Oct. 2016 12

1 2 3 4

5 6 7

Enrico Tapavicza, Ivano Tavernelli and Ursula Röthlisberger, École Polytéchnique Fédérale de Lauzanne, 2007

TDPBE/TDA with Landau-Zener surface hopping.

Hot reactions (~4000K): cyclic-CH

2OCH

2 -> CH

2CH

2O -> CH

3 + CHO or CH

4 + CO

NORDITA, 4 Oct. 2016 13

Goals :

Keep this presentation as simple as possible (Masters level).

But never be afraid to do something hard if it is also useful.

THIS IS A THEORY TALK

I want to give you an understanding of the mathematics and the physicsbehind the programs that do TD-DFT.

NORDITA, 4 Oct. 2016 14

OUTLINE

I. Ab Initio Quantum Chemistry

Pause

II. (TD-)DFT

NORDITA, 4 Oct. 2016 15

OUTLINE

I. Ab Initio Quantum Chemistry

A. Born-Oppenheimer ApproximationB. Hartree-Fock ApproximationC. Configuration InteractionD. Second QuantizationE. Equation-of-Motion/Superoperator Approach

NORDITA, 4 Oct. 2016 16

OUTLINE

I. Ab Initio Quantum Chemistry

A. Born-Oppenheimer ApproximationB. Hartree-Fock ApproximationC. Configuration InteractionD. Second QuantizationE. Equation-of-Motion/Superoperator Approach

NORDITA, 4 Oct. 2016 17

CONSIDER A MOLECULE ...

[Fe(H2O)

6]2+ [Fe(bpy)

3]2+[Fe(NH

3)]2+

M nuclei

R=( R1 , R2 ,⋯, RM )

N electrons

r=( r 1 , r 2 ,⋯, rM )

R=(XYZ )

x=( x1 , x2 ,⋯, xM )r=(

xyz ) x=(

xyzσ)

spin

Erwin SCHRÖDINGER

H ne(r ,R)Ψne(x ,R , t )=i ∂∂ tΨne(x ,R , t)

Hartree atomic units:

ℏ=me=e=1

NORDITA, 4 Oct. 2016 18

BORN-OPPENHEIMER SEPARATION

H ne(r ,R)Ψne(x ,R , t )=i ∂∂ tΨne(x ,R , t)

H ne(r ,R)=T n(R)+V nn(R)+ H e(r ;R)

(1)

(2)

A separation of time scales approximation:1) The electrons “see” fixed nuclei.2) The nuclei “see” the time-averaged field of the electrons.

H e(r ;R)Ψ Ie (x ;R)=E I

e(R)Ψ Ie (x ;R)

Time-independent orbital Schrödinger equation:

(3)

Born-Oppenheimer expansion:

Ψne(x ,R , t)=∑IΨ I

e(r ;R)Ψ In(R , t ) (5)

(4)H e(r ;R)=T e (r )+V ne(r ;R)+V ee(r )

NORDITA, 4 Oct. 2016 19

BORN-OPPENHEIMER GROUP EQUATION

−∑J {∑M

12mM

[∑K(δI , K ∇M+F I , K

(M ) (R))⋅(δK ,J ∇M+FK , J(M ) (R) ) ] }ΨJ

n(R , t )

+V I (R)Ψ In(R , t )=i ∂

∂ tΨ I

n(R , t ) (1)

(2)

(3)

Potential energy surface (PES)

V I (R)=E Ie(R)+V nn(R)

Derivative compling matrix (DCM)

F I , J (R)=⟨Ψ I

e(R)∣[ ∇M H e(R) ]∣Ψ I

e(R)⟩−δI , J ∇M E I

e(R)

EJe (R)−E I

e(R)

We may neglect DCM ifi. The force on the nuclei is sufficiently small.ii. The energy difference between different PESs is sufficiently big..

NORDITA, 4 Oct. 2016 20

BORN-OPPENHEIMER APPROXIMATION

We neglect the DCM.

( T n+V I (R) )Ψ I , vn (R , t )=i ∂

∂ tΨ I , v

n (R , t ) (1)

(2)

(3)

(4)

V I (R)=E Ie(R)+V nn(R)

H e(r ;R)Ψ Ie (x ;R)=E I

e(R)Ψ Ie (x ;R)

H e(r ;R)=T e (r )+V ne(r ;R)+V ee(r )

Ψ I , vne (x ,R , t )=Ψ I

e (x ;R)Ψ I , vn (R , t) (5)

NORDITA, 4 Oct. 2016 21

BUT PHOTOCHEMISTRY RELIES ON THE BREAK DOWN OF THE BORN-OPPENHEIMER APPROXIMATION!

NORDITA, 4 Oct. 2016 22

Let us concentrate on the electronic problem!

H e(r ;R)Ψ Ie (x ;R)=E I

e(R)Ψ Ie(x ;R)

H Ψ I (x)=E IΨ I (x)

(1)

(2)

NORDITA, 4 Oct. 2016 23

OUTLINE

I. Ab Initio Quantum Chemistry

A. Born-Oppenheimer ApproximationB. Hartree-Fock ApproximationC. Configuration InteractionD. Second QuantizationE. Equation-of-Motion/Superoperator Approach

NORDITA, 4 Oct. 2016 24

THE VARIATIONAL PRINCIPLE*

We want to solve

H I=E I I ; E0≤E1≤E2≤⋯ (1)

We will use the variational principle

E0≤W [Ψ]=⟨Ψ trial∣H∣Ψ trial⟩

⟨Ψ trial∣Ψ trial⟩(2)

Valid for every test function test

, satisfying the same boundary conditions as I.

* Chimie Quantique 101 ?

NORDITA, 4 Oct. 2016 25

THE PAULI PRINCIPLE

Ψ(1,2,⋯, j ,⋯, i ,⋯, N )=−Ψ(1,2,⋯, i ,⋯, j ,⋯, N )

Electrons are fermions:

(1)

where i= xi (2)

The simplest function satisfying equation (1) is aSlater determinant:

=∣1 ,2 ,⋯, N∣=1

N !det [

11 21 ⋯ N 112 22 ⋯ N 2⋮ ⋮ ⋱ ⋮

1N 2N ⋯ N N ] (3)

⟨ψi∣ψ j⟩=δi , j (4)where

NORDITA, 4 Oct. 2016 26

FOR A (FICTICIOUS) SYSTEM OF NONINTERACTING ELECTRONS

h0ψi=ϵi

0ψi ; ϵ1

0≤ϵ2

0≤ϵ3

0≤⋯ (1)

H 0=∑i=1,N

h0(i)

H 0Ψ0=E0Ψ0

(2)

Ψ0=∣ψ1 ,ψ2 ,⋯,ψN∣

(3)

E00=∑i=1,N

ϵi0

(4) (5)

NORDITA, 4 Oct. 2016 27

ROGUES GALLERY

Douglas RaynerHARTREE

VladimirFOCKWolfgang

PAULIJohn C. SLATER

Charlotte FROESE FISCHER*(applied mathematician and computer scientistwho worked on multiconfigurational Hartree-Fock)

* Lest we forget the heroinesamong our heroes!

NORDITA, 4 Oct. 2016 28

THE HARTREE-FOCK (HF) APPROXIMATION

The test function is a Slater determinant.

We minimize the variational energy subject to the condition that the orbitalsare orthonormal. The result is

f ψi=ϵiψi (1)

f=h+ v SCF (2)

The Fock operator:

Self-consistent field (SCF) :

v SCF=vH+Σx=J−K (3)

Physics notation Chemistry notation

NORDITA, 4 Oct. 2016 29

THE SELF-CONSISTENT FIELD

Density matrix γ(1,2)=∑iniψi(1)ψi

*(2) (1)

Density ρ(1)=∑ini∣ψi(1)∣

2 (2)

The Hartree (or Coulomb) potential

vH (1)=∫ρ(2)r12

d 2 (3)

The exchange self-energy (or exchange operator)

Σxϕ(1)=−∫γ(1,2)r12

ϕ(2)d 2 (4)

NORDITA, 4 Oct. 2016 30

HARTREE-FOCK (HF) ENERGY

E=∑ini ⟨ψi∣h∣ψi⟩+ESCF=∑i

ni ϵi−ESCF (1)

ESCF=EH+Ex (2)

EH=∫∫ρ(1)ρ(2)

r12

d 1d 2 (3) (4)Ex=−∫∫∣γ(1,2)∣2

r12

d 1d 2

NORDITA, 4 Oct. 2016 31

KOOPMANS' THEOREM*

* T. Koopmans, “Über die Zuordnung von Wellenfunktionen und Eigenwerten zu den EinzelnenElecktronen Eines Atomes'', Physica 1, 104 (1934).

In the HF approximation, in the absence of orbital relaxation:

−IPi=EN−EN−1i−1=i

Ionization potential (IP)(1)

The occupied orbitals “see” N-1 electrons.

−EAa=EN1a1−EN=a

Electron affinity (EA)

The unoccupied orbitals “see” N electrons.

M+IP→M++e-(v=0)

(2)

M+e-(v=0)→M-+EA (3)

(4)

Tjalling CharlesKOOPMANS

NORDITA, 4 Oct. 2016 32

f ψi=ϵiψi

Linear Combination of Atomic Orbitals (LCAO) Theory

Want to solve the molecular orbital (MO) equation,

(1)

Expand each MO in a basis of atomic orbitals (AOs),

ψi=∑νχνc ν , i (2)

Either (i) rigorously by finding the MO coefficients c,i by doing a variational minimization of the energy or (ii) heuristically by inserting Eq. (2) in Eq. (1) and projecting:

∑νf χνc ν ,i≈ϵi∑ν

χν cν ,i (3)

∑ν⟨χμ∣f∣χν⟩cν , i=ϵi∑ν

⟨χμ∣χν⟩cν , i (4a) N equations inN unknowns

f c i=ϵi s c i (4b)

NORDITA, 4 Oct. 2016 33

Per-Olov Löwdin's Solution

Solve s v i=λi v i (1) with v i+ v j=δi , j (2)

Construct s−1/2=∑iv i+ λi

−1/2 v i (3) and~f=s−1/2 f s−1/2

(4)

~f d i=ϵi di

may be obtained by solving(5)Then c i=s−1/2 di (6)

because ( s−1/2 f s−1/2 )⏟~f

( s+1/2 c i )⏟d i

=ϵi ( s+1/2 c i )⏟d i

(7)

Upsala,Sweden

Gainsville,Florida, USA

NORDITA, 4 Oct. 2016 34

OUTLINE

I. Ab Initio Quantum Chemistry

A. Born-Oppenheimer ApproximationB. Hartree-Fock ApproximationC. Configuration InteractionD. Second QuantizationE. Equation-of-Motion/Superoperator Approach

NORDITA, 4 Oct. 2016 35

We may finally start talking about excited states!!

NORDITA, 4 Oct. 2016 36

CONFIGURATION INTERACTION (CI)

All of the orthonormal orbitals, taken together, form a complete (infinite) basis set that canbe used to expand any one-electron wave function.

Similarly, all of the (infinitely many) N-electron Slater determinants, taken together, Constitute a complete basis set that may be used to expand any N-electron wave function.

This type of N-electron expansion is called “complete CI” and it is just a nice dream!

Ψ=C0Φ+∑i , aCaiΦi

a+∑i , j , a , bCabjiΦij

ab+⋯ (1)

An expansion in all of possible determinants made from a finite orbital basis is called “full CI”and is only possible for small basis sets (and hence for small molecules).

For normal sized orbital basis sets, we can only do “truncated CI”.

NORDITA, 4 Oct. 2016 37

TRUNCATED CI

test = truncated CI

Using this trial function in a variation calculation is a linear variation and specialtheorems apply.

H C I=E I C I

H=[H 0,0 H 0,1+ ⋯

H 0,1 H1,1 ⋯⋮ ⋮ ⋱

] C I=(C0

C2

⋮)

(C1 )ai=CaiH 0,0=⟨Φ∣H∣Φ⟩=EHF

( H 0,1 )ai=⟨Φia∣H∣Φ⟩=f a ,i=0

(1)

(2)

(3a)(4a)

(4b)

(H1,1 )ai , bj=δi , jδa , bEHF+Aai , bj

(3b)

(3c)

(3d)

(Brillouin's theorem)

NORDITA, 4 Oct. 2016 38

Only for linear variations (such as CI):

THE HYLLERAAS-UNDHEIM-MACDONALD or CAUCHY'S INTERLEAVING THEOREM

1 2 3 4 5 6 7 8 9 ...

The nth level is an upper bound to the true energyof the nth state.

NORDITA, 4 Oct. 2016 39

CONFIGURATION INTERACTION SINGLES (CIS)

[H 0,0 H 0,1+

H 0,1 H1,1] (C0

C1)=E (C0

C1) (1)

[EHF 0+

0 A+EHF] (C0

C1)=E (C0

C1) (2)

A C1=(E−EHF ) C1=ω C1 (3)

How do we solve such a big eigenvalue equation?

NORDITA, 4 Oct. 2016 40

NORDITA, 4 Oct. 2016 41

NORDITA, 4 Oct. 2016 42

Exact up to numerical convergence. Not an approximation!

NORDITA, 4 Oct. 2016 43

CONICAL (or diabolic) INTERSECTIONS

V I (R)=E Ie(R)+V nn(R)

If we only count internal degrees of freedom, then the PES is

F=3N−6 *

* F=3N−5 for a linear molecule.

dimensional in a 3N-5 dimensional space (3N-4 for a linear molecule).

What dimension is the intersection space of two PESs?

Conditions restricting the dimensionality:

E Ie(R)=EJ

e (R)i) ⇒ dimension F=3N−7

ii) H I , Je (R)=0 ⇒ dimension F=3N−8

(3N-6)

(3N-7)**

** Exception: This is not a restrictionwhen it is automatically guaranteedeither for symmetry reasons or becauseof Brillouin's theorem!

This is a diabolo.

NORDITA, 4 Oct. 2016 44

PHOTOCHEMICAL FUNNELS

Avoided crossing Conical intersection

Diatomics never have conicalintersections. However avoided crossings are possible betweenPESs corresponding to electronicstates with the same symmetry.

NORDITA, 4 Oct. 2016 45

OUTLINE

I. Ab Initio Quantum Chemistry

A. Born-Oppenheimer ApproximationB. Hartree-Fock ApproximationC. Configuration InteractionD. Second QuantizationE. Equation-of-Motion/Superoperator Approach

NORDITA, 4 Oct. 2016 46

(H1,1 )ai , bj=δi , jδa , bEHF+Aai , bj

HOW TO FIND A FORMULA FOR THE A MATRIX?

It is “easy” with Second Quantization.

Aai , bj=δi , jδa ,b (ϵa−ϵi )+[ai∣ jb ]−[ab∣ ji]

(1)

(2)

Answer:

NORDITA, 4 Oct. 2016 47

His car from the 1970s.

NORDITA, 4 Oct. 2016 48

For us, this is just a way to make it easier to do thealgebra needed when working with antisymmetricwave functions.

Only for Shaolin masters!?

(I don't think so.)

NORDITA, 4 Oct. 2016 49

FUNDAMENTAL IDEAS

The physical vacuum Φ=∣ψ1 ,ψ2,⋯,ψN∣ (1)

Creation operator ar+Φ=∣ψr ,ψ1 ,ψ2,⋯,ψN∣ (2)

Annihilation operator

ar ar+Φ=ar∣ψr ,ψ1 ,ψ2,⋯,ψN∣=∣ψ1 ,ψ2,⋯,ψN∣=Φ (3)

Anticommutation rules:

[r , s ]+=rs+sr=0 (6) [r+ , s+ ]+=r+ s++s+ r+

=0 (7)

[r , s+ ]=r s++s+ r=δr , s (8)

Lazy notation:

r=ar r+=ar+

(4) (5)

NORDITA, 4 Oct. 2016 50

SECOND-QUANTIZED OPERATORS

H=∑r , shr , sr

+ s+12∑p ,q , r , s

[ pr∣qs] p+q+ s r (1)

hr , s=⟨ψr∣h∣ψs ⟩

[ pq∣r s ]=∫∫ψp*(1)ψq(1)

1r12

ψr*(2)ψs(2)d 1d 2

(2)

Mullikan “charge cloud” notation:

(3)

NORDITA, 4 Oct. 2016 51

WICK'S THEOREM (I)

ab c⋯gh⏟unoccupied

i j k lmn⏟occupied

o pq⋯ y z⏟free

“FORTRAN index” notation

Normal order: A product of creation and annihilation operators is in normal order when(1) For occupied orbitals, all the creation operators are to the right of all of the annihilation

operators ( ijk … m+ n+ …)(2) For unoccupied orbitals, all the annihilation operators are to the right of all the creation

operators (a+ b+ c+ … d e ...)

Nonzero contractions:

i+ j=δi , j ab+=δa ,b

or

r+ s=nsδr , s

(occupied) (unoccupied)

s r+=(1−nr)δr , s=nrδr , s

NORDITA, 4 Oct. 2016 52

WICK'S THEOREM (II)

Wick's Theorem (simplified) :The expectation value with respect to the physical vacuum is just the sum of all possiblefully-contracted products multiplied by +1 (-1) if the number of intersections of thecontraction lines is even (odd).

Example :

NORDITA, 4 Oct. 2016 53

HARTREE-FOCK (HF)

EHF=∑ihi , i+

12∑i , j

([ii∣jj ]−[ ij∣ ji ] ) (1)

f=∑r , sf r , s r

+ s (2) f r , s=hr , s+∑i([rs∣ii ]−[ri∣is ]) (3)

If the orbital basis consists of HF orbitals, then

f r , s=ϵrδr , s f=∑rϵr r

+r(4) (5)

NORDITA, 4 Oct. 2016 54

EVALUATION OF Hia,jb

One-electron part

hai , bj=⟨Φia∣h∣Φ j

b⟩=∑p ,q

h p, q ⟨Φ∣i+ a p+ q b+ j∣Φ⟩ (1)

=∑p ,qhp , q (i

+ a p+ q b+ j+i+ a p+ q b+ j+i+ a p+ q b+ j ) (2)

=∑p , qhp ,q (δi , jδa , pδb ,q−δa ,bδi ,q δ j , p+δi , jδa ,bδp , qnp ) (3)

=δi , jha, b−δa , bh j , i+δi , jδa , b∑khk , k (4)

=δi , jha ,b−δa ,bh j , i+δi , jδa ,b ⟨Φ∣h∣Φ⟩ (5)

NORDITA, 4 Oct. 2016 55

EVALUATION OF Hia,jb

Two-electron part

v ai , bj=⟨Φia∣vee∣Φ j

b⟩=∑p ,q , r , s[ pr∣qs ]⟨Φ∣i+ a p+ q+ s r b+ j∣Φ⟩ (1)

=∑p ,q ,r , s[ pr∣qs] i+ a p+ q+ s r b+ j+∑p ,q , r , s

[ pr∣qs ] i+ a p+ q+ s r b+ j(Term A) (Term B)

(2)

NORDITA, 4 Oct. 2016 56

EVALUATION OF Hia,jb

Two-electron part (continued)

+∑p , q , r , s[ pr∣qs] i+ a p+ q+ s r b+ j+∑p ,q , r , s

[ pr∣qs ]i+ a p+ q+ s r b+ j(Term C) (Term D)

+∑p ,q , r , s[ pr∣qs] i+ a p+ q+ s r b+ j+∑p ,q , r , s

[ pr∣qs ]i+ a p+ q+ s r b+ j(Term E) (Term F)

(1)

(2)

NORDITA, 4 Oct. 2016 57

EVALUATION OF Hia,jb

Two-electron part (continued)

+∑p , q , r , s[ pr∣qs] i+ a p+ q+ s r b+ j+∑p ,q , r , s

[ pr∣qs ]i+ a p+ q+ s r b+ j(Term G) (Term H)

+∑p ,q , r , s[ pr∣qs] i+ a p+ q+ s r b+ j+∑p ,q , r , s

[ pr∣qs ]i+ a p+ q+ s r b+ j(Term I) (Term J)

(1)

(2)

NORDITA, 4 Oct. 2016 58

EVALUATION OF Hia,jb

Two-electron part (continued)

+∑p , q , r , s[ pr∣qs] i+ a p+ q+ s r b+ j+∑p ,q , r , s

[ pr∣qs ]i+ a p+ q+ s r b+ j(Term K) (Term L)

+∑p ,q , r , s[ pr∣qs] i+ a p+ q+ s r b+ j+∑p ,q , r , s

[ pr∣qs ]i+ a p+ q+ s r b+ j(Term M) (Term N)

(1)

(2)

NORDITA, 4 Oct. 2016 59

EVALUATION OF Hia,jb

Two-electron part

v ai , bj=⟨Φia∣vee∣Φ j

b⟩=∑p ,q , r , s[ pr∣qs ]⟨Φ∣i+ a p+ q+ s r b+ j∣Φ⟩ (1)

=δi , jδa ,b∑k ,l[kk∣l l ]−δi , jδa, b [kl∣lk ]

(Term A) (Term B)

which is simply δi , jδa ,b ⟨Φ∣vee∣Φ⟩

(2)

NORDITA, 4 Oct. 2016 60

EVALUATION OF Hia,jb

Two-electron part (continued)

+δi , j∑k[ab∣k k ]−δi , j∑k

[ak∣k b ]

(Terms C+E) (Terms D+F)

which is simply +δi , j va, bSCF

−δa ,b∑k[ ji∣k k ]+δi , j∑k

[ j k∣k i ]

(Terms H+I) (Terms G+J)

which is simply −δa ,b v j ,iSCF

(1)

(2)

NORDITA, 4 Oct. 2016 61

EVALUATION OF Hia,jb

Two-electron part (continued)

+[a i∣ j b ]−[ab∣ j i ]

(Terms L+M) (Terms K+N)

Putting it all together gives:

H ai , bj=Aai ,bj+δi , jδa ,bEHF

where

Aai , bj=δi , j f a ,b−δa ,b f j ,i+[ai∣ jb ]−[ab∣ji ]

orAai , bj=δi , jδa , b (ϵa−ϵi )+[ai∣ jb ]−[ab∣ji]

in terms of HF MOs.

(1)

(2)

(3)

(4)

NORDITA, 4 Oct. 2016 62

TWO­ORBITAL TWO­ELECTRON MODEL (TOTEM)*

ia

ia

ia

ia

ia

a i

a i

a i

a i ∣ii∣

∣ai∣

∣i a∣

∣i a∣

∣ai∣

Singlet

Triplets 1,0=∣i a∣

S , M

S

1,−1=∣ai∣

0,0=

1

2∣ai∣∣i a∣

1,0=

1

2∣ai∣−∣ia∣

* two-orbital two-electronmodel.

NORDITA, 4 Oct. 2016 63

TOTEM CIS

[ϵa−ϵi+[ai∣ia ]−[aa∣ii ] [ai∣ia ][ai∣ia ] ϵa−ϵi+[ai∣ia ]−[aa∣ii ] ] (

X aiα

X aiβ)=ω (X aiα

X aiβ)

(X aiα

X aiβ)=(11 ) ⇒ ω=ϵa−ϵi+2 [ai∣ia ]−[aa∣ii]

(X aiα

X aiβ)=( 1−1 ) ⇒ ω=ϵa−ϵi−[aa∣ii]

Singlet type excitation

Triplet type excitation

NORDITA, 4 Oct. 2016 64

A consequence of Koopmans' theorem is that the TOTEM formulae are almost never acccurate enough to be useful.

NORDITA, 4 Oct. 2016 65

HF DFT

NORDITA, 4 Oct. 2016 66

OUTLINE

I. Ab Initio Quantum Chemistry

A. Born-Oppenheimer ApproximationB. Hartree-Fock ApproximationC. Configuration InteractionD. Second QuantizationE. Equation-of-Motion/Superoperator Approach

NORDITA, 4 Oct. 2016 67

The idea here is to present a method which is relatively “easy”and “elegant” to understand but which gives us the same equations as response theory.

NORDITA, 4 Oct. 2016 68

EQUATION-OF-MOTION (EOM)

The EOMH O= [ H , O ]= O

has excitation-type solutions

[ H ,∣I0∣ ]=E I−E0∣I0∣

and de-excitation-type solutions

[ H ,∣0I∣ ]=E0−E I ∣0I∣

(1)

(2)

(3)

(it also has other solutions, but these are all we care about here!)

Let us seek a solution of the form

O+=∑aia+ i X ai+∑ai

i+aY ai (4)

NORDITA, 4 Oct. 2016 69

SUPEROPERATOR METRIC AND MATRIX EQUATION

H O= [ H , O ]= O (1)

O+=∑aia+ i X ai+∑ai

i+aY ai (2)

Inserting (2) into (1) and using the “metric”

( A∣B )=⟨Φ∣[ A+ , B ]∣Φ⟩ (3)

gives us

[ A BB∗ A∗ ] XY = [1 0

0 −1 ] XY (4)

with Aai , bj=δi , jδa ,b(ϵa−ϵi)+[ai∣ jb]−[ab∣ ji ]

Bai , bj=[ib∣ ja]−[ jb∣ia]

(5)

(6)

NORDITA, 4 Oct. 2016 70

PAIRED SOLUTIONS

[ A BB∗ A∗ ] XY = [1 0

0 −1 ] XY (1)

[ A BB∗ A∗ ] Y

X∗ =− [1 00 −1 ] Y

X∗ (2)

excitation

corresponding de-excitation

NORDITA, 4 Oct. 2016 71

THE TAMM-DANCOFF APPROXIMATION

[ A B=0B∗=0 A∗ ] XY = [1 0

0 −1 ] XY (1)

A X= X (2)

This is the same as CIS !!

NORDITA, 4 Oct. 2016 72

TOTEM

singlet

triplet

T= [a−iia∣f H∣ai−ii∣f H∣aa ] [a−i−ia∣f H∣ai−ii∣f H∣aa ]

S= [a−iia∣f H∣ai−ii∣f H∣aa ] [a−i3ia∣f H∣ai−ii∣f H∣aa ]

T will be imaginary when

ia∣f H∣aiii∣f H∣aaa−i

Triplet-type instability!

(1a)

(2a)

(3)

≈a−i−ii∣f H∣aa (1b)

≈a−i2ia∣f H∣ai−ii∣f H∣aa (2b)

NORDITA, 4 Oct. 2016 73

HF STABILITY ANALYSIS*

Given a spin-restricted solution (same orbitals for different spin), is it possible that there is alower energy solution with different orbitals for different spin?

rr =ei Ri I

rr

Look at an arbitrary unitary transformation of the orbitals:

E=E02 [ R A−B RI AB I ]O 3

We find

But another way to write the EOM matrix equation is

AB A−B ZI=

I2 Z

IConclusion : Symmetry breaking will occur if there are imaginary excitation energies.

* J. Cizek and J. Paldus, J. Chem. Phys. 47, 3976 (1967).

H H H H (1)

(2)

(3)

(4)

NORDITA, 4 Oct. 2016 74

IME OUT

LET'S TAKE ABREAK

NORDITA, 4 Oct. 2016 75

NORDITA, 4 Oct. 2016 76

OUTLINE

II. (TD-)DFT

A. DFT i. Formalism ii. Fonctionals B. TD-DFT iii. Formalism iv. Response theory v. The “Casida equation” vi. Absorption spectra vii. Closing remarks C. Some Problems

NORDITA, 4 Oct. 2016 77

OUTLINE

II. (TD-)DFT

A. DFT i. Formalism ii. Fonctionals B. TD-DFT iii. Formalism iv. Response theory v. The “Casida equation” vi. Absorption spectra vii. Closing remarks C. Some Problems

NORDITA, 4 Oct. 2016 78

Walter KOHN (9 March 1923 - 19 April 2016)

● Father of modern DFT● Theoretical physicist and a great friend to chemists● 1998 Nobel prize in Chemistry

NORDITA, 4 Oct. 2016 79

OUTLINE

II. (TD-)DFT

A. DFT i. Formalism ii. Fonctionals B. TD-DFT iii. Formalism iv. Response theory v. The “Casida equation” vi. Absorption spectra vii. Closing remarks C. Some Problems

NORDITA, 4 Oct. 2016 80

1st HK Theorem:  The external potential is determined up to an additive Constant by the ground­state charge density. 

Corollary:

N , vextC HC

I,

I=E

I−E

0

2nd HK Theorem: The ground­state energy and charge density may be obtained by minimizing the variational expression 

E=F []∫ vextr r d r

The functional F[] is "universel" in the sense of being independent of vext

.

The Two Hohenberg­Kohn (HK) Theorems[Phys. Rev. 136, B864 (1964)]

(1)

(2)

NORDITA, 4 Oct. 2016 81

The Levy­Lieb Constrained Search Formalism

Avoids the v­representablity problem.It assumes that the denisty is N­representable (and it really is!)The exact functional is actually known!  (but it is not practical!)

F []=min⟨∣TV ee∣ ⟩

⟨∣ ⟩

Ensemble formulation:

F []=minD tr [ TV ee D ]

D=∑IpI∣ I I∣

pI is the probability of observing the state I.

(1)

(2)

(3)

NORDITA, 4 Oct. 2016 82

Introducing N orthonormal Kohn­Sham orbitals allows us to give an exactdescription of an important part of the total energy. 

E=∑ini

i∣−

1

2∇ 2v

ext∣

i

1

2∫∫

r1r

2

∣r 1−r 2

∣d r 1

d r 2E

xc[]

wherer =∑i

ni∣

ir ∣2

Variational minimization subject to the orbital orthonormality contraint gives the Kohn­Sham equation.

[−1

2∇

2v

extr ∫

r '

∣r−r '∣d r 'v

xcr ] i

r = i ir

where the exchange­correlation (xc) potential is vxc[]r =

Exc[]

r

The Kohn­Sham Formulation[Phys. Rev. 140, A1133 (1965)]

(1)

(2)

(3)

(4)

NORDITA, 4 Oct. 2016 83

We need ...v

xc[]r =

Exc[]

r

Definition E xc[]−E xc []=∫

E xc []

r r d r

Functional Derivatives

f xc[]r 1 ,r 2=

2 E xc[]

r 1 r 2

gxc []r 1 ,r 2 ,r 3=

3 E xc []

r 1 r 2 r 3

and sometimes also 

(1)

(2)

(3)

(4)

NORDITA, 4 Oct. 2016 84

« Le but n'est pas toujours placé pour être atteint, mais pour servirde point de mire ou de direction. »* --- Joseph Joubert (p. 221 of Receuil des pensées de M. Joubert by F.-R. De Chateaubriand, 1838)

* The target is not always meant to be hit, but rather to show where we should aim.

?

NORDITA, 4 Oct. 2016 85

OUTLINE

II. (TD-)DFT

A. DFT i. Formalism ii. Fonctionals B. TD-DFT iii. Formalism iv. Response theory v. The “Casida equation” vi. Absorption spectra vii. Closing remarks C. Some Problems

NORDITA, 4 Oct. 2016 86

● Up to this point, we have only discussed pure DFT (depends only on .)● Today pure DFT is pure spin-DFT (depends on

and on

.)

● Today we really see pure DFT (pure Kohn-Sham). Most people use some type of generalized Kohn-Sham (GKS) that includes an orbital dependence.

Taking Stock

NORDITA, 4 Oct. 2016 87

“Jacob's Ladder”William Blake

watercolor1799-1800

NORDITA, 4 Oct. 2016 88

Size of the molecule

ab initio DFT

hybrid/OEP

mGGA*

GGA

LDA

“No computer land.”

HARTREE WORLD

THEORETICALCHEMISTRYHEAVEN** 

Jacob's Ladder

r

x r =∣∇ r ∣

r 4/3

r =∑in

i∣∇

i∣

2

KxHF

MBPT

* or include ∇2

pure DFT

** 1 kcal/mol or better precision

NORDITA, 4 Oct. 2016 89

Range­Separated Hybrids  (RSH*)

1r12

=erfc (γr12)

r12⏟SHORT RANGE

+erf (γ r12)

r12⏟LONGRANGE

Molecules: SR <-> DFT LR <-> WF (e.g., HF)

Solids: SR <-> WF LR <-> DFT

TD-DFT applications: Y. Tawada, T. Tsuneda, S. Yanagisawa, T. Yanai, and K. Hirao, J. Chem. Phys. 120, 8425 (2004). S. Tokura, T. Tsuneda, and K. Hirao, J. Theoretical and Computational Chem. 5, 925 (2006). O.A. Vydrov and G.E. Scuseria, J. Chem. Phys. 125, 234109 (2006). M.J.G. Peach, E.I. Tellgrent, P. Salek, T. Helgaker, and D.J. Tozer, J. Phys. Chem. A 111, 11930 (2007). E. Livshits and R. Baer, Phys. Chem. Chem. Phys. 9, 2932 (2007).

The favorite type of GKS is lessAnd less B3LYP and more andMore a RSH.

* because Nature is not always short-sighted.

The RSH idea came originally from Andreas Savin (Université Pierre et Marie Curie, Paris.)

NORDITA, 4 Oct. 2016 90

SEMI­EMPIRICAL DISPERSION CORRECTION*

* S. Grimme, J. Antony, S. Ehrlich, and H. Krieg, J. Chem. Phys. 132, 154104 (2010).“A consistent and accurate ab initio parameterization of density functional dispersioncorrection (DFT-D) for the 94 elements H-Pu”

Semi-empirical (molecular-modeling-type) formula designed to interpolate between twophysical limits: ''the asymptotic part that is described very accurately by [TD-DFT] (for atomic or molecular fragments) and the short-range regime for which standard [DFAs] often yield a rather accurate description of the exchange-correlation problem''.

Edisp=∑A ,B∑n=6,8,10,. ..sn

CnA , B

RA ,Bn [1+6 (

RAB

sr ,nR0A , B )

−αn

]−∑A ,B ,C

√C6ABC6

ACC6BC(3cosθacosθbcosθc+1)

(RABRBC RCA )3 [1+6 (

RABC

sr ,3R0A ,B )

−α3

]

NORDITA, 4 Oct. 2016 91

OUTLINE

II. (TD-)DFT

A. DFT i. Formalism ii. Fonctionals B. TD-DFT iii. Formalism iv. Response theory v. The “Casida equation” vi. Absorption spectra vii. Closing remarks C. Some Problems

NORDITA, 4 Oct. 2016 92

OUTLINE

II. (TD-)DFT

A. DFT i. Formalism ii. Fonctionals B. TD-DFT iii. Formalism iv. Response theory v. The “Casida equation” vi. Absorption spectra vii. Closing remarks C. Some Problems

NORDITA, 4 Oct. 2016 93

Given a system, initially in its ground state, exposed to a time­dependentperturbation: 

1st Theorem (RG1): vext

 (rt) is determined by (rt) up to an arbitrary additive function of time.

Corollary: r t N , vext r t C t H t C t t e−i∫t0

tC t ' dt '

2nd Theorem (RG2): The time­dependent charge­density is a stationary pointof the Frenkel­Dirac action

A[]=∫t0

t t ' ∣i ∂

∂ t '− H t ' ∣ t ' dt '

TIME­DEPENDENT DENSITY­FUNCTIONAL THEORY  (TD­DFT)[E. Runge and E.K.U. Gross, Phys. Rev. Lett. 52, 997 (1984)]

(1)

(2)

[RG2 suffers from a “causality paradox”.  But this difficulty now seems to have been solved by G. Vignale, Phys. Rev. A 77, 062511 (2008).  “Real­time solution of the causality paradox of time­dependent density­functional theory”]

(RG1 has only been proven for functions whose time dependence can be expanded in a Taylor series.)

NORDITA, 4 Oct. 2016 94

TIME-DEPENDENT KOHN-SHAM EQUATION

[−12∇

2vext r t ∫

r ' t ∣r−r '∣

d r 'vxc r t ] i r t =i ∂∂ t i r t

r t =∑ini∣ i r t ∣2where

and vxc r t = Axc []

r t

(1)

(2)

(3)

[E. Runge and E. K. U. Gross, Phys. Rev. Lett. 52, 997 (1984)]

Remark: If the initial state is not the ground state, then

v xc r , t =v xc [ , 0,0]r , t (4)

NORDITA, 4 Oct. 2016 95

TD-DFT ADIABATIC APPROXIMATION (AA)

t r =r t

This defines “conventional TD-DFT”.

v xc r t = A xc []

r t v xc r t =

Exc [t ]

t r

Assume that the exchange-correlation (xc) potential responds instantaneously and withoutmemory to any change in the time-dependent potential.

NORDITA, 4 Oct. 2016 96

OUTLINE

II. (TD-)DFT

A. DFT i. Formalism ii. Fonctionals B. TD-DFT iii. Formalism iv. Response theory v. The “Casida equation” vi. Absorption spectra vii. Closing remarks C. Some Problems

NORDITA, 4 Oct. 2016 97

ELECTRONIC POLARIZATION INDUCED BY THE APPLICATION OFA TIME-DEPENDENT ELECTRIC FIELD

Classical model of a photon

Induced dipole moment

t =−e 0∣r∣

0t

0 t ∣r∣

0

t = cos0t

v r t =e t ⋅r

HH

O

HH Hℏ

0

photon

t

(1)

(2)

(3)

NORDITA, 4 Oct. 2016 98

The linear response for a property a at

to the perturbation b b t = b cos0t

HI=E

I

I

[ Hb t ]0t =i ℏ ∂

∂ t

0t

0 t =[

0

0t ]e

−iE0t / ℏ

which allows us to deduce the 1st order equation

b t 0=i ℏ ∂

∂ t− HE

0

0t

REPONSE THEORY FOR CHEMISTS …

(1)

(2)

(3)

(4)

(5)

NORDITA, 4 Oct. 2016 99

ENERGY REPRESENTATION

Fourier transforms

f =∫−∞∞

ei t f t dt

f t =1

2∫−∞

e−i t f d

b t 0=i ℏ ∂

∂ t− HE

0

0t

Transform the 1st order equation

b 0=ℏ− HE

0

0

Solve by Rayleigh-Schrödinger perturbation theory!

0=∑I≠0

I

I∣b ∣

0

−I

ℏI=E

I−E

0where ℏ=1

Joseph FourierPrefect of the Isère

(Grenoble)1802-1815

(1)

(2)

(3)

(4)

(5) (6)

NORDITA, 4 Oct. 2016 100

TIME REPRESENTATION

∫−∞∞

ei t cos0t dt=[

0−

0]

Since

∫−∞∞

ei t sin0t dt=

i[

0−

0]

we find

0t =∑I≠0

I

I

I∣b∣

0

0

2−

I

2cos

0t

−i ∑I≠0

I

0

I∣b∣

0

0

2−

I

2sin

0t

(1)

(2)

(3)

NORDITA, 4 Oct. 2016 101

RESPONSE OF AN ARBITRARY PROPERTY

at =0∣a∣

0t

0t ∣a∣

0

at =∑I≠0

2Iℜe

0∣a∣

I

I∣b∣

0

0

2−

I

2cos

0t

∑I≠02

0ℑm

0∣a∣

I

I∣b∣

0

I

2−

0

2sin

0t

Electronic polarizabilities, NMR chemical shifts

Circular dichroïsme

(1)

(2)

NORDITA, 4 Oct. 2016 102

THE DYNAMIC POLARIZABILITY

it =

i∑ j

i , j

jcos t⋯

i , j =∑I≠0

2I 0∣r i∣ I I∣r j∣ 0

I2−

2

=∑I≠0

fI

I

2−

2

fI=

2

3

I∣

0∣x∣

I∣

2∣

0∣y∣

I∣

2∣

0∣z∣

I∣

2

Sum-over-states (SOS*)f

I

ωI

(1)

(2)

(3)

(4)

NORDITA, 4 Oct. 2016 103

OUTLINE

II. (TD-)DFT

A. DFT i. Formalism ii. Fonctionals B. TD-DFT iii. Formalism iv. Response theory v. The “Casida equation” vi. Absorption spectra vii. Closing remarks C. Some Problems

NORDITA, 4 Oct. 2016 104

FREQUENCY/ENERGY FORMULATION

[ AI BI

BI AI ] X I

Y I=I [1 0

0 −1 ] X I

Y I

Mark E. Casida in Recent Advances in Density Functional Methods, Part I, edited by D.P. Chong (Singapore, World Scientific, 1995), p. 155."Time­dependent density­functional response theory for molecules'' *

Aij ,kl = ,i ,k j , l i− jK ij , kl

K ij , kl=∫∫ ∗i r j r f Hxc ,r ,r ' ; k r ' ∗l r ' d r d r '

Bij ,kl =K ij , lk

“RPA” equation

(1)

or (2)

(3)

Coupling matrix

(4)

Remark: The general formulation does not make the adiabatic approximation.

* 1246 citations according to Google Scholar Citations, 1 Oct. 2016.

NORDITA, 4 Oct. 2016 105

NORDITA, 4 Oct. 2016 106

NORDITA, 4 Oct. 2016 107

OUTLINE

II. (TD-)DFT

A. DFT i. Formalism ii. Fonctionals B. TD-DFT iii. Formalism iv. Response theory v. The “Casida equation” vi. Absorption spectra vii. Closing remarks C. Some Problems

NORDITA, 4 Oct. 2016 108

From Mark E. Casida, Bhaarathi Natarajan, and Thierry Deutsch, http://arxiv.org/abs/1102.1849.in Fundamentals of Time-Dependent Density-Functional Theory, edited by Miquel Marques, Neepa Maitra, Fernando Noguiera, E.K.U. Gross, and Angel Rubio, Lecture Notes in Physics, Vol. 837 (Springer Verlag: 2011), p. 279. "Non-Born-Oppenheimer dynamics and conical intersections"

NORDITA, 4 Oct. 2016 109

BEER'S LAW*

I 0 I

cuvette

Absorbance (unitless)

A=C l

incident intensity transmitted intensity

A=log10

I 0

I

Beer's law

lwidth (cm)

Concentration (mol/L) C

Molar extinction coefficiient (L/mol.cm)

(1)

(2)

* August Beer (1825-1863). The term Beer's law appears to date from 1889.

NORDITA, 4 Oct. 2016 110

=

[Co(bpy)3]2+

From A. Vargas et al., J. Chem. Theory Comput. 2, 1342 (2006)

cobalt(II)tris(2,2’-biypiridine) 2E t2g6 eg

1

t2g

eg

experiment

theory

NORDITA, 4 Oct. 2016 111

STICK SPECTRUM

E1=h1

E2=h2

● Vertical transition (ground-state equilibrium geometry)

● Neglect vibrations (and rotations) ● Neglect other contributions to the

width of the peaks (e.g., solvent, spectrometer artifacts)

● Average over rotations● Dipole-dipole approximation

Spectral function (units: cm)

Oscillator strength (unitless)

S =∑If I − I

f I=2Ime

3ℏ∣⟨ 0∣r∣ I ⟩∣

2

=2

Angular frequency (rad/s)

h=ℏ

(1)

(2)

(3) (4)

NORDITA, 4 Oct. 2016 112

EXPERIMENTAL OSCILLATOR STRENGTHS

We assume that the photon absorption process is sudden and use Fermi's (2nd) GoldenRule to obtain (in Gaussian units):

f I=me c ln 10

N A e2 ∫ d

You must integrate over all of the spectra coming from the electronic transition!

f I=me c 40 ln 10

N A e2 ∫ d

In SI units, this becomes

f I=(1.44e-19 mol.cm.s/L)∫ ϵ(ν)d ν

which makes

(1)

(2)

(3)

ICBST*

* It Can Be Shown That

NORDITA, 4 Oct. 2016 113

THEORETICAL CALCULATION OF THE MOLAR EXTINCTION COEFFICIENT

f I=me c 40 ln 10

N A e2 ∫ d

For a single transition,

= N A e

2

mec 40 ln 10f I − I

For several transitions,

= N A e

2

mec 40 ln 10S

S =∑If I − I

(1)

(2)

(3)

(4)

NORDITA, 4 Oct. 2016 114

GAUSSIAN CONVOLUTION

Normalized gaussian function

g = e−2

Full width at half maximum (FWHM) FWHM=2 ln 2

= N A e

2

mec 40 ln 10

×∑If I g− I

experiment

theoryOne program:

spectrum.py By Pablo Baudin and M.E.C.

(1)

(2)

(3)

Normally experiment and properlynormalized theoretical spectra haveintensities which agree to within an order of magnitude: ● Imperfect spectrometers● Neglect of vibrational coupling● Neglect of solvent● ???

NORDITA, 4 Oct. 2016 115

CALCULATION OF THE ABSORPTION SPECTRUM OF [Ru(trpy)3]2+

Example prepared by Denis MAGERO with Gaussian09.

Optimisation beginning with theX-ray structure.

B3LYP/6-31G(d)+LANL2DZ ECP on Ru

NORDITA, 4 Oct. 2016 116

HOW GOOD IS THE GEOMETRY?

Gas phase B3LYP calculations give a geometry close to the one measuredin the crystal.

D. Magero

NORDITA, 4 Oct. 2016 117

Exp.

CIS

TD-HFTD-B3LYP

Only the TD-DFT spectrumis close enough to the experiment to allow anassignment.

D. Magero

CALCULATION OF THE ABSORPTION SPECTRUM OF [Ru(trpy)3]2+

NORDITA, 4 Oct. 2016 118

OUTLINE

II. (TD-)DFT

A. DFT i. Formalism ii. Fonctionals B. TD-DFT iii. Formalism iv. Response theory v. The “Casida equation” vi. Absorption spectra vii. Closing remarks C. Some Problems

NORDITA, 4 Oct. 2016 119

RECENT BOOKS ON TD-DFT

Book: Time-Dependent Density Functional Theory, Edited by M.A.L. Marques, C.A. Ullrich, F. Nogueira, A. Rubio, K. Burke, and E.K.U. Gross, Lecture Notes in Physics vol. 706 (Springer: Berlin, 2006).

Book: Fundamentals of Time-Dependent Density-Functional Theory, edited by M. Marques, N. Maitra, F. Noguiera, E.K.U. Gross, and A. Rubio, Lecture Notes in Physics, Vol. 837 (Springer: Berlin, 2011)

Book: Time-Dependent Density Functional Theory, C.A. Ullrich (Oxford: 2012)

NORDITA, 4 Oct. 2016 120

NORDITA, 4 Oct. 2016 121

OUTLINE

II. (TD-)DFT

A. DFT i. Formalism ii. Fonctionals B. TD-DFT iii. Formalism iv. Response theory v. The “Casida equation” vi. Absorption spectra vii. Closing remarks C. Some Problems

NORDITA, 4 Oct. 2016 122

Joseph Kosuth1987

NORDITA, 4 Oct. 2016 123

O. Valsson, C. Filippi, and M.E. Casida, J. Chem. Phys. 142, 144104 (2015)Regarding the use and misuse of retinal protonated Schiff base photochemistry as a test case for time-dependent density-functional theory

NORDITA, 4 Oct. 2016 124

http://www.chm.bris.ac.uk/motm/retinal/retinalv.htm

SOME BIOCHEMISTRY

ElizabethTaylor

11Z11Z

NORDITA, 4 Oct. 2016 125

CLASSIC SINGLET MECHANISM

NORDITA, 4 Oct. 2016 126

RECENT AB INITIO CALCULATIONS AND EXPERIMENT SAYS THAT THE PHOTOREACTION IS A BIT MORE COMPLICATED.

[GML+13] S. Gozem, F. Melaccio, R. Lindh, A.I. Krylov, A.A. Granovsky, C. Angeli, and M. Olivucci, J. Chem. Theory Comput. 9, 4495 (2013). “Mapping the excited state potential energy surface of a retinal chromophore model with multireference and equation-of-motioncoupled cluster methods”[SKS11] R. Send, V.R.I. Kaila and D. Sundholm, J. Chem. Theory Comput. 7, 2473 (2011). “Benchmarking the approximate second-order coupled cluster method on biochromophores”[VF10] O. Valsson and C. Filippi, J. Chem. Theory Comput. 6, 1275 (2010). “Photoisomerization of model retinal chromophores: Insight with Quantum Monte Carlo and multiconfiguration perturbation theory”[ZHC09] G. Zgrablić, S. Haacke, and M. Chergui, J. Phys. B 113, 4384 (2009). “Heterogeneityand relaxation dynamics of the photoexcited retinal Schiff base cation in solution”Warning: trans → cis

NORDITA, 4 Oct. 2016 127

RETINAL PROTONATED SCHIFF BASE (PSB)

56

7

8

9

10

1113

1214

15

hν11Z → 11E

12

3

416

11 12 11 12

11 12 11 12

NORDITA, 4 Oct. 2016 128

Bond Length Alternation (BLA)

BLA = < R(C-C) > - < R(C=C) >where C-C and C=C refer to the ground state.

NORDITA, 4 Oct. 2016 129

Classical mechanismonly accessed aftersurmounting anexcited-state activation barrier.

Dynamic correlation leads to a floppy excited state with more equal bond lengths and rotations around “double bonds”.

d

dd

d

d

dd

dd

d

dd

d

d

dd

d = double bond in ground state

NORDITA, 4 Oct. 2016 130

d

dd

d

d

dd

dd

d

dd

d

d

dd

d = double bond in ground state

θ=93o

γ=0o

θ=−10o

γ=43.6o

CASPT2

CASSCF

θ=8.1o

γ=112.7o

CASPT2

B3LYP---

CAM-B3LYP---

θ=−27.4o

γ=17.7o

LC-BLYP

B3LYP CAM-B3LYP

θ=−0.2o

γ=86.8o

LC-BLYP

θ=−2.4o

γ=85.4oθ=−1.2o

γ=85.2o

NORDITA, 4 Oct. 2016 131

-Criterion*

* M.J.G. Peach, P. Benfield, T. Helgaker, and D.J. Tozer, J. Chem. Phys. 128, 044118 (2008).“Excitation energies in density functional theory: An evaluation and a diagnostic test”

Λ=∑i , a

κia2 Oia

∑i , aκia

2(1)

κia=X ia+ Y ia (2)

Oia=∫∣ψi( r )∣∣ψa( r )∣d r (3)

Small values (< 0.3) of indicatea high likelihood of a “charge-transfer” underestimation.

NORDITA, 4 Oct. 2016 132

Scan aroundsingle bond

NORDITA, 4 Oct. 2016 133

(TD-)DFT is advancing faster than many users can keep up!

The first events in photobiological systems appear to be within reach. Techniques are now being formulated which promise to take us even

further.

NORDITA, 4 Oct. 2016 134

That's about it … … except for a few wise words.

NORDITA, 4 Oct. 2016 135

NORDITA, 4 Oct. 2016 136

Special thanks to Lars G.M. Pettersson, Patrick Norman,and to everyone else

TACK SÅ MYCKET

TACK SÅ MYCKET

NORDITA, 4 Oct. 2016 137

top related