Semi-Empirical Methods

CHEM 430
 Cost
100 fs, 10
atoms:
photochemistry
•  Hartree%Fock Correlated Method
scales as N4 (N=# (MP2, CCSD)
basis funcFons)
–  Due to two% Density
electron AtomisFc MM FuncFonal
integrals 10 ms, thousands Theory
Molecular detail

within Fock of atoms: 10 ps, 100

protein folding,
matrix drug binding
atoms: chemical
reactions
•  Semi%empirical cut
cost by reducing
number of Coarse%grain
integrals 1 ms+, 1 million
atoms: dynamics of
–  Typically scale large proteins, cell
as N2 membranes, viruses

2
ComputaFonal cost
 Semi-Empirical MO Methods
• the high cost of ab initio MO calculations is largely
due to the many integrals that need to be
calculated (esp. two electron integrals)
• semi-empirical MO methods start with the general
form of ab initio Hartree-Fock calculations, but
make numerous approximations for the various
integrals
• many of the integrals are approximated by
functions with empirical parameters
• these parameters are adjusted to improve the
agreement with experiment
 Semi-Empirical MO Methods
• core orbitals are not treated by semi-
empirical methods, since they do not
change much during chemical reactions
• only a minimal set of valence orbitals are
considered on each atom (e.g. 2s, 2px, 2py,
2pz on carbon)
• Extended Hückel, Zero Differential Overlap
and Neglect of Diatomic Differential Overlap
Really Big Personalities...

John Pople Michael Dewar

 Zero Differential Overlap (ZDO)
• two electron repulsion integrals are one of the
most expensive parts of ab initio MO
calculations
1
( µ% | #$ ) = & " µ (1) "% (1) " # (2) "$ (2)d! 1d! 2
r12
• neglect integrals if orbitals are not the same
( µ! | #$ ) = ( µµ | ## ) " µ! " #$
where " µ! = 1 if µ = ! , " µ! = 0 if µ % !
• approximate integrals by using s orbitals only
• CNDO, INDO and MINDO semi-empirical
methods
 Neglect of Diatomic Differential
Overlap (NDDO)
• fewer integrals neglected
1
( µ% | #$ ) = & " µ (1) "% (1) " # (2) "$ (2)d! 1d! 2
r12

• neglect integrals if µ and ! are not on the same atom

or ! and ! are not on the same atom
• integrals approximations are more accurate and have
more adjustable parameters than in ZDO methods
• parameters are adjusted to fit experimental data and ab
initio calculations
• MNDO, AM1 and PM3 semi-empirical methods
• recent improvements: PDDG and PM6
 Neglect of Diatomic DifferenFal
Overlap (NDDO)
•  No further approximaFons than ZDO
•  Integral approximaFons are more exact than ZDO
•  More adjustable parameters than ZDO
•  Introduced by John Pople

5
 Intermediate Neglect of DifferenFal
Overlap (INDO)
•  Neglects all two%center two%electron integrals which are not of the
Coulomb type
•  Preserves rotaFonal invariance
–  Total energy independent of rotaFon of coordinate system
!Some integrals made independent of orbital type (e.g., integral
involving a p%orbital must be the same as with an s%orbital)
!One%electron integrals with two different funcFons on the
same atom and a potenFal energy operator from another atom
disappear
•  Happy medium between NDDO and CNDO

6
 Complete Neglect of DifferenFal
Overlap (CNDO)
•  Only Coulomb one%center and two%center two%electron integrals
remain
•  Integrals and approximaFons the same as in INDO

CNDO vs INDO vs NDDO

Only differ in treatment of two%electron integrals

CNDO, INDO NDDO

Reduce two%electron integrals All one% and two%center
to two parameters, γAA and γAB integrals are kept

7
Approximation of 1-Electron Integrals

H µµ = U µµ - å VAB Uµµ from atomic spectra

B¹ A VAB value per atom pair

H µu = 0 µ,u on the same atom

H µu = b ABSµu b AB = 1 2 (b A + b B )

One b parameter per element

Neglect of 2-Electron Integrals
2-e- integral CNDO INDO NDDO
(µ A µ A || lA lA ) + + +
(µ A µ A || lBlB ) + + +
(µ Au A || lA lA ) - + +
(µ Au A || lBlB ) - - +
(µ Au B || lA lA ) - - -
(µ Au A || lAs A ) - + +
(µ Au A || lBs B ) - - +
(µ Au A || lAs B ) - - -
(µ Au B || lAs B ) - - -
 Modified NDDO model
•  Modified neglect of differenFal overlap (MNDO)
•  AusFn Model 1 (AM1)
•  Parametric Method Number 3 (PM3)

How to transform NDDO into working computaFonal model…

•  Remaining integrals can be calculated from funcFonal form of the
atomic orbitals
•  Remaining integrals can be made into parameters (and then fit to
experimental data)
•  Remaining integrals can be made into parameters (assigned values
based on experimental data)

8
 MNDO
•  InteracFons between O%H and N%H bonds treated differently
MNDO '
" e−α A RAH −α H RAH
%
Vnn (A, H ) = Z Z H
A sA sH sA sH $1+ +e '
# RAH &
•  Core%core repulsion treated differently
Fit the α values
VnnMNDO (A, B) = Z A' Z B' sA sB sA sB (1+ e−α A RAB + e−α B RAB )

•  Fit for elements H,B,C,N,O,F,Al,Si,P,S,Cl,Zn,Ge,Br,Sn,I,Hg,Pb

•  Parameters taken from atomic spectra and others fit to molecular
data

Limita0ons
•  Sterically crowded molecules, four membered rings, weak
interacFons, acFvaFon energies, ethers, sulfides 9
 AM1
•  MNDO repulsion between two atoms which are 2%3 Angstroms
apart is too high!acFvaFon energies too large (among other
errors)…due to too repulsive interacFon in the core%core potenFal
•  Modified core%core funcFon from MNDO by adding Gaussian
funcFons and reparameterized

Z A' Z B' # −bkA ( RAB −ckA )2 −bkB ( RAB −ckB )2

&
%∑α kA e +∑α kB e
AM1 MINDO
Vnn (A, B) = V
nn (A, B) + (
RAB $ k k '
•  k between 2 and 4 depending on the atom
•  ak,bk,ck fit to molecular data
LimitaFons
•  Geometries with hydrogen bonds, alkyl groups, nitro compounds,
peroxide bonds, phosphor compounds
10
 Some Limitations of AM1
• predicts hydrogen bond strengths approximately
correct (but geometry often wrong)
• activation energies much improved over MNDO
• hypervalent molecules improved over MNDO,
but still significant errors
• alkyl groups systematically too stable by ca 2
kcal/mol per CH2 group
• nitro groups too unstable
• peroxide bonds too short
 PM3
•  Same core%core repulsion as AM1 but only two Gaussian funcFons
added to each atom
•  Automated opFmizaFon process
–  All parameters opFmized simultaneously (MNDO, AM1 done by
hand)
–  Significantly larger training set used

LimitaFons
•  Hydrogen bonds too short, gauche of ethanol more stable than
trans, Si%X bonds too short, charge of nitrogen atoms ohen
incorrect sign and/or magnitude
11
 Some Limitations of PM3
• hydrogen bonds are too short by 0.1 A
• almost all sp3 nitrogens are pyramidal
• Si – halogen bonds too short
• structures for NH2NH2, ClF3 wrong
• charge on nitrogens unrealistic
ParameterizaFon
 Performance of
semi%empirical methods

13
More performance discussion

15
Nobel Prize in Chemistry for 1981

Roald Hoffmann Kenichi Fukui

orbital symmetry control frontier molecular orbital theory
(Woodward-Hoffmann Rules) (FMO control of organic reactions)
extended Huckel theory 1st Asian to win Chemistry Nobel
 Extended Hückel Theory
(Roald HoffmanQ, 1960’s; implemented in YAeHOMP)

Solve H Ci = S Ci Ei
!
H — Hamiltonian Matrix
!
Ci — column vector of molecular
orbital coefficients
!
Ei — orbital energies
!
S — overlap matrix
!
Hii — use valence shell ionization
potentials
!
Hij = K Sij (Hii + Hjj)/2 with K = 1.75
 Extended Hückel Theory (EHT)
• All core e- are ignored (all modern semiempirical methods use this approximation)
• Only valence orbitals are considered, represented as STOs (hydrogenic orbitals).
Correct radial dependence. Overlap integrals between two STOs as a functions of r are
readily computed. (HT assumed that the overlap elements Sij = dij). The overlap between
STOs on the same atom is zero.
• Resonance Integrals:
• For diagonal elements: Hµµ = negative of ionization potential (in the
appropriate orbital). Valence shell ionization potentials (VSIPs) have been
tabulated, but can be treated as an adjustable parameter.
• For off-diagonal elements: Hµn is approximated as (C is an empirical
constant, usually 1.75):
H µu = Cµn (H µµ + Hnn )S µn
1
2

RESULT: the secular equation can now be solved to determine MO energies and wave
functions.
** Matrix elements do NOT depend on the final MOs (unlike HF) ® the process is NOT
iterative. Therefore, the solution is very fast, even for large molecules.
TABLE 10-1 Cartesian Coordinates (in Angstroms) for Atoms of
Methane Oriented as Shown in Fig. 10-1

Atom x y z

C 0.0 0.0 0.0

Ha 0.0 0.0 1.1
Hb 1.03709 0.0 − 0.366667
Hc − 0.518545 0.898146 − 0.366667
Hd − 0.518545 − 0.898146 − 0.366667

Figure 10-1 Orientation of methane in a Cartesian axis system.

TABLE 10-2 Basis AOs for Methane

AO no. Atom Type na la ma exp

1 C 2s 2 0 0 1.625
2 C 2pz 2 1 0 1.625
3 C 2px 2 1 (1)b 1.625
4 C 2py 2 1 (1)b 1.625
5 Ha 1s 1 0 0 1.200
6 Hb 1s 1 0 0 1.200
7 Hc 1s 1 0 0 1.200
8 Hd 1s 1 0 0 1.200
a n, l, m are the quantum numbers described in Chapter 4.
b2p and 2p are formed from linear combinations of m= + 1 and m= − 1 STOs, and
x y
neither of these AOs can be associated with a particular value of m.
TABLE 10-3 Overlap Matrix for STOs of Table 10-2

1 2 3 4 5 6 7 8

1 1.0000 0.0 0.0 0.0 0.5133 0.5133 0.5133 0.5133

2 0.0 1.0000 0.0 0.0 0.4855 − 0.1618 − 0.1618 − 0.1618
3 0.0 0.0 1.0000 0.0 0.0 0.4577 − 0.2289 − 0.2289
4 0.0 0.0 0.0 1.0000 0.0 0.0 0.3964 − 0.3964
5 0.5133 0.4855 0.0 0.0 1.0000 0.1805 0.1805 0.1805
6 0.5133 − 0.1618 0.4577 0.0 0.1805 1.0000 0.1805 0.1805
7 0.5133 − 0.1618 − 0.2289 0.3964 0.1805 0.1805 1.0000 0.1805
8 0.5133 − 0.1618 − 0.2289 − 0.3964 0.1805 0.1805 0.1805 1.000
TABLE 10-4 The Extended Hückel Hamiltonian Matrix for CH4 a

1 2 3 4 5 6 7 8

1 − 0.7144 0.0 0.0 0.0 − 0.5454 − 0.5454 − 0.5454 − 0.5454

2 0.0 − 0.3921 0.0 0.0 − 0.3790 0.1263 0.1263 0.1263
3 0.0 0.0 − 0.3921 0.0 0.0 − 0.3573 0.1787 0.1787
4 0.0 0.0 0.0 − 0.3921 0.0 0.0 − 0.3094 0.3094
5 − 0.5454 − 0.3790 0.0 0.0 − 0.5000 − 0.1579 − 0.1579 − 0.1579
6 − 0.5454 0.1263 − 0.3573 0.0 − 0.1579 − 0.5000 − 0.1579 − 0.1579
7 − 0.5454 0.1263 0.1787 − 0.3094 − 0.1579 − 0.1579 − 0.5000 − 0.1579
8 − 0.5454 0.1263 0.1787 0.3094 − 0.1579 − 0.1579 − 0.1579 − 0.5000
a All energies in a.u.
TABLE 10-5 Energies for Methane by the
Extended Hückel Method

MO no. Energy (a.u.) Occ. no.

8 1.1904 0
7 0.2068 0
6 0.2068 0
5 0.2068 0
4 − 0.5487 2
3 − 0.5487 2
2 − 0.5487 2
1 − 0.8519 2
Figure 10-2 A drawing of the lowest-energy nondegenerate EHMO for methane. The AOs are
drawn as though they do not overlap. This is done only to make the drawing simpler. Actually, the
AOs overlap strongly.
TABLE 10-7 Coefficients for MOs φ2 , φ3 , φ4

φ2 φ3 φ4

2s 0.0 0.0 0.0

2pz 0.5313 0.0 0.0
2px 0.0 0.5313 0.0
2py 0.0 0.0 0.5313
1sa 0.5547 0.0 0.0
1sb − 0.1849 0.5228 0.0
1sc − 0.1849 − 0.2614 0.4529
1sd − 0.1849 − 0.2614 − 0.4529
Figure 10-3 The three lowest-energy degenerate MOs of methane.
Figure 10-7 Extended Hückel energy difference between staggered and eclipsed ethanes as a
function of K .
TABLE 10-11 Energy Barriers for Internal Rotation about Single Bondsa

Barrier (kcal/mole)b

Molecule Calculated Experiment

CH3 –CH3 3.04 2.88

CH3 –NH2 1.66 1.98
CH3 –OH 0.45 1.07
CH3 –CH2 F 2.76 3.33
CH3 –CHF2 2.39 3.18
CH3 –CF3 2.17 3.25
CH3 –CH2 Cl 4.58 3.68
CH3 –CHCH2 1.20 1.99
cis –CH3 –CHCHCl 0.11 0.62
CH3 –CHO 0.32 1.16
CH3 –NCH2 0.44 1.97
a Calculated barriers are for rigid rotation, where no bond length or angle changes occur
except for the torsional angle change about the internal axis.
bThe stable form for the fi t seven molecules has the methyl C–H bonds staggered with
respect to bonds across the rotor axis. For the last four molecules, the stable form has
a C–H methyl bond eclipsing the double bond.
TABLE 10-8 Mulliken Net AO and Overlap Populations for Methane as Computed by the
Extended Hückel Method

2s 2pz 2px 2py 1sa 1sb 1sc 1sd

2s 0.6827 0.0 0.0 0.0 0.2229 0.2229 0.2229 0.2229

2pz 0.5645 0.0 0.0 0.5723 0.0636 0.0636 0.0636
2px 0.5645 0.0 0.0 0.5087 0.1272 0.1272
2py 0.5645 0.0 0.0 0.3815 0.3815
1sa 0.6844 − 0.0491 − 0.0491 − 0.0491
1sb 0.6844 − 0.0491 − 0.0491
1sc 0.6844 − 0.0491
1sd 0.6844
TABLE 10-10 Gross AO Populations, Gross Atomic
Populations, and Net Atomic Charges for Methane

Gross AO Gross atom Net atomic

population population charge

C2s 1.128
3.966 + 0.0334
C2p a 0.946
Ha 1.008 1.008 − 0.0083
a All 2p AOs and all H AOs have identical values because they are
equivalent through symmetry.