Session 2

Basis Sets

CCCE 2008

Session 2: Basis Sets

Two of the major methods
(ab initio and DFT) require
some understanding of basis
sets and basis functions
This session describes the
essentials of basis sets:
What they are
How they are constructed
How they are used
Significance in choice
CCCE 2008

Running a Calculation
In performing ab initio and DFT computational chemistry calculations, the chemist has
to make several decisions of input to the code:
The molecular geometry (and spin state)
The basis set used to determine the wavefunction
The properties to be calculated
The type(s) of calculations and any accompanying
assumptions
CCCE 2008

Running a calculation
For ab initio or DFT
calculations, many
programs require a basis set
choice to be made
The basis set is an approximate representation of the
atomic orbitals (AOs)
The program then calculates
molecular orbitals (MOs)
using the Linear Combination of Atomic Orbitals
(LCAO) approximation
4

CCCE 2008

Computational Chemistry Map

Chemist Decides:
Computer calculates:
Starting Molecular
Geometry
AOs
Basis Set (with
ab initio and DFT)
Type of Calculation
(Method and
Assumptions)
Properties to
be Calculated

AOs determine the

wavefunction ()

O
H

LCAO
O
H

MOs

CCCE 2008

Critical Choices
Choice of the method (and basis set) used is
critical
Which method?
Molecular Mechanics, Ab initio, Semiempirical, or
DFT

Which approximation?
MM2, MM3, HF, AM1, PM3, or B3LYP, etc.

Which basis set (if applicable)?

Minimal basis set
Split-valence
Polarized, Diffuse, High Angular Momentum, ......
CCCE 2008

Why is Basis Set Choice Critical?

The basis set needs to be able to approximate
the actual wave function sufficiently well to
give chemically meaningful results
Also needs a reasonable computational cost
Integrals should be evaluated quickly and
accurately

Trade-offs
Choice always involves a balance between
accuracy and computational cost
More accurate methods using larger basis sets will
take more computer time
7

CCCE 2008

Theoretical Models
Goal of computational chemistry is to
mathematically represent chemical reality
Improving the basis set and the degree of electron
correlation improves the ability of the
computational model to approach reality

Ultimate goal is an exact solution of the

Schrdinger equation

CCCE 2008

Comparison of Some Methods for Accuracy

DFT Location??
HF
B
A
S
I
S
S
E
T

Minimal
STO-3G
Split Valence
3-21G
Polarized
6-31G(d)
6-311G(d,p)

MP2

MP3

MP4 QCISD(T) ...

Full CI

Electron Correlation

Diffuse
6-311+G(d,p)
High
ang.

Momentum
6-311+G(2d,p)
6-311++G(3df,3dp)

........

HF
Limit

Schrdinger
Equation
CCCE 2008

Possible Basis Functions

1. Hydrogen-like Orbitals
Derived for a one-electron atom
Not truly accurate for a many-electron atom

Form:

( r , , ) = R( r )Yl m ( , )
R( r ) = radial function
Yl m = spherical harmonic

+ Advantages: Mutually orthogonal

Disadvantages: Complex form is awkward for
calculations; Most atoms of interest > one
electron
CCCE 2008

10

Possible Basis Functions

True wavefunction should be antisymmetric to
electron interchange use spin orbitals
Antisymmetric linear combination of products of
spin orbitals used in an SCF calculation
HF-SCF calculation
Numerical methods were originally used to solve and
find the Hartree-Fock orbitals

Roothaan: Represent the HF orbitals as linear

combinations of a set of known (basis) functions
Commonly used set of basis functions for atomic HF
calculations is the set of Slater-type orbitals (STOs)
CCCE 2008

11

Possible Basis Functions

2. Slater Type Orbitals (STOs)
Normalized form:

( 2 / a0 ) n + 0.5 ( n 1) ( r / a0 ) m
r
e
Yl ( , )
[( 2n )!]0.5
where n, m, and l are integers and (orbital
exponent) is a variational parameter

( r , , ) =

Improve results by using a linear combination of

several STOs to represent each HF orbital
HF-SCF atomic calculations require lots of
computation

Hartree did this numerically in the 1930s

CCCE 2008

12

Slater Type Orbitals

+ Advantages: Have a complete set
Radial behavior closely matches hydrogenic
orbitals

Disadvantages:
No nodes, as with H-like orbitals
Not mutually orthogonal
For larger molecules, computer evaluation of the
many integrals involved is quite time consuming
Need to reduce the computational cost
13

CCCE 2008

Possible Basis Functions

3. Gaussian Type Orbitals (GTOs)
Proposed by S.F. Boys in 1950
GTO form:
0.5
2( 2 n + 1.5)
( 2 n + 1)/ 4 ( n 1) ( r 2 ) m
g( r , , ) =
r
e
Yl ( , )

( 2n 1)!
+ Advantages: Have a complete set
Computer evaluation of integrals much faster
Closed integrals; Integrated GTO gives a GTO

Disadvantages: Not mutually orthogonal

Representation of e- probability is poor near and
far away from the nucleus
14

CCCE 2008

Comparison
STO vs. GTO
1.2
1
0.8
0.6
0.4
0.2
0

STO: e-r
GTO: e-r2

STO
GTO

0.5

1.5

2.5

3.5

4.5

Radius
CCCE 2008

15

GTOs
Linear combinations of GTOs are used to
approximate STOs (which are themselves
approximations)
A single GTO basis function has significant errors
when compared to a STO, especially near the
nucleus (See previous slide)
If several GTOs are combined in a linear
combination, the basis function is greatly
improved
See next slide
16

CCCE 2008

Comparisons
STO vs. GTOs

0.9

0.8

Wavefunction

0.7

0.6

STO

0.5

GTO

0.4

STO-3G

0.3

0.2

0.1

0
0

Radius (Angstroms)
17

CCCE 2008

Use of GTOs
Individual GTOs not used as basis functions:
Use a normalized linear combination of a few
GTOs (called primitives), each with different
values to give a contracted Gaussian function
gc =

c g
i

where gc is a contracted gaussian,

g p is a primitive gaussian, and ci is
a contraction coefficient
A linear combination of these primitives (typically
1-7) is used to approximate the STO
CCCE 2008

18

Use of GTOs
Using contracted GTOs instead of primitive
GTOs as the basis set has advantages:
Number of variational coefficients to be
determined is reduced, which saves a lot of
computational time
Accuracy is NOT reduced significantly, as long as
the contraction coefficients (cis) are well chosen

Increasing the number of primitive GTOs used in

each contracted Gaussian improves the accuracy
Different types of basis sets use different numbers
and types of GTOs
19

CCCE 2008

Minimal Basis Sets

Jargon Used STO-NG (Single )
N is the number of primitive GTOs used
Example: STO-3G
Three primitive GTOs used per AO
Popular starting point for calculations
STO-3G basis functions have been developed for
most of the elements in the Periodic table

Minimal basis sets do not adequately describe

non-spherical (anisotropic) electron distribution in
molecules (as in polar covalent bonds)
20

CCCE 2008

Minimal Basis Sets

GTO representation of a STO for 1s AO:
s
1STO
( r ) = 0.5 1.5e( r )
3

where = 1
3

2
2
2
2 4
2 4
2 4
1sSTO-3G ( r ) = c1 1 e( 1r ) + c2 2 e( 2r ) + c3 3 e( 3r )

where c1 = 0.444615, c2 = 0.535336, c3 = 0154340

.
and 1 = 0109818
.
, 2 = 0.405771, 3 = 0.2.22766

c values are called the contraction coefficients

The exponents are the alpha () values
CCCE 2008

21

Split Valence Basis Sets

Jargon Used K-LMG (Double )
Differentiate between core and valence electrons
Developed to overcome problems of inadequate
description of anisotropic electron distributions
using minimal basis sets (Size is adjusted)

K = number of sp-type inner shell primitive GTOs

L = number of inner valence s- and p-type primitive
GTOs
M = number of outer valence s- and p-type
primitive GTOs
22

CCCE 2008

Split Valence Basis Sets

Each split valence atomic orbital is composed of a
variable proportion of two (or more) functions of
different size or radial extent
For a larger e- cloud
(longer bond)
a<b
For a smaller e- cloud
(shorter bond)
a>b

a and b are normalized and sum to 1

CCCE 2008

23

Split Valence Basis Sets

Examples:
3-21G (Used as the semiempirical basis set)
Three primitives for the inner shell (STO-3G); each
valence orbital is constructed with two sizes of
basis function (Two GTOs for contracted valence
orbitals; One GTO for extended valence orbitals)
6-311G
STO-6G for inner shell; Three sizes of basis
function for each valence orbital (Three GTOs for
contracted valence orbitals, and two different sizes
of GTO for extended valence orbitals)
CCCE 2008

24

Polarized Basis Sets

Jargon Used 6-31G(d) or 6-31G* (older)
Also have 6-31G(d,p) or 6-31G**
(d) or * type
d-type functions added to atoms with Z > 2
f-type functions added to transition metals

(d,p) or ** type
p-type functions added to H atoms
d-type functions added to atoms with Z > 2
f-type functions added to transition metals

6-31G(d) is another popular basis set choice

CCCE 2008

25

Polarized Basis Sets

In molecule formation, AOs become distorted
in shape (polarization)
Orbitals are influenced by other nuclei
Polarization accounts for these influences which
distort the orbital shape
+

CCCE 2008

26

Diffuse Basis Sets

Jargon Used 6-31+G(d) or 6-31++G(d)
6-31+G(d)
6-31G(d) basis set with an additional larger pfunction for atoms with Z > 2
6-31++G(d)
6-31+G(d) basis set with an additional larger sfunction for H atoms

Diffuse basis sets are useful for describing

anions, molecules with lone pairs, excited
states, and transition states (loosely held e-)
CCCE 2008

27

Basis Set Progression

Increasing number of GTOs Used:
Minimal < Split Valence < Polarized < Diffuse

Get an increasingly good approximation to the actual

wave function
The number of integrals increases as N4 where N
is the number of basis functions
During the minimization process, the orbital
exponents are adjusted to define a new basis set to
start another iteration

Computational cost has to be considered

28

CCCE 2008

Comparison of Some Methods for Accuracy

HF
B
A
S
I
S
S
E
T

Minimal
STO-3G
Split Valence
3-21G
Polarized
6-31G(d) ~ DZVP
6-311G(d,p) ~ TZVP

DFT Location??

MP2

MP3

MP4 QCISD(T) ...

Full CI

Electron Correlation

Diffuse
6-311+G(d,p)
High ang.
Momentum
6-311+G(2d,p)

6-311++G(3df,3dp)
........

Schrdinger
Equation

HF
Limit

29

CCCE 2008

Basis Set Choice and Expense

axial-methylcyclohexane on SGI Indigo2
(Spartan cpu time in sec.)
Method/Basis Set
AM1/STO-3G
HF/STO-3G
HF/ 3-21G(d)
HF/ 6-31G(d,p)

s.p.
~1
72
193
2632

opt.
10
983
2214
34655 (9.6 h)

As larger basis sets are used, the energy decreases

(Variational Principle)
CCCE 2008

30

Common Basis Sets

Brief Description of Standard Basis Sets
Basis Set
Description
STO-3G

Minimal basis; qualitative results - large systems

3-21G

Double ; more accurate results on large systems

6-31G(d)

Moderate set; Common use for medium systems

6-31G(d,p)

Used where H is site of interest; More accurate

6-31+G(d)

Used with anions, excited states, lone pairs, etc.

6-31+G(d,p) Used with anions, etc., where H is site of interest

6-311++G(d,p) Good for final, accurate energies, but expensive

Many other sets are in use. Existing sets can be modified

CCCE 2008

31

Optimization of a Basis Set

Variational Principle: Energy values are
bounded from below
The lower the calculated energy, the better
Procedure:
Vary the constants and exponents that describe the
Gaussian functions sequentially until the lowest
energy is obtained
Such a basis set may only apply to that
individual molecule, however
CCCE 2008

32

Lab: Gaussian Orbitals

Question: How are Gaussian orbitals used to
approximate a Slater Type Orbital?
Importance: Basis sets are approximations
based on mathematical use of two or more
Gaussian functions
Goal: Visualize an STO-3G basis set what does the resultant function look like?
Computational Tool: Spreadsheet
Refer to Basis Set Case Study handout for
detailed instructions
CCCE 2008

33