Summer School

„Molecular Methods in Energy Research“

Essen
2013

Lecture 3:
Introduction to the ORCA Program System
Frank Wennmohs

Max Planck Institute for

Chemical Energy Conversion
Stifstr. 34-36
D-45470 Mülheim an der Ruhr
Germany
 The Computational Environment

In order to run calculations with ORCA, three things are necessary:

1.Installation of the ORCA program

2.Using a text-editor to specify the calculation details, i. e. the input file

3.Running the ORCA program (in a cluster environment possibly controlled

by a batch system)

NOTE: ORCA is available for all popular platforms:

★ Windows,
★ MacOS,
★ Linux
The Computational Environment

★ ORCA is available for the popular Windows, Mac OS X, and Linux platforms.

★ ORCA is distributed as an archive for all platforms. In this archive resides a directory
with all executables. There are plenty of programs for extracting files from archives on
all platforms (gzip, tar, zip/unzip,...).

★ On the follwing slides it will be shown how to run a calculation on Windows Vista/7,
Mac OS X, and Linux.

★ It is assumed, that the orca executables reside in a directory ‘orca’

Configuring ORCA under Windows Vista / 7

Setting the PATH variable to the ORCA directory (with administrator rights)

The Path Variable

can be accessed via

Control Panel >

System >
Advanced system
Settings

Here it is assumed
that ORCA resides
in c:\orca
Configuring ORCA under Windows Vista / 7

Setting the PATH variable to the ORCA directory (as normal user)

The Path Variable

can be accessed via

Control Panel >

User Accounts
and Family Safety
> User Accounts >
Change my
environment
variables

Here it is assumed
that ORCA resides
in c:\orca
Editing Textfiles under Windows Vista / 7

Editing Text using Notepad

! RHF SVP Opt The Notepad

* xyz 0 1 application can be
C 0 0 0
O 0 0 1.13 found under
* Accessories

Of course all other

Text-editors can be
used on Windows.

(But not Word!)

Textfiles have to be
plain ASCII !
Running ORCA under Windows Vista / 7

Opening a command window:

Enter

‘cmd’

into the Search line,

then cmd.exe will be executed

upon pressing RETURN.
Running ORCA under Windows Vista / 7

Executing the ORCA program

Once ORCA is in
the PATH, you can
run jobs from any
directory you want

Using the sequence

orca myinput.inp
> myinput.out’
the output will be
put in a file named
like that.

The TaskManager
is a handy tool to
check if the
calculation is still
running.
Getting Started with ORCA under Mac OS X
TextEdit & Terminal

export PATH=%HOME/orca:$PATH

sagnix% orca myinput.inp >& myinput.out &

Configuring ORCA under Mac OS X
TextEdit & Terminal

export PATH=%HOME/orca:$PATH ORCA path entered in

the file

‘.bashrc’

using TextEdit
Running ORCA under Mac OS X

sagnix% orca simple.inp >& simple.out &

Configuring ORCA under Linux
Linux (Ubuntu 9.10): Setting the PATH variable

Setting the
PATH variable
is done in the
file

‘.bashrc’

You can use

frankw@MyUbuntu:~$ tail -5 .bashrc the
export PATH=$HOME/orca:$PATH
‘Text Editor’

to add the line

shown.
Running ORCA under Linux
Linux (Ubuntu 9.10): Text Editor (gedit) & ORCA execution

Especially
under Linux
there is an
uncountable
frankw@MyUbuntu:~/simple$ orca simple.inp >& simple.out & number of
frankw@MyUbuntu:~/simple$ tail -f simple.out text-editors.

Very popular
ones are
Vim, Emacs,
NEdit, Kate,
etc.
 The ORCA Project

Hartree-Fock Density Functional

Semiempirical
LDA, GGA, Hybrid Functionals
RHF,UHF,ROHF,CASSCF INDO/S,MNDO,AM1,PM3,NDDO/1 Double hybrid functionals,
Direct, Semidirect, Conventional,
RI-Approx., Newton-Raphson
RI-Approx., Newton-Raphson
RKS,UKS,ROKS

Electron Correlation Join 10000 users Excited States

MP2/RI-MP2
FREE Download
TD-DFT/CIS+gradients
CCSD(T),QCISD(T),CEPA,CPF http:// MR-CI/DDCI/SORCI
(all with and without RI, Local)
MR-MP2, MR-MP3, MR-MP4(SD)
www.cec.mpg.de/
MR-CI, MR-ACPF, MR-AQCC forum/portal.php
Relativistic Methods Molecular Properties
Analytical Gradients(HF,DFT,MP2) + Geometries + Trans. States
1st-5th Order Douglas-Kroll-Hess Polarizabilities, Magnetizabilities (Coupled-Perturbed HF/KS)
Zero‘th Order Regular Approximation (ZORA) COSMO Solvation Model Throughout
Infinite Order Regular Approximation (IORA) IR, Raman and Resonance Spectra (Numerical Frequencies)
Picture Change Effects, All electron basis sets, EPR-Parameters (g,A,D,J,Q)
(Effective core potentials) Mössbauer-Parameters (δ,ΔEQ)
ABS,CD,MCD Spectra
Population Analysis, NBOs, Localization, Multipole Moments,...
 The ORCA Project
Just Released - ORCA 3.0.0 Density Functional
Hartree-Fock Semiempirical
‣ DLPNO Method: Chemical Accuracy for Large Molecules
‣ Explicitly Correlated Wavefunctions: F12 (DKH, COSMO, ZORA, IORA)
RHF,UHF,ROHF,CASSCF INDO/S,MNDO,AM1,PM3,NDDO/1
LDA, GGA, Hybrid Functionals
Double hybrid functionals,
‣Direct,
Analytical Hessian
Semidirect, Conventional,
RI-Approx., Newton-Raphson
‣ RI-Approx.,
ROCIS: Newton-Raphson
L-Edges, Open-Shell Excitations RKS,UKS,ROKS
‣ Highly Increased Performance (New LibInt 2.0 Integrals, Integral-Handling)
‣ Completely
Electron Correlation
Parallelized Join 10000 users Excited States
‣ DMRG Interface (Linux only) FREE Download
MP2/RI-MP2
‣ NBO 6.0 Interface
CCSD(T),QCISD(T),CEPA,CPF http:// TD-DFT/CIS+gradients
(all ‣withHF-3c
MR-CI/DDCI/SORCI
and without RI, Local)
‣ Long-range
MR-MP2,
www.cec.mpg.de/
separated DFT & TDDFT
MR-MP3, MR-MP4(SD)
‣ New
MR-CI, Coupled
MR-ACPF, Cluster Methods forum/portal.php
MR-AQCC
(OO, RelaxedMethods
Relativistic Densities, Brückner Orbitals, EOM) Molecular Properties
‣ VdW DFT-SCNL Analytical Gradients(HF,DFT,MP2) + Geometries + Trans. States
‣1st-5
Modern DFT-Functionals
th Order Douglas-Kroll-Hess Polarizabilities, Magnetizabilities (Coupled-Perturbed HF/KS)
Zero‘th Order RegularSOC-CAS
Approximation (ZORA) COSMO Solvation Model Throughout
‣ Variational
Infinite Order Regular Approximation (IORA) IR, Raman and Resonance Spectra (Numerical Frequencies)
Picture‣Change Effects, All electron basis sets,
Improved RIJ-COSX, numerical overlap fitting EPR-Parameters (g,A,D,J,Q)
‣ Geometrical
(Effective coreCounterpoise
potentials) Correction Mössbauer-Parameters (δ,ΔEQ)
‣ New Basis Sets: MINIX, MINIS, Rappaport, SARC completeABS,CD,MCD Spectra
Population Analysis, NBOs, Localization, Multipole Moments,...
Philosophy of the ORCA Project
General goal: Create a powerful tool to allow connection between theory and
experiment. → Observables! (Spectra, Structures, Energies)
Design principles:
1. The program should be as flexible as possible
2. The program should be as efficient as possible (parallel, efficient
algorithms, BLAS libraries)
3. The program should be as comprehensive as possible
4. The program should be as user friendly as possible
5. The program should be easily extendable (highly modular)
6. The source code should be as clean and well structured as possible (C
++ rather than Fortran).
7. The program should be platform independent
Tasks to be Performed
• Calculation of single point energies
– Hartree-Fock calculations: RHF, UHF and ROHF
– Density functional calculations
• Optimization of molecular structures
– Equilibrium geometries
– Transition states and reaction rates
• Calculation of vibrational frequencies
– Characterization of stationary points
– Thermodynamic properties
– Vibrational spectra
• Calculation of ground state properties
– Looking at charge distributions and orbitals
– IR+Raman spectra
– NMR spectra
– EPR spectra and exchange couplings
– Mössbauer spectra
• Calculation of excited states and their properties
– DFT calculation of absorption and CD spectra
– Ab initio calculation of absorption and CD spectra
– Advanced: Resonance-Raman, X-Ray absorption, forbidden transitions, MCD, Excited state
geometry optimizations, ...
Common Jobs in Computational Chemistry
Common Jobs in Computational Chemistry

• Energy Calculations
Common Jobs in Computational Chemistry

• Energy Calculations

• Geometry Optimizations
Common Jobs in Computational Chemistry

• Energy Calculations

• Geometry Optimizations

• Frequency Calculations
Common Jobs in Computational Chemistry

• Energy Calculations

• Geometry Optimizations

• Frequency Calculations

• Transition States
Common Jobs in Computational Chemistry

• Energy Calculations

90%
• Geometry Optimizations

Of •all Calculations!
Frequency Calculations

• Transition States
First Stop: Energy Calculations
A First ORCA Job

Request tight
DFT Functional The basis set convergence
A comment line

# A simple single point DFT calculation

Start a keyword line ! RKS B3LYP SV(P) TightSCF

* xyz 0 1
Restricted (closed-shell) C 0 0 0
Calculation O 0 0 1.128
*

Start of coordinate definition

Coordinates are defined as Total Charge is zero and Multiplicity
cartesian coordinates (2*S+1) is one (no unpaired
electrons)
End of coordinate definition
One atom each line:
atomic symbol x y z coordinates in Angström
 Output of Single Points
We first echo the input file and some references to the basis sets used. Next the SCF program is taking over and commences with giving
Then you get information on the job-type, the input coordinates in all details about the SCF settings
various formats and the basis set ------------
**************************** SCF SETTINGS
* Single Point Calculation * ------------
**************************** Hamiltonian:
Density Functional Method .... DFT(GTOs)
--------------------------------- Exchange Functional Exchange .... B88
CARTESIAN COORDINATES (ANGSTROEM) X-Alpha parameter XAlpha .... 0.666667
--------------------------------- Becke's b parameter XBeta .... 0.004200
C 0.000000 0.000000 0.000000 Correlation Functional Correlation .... LYP
O 0.000000 0.000000 1.130000 LDA part of GGA corr. LDAOpt .... VWN-5
Gradients option PostSCFGGA .... off
--------------------- Hybrid DFT is turned on
BASIS SET INFORMATION Fraction HF Exchange ScalHFX .... 0.200000
--------------------- Scaling of DF-GGA-X ScalDFX .... 0.720000
There are 2 groups of distinct atoms Scaling of DF-GGA-C ScalDFC .... 0.810000

Group 1 Type C : 7s4p1d contracted to 3s2p1d pattern {511/31/1}

Group 2 Type O : 7s4p1d contracted to 3s2p1d pattern {511/31/1} General Settings:
Integral files IntName .... JOB-01
Hartree-Fock type HFTyp .... RHF
Total Charge Charge .... 0
Multiplicity Mult .... 1
Number of Electrons NEL .... 14
Basis Dimension Dim .... 28
Next the one-electron integrals are calculated (and perhaps also the two- Nuclear Repulsion ENuc .... 22.4778902655 Eh
electron integrals if conv if requested for “conventional SCF”)
------------------------------------------------------------------------
ORCA GTO INTEGRAL CALCULATION Convergence Tolerance:
------------------------------------------------------------------------ Energy Change TolE .... 1.000e-008 Eh
Max Density Change TolMaxP .... 1.000e-007
BASIS SET STATISTICS AND STARTUP INFO RMS Density Change TolRMSP .... 1.000e-008
DIIS Error TolErr .... 1.000e-007
# of primitive gaussian shells ... 24
# of primitive gaussian functions ... 48
# of contracted shell ... 12 Diagonalization of the overlap matrix:
# of contracted basis functions ... 28 Smallest eigenvalue ... 1.911e-002
Highest angular momentum ... 2 Time for diagonalization ... 0.010 sec
Maximum contraction depth ... 5 Time for construction of square roots ... 0.030 sec
Integral threshhold Thresh ... 1.000e-010 Total time needed ... 0.040 sec
Primitive cut-off TCut ... 3.000e-012
The integration grid is produced and the initial guess performed The total energy and its components are printed
------------------- ----------------
DFT GRID GENERATION TOTAL SCF ENERGY
------------------- ----------------
General Integration Accuracy IntAcc ... 4.010 Total Energy : -113.17273851 Eh -3079.50944 eV
Total number of grid points ... 5057 Components:
Nuclear Repulsion : 22.47789027 Eh 611.63913 eV
------------------------------ Electronic Energy : -135.65062877 Eh -3691.14856 eV
INITIAL GUESS: MODEL POTENTIAL
------------------------------ One Electron Energy: -198.09397475 Eh -5390.27572 eV
Loading Hartree-Fock densities ... done Two Electron Energy: 62.44334598 Eh 1699.12715 eV
....etc
Now organizing SCF variables ... done Virial components:
------------------ Potential Energy : -225.62136955 Eh -6139.31540 eV
INITIAL GUESS DONE Kinetic Energy : 112.44863104 Eh 3059.80596 eV
------------------ Virial Ratio : 2.00643945

Now we are ready to start the SCF iterations

--------------
SCF ITERATIONS
Next you get the orbital energies printed:
----------------
--------------
ORBITAL ENERGIES
*** Starting incremental Fock matrix formation ***
----------------
***Turning on DIIS***
NO OCC E(Eh) E(eV)
----------------------------
0 2.0000 -19.243280 -523.6231
! ITERATION 0 !
1 2.0000 -10.301611 -280.3141
----------------------------
2 2.0000 -1.152619 -31.3636
Total Energy : -112.951951547431 Eh
3 2.0000 -0.568518 -15.4698
Energy Change : -112.951951547431 Eh
4 2.0000 -0.476766 -12.9731
MAX-DP : 0.674569966353
5 2.0000 -0.476766 -12.9731
RMS-DP : 0.053899776162
6 2.0000 -0.373742 -10.1698
Actual Damping : 0.0000
7 0.0000 -0.026298 -0.7156
Int. Num. El. : 14.00002071 (UP= 7.00001035
8 0.0000 -0.026298 -0.7156
Exchange : -10.98287769
9 0.0000 0.221613 6.0303
Correlation : -0.58429246
10 0.0000 0.400346 10.8937
DIIS-Error : 0.450135783168
11 0.0000 0.457384 12.4458
Which hopefully eventually converge and: 12 0.0000 0.457384 12.4458
...
*****************************************************  
* SUCCESS *
* SCF CONVERGED AFTER 11 CYCLES *
*****************************************************
Now comes the “soft science” (e.g. population The Löwdin analysis contains a detailed breakdown of the MOs in terms of
analysis of the SCF density matrix angular momentum components of each atom. This helps a lot when you
select the orbitals for plotting.
*******************************
* LOEWDIN POPULATION ANALYSIS *
******************************** *******************************
* MULLIKEN POPULATION ANALYSIS *
******************************** ----------------------
LOEWDIN ATOMIC CHARGES
----------------------- ----------------------
MULLIKEN ATOMIC CHARGES 0 C : -0.050462
----------------------- 1 O : 0.050462
0 C : 0.016669
1 O : -0.016669 etc.
Sum of atomic charges: -0.0000000 ------------------------------------------
LOEWDIN REDUCED ORBITAL POPULATIONS PER MO
-------------------------------- -------------------------------------------
MULLIKEN REDUCED ORBITAL CHARGES THRESHOLD FOR PRINTING IS 0.1%
-------------------------------- 0 1 2 3 4 5
0 C s : 3.834569 s : 3.834569 ORB-EN -19.24328 -10.30161 -1.15262 -0.56852 -0.47677 -0.47677
pz : 1.009723 p : 2.073008 OCC 2.00000 2.00000 2.00000 2.00000 2.00000 2.00000
-------- -------- -------- -------- -------- --------
px : 0.531642 0 C s 0.0 99.5 18.6 11.5 0.0 0.0
py : 0.531642 0 C pz 0.1 0.0 16.7 4.8 0.0 0.0
dz2 : 0.027088 d : 0.075754 0 C px 0.0 0.0 0.0 0.0 6.9 20.9
dxz : 0.024333 0 C py 0.0 0.0 0.0 0.0 20.9 6.9
dyz : 0.024333 0 C dz2 0.1 0.0 2.4 0.0 0.0 0.0
0 C dxz 0.0 0.0 0.0 0.0 0.5 1.5
dx2y2 : 0.000000 0 C dyz 0.0 0.0 0.0 0.0 1.5 0.5
dxy : 0.000000 1 O s 99.8 0.1 54.7 20.8 0.0 0.0
1 O s : 3.737862 s : 3.737862 1 O pz 0.0 0.4 7.3 62.3 0.0 0.0
pz : 1.381818 p : 4.256855 1 O px 0.0 0.0 0.0 0.0 17.4 52.4
px : 1.437518 1 O py 0.0 0.0 0.0 0.0 52.4 17.4
1 O dz2 0.0 0.1 0.3 0.6 0.0 0.0
py : 1.437518 1 O dxz 0.0 0.0 0.0 0.0 0.1 0.3
dz2 : 0.008940 d : 0.021953 1 O dyz 0.0 0.0 0.0 0.0 0.3 0.1
dxz : 0.006506
dyz : 0.006506 6 7 8 9 10 11
dx2y2 : 0.000000 -0.37374 -0.02630 -0.02630 0.22161 0.40035 0.45738
2.00000 0.00000 0.00000 0.00000 0.00000 0.00000
dxy : 0.000000 -------- -------- -------- -------- -------- --------
0 C s 49.2 0.0 0.0 33.5 60.9 0.0
------------------------ 0 C pz 39.3 0.0 0.0 46.5 35.3 0.0
MULLIKEN OVERLAP CHARGES 0 C px 0.0 15.2 54.7 0.0 0.0 15.5
------------------------ 0 C py 0.0 54.7 15.2 0.0 0.0 81.4
0 C dz2 0.8 0.0 0.0 7.8 1.2 0.0
B( 0-C , 1-O ) : 1.3340 0 C dxz 0.0 0.4 1.5 0.0 0.0 0.1
0 C dyz 0.0 1.5 0.4 0.0 0.0 0.3
1 O s 2.0 0.0 0.0 12.0 0.6 0.0
1 O pz 8.8 0.0 0.0 0.2 1.0 0.0
1 O px 0.0 5.9 21.2 0.0 0.0 0.3
1 O py 0.0 21.2 5.9 0.0 0.0 1.5
1 O dz2 0.0 0.0 0.0 0.0 0.9 0.0
1 O dxz 0.0 0.2 0.9 0.0 0.0 0.1
1 O dyz 0.0 0.9 0.2 0.0 0.0 0.7
Looking at Orbitals
There is a utility program orca_plot which lets you generate graphics information. The
information about the MOs, the geometry and the basis set is stored in the so called
gbw-File (‚geometry-basis-wavefunction‘).

To generate the plot information interactively use:

orca_plot myjob.gw -i

You will get a „stone-age“ menu which you can use to generate the necessary files.
1. Press 5 ENTER to choose the output formation. (press 7 ENTER for
gaussian cube, the preferred format)
2. Press 4 ENTER to choose the number of grid intervals. Something like 40
will be o.k. For high resolution on larger molecules choose 65-75.
3. To plot an orbital from a closed shell calculation press 3 ENTER and
choose 0 ENTER. For spin-up from UHF/UKS the same. For spin down from
UHF/UKS choose 1 ENTER.
4. Press 2 ENTER and enter the number of the MO that you want to plot.
NOTE THAT COUNTING STARTS WITH 0!!!
5. Press 10 ENTER to generate the output file.
6. Repeat steps 4 and 5 until all desired files are produced
The orca_plot program automatically produces an ‚xyz‘ file which contains the molecular
coordinates.
We presently like the Chimera program for vizualization. There are many free alternatives such
as Molekel, gOpenMol, Molden,... any program that reads .xyz and .cube files

With Chimera, do the following:

1. Start Chimera
2. Choose „File Open“ and navigate to open the .xyz file
of interest HOMO
3. Choose „Presets → Publication 1“ (MO#6)
4. Choose „Actions → Atoms&Bonds → ball & stick“ of CO
5. Choose „Actions → Color → By element“
6. Choose „Tools → Volume Data → Volume Viewer“
7. Choose „File → Open Map“ and Gaussian Cube
format and open the cube file of interest
8. Enter in „Level“ 0.03 and press ENTER. Then click on
Color and choose red (or whatever you like) LUMO
9. Hold the Ctrl-key and click on the negative part of the (MO#7)
contours. Enter in „Level“ -0.03 and press ENTER, then of CO
go to color and choose yellow (or whatever you like)
10. Orient the contour in the way you like and go to „File →
Save Image“. Choose PovRay „true“ and go ahead.
What You Need for an Energy Calculation

Energy calculations make up the simplest

computational task available

To actually start a calculation three things have to be determined before:

What You Need for an Energy Calculation

Energy calculations make up the simplest

computational task available

To actually start a calculation three things have to be determined before:

• Method
What You Need for an Energy Calculation

Energy calculations make up the simplest

computational task available

To actually start a calculation three things have to be determined before:

• Method
• Basis set
What You Need for an Energy Calculation

Energy calculations make up the simplest

computational task available

To actually start a calculation three things have to be determined before:

• Method
• Basis set
• Coordinates
Method: Ab initio wavefunction based or DFT?
Accuracy versus Computing Time Tradeoff

Fast, but not accurate enough.

Hartree Fock
Usually only as a basis for higher methods
Fast, usually good accuracy.
DFT But is the accuracy also good in my case?
Very dependent on the Functional

Very high accuracy,

Correlated, with Coupled Pair Theories up to Chemical Accuracy.
wavefunction based BUT: Calculation scales formally at least
methods with the size of the basis set to N5
Method: Ab initio wavefunction based or DFT?
Roughly there are three main branches available:

Method Keyword

Hartree Fock RHF, UHF, ROHF

But
DFT
I just want CCSD(T),
RKS, UKS, ROKS
because it’s the best, MP2

Correlated,
wavefunction based
right? Coupled Pair Theories
(Single & Double Excitations, CEPA, (N)CPF, CCSD)
methods Coupled Pair Theories + Linearized Triple Excitations
( CCSD(T) )
The Functional Zoo
Name Type Comments

BLYP GGA One of the earliest GGA functionals. Usually inferior to BP86 and PBE. Predicts
too long bonds.
BP86 GGA Excellent geometries and vibrational frequencies. Energetics is usually not highly
accurate but performs often well in spectroscopic investigations.
PW91 GGA One of the older GGA functionals with excellent accuracy for exchange
couplings.
PBE GGA A GGA version designed to replace PW91. Very popular in physics. Often similar
to BP86.
OLYP GGA Violates the uniform electron gas limit but gives improved results for molecules

B3LYP Hybrid De facto standard in chemistry for structures, energies and properties. See
discussion in the text.
PBE0 Hybrid Excellent accuracy; competitive with B3LYP
TPSS Meta-GGA Improvement over PBE. Includes the kinetic energy density and obeys more
constraints known from rigorous theory.
TPSSh Hybrid meta- Probably improvement over PBE0; perhaps increase fraction of HF to 25%
GGA (TPSS0)
B2PLYP Double hybrid First (and prototypical) member of the double hybrid class of functionals. So far
been proven excellent for energies and geometries. More exploration needed.
 Efficient DFT: The RI Approximation
As long as there is NO Hartree-Fock exchange present (no hybrid functionals), a very
efficient approximation can be used to speed up (factor 10-100) DFT calculations: the RI
approximation (also called density fitting method)
In this method the electron density is fit to an auxiliary basis set which must be provided
by the user. The effect of the approximation on structures and frequencies is barely
visible. Absolute energies are affected to a few kcal/mol, relative energies much less.
! RKS BP86 RI SV(P) SV/J TightSCF Opt

* int 0 1
C 0 0 0 0 0 0 Auxiliary basis appropriate for SV(P) (equivalently
O 1 0 0 1.15 0 0 Def-2 uses SV(P) and SV/J together; Def-3 is
H 1 2 0 1.00 115 0 equivalent to TZVP and TZV/J)
H 1 2 3 1.00 115 180
*

The job leads to an energy of -114.37494125 Eh Use the RI approximation

Without RI and SV/J -114.37466595 Eh

Error 0.00028 Eh = 0.17 kcal/mol

TIP: For geometry and frequency calculations the BP86 and PBE functionals together with the RI approximation is recommended. Basis sets of TZVP quality
are appropriate for good accuracy! SV(P) is already good enough for a first orientation. Use keywords QuickOpt, NormalOpt or GoodOpt!
For energy calculations I recommend the B3LPY or PBE0 functionals and larger basis sets (TZVPP if possible or even aug-TZVPP). Use Keyword DFTEnergy
Efficient HF and Hybrid DFT: RIJCOSX
Unfortunately, RI does not smoothly carry over to Hartree-Fock and hybrid DFT
calculations. One attempt to do so is the RI-JK approximation that needs to be invoked
together with ,JK‘ fitting bases. RI-JK cannot be used for optimizations

! B3LYP RI-JK def2-SVP def2-SVP/JK TightSCF

A more efficient approximation that leads to large speedups is the RIJCOSX

approximation. it uses RI-J for the Coulomb part and a special approximation ,COSX‘
for the exchange part. It is available throughout the program and leads to large
speedups at very little loss in accuracy

! B3LYP RIJCOSX def2-SVP def2-SVP/J TightSCF

If combined with RI-MP2 you need to give two auxiliary basis sets for optimal speed
and accuracy:
! RI-MP2 RIJCOSX def2-SVP def2-SVP/J def2-SVP/C TightSCF

Neese, F.; Wennmohs, F.; Hansen, A.; Becker, U. (2009) Chem. Phys., 356, 98–109
The Crux of DFT: Which Functional to Choose?
Always first thing to do:

RI BP86

The BP86 functional in combination with the Resolution of Identity approximation (RI) is
the fastest way to calculate by far, with reasonable results with respect to geometries,
etc.

For a more quantitative result, for example, the calculation of energy differences, the
following hybrid functionals are known to provide good results:

B3LYP RIJCOSX
( PBE0 RIJCOSX )
To keep in mind:
Usually the results are good, but this does not rule out the possibility, that it might produce
wrong results in other cases!
Modern DFT: Double Hybrid Functionals
★ A Special feature of ORCA are ,double hybrid functionals‘ proposed in 2006 by
Stefan Grimme. Here part of the DFT correlation is replaced by a MP2 like term and
part of the DFT exchange by Hartree-Fock exchange.
★ For main group thermodynamics and kinetics it is more accurate than standard
functionals
★ ORCA is the only program to have analytic gradients and excited states for these
functionals
★ The prototypical member is „B2PLYP“

! RI-B2PLYP def2-SVP def2-SVP/C TightSCF Opt

★ It can be combined with RIJCOSX

! RI-B2PLYP RIJCOSX def2-SVP def2-SVP/J def2-SVP/C TightSCF Opt

★ And together with empirical van der Waals corrections (also proposed by Stefan
Grimme) it is particularly accurate:

! RI-B2PLYP D3BJ RIJCOSX def2-SVP def2-SVP/J def2-SVP/C

TightSCF Opt
Choice of Basis Set

6-31G
6-31G*
6-31G**
6-31++G**
6-3111++G** Lots of basis sets to
DZP
choose from
TZP

Ahlrichs SVP Which one is the best for

Ahlrichs TZVP
one’s purpose?
cc-pVDZ
cc-pVTZ
aug-cc-pVTZ
Choice of Basis Set
polarization functions on heavy atoms
Pople basis sets: (most popular)

Ncore – valence – diffuse G(polar. heavy, polar. hydrogen)

+ for additional
# of primitives diffuse orbitals on
in core orbitals polarization functions on
heavy atoms hydrogen atoms
contraction pattern ++ for additional
for valence orbitals diffuse orbitals on hydrogen
example: 6-31++G(d,p): atoms ( anions!)

Core orbitals are represented by a contraction of six primitive gaussian

functions. The valence shell is double- ζ with 3 primitives in first contraction,
one in second.
There is a set of diffuse orbitals (p,d) on heavy atoms as well as on
hydrogen atoms (diffuse s-orbitals).There is one additional set of d-functions
on heavy atoms and one additional set of p-polarization functions on
hydrogen atoms.
note: Core orbitals have the lowest energy. Thus, an accurate description is necessary
(otherwise the variational principle makes the valence orbitals “core like“).
 Deep contraction for core orbitals, but only one function per AO
(because the core-radial function does not change much in the molecule)!
Choice of Basis Set
These Pople style basis sets are very
6-31G popular in the English-speaking
6-31G* community. They are also very popular
6-31G** due to their availability in the Gxx series
6-31++G** of programs.
6-3111++G**

DZP
TZP
Better:
Ahlrichs SVP The Ahlrichs basis sets provide better results at the
Ahlrichs TZVP same computational cost.
cc-pVDZ
cc-pVTZ
aug-cc-pVTZ Tip:
Always start calculations at the SVP level,
especially geometry optimizations. Only
when precise energies are requested, switch
to TZVP (or even: def2-TZVPP)
Generation of Coordinates

In principle:
Only a texteditor is needed to enter a molecule’s coordinates

Small molecules can be entered by specification of

internal coordinates and their values. This is usually
called a

Z-Matrix

But this is only feasible for molecules of a few (<5)

atoms.

Today
Molecules are created with full specification in an xyz space
using common programs.
Generation of Coordinates

Today
Molecules are created with full specification in an xyz space
using common programs.

Building molecules
using PyMol
Generation of Coordinates
Today
Molecules are created with full specification in an xyz space
using common programs.
(Molden, PyMol, GaussView, etc. )

There are two ways to get coordinates into ORCA, once those
have been created and exported into an .xyz file:
Generation of Coordinates
Today
Molecules are created with full specification in an xyz space
using common programs.
(Molden, PyMol, GaussView, etc. )

There are two ways to get coordinates into ORCA, once those
have been created and exported into an .xyz file:

1. Copy the contents of the .xzy file into the ORCA input file, using copy/
paste facilities of the texteditor or operating system
Generation of Coordinates
Today
Molecules are created with full specification in an xyz space
using common programs.
(Molden, PyMol, GaussView, etc. )

There are two ways to get coordinates into ORCA, once those
have been created and exported into an .xyz file:

1. Copy the contents of the .xzy file into the ORCA input file, using copy/
paste facilities of the texteditor or operating system

2. Read in the .xyz file via ORCA input, using the *xyzfile command
Generation of Coordinates
Today
Molecules are created with full specification in an xyz space
using common programs.

There are two ways to get coordinates into ORCA, once those
have been created and exported into an .xyz file:

1. Copy the contents of the .xzy file into the ORCA input file, using copy/
paste facilities of the texteditor or operating system
2. Read in the .xyz file via ORCA input, using the *xyzfile command

Although the first method looks somewhat more cumbersome at first,

it usually is the safer way:
By this, all information reside in your input file, reducing the error margin and making
the file more easily transportable.
Other Sources of Molecular Structures
Today there are many more sources for acquiring the molecular structure of
molecules. It is quite easy, to generate a small organic molecule from scratch.
This is definitely not the case with large biomolecular structures like proteins
or DNA !
The structures deposited in publicly available databanks might have different
origins:
• Crystal Structures:
Structures generated by analyzing X-Ray diffraction patterns.
Very good resolution with organic/inorganic molecules, varying with large
biomolecules
• NMR Structures:
Structures, or usually an ensemble of structures, by applying NMR structural
information into the generation of structures.
Very good for accessing structures in solution.
• Theoretical Models
Sometimes there are neither X-Ray nor NMR information available. In that case
people try to come up with a theoretical model, which is usually refined by
techniques like simulated annealing etc. to make certain the overall structure
reaches a state of low energy.
It can be hard to assess the quality of these structures at time. The difficulty of
generating good theoretical models gets worse with increasing molecular size.
SCF Convergence

Sooner or later it happens:

The SCF cycles does not converge!

What now?
There are serveral reasons causing the SCF not to converge.
• Molecular Structure contains errors
For example, missing atoms, the electronic state is wrong, etc.
• Very Large Systems
Due to the accumulation of numerical noise the SCF might not converge
• Difficult Chemical Bonding
Bonding situations containing transition metals might be hard to converge
due, especially if the specified state is far away from the ground state
SCF Convergence

What now?
There are many ways to influence the SCF
convergence, for example:
SCF Convergence

What now?
There are many ways to influence the SCF
convergence, for example:

• Damping
SCF Convergence

What now?
There are many ways to influence the SCF
convergence, for example:

• Damping
• Level Shift
SCF Convergence

What now?
There are many ways to influence the SCF
convergence, for example:

• Damping
• Level Shift
• DIIS
SCF Convergence

What now?
There are many ways to influence the SCF
convergence, for example:

• Damping
• Level Shift
• DIIS
• SOSCF
SCF Convergence

What now?
There are many ways to influence the SCF
convergence, for example:

• Damping
• Level Shift
• DIIS
• SOSCF
• Newton - Raphson
 SCF Convergence

What now?
There are many ways to influence the SCF
convergence, for example:

• Damping First thing to try:

• Level Shift
Specifying !slowconv
• DIIS
• SOSCF
• Newton - Raphson Otherwise:
• Increase Level Shift
• Increase DIIS Length
( For the whole list, see ORCA manual )
 Transition Metal Calculations
For open-shell transition metals convergence of the SCF is often a problem. There are a
few ways in ORCA to help this process:

# Help a transition metal calculation to converge

! UKS B3LYP SV(P) TightSCF SlowConv
This tells the program that you
expect slow-convergence
%scf shift shift 0.1 erroff 0 end
damp fac 0.80 erroff 0.001 end
end
Use level-shifting. Never turn it off
* xyz –2 2
Cu 0 0 0 U s e d a m p i n g . Tu r n i t o ff a f t e r
F 1.9 0 0 convergence of the DIIS procedure to
F –1.9 0 0 0.001 has been achieved
F 0 1.9 0
F 0 -1.9 0
*

... Actually, this particular job converges without any damping or shifting. However, in many situations, the damping and shifting is
necessary as you will undoubtedly find out yourself once you start calculations in the „real world“.

In the present example the convergence „aids“ acctually strongly slow down convergence. It is quite uniform since the energy is always
decreasing. Thus, these „helpers“ more or less „babysit“ the job to a converged solution.
 Restarting Calculations
In many cases it is a very good idea to start from the convergence MOs of a previous job.

# Start calculation from MOs of a previous one

! UKS B3LYP TZVP TightSCF SlowConv
! moread
%moinp “JOB-02.gbw”
Use previous MOs Large basis set calculation
%scf guessmode cmatrix started from MOs of a small
Name of the GBW file end basis set calculation
NOTE: must not have the same
name as your present input file! * xyz –2 2
Cu 0 0 0
F 1.9 0 0
„Play safe“ if restarting
F –1.9 0 0
calculations on anions
F 0 1.9 0
F 0 -1.9 0
*

Note that the calculation can be started from a gbw file which is from a nearby
geometry, uses a different basis set of theoretical method or HFType (UHF,RHF or
ROHF). The program will take care to translate the orbitals to the present situation.
Second Stop: Geometry Optimizations
Geometry Optimizations
From a formal point of view:
A geometry optimizations minimizes the overall energy of the system by
changing the atomic coordinates.

Minimum: All gradients have to vanish


 Problem:
Leads to the nearest local minimum,
not the global minimum of the system!



Geometry Optimizations

Things to keep in mind when performing geometry

optimizations:
Geometry Optimizations

Things to keep in mind when performing geometry

optimizations:

• The SCF wavefunction has to be quite precise to generate a smooth potential surface
without ‘bumps’
Geometry Optimizations

Things to keep in mind when performing geometry

optimizations:

• The SCF wavefunction has to be quite precise to generate a smooth potential surface
without ‘bumps’
• Geometry optimizations might be tough if the molecule contains many soft potentials,
for example dihedrals with very low barriers
Geometry Optimization
To optimize the geometry of the molecule, simply include the keyword Opt
! RKS SV(P) B3LYP TightSCF Opt

* int 0 1
C 0 0 0 0 0 0
O 1 0 0 1.15 0 0
H 1 2 0 1.00 115 0
H 1 2 3 1.00 115 180
*

The program will first produce a set of „redundant internal coordinates“ which are used
in the calculation.
--------------------------------------------------------------
Redundant Internal Coordinates

--------------------------------------------------------------
Defintion Initial Value Approx d2E/dq
--------------------------------------------------------------
1. B(O 1,C 0) 1.1500 1.351281
2. B(H 2,C 0) 1.0000 0.501167
3. B(H 3,C 0) 1.0000 0.501167
4. A(H 2,C 0,O 1) 115.0000 0.425466
5. A(H 3,C 0,O 1) 115.0000 0.425466
6. A(H 3,C 0,H 2) 130.0000 0.323418
7. I(O 1,H 3,H 2,C 0) 0.0000 0.151694
--------------------------------------------------------------

TIP: Always use TightSCF or VeryTightSCF in geometry optimizations. Otherwise the gradients are somewhat noisy.
After calculating the SCF energy and the gradient of the energy, a And a new geometry is proposed:
relaxation is step is carried out: ------------------------------------------------------------------------
Redundant Internal Coordinates
------------------------------------------------------------------------------ (Angstroem and degrees)
ORCA GEOMETRY RELAXATION STEP
------------------------------------------------------------------------------
Definition Value dE/dq Step New-Value
Number of atoms .... 4
Number of internal coordinates .... 7 ------------------------------------------------------------------------
Current Energy .... -114.317745134 Eh 1. B(O 1,C 0) 1.1500 -0.142336 0.0535 1.2035
Current gradient norm .... 0.207887808 Eh/bohr
Maximum allowed component of the step .... 0.300 2. B(H 2,C 0) 1.0000 -0.103355 0.0980 1.0980
Current trust radius .... 0.300 3. B(H 3,C 0) 1.0000 -0.103355 0.0980 1.0980
Evaluating the initial hessian .... (Almloef) done
Projecting the Hessian .... done 4. A(H 2,C 0,O 1) 115.00 -0.017685 2.44 117.44
Forming the augmented Hessian .... done 5. A(H 3,C 0,O 1) 115.00 -0.017685 2.44 117.44
Diagonalizing the augmented Hessian .... done
Last element of RFO vector .... 0.957975075 6. A(H 3,C 0,H 2) 130.00 0.035370 -4.89 125.11
Lowest eigenvalues of augmented Hessian: 7. I(O 1,H 3,H 2,C 0) 0.00 -0.000000 0.00 0.00
-0.057174708 0.151693870 0.360949845 0.425465740 0.501166791
Length of the computed step .... 0.299435193 ------------------------------------------------------------------------
The final length of the internal step .... 0.299435193
Converting the step to cartesian space:
Transforming coordinates: *************************************************************
Iter 0: RMS(Cart)= 0.0824560429 RMS(Int)= 0.1127292561 * GEOMETRY OPTIMIZATION CYCLE 2 *
Iter 1: RMS(Cart)= 0.0024951498 RMS(Int)= 0.0033197589
Iter 2: RMS(Cart)= 0.0002432861 RMS(Int)= 0.0003067902 *************************************************************
Iter 3: RMS(Cart)= 0.0000185328 RMS(Int)= 0.0000231161
Iter 4: RMS(Cart)= 0.0000012690 RMS(Int)= 0.0000015779
Iter 5: RMS(Cart)= 0.0000000832 RMS(Int)= 0.0000001034
Iter 6: RMS(Cart)= 0.0000000054 RMS(Int)= 0.0000000067 Then the next SCF is done and the next gradient calculated, a new
done
Storing new coordinates .... Done
geometry is proposed until (hopefully) finally:

The status of the geometry convergence is printed:

.--------------------.
----------------------|Geometry convergence|---------------------
Item value Tolerance Converged
-----------------------------------------------------------------
***********************HURRAY********************
RMS gradient 0.07883145 0.00010000 NO *** THE OPTIMIZATION HAS CONVERGED ***
MAX gradient 0.14233649 0.00030000 NO *************************************************
RMS step 0.11317586 0.00200000 NO
MAX step 0.18511086 0.00400000 NO
....................................................
Max(Bonds) 0.0980 Max(Angles) 4.89
Max(Dihed) 0.00 Max(Improp) 0.00
----------------------------------------------------------------- Following this statement one more energy calculation is performed
The optimization has not yet converged - more geometry cycles are needed
in order to make sure that the energy and properties are really done
at the stationary point of the PES.
Constraints and Relaxed Scans
You can „freeze“ certain geometrical parameters in an optimization:

! RKS B3LYP/G SV(P) TightSCF Opt

%geom Constraints
{ B 0 1 1.25 C }
{ A 2 0 3 120.0 C }
end
end
* int 0 1
C 0 0 0 0.0000 0.000 0.00
O 1 0 0 1.2500 0.000 0.00
H 1 2 0 1.1075 122.016 0.00
H 1 2 3 1.1075 122.016 180.00
*

... Or freeze some and vary others (one frequently used possibility is to only optimize
hydrogen positions OptimizeHydrogens true). Constrained surfaces are calculated
as:

%geom Scan
B 0 1 = 1.35, 1.10, 12 # C-O distance that will be scanned
end
end
Third Stop: Frequency Calculations
Frequencies? What Frequencies?
Behind the simple term ‘Frequency Calculation’ quite a few useful properties are hidden.

Calculation of the mass weighted

Hessian matrix, i. e. the second
derivative of the energy with
respect to all coordinates Following the strict meaning, it describes the
calculation of frequencies, usually in the domain of
Infrared, in the harmonic approximation of all
coordinates of the molecule.

Upon diagonalization we get the

frequencies of the normal modes: ?


Transition State?
At times the geometry optimization might not lead to a minimum on the PES, but to a
transition state.


 

Two-­‐dimensional  poten0al  energy  surface  for  the  system  H-­‐H-­‐H  in  a  linear  arrangement.  A  transi0on  state  is  observed  
around  the  H1-­‐H2  and  H2-­‐H3  distances  being  both  1.0  Angström.

A transition state is defined as the point, where all

coordinates are at a minimum level, with the exception of
one, which is at its maximum.
Transition State?
At times the geometry optimization might not lead to a minimum on the PES, but to a
transition state.

Analyzing the Normal

Modes and Frequencies:

A Transition State has

exactly one negative
frequency!


 

Two-­‐dimensional  poten0al  energy  surface  for  the  system  H-­‐H-­‐H  in  a  linear  arrangement.  A  transi0on  state  is  observed  
around  the  H1-­‐H2  and  H2-­‐H3  distances  being  both  1.0  Angström.

A transition state is defined as the point, where all

coordinates are at a minimum level, with the exception of
one, which is at its maximum.
Zero Point Energy (ZPE)
As known from basic Physical Chemistry, even at 0K a
molecule still has some vibrational energy.

Having calculated all normal modes of the

molecule, as well as the frequency of vibration, it For the translational and
is fairly easy to calculate the zero point energy rotational degrees of freedom,
using the harmonic approximation. the statistical equipartition of
energy is applied, meaning
In addition, taking into account a finite
Temperature, the vibrational part of the ZPE Htrans  =  Hrot  =  3/2  RT  
accounts to



Frequency Calculations
There are several good reasons for calculating the harmonic frequencies:
1. Characterize stationary points as minima (no negative frequencies), transition
states (one negative frequency) or higher-order saddle point (more negative
frequencies
2. Predict vibrational spectra (IR, Raman)
3. Calculate thermodynamic properties (zero-point energy, finite temperature
correction)

ORCA presently calculates harmonic frequencies through (one- or two-sided) numeric

differentiation of analytic frequencies

! RKS BP86 RI SV(P) SV/J Grid4 TightSCF SmallPrint

! TightOpt NumFreq

%freq CentralDiff true

Run a numerical frequency calculation
Increment 0.005
end
Two sided differences (twice as expensive but
more accurate! Be careful – numerical
* xyz 0 1 frequencies can be quite noisy)
C 0.000000 0.000000 -0.533905
O 0.000000 0.000000 0.682807 Increment for displacements (in Bohrs)
H 0.000000 0.926563 -1.129511
H 0.000000 -0.926563 -1.129511
*
The first thing that is printed are the vibrational frequencies. The first
The IR spectrum can be plotted through the orca_mapspc utility
six modes are translations and rotations and these are zero because
(orca_mapspc BaseName.out IR)
they are projected out.
-----------------------
VIBRATIONAL FREQUENCIES
-----------------------

0: 0.00 cm**-1
1: 0.00 cm**-1
2: 0.00 cm**-1
3: 0.00 cm**-1
4: 0.00 cm**-1
5: 0.00 cm**-1
6: 1140.72 cm**-1
7: 1230.49 cm**-1
8: 1498.92 cm**-1
9: 1812.75 cm**-1
10: 2773.72 cm**-1 Finally, the thermodynamic properties at 298.15 K are printed
11: 2805.59 cm**-1
(assuming ideal gas behaviour)
Electronic energy ... -114.41435858 Eh
Zero point energy ... 16.10 kcal/mol
Then the program prints the normal modes. This is usually not very Thermal vibrational correction ... 0.03 kcal/mol
Thermal rotational correction ... 0.89 kcal/mol
revealing. The program produces a BaseName.hess file which you Thermal translational correction ... 0.89 kcal/mol
can run through orca_vib to get additional information. ------------------------------------------------------
XYZ files to be used for animation of vibrational modes are produces Total thermal energy -114.38582646 Eh
by the program orca_pltvib.
Then you get the IR spectrum: Then enthalpy+entropy and finally the free energy:

Total enthalpy -114.38488225 Eh

Total entropy correction -15.36 kcal/mol
------------------------------------------
----------- Final Gibbs free enthalpy -114.40935858 Eh
IR SPECTRUM G-E(el) = 0.00500000 Eh = 3.14 kcal/mol
-----------

Mode freq (cm**-1) T**2 TX TY TZ

-------------------------------------------------------------------
6: 1140.72 1.465985 ( -1.210739 -0.008963 -0.004000)
7: 1230.49 10.082152 ( 0.004428 -3.175237 -0.001184)
8: 1498.92 4.648016 ( -0.000197 0.001760 -2.155925)
9: 1812.75 105.821353 ( -0.000043 0.002847 -10.286950)
10: 2773.72 73.358541 ( 0.000042 -0.020336 -8.564936)
11: 2805.59 205.230431 ( -0.000359 -14.325864 0.007790)
 Solvation
★ The effects of solvation can be quite important for many properties (and in particular for
charged molecules – for cations less than for anions).
★ Solvation effects can be (relatively crudely) modelled by polarizable continuum models.
In ORCA the Conductor Like Screening (COSMO) model by Klamt is implemented in
an efficient way for energies and structures (don‘t do frequencies!)

! RKS DFTEnergy COSMO(water)

%cosmo epsilon 80
refrac 1.33
end
Shortcut is COSMO(solvent)
* int 0 1
C 0 0 0 0.000000 0.000 0.000 Alternatively: the dielectric constant and
O 1 0 0 1.200371 0.000 0.000
refractive index can be input as needed
H 1 2 0 1.107372 121.941 0.000
H 1 2 3 1.107372 121.941 180.000
*

TIP: Solation can be quite important in the prediction of optical and magnetic spectroscopic properties. Consider using it!
The End