Instructor: P. Piecuch. Laboratory No. 2 (February 10, 2010)

Exploring the excited states, geometry, and harmonic vibrational frquencies of the water molecule

We are going to learn how to use Gaussian to optimize geometries and calculate harmonic vibrational frequencies in a single two-step calculation at the restricted Hartree-Fock (RHF), second-order many-body perturbation theory [MBPT(2) or MP2], and configuration interaction singles and doubles (CISD) levels. We will also learn how to optimize geometries and how to calculate harmonic vibrational frequencies with GAMESS using MP2 and the completely renormalized coupled-cluster approach with singles, doubles, and a non-iterative treatment of triples termed CR-CC(2,3). We will use the configuration interaction with singles (CIS) approach in Gaussian and the equation-of-motion coupled-cluster singles and doubles (EOMCCSD) method in GAMESS to obtain information about excitation energies. We will also use the CIS approach in Gaussian to optimize the geometry of a molecule in the excited state.

You will have an opportunity to compare the results of geometry and frequency calculations with different methods, including RHF, MP2=MBPT(2), CISD, and CR-CC(2,3). You will also see how accurate the CIS method is relative to the more sophisticated, yet very practical, EOMCCSD approach in predicting vertical excitation energies, and how the computed excitation energies may depend on the basis set. The knowledge and experience acquired by performing the above calculations will be very useful for your homework, so please pay attention to details. You can always come back to this web site after the lab to re-examine the calculations we will perform during the lab.

All of the relevant files reside in the directory ~piecuch/cem888_SS10/students2/. The subdirectories 'set1' and 'set2' correspond to two sets of calculations we are interested in. Use 'cp -r ~piecuch/cem888_SS10/students2/* ./' to copy the files relevant to today's lab to the appropriate working subdirectory in your home area.

I. Problem Set 1 (subdirectory 'set1'): Gaussian CIS and GAMESS EOMCCSD calculations

All calculations in this problem set (subdirectory 'set1') deal with the water molecule. With an exception of two GAMESS runs, all of the calculations use the 6-311++G(d,p) basis set. The appropriate input files are described below. Run the rung98 script to perform the Gaussian calculations and use 'gmssub' or 'gmssub -b basis' to submit the GAMESS calculations. Before running the calculations, please examine the input files and try to figure out what each of the keywords means. The appropriate manuals can be found here (Gaussian) and here (GAMESS).

You will find the following input files in the subdirectory 'set1':

h2o_rhf_opt.com - the Gaussian input file for the geometry optimization
                  at the RHF level (the initial values of
                  the variables roh and ang are set at the
                  experimental values of the O-H bond length
                  and the H-O-H angle, respectively). As you know,
                  the CIS ground state is represented by
                  the Hartree-Fock (in this case, RHF) wave
                  function.

h2o_cis_exp.com - the Gaussian input file for the CIS calculations
                  of the 5 lowest singlet and 5 lowest triplet
                  excited states. The value of Nstate equals
                  the number of excited states you would like
                  to calculate. Since water is a closed-shell
                  molecule, the CIS calculation generates only
                  singlet states as default. If you would like to
                  learn about triplet states, you must add
                  the 50-50 option, as shown in the input file.
                  The experimental geometry of water in the
                  ground electronic state is used. The CIS
                  approach can only give singlet and triplet
                  excited states for closed-shell
                  molecules. Do you know why?

h2o_cis_rhfgeom.com - same as above for the RHF-optimized
                  geometry of water.

h2o_cis_2opt.com - the Gaussian input file for the optimization of the
                  geometry of water in the first-excited SINGLET
                  state (triplet states are not even considered
                  in this calculation, since we do not use the
                  50-50 option). The value of Root equals to
                  the excited-state number whose geometry is
                  being optimized (in our case, Root=1, which
                  means that we optimize geometry of water
                  in the first-excited singlet state).

h2o_eomccsd_exp.inp - the GAMESS input file for the EOMCCSD calculations
                  of the 5 lowest SINGLET excited states using
                  the 6-311++G(d,p) basis set (36 functions). As indicated
                  by the 'nstate(1)=2,1,0,2' option in the $eominp input
                  group (and 'CNV 2' in $data), GAMESS uses point symmetry
                  to help the calculations. The five low-lying excited
                  states that we are interested in determining are two
                  A1 singlets, one A2 singlet, and two B2 singlets;
                  note that due to different conventions used by Gaussian
                  and GAMESS, the B1 states in Gaussian are listed as the
                  B2 states in GAMESS and vice versa.

h2o_eomccsd_exp_acct.inp - same as above for the aug-cc-pVTZ basis set
                 (92 functions), which is considerably larger than the
                 6-311++G(d,p) basis. In order to save time
                 during the lab, the output is provided
                 for your convenience. Compare the results with the
                 earlier EOMCCSD calculation.

h2o_eomccsd_exp_diff.inp - same as above for the modifed variant of the
                 cc-pVTZ basis set, in which we added the suitably
                 optimized diffuse functions to obtain a better description
                 of some excited states. The basis set used in this
                 EOMCCSD calculation (68 functions) is smaller than aug-cc-pVTZ,
                 but it yields considerable improvements in the results for
                 the second singlet B1 and second singlet A1 excited states.
In each Gaussian CIS calculation, the lowest orbital (more or less, the 1s orbital of the oxygen atom) is frozen (cf., '2,0' in the window and the rw option). The same is true in the GAMESS EOMCCSD calculations ('ncore=1'). The lowest-energy molecular orbital of water is almost identical to the 1s orbital in the oxygen atom and can be frozen.

Things to do:

1. Compare the RHF-optimized geometry with the experimental geometry of water
   (the initial values of roh and ang).

2. Look at the vertical excitation energies in h2o_cis_exp.log,
   h2o_cis_rhfgeom.log, h2o_eomccsd_exp.log, h2o_eomccsd_exp_acct.log, and
   h2o_eomccsd_exp_diff.log, and compare them to the experimental values of these
   energies, which are:
   3B1: 7.0 or 7.2 eV,
   1B1: 7.42 eV,
   3A2: 8.9 eV,
   1A2: 9.1 eV,
   3A1: 9.3 eV,
   1A1: 9.67 eV,
   3A1: 9.81 eV,
   3B1: 9.98 eV,
   1B1: 10.01 eV,
   1A1: 10.17 eV.
   Any comments?
   Remember that the B1 states in Gaussian and above are listed as the
   B2 states in GAMESS and vice versa.

3. Compare the geometries of water in the ground and excited states.

II. Problem Set 2 (subdirectory 'set2'): Geometry optimizations and vibrational frequency calculations

All calculations in this problem set deal with the water molecule described by the 6-31++G(d,p) basis set. The first goal is to learn how to combine geometry optimization and harmonic vibrational frequency calculation by creating a single Gaussian input. The harmonic frequency calculations cannot be performed if we do not first determine the equilibrium geometry. Clearly, we can always run a geometry optimization first, read the results from the output, and then use the optimized geometry to prepare the input for the subsequent frequency calculation. Well, there is a nice way by using the so called checkpoint files and --Link-- lines in the Gaussian inputs. The second goal is to perform the analogous geometry optimizations and frequency calculations for the water molecule using the MP2 and CR-CC(2,3) methods in GAMESS. The MP2 calculations with Gaussian and GAMESS should provide the same results. CR-CC(2,3) is a higher-level theory compared to MP2 (and CISD), so we expect improvements in the calculated geometry and frequencies. The appropriate input files are described below. As in the case of Set 1, run the rung98 script to perform the Gaussian calculations and use 'gmssub' to submit the GAMESS calculations. Again, before running the calculations, please examine the input files and try to figure out what each of the keywords means. The appropriate manuals can be found here (Gaussian) and here (GAMESS). You will find the following input files in the subdirectory 'set2':

Gaussian input files

h2o_rhf_freq.com
----------------
This is an input file for the geometry optimization and frequency
calculation at the RHF level.  The geometry optimization is performed
first using analytic gradients:

# rhf/6-31++G(d,p) UNITS=angs SCF=Direct Opt

The most essential information, including the optimum geometry
obtained in the first calculation, is stored in the checkpoint file
rhffreq (Chk=rhffreq). The geometry stored in file rhffreq is read
into the second calculation, in which frequencies are calculated at
the RHF level. The following line describes the
frequency calculation, which is also done analytically:

# rhf/6-31++G(d,p) Geom=Allcheckpoint Guess=Read UNITS=angs SCF=Direct Freq

The --Link1-- line is essential for linking the two calculations
together.

h2o_mp2_freq.com
----------------
This is an input file for the geometry optimization and frequency
calculation at the MBPT(2)=MP2 correlated level. The structure of this
input file is essentially the same as in the RHF case. The only
difference is the method employed (MP2) and information about the orbital
frozen in the calculation (the 2,0 window). Analytic energy derivatives of
MP2 are employed.

h2o_cisd_freq.com
-----------------
This is an input file for the geometry optimization and frequency
calculation at the CISD level.  The structure of this input file is
essentially the same as in the MP2 case. The only difference is the
method employed (CISD). Analytic energy gradients are followed by
numerical differentiation of analytically computed first derivatives.
Please note that we use different names for the checkpoint files produced and used by different calculations (rhffreq.chk, mp2freq.chk, cisdfreq.chk) to avoid confusion and accidental overwrite.

GAMESS input files

h2o_mp2_opt_GAMESS.inp
-----------------------
This is a GAMESS input file for the geometry optimization
at the MP2 level. Analytic gradients of MP2 are exploited.

h2o_mp2_freq_GAMESS.inp
-----------------------
This is a GAMESS input file for the frequency calculations
at the MP2 level. Analytic energy gradients are followed by
a numerical differentiation of analytically computed first
derivatives.

h2o_crcc23_opt_GAMESS.inp
-------------------------
This is a GAMESS input file for the geometry optimization
at the CR-CC(2,3) level. Numerical energy derivatives
are exploited. If the time allocated to the lab becomes
a constraint, we will examine the output provided by the
instructor.

h2o_crcc23_freq_GAMESS.inp
-------------------------
This is a GAMESS input file for the frequency calculations
at the CR-CC(2,3) level. Numerical energy derivatives
are exploited. If the time allocated to the lab becomes
a constraint, we will examine the output provided by the
instructor.

Things to do:

Compare the geometries and harmonic frequencies obtained in your
calculations with the experimental values which are as follows:
roh=0.957 Angs,
ang=104.5 Degrees,
omega1 = 1649 cm-1, omega2 = 3832 cm-1, omega3 = 3943 cm-1.

Any thoughts?

There is a useful program for visualization of molecules
and for animating the normal-mode vibrations.
It is called viewmol. A nice feature of this program
is that it shows the theoretical harmonic spectrum (peaks
with intensities corresponding to calculated
values). You can try this program by typing
'viewmol h2o_rhf_freq.log',
where h2o_rhf_freq.log is the output of your RHF frequency calculation.
There are other tools in this category.