X-Ray Absorption from ΔSCF
In this exercise we are going to compute near-edge X-ray absorption spectra of bulk MgS and MgO, performing all-electron calculations with the GAPW method, using the Transition Potential and \(\Delta SCF\) approaches. Our goal is to identify differences in the electronic structure, and as a consequence in the K-edge absorption spectrum, of the magnesium due to the different anions it is bounded to. We are also going to analyze the influence of basis set quality in the calculations.
Before starting, it is recommended to create one directory for each system (MgO and MgS) and, within the system’s directory, create the subfolders ‘optimization’, ‘dscf’ and ‘xas’.
Part 1: optimizing geometry
The first step of the calculation is to optimize the geometry of the systems you are going to work with. It is also possible to use experimental geometries if available.
MgO
To start the calculation, download or copy the input MgO_opt.inp to the optimization folder in the
work directory of MgO.
&GLOBAL
PROJECT_NAME MgO
RUN_TYPE GEO_OPT
PRINT_LEVEL LOW
FLUSH_SHOULD_FLUSH .TRUE.
&END GLOBAL
&MOTION
&GEO_OPT
TYPE MINIMIZATION
OPTIMIZER BFGS
MAX_ITER 200
&END GEO_OPT
&END MOTION
&FORCE_EVAL
METHOD QS
STRESS_TENSOR ANALYTICAL
&DFT
! in the geometry optimization there is no need to run an all-electron calculation, so we are
! going to make use of the GTH pseudopotentials for the core electrons.
BASIS_SET_FILE_NAME GTH_BASIS_SETS
POTENTIAL_FILE_NAME GTH_POTENTIALS
&MGRID
NGRIDS 5
CUTOFF 400
REL_CUTOFF 60
&END MGRID
&QS
METHOD GPW ! to optimize the geometry the GPW method will be used
&END QS
&SCF
MAX_SCF 200
EPS_SCF 1.0E-6
SCF_GUESS ATOMIC
&OT
MINIMIZER DIIS
PRECONDITIONER FULL_ALL
&END OT
&END SCF
&XC
&XC_FUNCTIONAL PBE ! PBE exchange-correlation functional
&END XC_FUNCTIONAL
&XC_GRID
XC_SMOOTH_RHO NN50
XC_DERIV NN50_SMOOTH
&END XC_GRID
&END XC
&END DFT
&SUBSYS
&COORD
O 3.010000 1.737824 1.228827
Mg 0.000000 0.000000 0.000000
&END COORD
&CELL
PERIODIC XYZ ! we are considering the system periodic in the three directions
ALPHA_BETA_GAMMA 60 60 60
ABC 3.010 3.010 3.010
&END CELL
&KIND Mg
ELEMENT Mg
BASIS_SET DZVP-GTH
POTENTIAL GTH-PBE-q10
&END KIND
&KIND O
ELEMENT O
BASIS_SET DZVP-GTH
POTENTIAL GTH-PBE-q6
&END KIND
&END SUBSYS
&END FORCE_EVAL
Since both systems have only two atoms in their unit cells it is not necessary to have a separate
.xyz file with the atomic positions. To make it simple we are going to write the coordinates in the
&COORD subsection of the input file.
Do not forget to put in your work directory the files GTH_POTENTIALS and GTH_BASIS_SETS, which
contain the parameters for the pseudopotentials and basis sets used in the calculations.
After the calculation is finished, you can check the files created in your directory. First open the
output file MgO_opt.out and search for the following banner:
*******************************************************************************
*** GEOMETRY OPTIMIZATION COMPLETED ***
*******************************************************************************
If you found it, it means that the optimization of the geometry is done, and you can find the final
atomic coordinates in the file MgO-pos-1.xyz. You can visualize the optimized geometry using
Avogadro or VESTA programs, for example.
cp2k prints out the coordinates for each step of the calculation (they are indicated in the file by the index i, right below the number of atoms), so in order to use the optimized geometry in the following calculations, you should use the positions corresponding to the last iteration.
It is also important to check for warnings in your output file. In the end of the file, you can find the following banner:
-------------------------------------------------------------------------------
The number of warnings for this run is : 0
-------------------------------------------------------------------------------
which means that the calculation ran without problems. If the number is different than 0, search for the warning messages throughout the output file.
MgS
Now we are going to perform the same calculation, but for the MgS system. In order to do so, let’s
make some changes to the input file MgO_opt.inp. You can either download the input file above
again, and change its name no MgS_opt.inp, or type in your terminal:
cp MgO_opt.inp MgS_opt.inp
This will create a copy of the previous input file with the name MgS_opt.inp. Move the new file to
the optimization folder of the MgS work directory. Now we need to make some modifications to the
input in order to perform the calculation for the MgS system. Let’s start with the project name:
change it to MgS.
Now, in the &COORD subsection, we are going to give the initial atomic coordinates as multiples of
the lattice vectors. In order to do it, delete the two lines with the coordinates of the previous
system, and add the following:
SCALED
S 0.5 0.5 0.5
Mg 0.0 0.0 0.0
We also need to change the lattice parameters in the &CELL subsection, since the vectors have
different lengths now. Delete the numbers that follow the ABC keyword and type:
3.697 3.697 3.697
In order to deal with a smaller number of atoms, we are declaring the structures of MgO and MgS using the rhombohedral unit cell, so the lengths of the lattice vectors a, b, and c, so as the angles \(\alpha\) , \(\beta\) and \(\gamma\), are the same.
The last modification that needs to be done is regarding the atomic types. In this case, we do not
have oxygen in the system anymore, so the subsection &KIND O can be renamed &KIND S. The only
modification that needs to be done is in the keyword ELEMENT, where O has to be replaced by S.
Even though we are not changing the name of the basis set or pseudopotential used, cp2k will use the
parameters for the sulfur atom now, since the ELEMENT type is different. However, it is important
to check in the GTH_BASIS_SET and GTH_POTENTIALS files whether the names are the same for
different atoms.
Now the input is ready, and it can be run in the same way as before, just remember to change the
file cp2k.sh.
After the calculation is finished, open the output file MgS_opt.out and look for the same banner
as before. The optimized atomic positions are written in the file MgS-pos-1.xyz.
Part 2: XAS calculations
To compute the absorption spectra, download or copy the input file bellow to the working directory. It is a general input that needs to be edited depending on which system you are working with.
&GLOBAL
PROJECT_NAME MgX ! TASK: change X to O or S
RUN_TYPE ENERGY
PRINT_LEVEL LOW
FLUSH_SHOULD_FLUSH .TRUE.
&END GLOBAL
&FORCE_EVAL
METHOD QS
&DFT
!where to find all-electron basis sets and potentials
BASIS_SET_FILE_NAME EMSL_BASIS_SETS
POTENTIAL_FILE_NAME POTENTIAL
UKS
&MGRID
NGRIDS 5
CUTOFF 400
REL_CUTOFF 60
&END MGRID
&QS
METHOD GAPW ! using GAPW for all-electron calculations
EXTRAPOLATION ASPC
EXTRAPOLATION_ORDER 3
MAP_CONSISTENT
EPS_DEFAULT 1.0E-10
! algorithm to construct the atomic radial grid for GAPW
QUADRATURE GC_LOG
! parameters needed for the GAPW method, look at the manual for more details
EPSFIT 1.E-4 ! precision to give the extension of a hard gaussian
EPSISO 1.0E-12
EPSRHO0 1.E-8
LMAXN0 4
LMAXN1 6
ALPHA0_H 10 ! Exponent for hard compensation charge
&END QS
&SCF
MAX_SCF 50
EPS_SCF 1.0E-5
SCF_GUESS ATOMIC
ADDED_MOS 8
&MIXING
METHOD BROYDEN_MIXING
ALPHA 0.5
&END MIXING
&END SCF
&XC
&XC_FUNCTIONAL PBE
&END XC_FUNCTIONAL
&XC_GRID
XC_SMOOTH_RHO NN50
XC_DERIV NN50_SMOOTH
&END XC_GRID
&VDW_POTENTIAL
POTENTIAL_TYPE PAIR_POTENTIAL
&PAIR_POTENTIAL
PARAMETER_FILE_NAME dftd3.dat
TYPE DFTD3
REFERENCE_FUNCTIONAL PBE
R_CUTOFF [angstrom] 16
&END PAIR_POTENTIAL
&END VDW_POTENTIAL
&END XC
&XAS
RESTART .FALSE.
METHOD TP_HH ! transition potential half core hole
DIPOLE_FORM VELOCITY
STATE_TYPE 1s ! excitation from 1s orbital (K-edge calculation)
ATOMS_LIST 1 2 ! calculate absorption for 1st and 2nd atoms in the &COORD subsection
ADDED_MOS 8
&SCF
EPS_SCF 1.0E-5
MAX_SCF 200
&MIXING
METHOD BROYDEN_MIXING
ALPHA 0.5
&END MIXING
&SMEAR
ELECTRONIC_TEMPERATURE [K] 300
METHOD FERMI_DIRAC
&END SMEAR
&END SCF
&LOCALIZE
&END LOCALIZE
&PRINT
&PROGRAM_RUN_INFO
&END PROGRAM_RUN_INFO
&RESTART
FILENAME ./MgX ! TASK: change X to O or S
&EACH
XAS_SCF 20
&END EACH
ADD_LAST NUMERIC
&END RESTART
&XAS_SPECTRUM
FILENAME ./MgX ! TASK: change X to O or S
&END XAS_SPECTRUM
&XES_SPECTRUM
FILENAME ./MgX ! TASK: change X to O or S
&END XES_SPECTRUM
&END PRINT
&END XAS
&END DFT
&SUBSYS
&COORD
X x(X) y(X) z(X)
Mg x(Mg) y(Mg) z(Mg)
&END COORD
&CELL
PERIODIC XYZ
ALPHA_BETA_GAMMA 60 60 60
ABC A B C
&END CELL
&KIND Mg
ELEMENT Mg
BASIS_SET Ahlrichs-pVDZ
POTENTIAL ALL ! all-electron calculations
LEBEDEV_GRID 80
RADIAL_GRID 200
&END KIND
&KIND X ! TASK: change X to O or S
ELEMENT X ! TASK: change X to O or S
BASIS_SET Ahlrichs-pVDZ
POTENTIAL ALL ! all-electron calculations
LEBEDEV_GRID 80
RADIAL_GRID 200
&END KIND
&END SUBSYS
&END FORCE_EVAL
MgS
To compute the absorption spectra for the bulk MgS, first rename the input file changing the X to
S. It can be done by typing in the terminal:
cp MgX_xas.inp MgS_xas.inp
Now change all the Xs in the input file to Ss. Move the new input file to the correct working
directory. The next step is to add the optimized coordinates of the system, which you can find in
the .xyz file written by the program after the geometry optimization. Use the last iteration step
values and write them in the &COORD subsection. The final step is to add the correct values for
the lattice vectors. You can copy it from the geometry optimization input file.
At this time we are using cartesian coordinates to indicate the position of the atoms, so the
keyword SCALED should be removed. To run this calculation, proceed as you did before.
This calculation should take longer than the geometry optimization to run. Once finished, check the
number of warnings and if the calculation converged. Sometimes it does not converge within the
maximum number of iterations we set in the input file. If this is the case, you can increase the
number using the keyword MAX_SCF.
You can check in the working directory that some files were created. The absorption energies and
intensities (oscillator strength) are written in the files named MgS-xas_at1_st1.spectrum and
MgS-xas_at2_st1.spectrum, where the first one corresponds to the atom 1 in your input file, and
the second one to atom number 2.
The file looks like
Absorption spectrum for atom 1, index of excited core MO is 2, # of lines 9
11 531.57449433 0.00000000 0.00000019 -0.00000002 0.00000000 0.00000
12 549.96927153 0.31337224 0.18092555 0.12793369 0.14730324 0.00000
13 550.01480014 -0.22208298 0.24828653 0.19285978 0.14816194 0.00000
14 550.01480014 -0.00815280 0.23149701 -0.30741602 0.14816194 0.00000
15 574.27304606 -0.84466734 0.95907128 0.71266966 2.14117868 0.00000
16 574.27304607 -0.01626535 0.86344807 -1.18125846 2.14117868 0.00000
17 574.27591527 1.19525241 0.69008026 0.48796033 2.14294438 0.00000
18 694.86428215 0.00000000 -0.00000010 -0.00000012 0.00000000 0.00000
and the first column corresponds to the index of the KS virtual state, the second to the energy in eV, the third, fourth, and fifth to the intensities projected onto x, y and z, respectively, and in the sixth column you can find the norm of the absorption intensity, which is the quantity we are interested at.
To convolute the spectra with gaussian functions, download the files {{exercises:2019_conexs_newcastle:lib_tools.zip}} and extract them in the same directory as the output files. Now run the script typing in the terminal:
./get_average_spectrum.sh
As an output, you are going to get two files: spectrum.inp and spectrum.out. The first one
contains the same information as the Mgs-xas_at1_st1.spectrum file, and in the second one you will
find you absorption spectrum for atom 1. Change the name of the files to S_K-edge.inp and
S_K-edge.out, for example. You can now plot both absorption intensities from the file
S_K-edge.inp and the convoluted spectrum from the file S_K-edge.out. From the first one only the
second and sixth columns need to be plotted.
In order to obtain the spectrum for atom 2, you can open the file get_average_spectrum.sh and
replace at1 by at2 in the line for i in $(ls $\{DIR}/*xas_at2*spectrum). Run the script again
and you will obtain the same two files again, but now with the absorption intensities and spectrum
of atom 2. Change their names to Mg_K-edge.inp and Mg_K-edge.out, and plot the absorption
spectrum.
Part 3: \(\Delta SCF\) calculations
Now, to finally finish the calculation, we need to get an accurate energy for the first transition.
In order to do that, we need to perform a \(\Delta SCF\) calculation. Copy the input file of the
previous step to the ‘dscf’ directory. Change its name to MgX_dscf.inp, where X can be again S
or O. The only thing that needs to be changed in the input file is the keyword METHOD in the
&XAS section. Use now
METHOD DSCF
instead of TP_HH, and you can run the calculation in the same way as you did before.
After the calculation is done, look for the message
Ionization potential of the excited atom: -92.73815588900608
in the output file. The energy is given in Hartree, and to convert it to electron volts multiply the value by 27.211. This is the energy of the first transition, and you can use this value to rigidly shift your absorption spectrum.
Part 4: Changing basis set
Before performing the XAS calculations for the MgO system and comparing the Mg absorption spectra, you can try to change the basis set you are using to run the absorption calculations to analyze differences it can bring to the description of the process. Try to perform the calculations using:
pc-0 (smaller basis set)
pob-TZVP (basis set for solid-state calculations)
DZVP-all
Ahlrichs-def2-SVP
In this exercise, we have obtained absorption spectra using a simple basis set to perform the calculations on small machines and using a limited time. Therefore, careful tests on the basis set size, XC functional, etc have to be carried out for production runs to get more reliable spectra.