Transition State Search with a Partial Hessian

  • calculate a partial Hessian
  • do a constrained TS search using the partial Hessian

In this “advanced” tutorial we consider a slightly more realistic system. Some of the calculations may require 20 minutes to run on a two core machine.

Step 1: Create the system

We are going to make a one layer Li (001) slab with a 2x2 unit cell, assuming familiarity with the build tools

Start ADFinput and switch to BAND mode.
From the structure tool select ‘Cubic’ and ‘bcc’
Set ‘Element’ to Li
Set the lattice parameter to 3.49
Press ‘OK’
Invoke the Slice tool
Set the Miller indices to 001, select ‘Cartesian’, and enter 1 layer.
Press ‘OK’
Use the Edit → Crystal → Generate Super Cell... command
Select the preset ‘2x2’ and press ‘OK’
Panel bar Model → Coordinates

Your screen should look like this:

../_images/LiSlab.png
Add with the mouse two hydrogen atoms anywhere in the screen
Set in the table the coordinates of the first hydrogen atom to (0, -0.5, 2)
Set the second H atom coordinates to (0, 0.5, 2)

The final geometry looks like this

../_images/H2onLiSlab.png

Step 2: Calculate a partial Hessian

Select the ‘BAND Main’ panel
Select the Frequencies preset
Set ‘Basis Set’ to ‘SZ’
../_images/FreqMainOptions.png
Go to the ‘Frequencies’ details panel (click on the ‘...’ button)
Select with the mouse the two tiny Hydrogen atoms (zoom in to make selecting them easier)
Click on the ‘+’ button next to ‘Partial Hessian For:’
../_images/FreqSelectH2.png
File → Run, name the job H2onLi_freq
Say ‘No’ when asked to update the coordinates

Let us examine the eigen-modes that we have found for the Hydrogen molecule

Select the SCM → Spectra menu command
../_images/FreqSpectra.png

Now you will see that there is an eigen-mode at 465 cm-1 and one at 2428 cm-1. Convince yourself that the 465 mode moves the H2 perpendicular to the service and that the 2428 mode is essentially an H2 stretch mode. The lowest mode looks like a promising start to find the transition state for dissociation over the Li surface.

Step 3: Transition state search with a frozen substrate

We have just found the vibrational modes of the Hydrogen molecule, assuming that the Li substrate remains fixed. Let us now find the transition state under the same assumption.

Close the ‘ADFspectra’ window and go back to BANDinput.
Select the ‘BAND Main’ panel and select the ‘Transition State Search’ preset
../_images/TSMainOptions.png
Panel bar Model → Geometry Constraints
The two hydrogen atoms are still selected, if not select them again
Use the Select → Invert Selection command
Click on the ‘+’ button next to ‘freeze selected atoms’
../_images/TSConstraints.png
Panel bar Details → Geometry Convergence
Set the initial Hessian to ‘H2onLi_freq.runkf’
../_images/TSInitialHessian.png
File → Save As..., use name H2onLi_ts
File → Run

After the run has completed:

SCM → Movie
In the movie window: Graph → Energy
View → View Direction → Along x axis
Reorient to get a good look

It should look like

../_images/TSMovie.png