Transition state search from frequency calculation


Click here for full thread
Clicked A Few Times
Hi All

I am trying to perform transition state searches for oxidations of small molecules. I am very familiar with how this is done in the program Turbomole. I am now in a position where I am using NWchem for the same task and I have problems with the syntax of the transition state search.
The example here is a phenolic hydrogen abstraction of acetaminophen. I am using a methoxy radical as the reactive specie.
My workflow is:

constrained geometry optimization
geometry
 load geometry.xyz
end
memory 1000 mb
basis
 * library 6-31g
end
dft
 xc b3lyp
 mult 2
 iterations 300
 direct
end
geometry adjust
 zcoord
  bond 13 21 1.21 constant
  bond  9 13 0.94 constant
 end
end
driver
 xyz test
 maxiter 50
end
task dft optimize


frequency calculation
restart consgeo
memory 1000 mb
geometry adjust
 zcoord
  bond 13 21 1.21
  bond  9 13 0.94
 end
end
freq
 animate
end
task dft freq


transition state search
restart consgeo
memory 1000 mb
task dft saddle


This is the first line of output frequencies:
Frequency       -339.75      -54.43        5.32        9.09       12.46       15.11


The first frequency is along the reaction coordinates, however, these frequencies does not match what I expect for a hydrogen abstraction - is there a way to increase precision?

My transition state search finishes just fine however the geometry and energy does not match the ones I used to get with Turbomole using the same functional and basis. I am not sure that the saddle task uses the information from the frequency calculation. How can I specify to the program that it is the first imaginary frequency on which I want to perform the saddle point search?

Thanks and best regards,
Rasmus