Hi all,
I'm doing an optimization for ZnCl2 in acetonitrile solvent using Gaussian DFT method. The command line is below:
#p M062x/6-31+G* opt SCF=QC freq scrf=(smd,solvent=Aceton
itrile) integral=ultrafine
The calculation always ends with 2 negative frequency with many tries. I also tried to use 6-31++G** basis set and opt=tight, but the negative frequencies kept showing up. The negative frequencies are up-down and back-forth bending models.
This is the convergence result when using 6-31++G** basis set and opt=tight command:
Item Value Threshold Converged?
Maximum Force 0.000000 0.000015 YES
RMS Displacement 0.000000 0.000010 YES
Maximum Displacement 0.000002 0.000060 YES
RMS Displacement 0.000001 0.000040 YES
The convergence parameters turned to be reliable, but the frequency data tells me there must be something wrong in the result.
Please let me know if I did something wrong in the calculation that lead to this result. Your suggestions and help are much appreciated!
Dear Wenda Wu ,
I assume that you are starting from a linear Cl-Zn-Cl arrangement. According to those imaginary frequencies, it seems that M06-2X predicts that this triatomic molecule is not linear, but bent, like water. If you try to optimize water as a linear structure, it will yield the same imaginary frequencies. Try to optimize ZnCl2 starting from a clearly bent structure, so it does not fall to that 2nd-order saddle point.
Hope you find it helpful
Dear Wenda Wu
In general, for such problems, you must first make sure of the initial structure, method, basis set.
There is no correct answer to the question “Which is the best DFT functional?”, since functionals that give good results for organic compounds may give inferior results for inorganic compounds, and functionals that give good results for energy changes in reactions may give inferior results for activation energies of reactions.
for example
To calculate the electrostatic charges (ZnCl4 ^-2), I first used the following semi-empirical method :
#p opt freq am1 scf=qc
Then from the following command:
#p pbepbe/6-31g(d,p) pop=mk
Good luck
Hello Wenda,
Assuming this isn't a problem with the initial geometry (e.g., symmetry) or the DFT method, you can try distorting the (final) geometry along the same direction as the negative frequency.
What type of motions are the two frequencies?
Meisam Karimi Rad
Hi Meisam,
Thanks for your answer. You are right about setting siutable calculation level before running any simulation. I have checked some literatures and I found that Zn is recognized more like main group elements instead of transition metals in DFT calculation. So the combination of M062x/6-31+G* should work for it. I found that my initial geometry was incorrect so it made the imaginary frequency. After changed the initial geometry, I resolved the imarginary frequency. You example for [ZnCl4]2- DFT calculation is very interesting. I may also try this level and see what the result would be.
Mark James Watkins
Hi Mark,
Thanks for your answer. The two imaginary frequnceies are both bending model. One is up and down and the other one is back and forth. By bended the two Zn-Cl bonds a little bit as initial stage for calculation, the new geometry could reach convergence with positive frequencies.
Dear Wenda,
For rectifying imaginary frequencies, use opt= vertight option in your route section and again reoptimize the structure..
Thanks in advance
Hello Wenda,
In such a case you should always reset your structure by making it more symmetrical in nature or sometimes you need to check the orientation of the structure that could require to be changed. Then again you should optimise your structure. The imaginary frequency means the obtained optimised structure is not the most stable structure on the potential energy surface. Therefore other structures must be tried.
Hello Wenda,
Yes, I thought the two frequencies would be bending modes, and I strongly suspect are degenerate. Saying this, I've just noticed Enrique Arpa's answer which effectively states what I was going to say next.
The two imaginary frequencies are the transition state geometries between the bent (minima) geometries.
At the risk of suggesting what you've already done, this structure should likely be C2v symmetry (just to check it isn't on a C1 potential surface).
Mark James Watkins
Hi Mark,
Many thanks for your explanation. I do learn a lot from yours and Enrique's answers.
The two imarginary frequencies are indeed degenerate bending models that sharing same frequency value. The final optimized geometry is a c2v symmetry just as water but with different angle and bond length.
Thanks for your answer again. :)
Manonmani Gunasekaran
Hi Gunasekaran,
'opt=verytight' was one of the options I tried previously before I came to researchgate for help. As you can see in my description, the geometry that has imaginary frequencies has very small force and displacement values. So lower optimization threshold didn't help. I guess this natural made Gaussian thought it is the optimal geometry and just stopped at this point.
Thanks for Enrique's and Mark's explanation, I found such geometry suppose to be an unstable transition state or minimax energy stage that 'deceived' Gaussian. By changing geomtry to a c2v symmetry geometry, the new optimized structure could reach optimization with a lower energy and also no more imaginary frequencies.
Ankit Kargeti
Hi Ankit,
Thanks for your suggestion. The problem has been resolved by bending the Cl-Zn-Cl bond a little to make it as a C2v symmetry instead of D2h symmetry. At first I thought ZnCl2 would be a linear structure because the optimization in vacuum gave such geometry. In fact the solvation calculation makes it lot more different than expected.
- Badges
- Science topic
- Similar topics
- Biological Science
- Bioecology
- Systems Ecology
- Landscape Ecology
- Connectivity