Electron transfer between two different conjugated molecules

1. Dear NWchem people,
   

1. I am new to NWchem and I would like to learn more about how to use the electron transfer module. I find some posts on electron transfer but I didn't find then really clear for the specific case of (localized) charge transfer between two different organic molecules.
   
2. What I really when to do is to compute charge transfer integrals (for both positive and negative charges) between two different Pi-conjugated molecules. To do so, we need to localize the charge on a give molecular fragment to generate the reactants and products states required to compute the transfer integral. To illustrate this post let me consider two face to face anthracene molecules. In the second molecule one C-C distance has been slightly increase to get two "different" molecular fragments.
   

1. Lets define a small basis set.
   

* library 6-31g

1. To use the electron transfer module we have first to prepare the reactants and products states. To do so we have to use "vector" to merge together the MOs of the two molecular fragments. First we prepare the MOs of each molecular fragment in the charged and neutral states.
   

uhf             

singlet         

vectors input atom output Ant1_N.mo

uhf             

doublet         

vectors input atom output Ant1_P.mo

uhf             

doublet         

vectors input atom output Ant1_M.mo

uhf             

singlet         

vectors input atom output Ant2_N.mo

uhf             

doublet         

vectors input atom output Ant2_P.mo

uhf             

doublet         

vectors input atom output Ant2_M.mo

1. Then we have to merge together the MOs of the molecular fragments to prepare the reactants and the products states. At this point the problem is that with "vector" we just set the initial guess for the SCF procedure and then a SCF calculation is perform. Therefore, after this new electronic optimization the charge is not anymore localized on one molecule (Am I right?).
   

doublet             

uhf                 

vectors input fragment Ant1_N.mo Ant2_P.mo output Ant1_N_Ant2_P.mo

maxiter 100                                                       

doublet             

uhf                 

vectors input fragment Ant1_P.mo Ant2_N.mo output Ant1_P_Ant2_N.mo

maxiter 100                                                       

doublet             

uhf                 

vectors input fragment Ant1_N.mo Ant2_M.mo output Ant1_N_Ant2_M.mo

maxiter 100                                                       

doublet             

uhf                 

vectors input fragment Ant1_M.mo Ant2_N.mo output Ant1_M_Ant2_N.mo

maxiter 100                                                       

1. Is there a way to contrain the charge to be localized on one molecule? I have tryed the following (set maxiter to 0 in the scf calculation) but then I am in trouble to compute the tranfer integral using the electron transfer module.
   

nopen 1             

print "initial vector analysis"

uhf                            

vectors input fragment Ant1_N.mo Ant2_P.mo output Ant1_N_Ant2_P.mo

maxiter 0

print mulliken

nopen 1

print "initial vector analysis"

uhf

vectors input fragment Ant1_P.mo Ant2_N.mo output Ant1_P_Ant2_N.mo

maxiter 0

print mulliken

1. Once we have these "constrained" reactants and products states I try to calculate the transfer integral as follow:
   

vectors reactants Ant1_N_Ant2_P.mo

vectors products  Ant1_P_Ant2_N.mo

1. Here the problem is the at the end of the SCF calculations for which we set maxiter to 0 the energy of the system is not saved because the calculation doesn't converged (I have added the "ignore" keyword to avoid error messages when the calculation doesn't converged).
   

1. I have also tryed to use the constrained DFT to localized the charge but it seems that the "et" module for cdft desn't have been yet implemented in nwchem.
   

1. Do you think that it is feasible to do what I am trying to do with nwchem ?
   
2. Thank you in advance.
   

1. Julien