Metal convergence problem

[sdTurb]

I am currently running DFT calculations on gold contacts with a CRENBS ECP basis set and PBE functional with grid size m3.
Trying to converge the calculation without any hole states is proving difficult.


 When using Fermi smearing i obtain a  negative HOMO-LUMO energy gap with a hole state as below
                                                                          

  
Total energy =   -143.9210604175 H =     -3916.2935434 eV

 

 HOMO-LUMO Separation

   HOMO:   272.   271 a     -0.08983662 H =     -2.44458 eV

   LUMO:   270.   272 a     -0.09064732 H =     -2.46664 eV

   Gap :                    -0.00081070 H =     -0.02206 eV

        !! WARNING: HOMO-LUMO Gap is negativ !!

 

Number of MOs=    542, Electrons=    542.00, Symmetry: c1  

 

   Nr.   Orbital    Occupation       Energy

  287.   287 a                     -0.075128 H =        -2.044 eV

  286.   286 a                     -0.075138 H =        -2.045 eV

  285.   285 a                     -0.079585 H =        -2.166 eV

  284.   284 a                     -0.079626 H =        -2.167 eV

  283.   283 a                     -0.081439 H =        -2.216 eV

  282.   282 a                     -0.081542 H =        -2.219 eV

  281.   281 a                     -0.084276 H =        -2.293 eV

  280.   280 a                     -0.084327 H =        -2.295 eV

  279.   279 a                     -0.084414 H =        -2.297 eV

  278.   278 a                     -0.087521 H =        -2.382 eV

  277.   277 a                     -0.088287 H =        -2.402 eV

  276.   276 a                     -0.088885 H =        -2.419 eV

  275.   275 a                     -0.089466 H =        -2.434 eV

  274.   274 a                     -0.089691 H =        -2.441 eV

  273.   273 a                     -0.089834 H =        -2.445 eV

  272.   271 a        2.000        -0.089837 H =        -2.445 eV

  271.   270 a        2.000        -0.090052 H =        -2.450 eV

  270.   272 a                     -0.090647 H =        -2.467 eV

  269.   269 a        2.000        -0.090737 H =        -2.469 eV

  268.   268 a        2.000        -0.095203 H =        -2.591 eV

  267.   267 a        2.000        -0.096041 H =        -2.613 eV

  266.   266 a        2.000        -0.096473 H =        -2.625 eV

  265.   265 a        2.000        -0.098099 H =        -2.669 eV

  264.   264 a        2.000        -0.099244 H =        -2.701 eV

  263.   263 a        2.000        -0.100215 H =        -2.727 eV

  262.   262 a        2.000        -0.100449 H =        -2.733 eV

  261.   261 a        2.000        -0.100481 H =        -2.734 eV

  260.   260 a        2.000        -0.101488 H =        -2.762 eV

  259.   259 a        2.000        -0.101563 H =        -2.764 eV

  258.   258 a        2.000        -0.103084 H =        -2.805 eV

  257.   257 a        2.000        -0.104116 H =        -2.833 eV

  256.   256 a        2.000        -0.104190 H =        -2.835 eV

  255.   255 a        2.000        -0.105680 H =        -2.876 eV




On another thread i read about the $lastdiag option as a possible solution.
 When i implement the $lastdiag option it eliminates the negative HOMO-LUMO gap but shifts the LUMO orbitals resulting in a large positive HOMO-LUMO gap too big for a metal contact as can be seen below. (I have more severe examples of this even up to 5-6ev gap)
I cant seem to solve this problem with either approach.

Total energy =   -143.9209569997 H =     -3916.2907293 eV

 

 HOMO-LUMO Separation

   HOMO:   271.   271 a     -0.08962911 H =     -2.43893 eV

   LUMO:   272.   272 a     -0.02712896 H =     -0.73822 eV

   Gap :                    +0.06250014 H =     +1.70072 eV

 

Number of MOs=    542, Electrons=    542.00, Symmetry: c1  

 

   Nr.   Orbital    Occupation       Energy

  288.   288 a                     -0.004102 H =        -0.112 eV

  287.   287 a                     -0.011500 H =        -0.313 eV

  286.   286 a                     -0.011515 H =        -0.313 eV

  285.   285 a                     -0.015932 H =        -0.434 eV

  284.   284 a                     -0.015986 H =        -0.435 eV

  283.   283 a                     -0.017821 H =        -0.485 eV

  282.   282 a                     -0.017978 H =        -0.489 eV

  281.   281 a                     -0.020789 H =        -0.566 eV

  280.   280 a                     -0.020849 H =        -0.567 eV

  279.   279 a                     -0.020876 H =        -0.568 eV

  278.   278 a                     -0.023880 H =        -0.650 eV

  277.   277 a                     -0.024737 H =        -0.673 eV

  276.   276 a                     -0.025506 H =        -0.694 eV

  275.   275 a                     -0.026002 H =        -0.708 eV

  274.   274 a                     -0.026009 H =        -0.708 eV

  273.   273 a                     -0.026191 H =        -0.713 eV

  272.   272 a                     -0.027129 H =        -0.738 eV

  271.   271 a        2.000        -0.089629 H =        -2.439 eV

  270.   270 a        2.000        -0.090226 H =        -2.455 eV

  269.   269 a        2.000        -0.090767 H =        -2.470 eV

  268.   268 a        2.000        -0.095126 H =        -2.589 eV

  267.   267 a        2.000        -0.095983 H =        -2.612 eV

  266.   266 a        2.000        -0.096415 H =        -2.624 eV

  265.   265 a        2.000        -0.098040 H =        -2.668 eV

  264.   264 a        2.000        -0.099593 H =        -2.710 eV

  263.   263 a        2.000        -0.100158 H =        -2.725 eV

  262.   262 a        2.000        -0.100397 H =        -2.732 eV

  261.   261 a        2.000        -0.100429 H =        -2.733 eV

  260.   260 a        2.000        -0.101295 H =        -2.756 eV

  259.   259 a        2.000        -0.101624 H =        -2.765 eV

  258.   258 a        2.000        -0.103021 H =        -2.803 eV

  257.   257 a        2.000        -0.104154 H =        -2.834 eV

  256.   256 a        2.000        -0.104294 H =        -2.838 eV




I am using Turbomole 6.1. Any insight would be greatly appreciated.

[uwe]

Hi,

for such a tiny gap the $lastdiag should indeed solve the problem with negative gaps, but you could also consider to increase the gridsize for DFT and/or SCF and density convergence - there are a LOT of unoccupied orbitals with more or less the same energy as the HOMO...

Please note that the 'negative' gap for such small differences is not really a problem but just a numerical artifact - the total energy is correct. It is, however, a hint for a possible multi-reference case. And for TDDFT it is awkward since escf and egrad will complain in such cases.

The really puzzling thing is the huge gap if you use $lastdiag. It looks like an output error - could be that the orbital shift has not been undone in your case. What is the entry of the $scforbitalshift keyword in the control file?

Also, how does your CRENBS ECP basis set look like? I have just checked it in EMSL basis set library and it seems that the basis set is not correct there - just a dummy set for the ECPs and not an optimized one... So if you took CRENBS from there, the results will be very strange anyway.

Regards,

Uwe