Author Topic: Negative excitation energies in ADC(2)  (Read 5580 times)

dcav

  • Newbie
  • *
  • Posts: 2
  • Karma: +0/-0
Negative excitation energies in ADC(2)
« on: January 18, 2019, 07:51:58 PM »
Dear all,
    I am trying to obtain a relaxed scan of furan and some other molecules in the S1 state, using sos-ri-adc(2)/def2-tzvp, and increasing C-O distance. Several of the initial geometries appear to approach conical intersections well documented in the literature. However, a couple of these geometries have a negative S1 excitation energy. How should I interpret this? Is this calculation still reliable?

   Here is a fragment of the control file for the ricc2 calculation:

$denconv     0.10000000E-06
$freeze
 implicit core=    5 virt=    0
$cbas    file=auxbasis
$ricc2
  adc(2)
  geoopt model=adc(2)    state=(a 1)
  sos   cos= 1.30000
  maxiter=60
$excitations
  irrep=a  multiplicity=  1  nexc=  4  npre=  5  nstart=  6
$closed shells
 a       1-18                                   ( 2 )


   Also, some of the output of ricc2.

 +==================================================================+
 | sym | multi | state |          CCS excitation energies           |
 |     |       |       +--------------------------------------------+
 |     |       |       |   Hartree    |     eV       |     cm-1     |
 +==================================================================+
 | a   |   1   |   1   |    0.0077019 |      0.20958 |     1690.370 |
 | a   |   1   |   2   |    0.1500643 |      4.08346 |    32935.306 |
 | a   |   1   |   3   |    0.1510398 |      4.11000 |    33149.398 |
 | a   |   1   |   4   |    0.2295911 |      6.24749 |    50389.415 |
 | a   |   1   |   5   |    0.2469002 |      6.71850 |    54188.332 |
 +==================================================================+
 

+================================================================================+
 | sym | multi | state |          ADC(2) excitation energies    |  %t1   |  %t2   |
 |     |       |       +----------------------------------------+--------+--------+
 |     |       |       |   Hartree    |    eV      |    cm-1    |    %   |    %   |
 +================================================================================+
 | a   |   1   |   1   |   -0.0086017 |   -0.23406 |  -1887.848 |  89.97 |  10.03 |
 | a   |   1   |   2   |    0.0606178 |    1.64949 |  13304.069 |  82.25 |  17.75 |
 | a   |   1   |   3   |    0.1137216 |    3.09452 |  24959.007 |  90.70 |   9.30 |
 | a   |   1   |   4   |    0.1630437 |    4.43665 |  35783.961 |  87.32 |  12.68 |
 +================================================================================+

   Thanks in advance for any help,

   Daniel.

Arnim

  • Developers
  • Sr. Member
  • *
  • Posts: 253
  • Karma: +0/-0
Re: Negative excitation energies in ADC(2)
« Reply #1 on: January 21, 2019, 10:12:28 AM »
Hello Daniel,

it could be that the ground state of these geometries can not be decribed with a single-reference wave function. Then you would have to apply multi-reference methods to sudy them.
However, having not looked at the systems in detail, this is only a guess.

Cheers,

Arnim

christof.haettig

  • Global Moderator
  • Sr. Member
  • *****
  • Posts: 291
  • Karma: +0/-0
    • Hattig's Group at the RUB
Re: Negative excitation energies in ADC(2)
« Reply #2 on: November 18, 2019, 11:58:07 AM »
Arnim is right. The CCS calculations shows that the system is close to a singlet instability or even has a singlet instability.
This is not unusual in excited state geometry optimizations: If you break along the relaxation pathway a bond, the ground state energy increases a lot and the gap between the excited and the ground states gets small.

The CI between an excited and the ground state can not be described with response methods. For that you have to go the multireference methods.
If you are not interested in the details of the CI, but only on its approximate position you can ignore the problem.
To model the deactivation after reaching the CI you can continue the geometry optimization (or IRC or DRC run) on the ground state potential energy surface (for ADC(2) that is MP2).