Error while running frequency calculation

Questions
#1

Hi everyone,

I am running an Opt + Frequency calculation in Q-Chem, but the frequency step fails with the following error:

Direct stationary perturbation theory relativistic correction:



  ***ERROR*** Hessian Appears to have all zero or negative eigenvalues

 Q-Chem fatal error occurred in module /home/data/root/qchem6.4/libopt/optexit.F, line    9:

 OPTIMIZE fatal error

input:

$molecule
0 1
  C   -0.9623647    1.5781618   -0.0410529
  C    0.4581597    1.0619433   -0.0153521
  C    0.5442678   -0.4448678    0.0247855
  C   -0.6764020   -1.2762843    0.0360213
  C   -1.8854054   -0.7038591    0.0106625
  C   -2.0318571    0.7662997   -0.0290223
  H   -3.0288145    1.1952117   -0.0488012
  H   -1.1189404    2.6521662   -0.0704721
  H    0.9663238    1.4739760    0.8819976
  H    0.9818445    1.4267525   -0.9241461
  H   -0.5937187   -2.3588371    0.0650279
  H   -2.7750478   -1.3256808    0.0192648
  O    1.6391574   -0.9844674    0.0483111           
$end

$rem
JOBTYPE Opt
method bhhlyp
BASIS aug-cc-pVTZ
scf_max_cycles  600
geom_opt_max_cycles     600
$end

@@@

$molecule
read
$end

$rem
JOBTYPE frequency
method bhhlyp
BASIS aug-cc-pVTZ
$end
#2

If you want someone to take a look at this, please post the optimized geometry so that we don’t have to run the whole optimization to get it.

#3

Please find below optimized coordinates:

******************************
	**  OPTIMIZATION CONVERGED  **
	******************************
 ----------------------------------------------------------------
             Standard Nuclear Orientation (Angstroms)
    I     Atom           X                Y                Z
 ----------------------------------------------------------------
    1      C       1.1919548763     1.1509006480    -0.0000023516
    2      C      -0.2900542472     1.2656941525     0.0000020738
    3      C      -1.0676089987    -0.0327755872    -0.0000100274
    4      C      -0.2975011773    -1.2746490845     0.0000023236
    5      C       1.0354191870    -1.2570651448     0.0000010565
    6      C       1.8024543264    -0.0249502030    -0.0000039407
    7      H       2.8742459597    -0.0890023437    -0.0000055912
    8      H       1.7616463135     2.0627508619    -0.0000022637
    9      H      -0.6235134410     1.8406038817    -0.8604153977
   10      H      -0.6235078384     1.8405864518     0.8604337758
   11      H      -0.8587356428    -2.1897946278     0.0000077012
   12      H       1.5775374045    -2.1860365025     0.0000045786
   13      O      -2.2710841173    -0.0249484903     0.0000057512
#4

I can reproduce this with Q-Chem v. 6.4 It appears to be a bug related to the new integrals code. A workaround is to turn that off. Set:

use_libqints false
mem_total 110000

(The 2nd one is because the code told me I needed more memory when I turned libqints off.) I will submit this bug to Q-Chem.

UPDATE: It turns out that it’s a memory issue rather than a libqints issue per se, it’s just that the old integrals code asks you for more memory whereas the new code just crashes. Adding

mem_total 110000

was enough to get this job to run for me.