Cannot get rid of Imaginary Frequency

[Hridita_Purba_Saha]

Hi, I am trying to do geometry optimization and frequency calculation of aniline and could not get rid of the imaginary frequency part. I have tried with higher grid but it’s still there. Could anyone help me with this?
$molecule
0 1
H -0.0000000 -2.1476227 1.7123822
C -0.0000000 -1.2107108 1.1704369
C -0.0000000 0.0000000 1.8693220
H -0.0000000 0.0000000 2.9516701
C 0.0000000 1.2107108 1.1704369
H 0.0000000 2.1476227 1.7123822
C 0.0000000 1.2125266 -0.2282863
H 0.0000000 2.1559776 -0.7596931
C 0.0000000 -0.0000000 -0.9354220
C -0.0000000 -1.2125266 -0.2282863
H -0.0000000 -2.1559776 -0.7596931
N 0.0000000 -0.0000000 -2.3519793
H 0.0000000 0.8862698 -2.8778829
H 0.0000000 -0.8862698 -2.8778829
$end

$rem
BASIS = 6-31G(d,p)
GUI = 2
JOB_TYPE = Optimization
METHOD = CAMB3LYP
SCF_CONVERGENCE = 8
XC_GRID = 2
SOLVENT_METHOD = PCM
SYMMETRY = FALSE
SYM_IGNORE = TRUE
MEM_TOTAL = 248000
MEM_STATIC = 4000
$end

$pcm
THEORY IEFPCM
$end

$solvent
DIELECTRIC 35.688
OPTICALDIELECTRIC 1.8068
$end

@@@

$molecule
read
$end

$rem
BASIS = 6-31G(d,p)
GUI = 2
JOB_TYPE = Frequency
METHOD = CAMB3LYP
SCF_CONVERGENCE = 8
SOLVENT_METHOD = PCM
IDERIV = 1
XC_GRID = 2
MEM_TOTAL = 248000
MEM_STATIC = 4000
SYMMETRY = FALSE
SYM_IGNORE = TRUE
$end

[jherbert]

These are occasionally grid artifacts but that’s relatively rare because the default grids are reasonable. Better suggestion is to tighten the optimization convergence criteria. For this and other suggestions, see

How to troubleshoot artifical negative frequencies

I am doing rather routine calculation: TD-DFT, S1 state, optimization and frequencies. Everything converges smoothly, but then I end up with one imaginary frequency, which I am sure is an artefact. I tried several things at random with no effect. Question to the experts: What are standard steps of troubleshooting this sort of a problem? What are the first thing to tighten in terms of thresholds/convergence/grids? Here is my basic input, using defaults for everything: Blockquote $comment OPT&…

It can also be helpful to manually displace the system along the negative-curvature mode, using the normal mode information that is printed by the frequency job.

[Hridita_Purba_Saha]

Hi @jherbert,

I was following the workflow you suggested in the earlier discussion—running a frequency calculation first, then performing a geometry optimization using GEOM_OPT_HESSIAN = READ, and finally running another frequency calculation at the optimized structure. But still I am getting one imaginary freq.

This is my input file:
$molecule
0 1
H -0.0000000 -2.1476227 1.7123822
C -0.0000000 -1.2107108 1.1704369
C -0.0000000 0.0000000 1.8693220
H -0.0000000 0.0000000 2.9516701
C 0.0000000 1.2107108 1.1704369
H 0.0000000 2.1476227 1.7123822
C 0.0000000 1.2125266 -0.2282863
H 0.0000000 2.1559776 -0.7596931
C 0.0000000 -0.0000000 -0.9354220
C -0.0000000 -1.2125266 -0.2282863
H -0.0000000 -2.1559776 -0.7596931
N 0.0000000 -0.0000000 -2.3519793
H 0.0000000 0.8862698 -2.8778829
H 0.0000000 -0.8862698 -2.8778829
$end

$rem
BASIS = 6-31G(d,p)
GUI = 2
JOB_TYPE = Frequency
METHOD = CAMB3LYP
SCF_CONVERGENCE = 8
SCF_MAX_CYCLES = 100
SOLVENT_METHOD = PCM
SYMMETRY = FALSE
SYM_IGNORE = TRUE
MEM_TOTAL = 248000
MEM_STATIC = 4000
$end

$pcm
THEORY IEFPCM
$end

$solvent
DIELECTRIC 35.688
OPTICALDIELECTRIC 1.8068
$end

$molecule
read
$end

$rem
BASIS = 6-31G(d,p)
GUI = 2
JOB_TYPE = Optimization
GEOM_OPT_HESSIAN = READ
GEOM_OPT_MAX_CYCLES = 1000
METHOD = CAMB3LYP
SCF_CONVERGENCE = 8
SCF_MAX_CYCLES = 100
SOLVENT_METHOD = PCM
SYMMETRY = FALSE
SYM_IGNORE = TRUE
MEM_TOTAL = 248000
MEM_STATIC = 4000
$end

$pcm
THEORY IEFPCM
$end

$solvent
DIELECTRIC 35.688
OPTICALDIELECTRIC 1.8068
$end

@@@

$molecule
read
$end

$rem
BASIS = 6-31G(d,p)
GUI = 2
JOB_TYPE = Frequency
METHOD = CAMB3LYP
SCF_CONVERGENCE = 8
SCF_MAX_CYCLES = 100
SOLVENT_METHOD = PCM
MEM_TOTAL = 248000
MEM_STATIC = 4000
SYMMETRY = FALSE
SYM_IGNORE = TRUE
$end

$pcm
THEORY IEFPCM
$end

$solvent
DIELECTRIC 35.688
OPTICALDIELECTRIC 1.8068
$end

[jherbert]

Did you try the other suggestions?
(1) tighten up the geom opt stopping criteria
(2) displace along the negative-curvature mode