## Firefly and PC GAMESS-related discussion club

Learn how to ask questions correctly

Re: IRC run: please, explain how to get result

Alex Granovsky
gran@classic.chem.msu.su

Hello,

most likely you have some problems with the geometry of TS or its
hessian, or with the atomic numbering scheme used in hessian and
in other runs.

Below, I have attached archive with four input and output files for
ammonia, which I hope will be of some help to you.

First, the geometry optimization was performed in C3v group.
For simplicity, to impose planarity on TS I used cartesian coordinates
for optimization, with Z coordinates of all atoms frozen.
The more straightforward way to do this would be just to run

Second, the hessian was computed at the found stationary point using
C1 group, because I wanted to use C1 group also for IRC. (Technically
it is possible to compute hessian using higher group to speed things
up but then one needs to follow the atom numbering scheme dictated
by this group. Otherwise atomic coordinates and hessian would be
inconsistent.) Unlike you, I did not find any large "zero" frequencies.

Third and fourth, I extracted the computed hessian and ran IRC job in
two directions leaving most of the parameters at their defaults.
Both IRC jobs terminated at the local minimum and voila.

Hope this helps.

Kind regards,
Alex Granovsky

On Wed Aug 19 '15 3:38am, olgerdovich wrote
-------------------------------------------
>Dear colleagues,

>To get some practice in tracing reaction path from TS to products, I tried some IRC-runs on one of the simplest model reaction, ammonia inversion, but I faced with problem even in this very simple case. Unfortunately, manual gave me no hints, so I ask you to instruct me how to do it in right way.

>First, I optimized flat NH3-molecule (MP2/6-311++G**) by freezing dihedral at 180° in z-matrix coordinate input. It gave me flat (D3h) NH3 with r(N-H) = 0.99890 A with one imaginary frequency (850.33 I), six rotational/translation freedom degrees (five not exceeding 18 cm-1 and one rather high, 125 cm-1), and five strong vibrations (1590...3910 cm-1)

>Second, I put this geometry as starting and start IRC run by changing RUNTYP for IRC, adding "\$irc forwrd=.t. npoint=1000 MXOPt = 1000 \$end" string and \$HESS group from flat molecule optimization run, leaving all the other parameters constant, and problems began

>I was forced to add MXOPT explicitly to defaults since otherwise firefly stopped at the second IRC-iteration (TOO MANY STEPS TAKEN IN CONSTRAINED OPTIMIZATION). Being thus unleashed, Firefly found another reason to stop at the same second iteration after 226 steps, claiming:

> RADIUS IN CIRCLE OPTIMIZATION 0.1860667 DEVIATES SIGNIFICANTLY FROM CONSTRAIN CONDITION 0.1500000
> IT IS POSSIBLE THAT THE NEXT IRC POINT IS CLOSE TO A MINIMUM

>(actually, it was still very far from pyramidal ammonia)

>adding "pace=linear stride=0.10" to \$IRC group rapidly (just on POINT   1 ON THE REACTION PATH) resulted in
> YOU ARE NO LONGER ON A STEEPEST DESCENT PATH!
> THIS IRC SHOULD BE RUN WITH SMALLER STEPS AND/OR A BETTER INTEGRATION METHOD.

>and lowering STRIDE to stride=0.01 and adding elbow=180 changed nothing, as well as switching to pace=AMPC4 or pace = RK4 .

>Here is one of the tested inputs:

> \$BASIS GBASIS=N311 NGAUSS=6 NDFUNC=1 NPFUNC=1 DIFFSP=.T. DIFFS=.T. \$END
> \$CONTRL
>  SCFTYP=RHF RUNTYP=IRC maxit=3000 ICHARG=0 MULT=1 MPLEVL = 2
> \$END
> \$CONTRL INTTYP=HONDO, ICUT=11, ITOL=30 \$END
> \$SCF dirscf=.t. FDIFF=.FALSE. DIIS = .T. SOSCF = .F. \$END
> \$STATPT OPTTOL=0.00001 NSTEP=100 HSSEND=.TRUE. \$END
> \$irc forwrd=.t. pace=RK4 stride=0.01 npoint=1000 MXOPt = 1000 \$end

> \$DATA
> NH3
>C1
>NITROGEN    7.0      0.000000175        -0.000000255        -0.000000001
>HYDROGEN    1.0      0.865069400         0.499451188         0.000000006
>HYDROGEN    1.0     -0.865070520         0.499447858         0.000000006
>HYDROGEN    1.0     -0.000001304        -0.998895509         0.000000006
> \$END