From owner-chemistry@ccl.net Tue Apr 30 03:38:02 2019 From: "Muhammed mbtemiz3(_)gmail.com" To: CCL Subject: CCL: Locating shallow local minimum geometries Message-Id: <-53719-190430033650-17759-lpniXhCdMNwtUyzmWHAH3g * server.ccl.net> X-Original-From: Muhammed Content-Type: multipart/alternative; boundary="00000000000070f8690587ba766d" Date: Tue, 30 Apr 2019 10:36:32 +0300 MIME-Version: 1.0 Sent to CCL by: Muhammed [mbtemiz3-$-gmail.com] --00000000000070f8690587ba766d Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable Dear Eric, Thank you for the response. How different are the structures? Relatively small differences are to be > expected. How are you judging this to be a different structure in the > reaction mechanism? Most of the intermediates have small differences. It may be helpful to think of them as different conformers of an enzyme center. I also ran IRCs on the previous TSs to confirm that it is a different point on the PES. Additionally, I have located this minimum using a different functional, M06L, other structures are computed with B3LYP. I have used M06L geometry and re-optimized it with B3LYP. This didn't work either. As a secondary approach, I tried running single point calculation on the M06L geometry with B3LYP, but the energies are not sensible, the intermediate has higher energy than the TSs. I have also checked the electronic structures since the two functionals may collapse to different electronic structures. I now think maybe I should have used a different functional that is similar to B3LYP? If you have a very flat PES and the optimization is indeed taking you away > from the desired minimum, I would recommend adding maxstep=3D1 to your op= t > keywords; i.e. opt=3D(maxstep=3D1,=E2=80=A6,=E2=80=A6). This reduces the= step size to the > minimum value (which is 1/30th of the default step size). I have found > this to be quite helpful in locating stationary points on flat surfaces. > You may need to combine it with those other things you=E2=80=99ve already= tried > (finer grid, calcall, etc). I have tried decreasing maxstep and running with calcall. But no luck there either. Are there any approaches I should be considering? I think that the structure I am looking for should exist theoretically. After all there should be a minimum on the pes connecting TS1, (int) and TS2. When should I conclude I couldn't find this structure? Thank you, Bests. On Tue, Apr 30, 2019 at 12:06 AM Eric V. Patterson eric.patterson]-[ stonybrook.edu wrote: > Dear Muhammed, > > On Apr 29, 2019, at 8:45 AM, Muhammed Buyuktemiz mbtemiz3- -gmail.com < > owner-chemistry(-)ccl.net> wrote: > > But when I re-optimize the geometry I got from IRC > calculations, they collapse to a different structure in the reaction > mechanism. > > > How different are the structures? Relatively small differences are to be > expected. How are you judging this to be a different structure in the > reaction mechanism? > > > Should I just use the geometry found with IRC? > > > No. The structures from IRCs are not fully-optimized stationary points > (except maybe in rare cases when the IRC stops perfectly at the stationar= y > point). > > > I've tried this approach for a different intermediate/TS pair and found > that the energies are not quite the same - there is a ~4 kcal/mol > difference. > > > That is a very small difference, and quite reasonable for the difference > between a =E2=80=9Cfinished=E2=80=9D IRC structure and the actually stati= onary point. > > If you have a very flat PES and the optimization is indeed taking you awa= y > from the desired minimum, I would recommend adding maxstep=3D1 to your op= t > keywords; i.e. opt=3D(maxstep=3D1,=E2=80=A6,=E2=80=A6). This reduces the= step size to the > minimum value (which is 1/30th of the default step size). I have found > this to be quite helpful in locating stationary points on flat surfaces. > You may need to combine it with those other things you=E2=80=99ve already= tried > (finer grid, calcall, etc). > > Cheers, > Eric > > ---------------------------------------------------- > Eric V. Patterson, PhD > Director of Undergraduate Laboratories > Senior Lecturer > > Stony Brook University > Department of Chemistry > 3400 SUNY > Stony Brook, NY 11794-3400 > > 465 Chemistry > eric.patterson(-)stonybrook.edu > https://www.stonybrook.edu/commcms/chemistry/faculty/patterson.eric.html > https://sites.google.com/a/stonybrook.edu/evpatterson > > > voice: (631) 632-7449 > FAX: (631) 632-7960 > > > --=20 Muhammed Buyuktemiz Chemistry Department, Gazi University +90 554 844 11 25 --00000000000070f8690587ba766d Content-Type: text/html; charset="UTF-8" Content-Transfer-Encoding: quoted-printable
Dear Eric,
Thank you for the response.=C2=A0

<= blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-l= eft:1px solid rgb(204,204,204);padding-left:1ex">How different are the stru= ctures?=C2=A0 Relatively small differences are to be expected. How are you = judging this to be a different structure in the reaction mechanism?
=C2=A0
Most of the intermediates have small differences. It = may be helpful to think of them as different conformers of an enzyme center= . I also ran IRCs on the previous TSs to confirm that it is a different poi= nt on the PES.

Additionally, I have located this minimum using a di= fferent functional, M06L, other structures are computed with B3LYP. =C2=A0<= br>I have used M06L geometry and re-optimized it with B3LYP. This didn'= t work either.
As a secondary approach, I tried running single point ca= lculation on the M06L geometry with B3LYP, but the energies are not sensibl= e, the intermediate has higher energy than the TSs. I have also checked the= electronic structures since the two functionals may collapse to different = electronic structures.

I now think maybe I should have used a differ= ent functional that is similar to B3LYP?


=C2=A0If you have a very flat PES and the optimi= zation is indeed taking you away from the desired minimum, I would recommen= d adding maxstep=3D1 to your opt keywords; i.e. opt=3D(maxstep=3D1,=E2=80= =A6,=E2=80=A6).=C2=A0 This reduces the step size to the minimum value (whic= h is 1/30th of the default step size).=C2=A0 I have found this to be quite = helpful in locating stationary points on flat surfaces.=C2=A0 You may need = to combine it with those other things you=E2=80=99ve already tried (finer g= rid, calcall, etc).
=C2=A0
I have tried decreasing ma= xstep and running with calcall. But no luck there either.

Are there= any approaches I should be considering?
I think that the structure I a= m looking for should exist theoretically. After all there should be a minim= um on the pes connecting TS1, (int) and TS2.

When should I conclude= I couldn't find this structure?

Thank you,
Bests. =C2=A0

On Tue, Apr 30, 2019 at 12:06 AM Eric V. Patterson eric.patterson]-[stonybrook.edu <owner-chemistry]_[ccl.net> wrote:
Dear Muhammed,

=
On Apr 29, 2019, at 8:45 AM, Muha= mmed Buyuktemiz mbtemiz3- -g= mail.com <owner-chemistry(-)ccl.net> wrote:

=C2=A0But when I re-optimize the geometry I got from IRC=C2=A0<= /span>
c= alculations, they collapse to a different structure in the reaction=C2=A0
mechan= ism.

How different are the structu= res?=C2=A0 Relatively small differences are to be expected.=C2=A0 How are y= ou judging this to be a different structure in the reaction mechanism?


Should I just use the geometry found with IRC?= =C2=A0

No.=C2=A0 Th= e structures from IRCs are not fully-optimized stationary points (except ma= ybe in rare cases when the IRC stops perfectly at the stationary point).


I've tried this approach for a different= intermediate/TS pair and found=C2=A0
that the energies are not quite the same -= there is a ~4 kcal/mol=C2=A0
difference. =C2=A0

That is a very small difference, and quite reasonable for the diff= erence between a =E2=80=9Cfinished=E2=80=9D IRC structure and the actually = stationary point.

If you have a very flat PES and = the optimization is indeed taking you away from the desired minimum, I woul= d recommend adding maxstep=3D1 to your opt keywords; i.e. opt=3D(maxstep=3D= 1,=E2=80=A6,=E2=80=A6).=C2=A0 This reduces the step size to the minimum val= ue (which is 1/30th of the default step size).=C2=A0 I have found this to b= e quite helpful in locating stationary points on flat surfaces.=C2=A0 You m= ay need to combine it with those other things you=E2=80=99ve already tried = (finer grid, calcall, etc).

Cheers,
Eric=

----------------------------------------------------
Eric V. Patt= erson, PhD
Director of Undergraduate Laboratories
Senior Lecturer

Stony Broo= k University
Department of Chemistry
3400 SUNY
Stony =C2=A0Brook, = NY 11794-3400

465 Chemistry
eric.patterson(-)stonybrook.edu<= /div>

voice: (631) 632-7449
FAX: (631) 632-79= 60




--
Muhammed Buyuktemiz
Chemistry Department,=C2=A0Gazi University=C2=A0
+90 554 844 11 2= 5=C2=A0
--00000000000070f8690587ba766d-- From owner-chemistry@ccl.net Tue Apr 30 09:28:00 2019 From: "Eric V. Patterson eric.patterson!^!stonybrook.edu" To: CCL Subject: CCL: Locating shallow local minimum geometries Message-Id: <-53720-190430092640-17116-hH5AkmdZw1b344wN/CpgZQ^server.ccl.net> X-Original-From: "Eric V. Patterson" Content-Transfer-Encoding: 8bit Content-Type: text/plain; charset=utf-8 Date: Tue, 30 Apr 2019 09:26:33 -0400 Mime-Version: 1.0 (Mac OS X Mail 12.4 \(3445.104.8\)) Sent to CCL by: "Eric V. Patterson" [eric.patterson^^^stonybrook.edu] Dear Muhammed, > On Apr 30, 2019, at 3:36 AM, Muhammed mbtemiz3(_)gmail.com wrote: > > Most of the intermediates have small differences. It may be helpful to think of them as different conformers of an enzyme center. I also ran IRCs on the previous TSs to confirm that it is a different point on the PES. Ah. An IRC will not follow a conformational change unless the barrier is very, very low. There will be a separate TS for each conformational change. If the barrier for each conformational interconversion is small, they won’t matter in the overall analysis. Of course, the trick is knowing for sure that the barrier is small… I think, however, that is often a reasonable assumption. In that case, the important points of the PES might be defined as follows: TS1 —> INT (conformer 1 by IRC from TS1) [<— unexplored interconversion —>] INT (conformer 2 by IRC from TS2) —> TS2. > > Additionally, I have located this minimum using a different functional, M06L, other structures are computed with B3LYP. > I have used M06L geometry and re-optimized it with B3LYP. This didn't work either. > As a secondary approach, I tried running single point calculation on the M06L geometry with B3LYP, but the energies are not sensible, the intermediate has higher energy than the TSs. I have also checked the electronic structures since the two functionals may collapse to different electronic structures. > > I now think maybe I should have used a different functional that is similar to B3LYP? I recommend exactly the opposite. B3LYP has many well-known flaws. M06L probably gives the more reliable PES. In the end, all structures should be optimized with the same level of theory. Single-point energies will give you a good idea of how sensitive the PES is to changing the theory. > I have tried decreasing maxstep and running with calcall. But no luck there either. > > Are there any approaches I should be considering? > I think that the structure I am looking for should exist theoretically. After all there should be a minimum on the pes connecting TS1, (int) and TS2. > > When should I conclude I couldn't find this structure? > After you have used one or more of the good, modern density functionals with a suitably-large basis set. You haven’t mentioned solvent, so make sure you are also modeling that appropriately. It is possible that the mechanism is concerted rather than two steps. It is possible that the PES is highly sensitive to the level of theory and a stationary point that exists at one level is not found at another (definitely seems to be the case here). It is possible that a single-reference method is inappropriate for your transformation. It is possible that the Born-Oppenheimer approximation breaks down for one reason or another for your reaction. In other words, there are many things that you might still need to check before you can conclude you have modeled the correct mechanism. Cheers, Eric