From owner-chemistry@ccl.net Thu Aug 27 04:12:01 2015 From: "Adedapo Adeyinka u11335132!A!tuks.co.za" To: CCL Subject: CCL: Empirical Dispersion Correction Message-Id: <-51624-150827040850-15081-NFaZS08/glcTn1Y+TXyF/Q-,-server.ccl.net> X-Original-From: Adedapo Adeyinka Content-Type: multipart/alternative; boundary=089e013d1d5ec3a8f4051e467979 Date: Thu, 27 Aug 2015 10:08:42 +0200 MIME-Version: 1.0 Sent to CCL by: Adedapo Adeyinka [u11335132++tuks.co.za] --089e013d1d5ec3a8f4051e467979 Content-Type: text/plain; charset=UTF-8 Dear all, I am trying to predict the protonation constants of some amine compounds using DFT methods. In order to describe solvation properly, I have used a number of explicit water molecules to represent the first solvation shell around the solute, then embedded this in the PCM continuum solvation model). Also I have used B3LYP, B3LYP-D3 and B97D functionals (and the 6-311++Gdp basis set) to carry out my structure optimizations since the importance of including dispersion corrections has been emphasized a lot in literature recently. Contrary to my expectations, I got the best predicted protonation constants with the B3LYP functional followed by B97D. The results that deviated most from experiment was that obtained with the B3LYP-D3. I have tried to find out why this was the case and so far the only lead I have got is that the empirical dispersion correction was designed to yield accurate isolated molecule energies and geometries. Does anyone know why my results followed this trend? Does this mean that since I have used explicit water molecules there is no need for dispersion correction to be included with the B3LYP functional? Thanks in advance for your contributions. -- Adedapo Adeyinka -- This message and attachments are subject to a disclaimer. Please refer to http://www.it.up.ac.za/documentation/governance/disclaimer/ for full details. --089e013d1d5ec3a8f4051e467979 Content-Type: text/html; charset=UTF-8 Content-Transfer-Encoding: quoted-printable
Dear all,
=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 = =C2=A0 I am trying to predict the protonation constants of some amine compo= unds using DFT methods. In order to describe solvation properly, I have use= d a number of explicit water molecules to represent the first solvation she= ll around the solute, then embedded this in the PCM continuum solvation mod= el). Also I have used B3LYP, B3LYP-D3 and B97D functionals (and the 6-311++= Gdp basis set) to carry out my structure optimizations since the importance= of including dispersion corrections has been emphasized a lot in literatur= e recently. Contrary to my expectations, I got the best predicted protonati= on constants with the B3LYP functional followed by B97D. The results that d= eviated most from experiment was that obtained with the B3LYP-D3.=C2=A0
I have tried to find out why this was the case and so far the only l= ead I have got is that the empirical dispersion correction was designed to = yield accurate isolated molecule energies and geometries. Does anyone know = why my results followed this trend? Does this mean that since I have used e= xplicit water molecules there is no need for dispersion correction to be in= cluded with the B3LYP functional?

Thanks in advanc= e for your contributions.

--
Adedapo Adeyinka


This message and attachments are subject to= a disclaimer. Please refer to=20 http://www.it.u= p.ac.za/documentation/governance/disclaimer/ for full details. --089e013d1d5ec3a8f4051e467979-- From owner-chemistry@ccl.net Thu Aug 27 06:42:00 2015 From: "Ramon Crehuet rcsqtc#,#iqac.csic.es" To: CCL Subject: CCL:G: H-bond distance with different solvation models Message-Id: <-51625-150827064015-16586-Smh/hLSx4idhQLvyMrxJhA : server.ccl.net> X-Original-From: Ramon Crehuet Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=utf-8; format=flowed Date: Thu, 27 Aug 2015 12:39:57 +0200 MIME-Version: 1.0 Sent to CCL by: Ramon Crehuet [rcsqtc_+_iqac.csic.es] Dear all, I would like to study the dependence of an H-bond distance with the dielectric constant of water (which depends on temperature, pressure, etc). My system consists of a water molecule and a small molecue containg a donating NH group. I have run the same calculations with Gaussian, using the PCM model and with Orca, using the Cosmo model, obtaining different results. For an epsilon range of 80-95, the Cosmo in Orca changes the H-bond from 1.85A to 1.95A. The same model with Gaussian PCM gives a distance that is constant to 3 significant digits: 1.897A. Gaussian seems to be correctly reading the epsion, as the log file states: Solvent : Water, Eps= 95.000000 Eps(inf)= 1.777849 Being electrostatic in character, I would expect the H-bond distance to depend on the dielectric constant. Beyond that, are there any reasons to prefer one of the methods? Thanks in advance, Ramon -- Ramon Crehuet Cientific Titular (Assistant Professor) Institute of Advanced Chemistry of Catalonia IQAC - CSIC http://www.iqac.csic.es/qteor https://twitter.com/rcrehuet http://ramoncrehuet.wordpress.com/ Tel. +34 934006116 Jordi Girona 18-26 08034 Barcelona (Spain) From owner-chemistry@ccl.net Thu Aug 27 07:36:01 2015 From: "Stefan Grimme grimme : thch.uni-bonn.de" To: CCL Subject: CCL: Empirical Dispersion Correction Message-Id: <-51626-150827060524-13640-+soOQL+WlkHqDur064wvRA,,server.ccl.net> X-Original-From: "Stefan Grimme" Date: Thu, 27 Aug 2015 06:05:22 -0400 Sent to CCL by: "Stefan Grimme" [grimme##thch.uni-bonn.de] Dear Adedapo, dispersion corrections to B3LYP are essential even in "non-obvious" cases because they correct for fundamental correlation energy deficiencies in the functional at medium- as well as long interatomic distances. This is not directly related to the issue of explicit solvation. In order to properly compute protonation constants (I assume a free energy property) its essential that you inclcude ALL free energy contributions (internal as well as solvation) accurately. In most (electronically simple) cases when B3LYP is better than B3LYP-D3 there is a hidden error compensation and the good result you get is based on the wrong reason. In summary: forget B3LYP, use always B3LYP-D3 or other dispersion corrected versions of B3LYP (or even more modern dispersion-inclusive functionals). Hope this helps Stefan From owner-chemistry@ccl.net Thu Aug 27 10:57:01 2015 From: "Andreas Klamt klamt-*-cosmologic.de" To: CCL Subject: CCL:G: H-bond distance with different solvation models Message-Id: <-51627-150827100924-12716-1hLxVvmIaQs7G2aNu4ygWg]-[server.ccl.net> X-Original-From: Andreas Klamt Content-Transfer-Encoding: 8bit Content-Type: text/plain; charset=utf-8 Date: Thu, 27 Aug 2015 16:09:18 +0200 MIME-Version: 1.0 Sent to CCL by: Andreas Klamt [klamt+*+cosmologic.de] Dear Ramon, basically one should expect a dependence on the dielectric constant, but I would not expect a significant change from eps=80 to 95. Hence, let me say, I am surprised by the COSMO result. Perhaps contact the ORCA team, send them your input and let them check whether you made any mistake. But there cannot be a major preference between PCM or COSMO. Depending on which PCM variant you have chosen, IEFPCM or CPCM, you may already have used essentially COSMO. And even IEFPCM is almost indistinguishable > from COSMO (see our recent paper: A Comprehensive Comparison of the IEFPCM and SS(V)PE Continuum Solvation Methods with the COSMO Approach http://pubs.acs.org/doi/abs/10.1021/acs.jctc.5b00601 ) Major difference may only arise from the different cavity constructions chosen in the different implementations. But again, let me point out: The change by 0.1 A when changig eps from 80 to 95 is very strange! Andreas Am 27.08.2015 um 12:39 schrieb Ramon Crehuet rcsqtc#,#iqac.csic.es: > > Sent to CCL by: Ramon Crehuet [rcsqtc_+_iqac.csic.es] > Dear all, > I would like to study the dependence of an H-bond distance with the > dielectric constant of water (which depends on temperature, pressure, > etc). My system consists of a water molecule and a small molecue > containg a donating NH group. I have run the same calculations with > Gaussian, using the PCM model and with Orca, using the Cosmo model, > obtaining different results. > For an epsilon range of 80-95, the Cosmo in Orca changes the H-bond from > 1.85A to 1.95A. The same model with Gaussian PCM gives a distance that > is constant to 3 significant digits: 1.897A. > Gaussian seems to be correctly reading the epsion, as the log file states: > > Solvent : Water, Eps= 95.000000 Eps(inf)= 1.777849 > > Being electrostatic in character, I would expect the H-bond distance to > depend on the dielectric constant. Beyond that, are there any reasons to > prefer one of the methods? > > Thanks in advance, > > Ramon > -- -------------------------------------------------- Prof. Dr. Andreas Klamt CEO / Geschäftsführer COSMOlogic GmbH & Co. KG Imbacher Weg 46 D-51379 Leverkusen, Germany phone +49-2171-731681 fax +49-2171-731689 e-mail klamt~~cosmologic.de web www.cosmologic.de [University address: Inst. of Physical and Theoretical Chemistry, University of Regensburg] HRA 20653 Amtsgericht Koeln, GF: Prof. Dr. Andreas Klamt Komplementaer: COSMOlogic Verwaltungs GmbH HRB 49501 Amtsgericht Koeln, GF: Prof. Dr. Andreas Klamt From owner-chemistry@ccl.net Thu Aug 27 11:32:00 2015 From: "Adedapo Adeyinka u11335132|a|tuks.co.za" To: CCL Subject: CCL: Empirical Dispersion Correction Message-Id: <-51628-150827100352-2455-JjU6Jq1y6KP6D0u2BCVKlg^^^server.ccl.net> X-Original-From: Adedapo Adeyinka Content-Type: multipart/alternative; boundary=001a11c3711a45e52c051e4b6fb8 Date: Thu, 27 Aug 2015 16:03:41 +0200 MIME-Version: 1.0 Sent to CCL by: Adedapo Adeyinka [u11335132 ~ tuks.co.za] --001a11c3711a45e52c051e4b6fb8 Content-Type: text/plain; charset=UTF-8 Dear Stefan, Thanks a lot for your explanation. Could you please shed more light on the likely sources (technical or personal) of this hidden error compensation that could have led to my obtaining good results with B3LYP as against that of B3LYP-D3? And is there any step I can take computationally to correct for this error or this is just system dependent? Once again, thank you very much for your help. On 27 August 2015 at 12:05, Stefan Grimme grimme : thch.uni-bonn.de < owner-chemistry.:.ccl.net> wrote: > > Sent to CCL by: "Stefan Grimme" [grimme##thch.uni-bonn.de] > Dear Adedapo, > dispersion corrections to B3LYP are essential even in "non-obvious" > cases because they correct for fundamental correlation energy deficiencies > in the functional at medium- as well as long interatomic distances. > This is not directly related to the issue of explicit solvation. > > In order to properly compute protonation constants (I assume a free > energy property) its essential that you inclcude ALL free energy > contributions (internal as well as solvation) accurately. > > In most (electronically simple) cases when B3LYP is better than > B3LYP-D3 there is a hidden > error compensation and the good result you get is based on the > wrong reason. In summary: forget B3LYP, use always B3LYP-D3 or other > dispersion corrected versions of B3LYP (or even more modern > dispersion-inclusive functionals). > > Hope this helps > Stefan> > > -- Adedapo Adeyinka -- This message and attachments are subject to a disclaimer. Please refer to http://www.it.up.ac.za/documentation/governance/disclaimer/ for full details. --001a11c3711a45e52c051e4b6fb8 Content-Type: text/html; charset=UTF-8 Content-Transfer-Encoding: quoted-printable
Dear Stefan,
=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0= =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0Thanks a lot for your explanation. Could= you please shed more light on the likely sources (technical or personal) o= f this hidden error compensation that could have led to my obtaining good r= esults with B3LYP as against that of B3LYP-D3? And is there any step I can = take computationally to correct for this error or this is just system depen= dent?

Once again, thank you very much for your hel= p.




On 27 August 2015 at 12:05, Stefa= n Grimme grimme : thch.uni-bonn.de = <owner-chemistry.:.ccl.net> wrote:

Sent to CCL by: "Stefan=C2=A0 Grimme" [grimme##thch.uni-bonn.de= ]
Dear Adedapo,
dispersion corrections to B3LYP are essential even in "non-obvious&quo= t;
cases because they correct for fundamental correlation energy deficiencies<= br> in the functional at medium- as well as long interatomic distances.
This is not directly related to the issue of explicit solvation.

In order to properly compute protonation constants (I assume a free
energy property) its essential that you inclcude ALL free energy
contributions (internal as well as solvation) accurately.

In most (electronically simple) cases when B3LYP is better than
B3LYP-D3 there is a hidden
error compensation and the good result you get is based on the
wrong reason. In summary: forget B3LYP, use always B3LYP-D3 or other
dispersion corrected versions of B3LYP (or even more modern
dispersion-inclusive functionals).

Hope this helps
Stefan



-=3D This is automatically added to each message by the mailing script =3D-=
E-mail to subscribers: CHEMISTRY.:.ccl.n= et or use:
=C2=A0 =C2=A0 =C2=A0 http://www.ccl.net/cgi-bin/ccl/s= end_ccl_message

E-mail to administrators: CHEM= ISTRY-REQUEST.:.ccl.net or use
=C2=A0 =C2=A0 =C2=A0 http://www.ccl.net/cgi-bin/ccl/s= end_ccl_message

Subscribe/Unsubscribe:
=C2=A0 =C2=A0 =C2=A0 http://www.ccl.net/chemistry/sub_un= sub.shtml

Before posting, check wait time at: http://www.ccl.net

Job: http://www.ccl.net/jobs
Conferences: http://server.ccl.net/chemist= ry/announcements/conferences/

Search Messages: http://www.ccl.net/chemistry/sear= chccl/index.shtml
=C2=A0 =C2=A0 =C2=A0 http://www.ccl.net/spammers.txt

RTFI: http://www.ccl.net/chemistry/aboutccl/ins= tructions/





--
Adedapo Adeyinka

This message and attachments are subject to= a disclaimer. Please refer to=20 http://www.it.u= p.ac.za/documentation/governance/disclaimer/ for full details. --001a11c3711a45e52c051e4b6fb8-- From owner-chemistry@ccl.net Thu Aug 27 14:31:00 2015 From: "John McKelvey jmmckel,+,gmail.com" To: CCL Subject: CCL:G: H-bond distance with different solvation models Message-Id: <-51629-150827105905-2312-GwDn8pSRaDddsS66IGBOgQ##server.ccl.net> X-Original-From: John McKelvey Content-Type: multipart/alternative; boundary=047d7b86eeaebbdcf2051e4c3414 Date: Thu, 27 Aug 2015 09:58:54 -0500 MIME-Version: 1.0 Sent to CCL by: John McKelvey [jmmckel^-^gmail.com] --047d7b86eeaebbdcf2051e4c3414 Content-Type: text/plain; charset=UTF-8 Ramon, I am not up to date on the current diel-constant containing formula is in Gaussian, but in ORCA it is (eps-1)/(eps+1/2) = .984 for eps=95. I suspect it is similar in magnitude for Gaussian. An interesting exercise is to plot (eps-1)/(eps+0.5) vs eps. For eps=2.02, cyclohexane, the value .40 is obtained. So for the simplest solvent 0.4 of the dielectric effect is already found. For water the value is 0.98. This fn rapidly approaches unity as the dielectric is increased... Hope this is useful... John On Thu, Aug 27, 2015 at 5:39 AM, Ramon Crehuet rcsqtc#,#iqac.csic.es < owner-chemistry a ccl.net> wrote: > > Sent to CCL by: Ramon Crehuet [rcsqtc_+_iqac.csic.es] > Dear all, > I would like to study the dependence of an H-bond distance with the > dielectric constant of water (which depends on temperature, pressure, etc). > My system consists of a water molecule and a small molecue containg a > donating NH group. I have run the same calculations with Gaussian, using > the PCM model and with Orca, using the Cosmo model, obtaining different > results. > For an epsilon range of 80-95, the Cosmo in Orca changes the H-bond from > 1.85A to 1.95A. The same model with Gaussian PCM gives a distance that is > constant to 3 significant digits: 1.897A. > Gaussian seems to be correctly reading the epsion, as the log file states: > > Solvent : Water, Eps= 95.000000 Eps(inf)= 1.777849 > > Being electrostatic in character, I would expect the H-bond distance to > depend on the dielectric constant. Beyond that, are there any reasons to > prefer one of the methods? > > Thanks in advance, > > Ramon > > -- > Ramon Crehuet > Cientific Titular (Assistant Professor) > Institute of Advanced Chemistry of Catalonia > IQAC - CSIC > http://www.iqac.csic.es/qteor > https://twitter.com/rcrehuet > http://ramoncrehuet.wordpress.com/ > Tel. +34 934006116 > Jordi Girona 18-26 > 08034 Barcelona (Spain)http://www.ccl.net/chemistry/sub_unsub.shtmlConferences: > http://server.ccl.net/chemistry/announcements/conferences/> > > -- John McKelvey 545 Legacy Pointe Dr O'Fallon, MO 63376 636-294-5302 jmmckel a gmail.com --047d7b86eeaebbdcf2051e4c3414 Content-Type: text/html; charset=UTF-8 Content-Transfer-Encoding: quoted-printable
Ramon,

I am not up to date on = the current diel-constant containing formula is in Gaussian, but in ORCA it= is (eps-1)/(eps+1/2) =3D .984 for eps=3D95.=C2=A0 I suspect it is similar = in magnitude for Gaussian.=C2=A0 An interesting exercise is to plot (eps-1)= /(eps+0.5) vs eps.=C2=A0 For eps=3D2.02, cyclohexane, the value .40 is obta= ined.=C2=A0 So for the simplest solvent 0.4 of the dielectric effect is alr= eady found.=C2=A0 For water the value is 0.98.=C2=A0 This fn rapidly approa= ches =C2=A0 unity as the dielectric is increased...

Hope this = is useful...

John

On Thu, Aug 27, 2015 at 5:39 AM, Ramon Crehuet rcsqt= c#,#iqac.csic.es <= owner-chemistr= y a ccl.net> wrote:

Sent to CCL by: Ramon Crehuet [rcsqtc_+_iqac.csic.es]
Dear all,
I would like to study the dependence of an H-bond distance with the dielect= ric constant of water (which depends on temperature, pressure, etc). My sys= tem consists of a water molecule and a small molecue containg a donating NH= group. I have run the same calculations with Gaussian, using the PCM model= and with Orca, using the Cosmo model, obtaining different results.
For an epsilon range of 80-95, the Cosmo in Orca changes the H-bond from 1.= 85A to 1.95A. The same model with Gaussian PCM gives a distance that is con= stant to 3 significant digits: 1.897A.
Gaussian seems to be correctly reading the epsion, as the log file states:<= br>
Solvent=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 : Water, Eps=3D=C2= =A0 95.000000 Eps(inf)=3D=C2=A0 =C2=A01.777849

Being electrostatic in character, I would expect the H-bond distance to dep= end on the dielectric constant. Beyond that, are there any reasons to prefe= r one of the methods?

Thanks in advance,

Ramon

--
Ramon Crehuet
Cientific Titular (Assistant Professor)
Institute of Advanced Chemistry of Catalonia
IQAC - CSIC
http://www.iqac.csic.es/qteor
https://twitter.com/rcrehuet
http://ramoncrehuet.wordpress.com/
Tel. +34 934006116
Jordi Girona 18-26
08034 Barcelona (Spain)



-=3D This is automatically added to each message by the mailing script =3D-=
E-mail to subscribers: CHEMISTRY a ccl.net or use:
=C2=A0 =C2=A0 =C2=A0http://www.ccl.net/cgi-bin/ccl/se= nd_ccl_message

E-mail to administrators: CHEMISTRY-REQUEST a ccl.net or use
=C2=A0 =C2=A0 =C2=A0http://www.ccl.net/cgi-bin/ccl/se= nd_ccl_message

Subscribe/Unsubscribe:=C2=A0 =C2=A0 =C2=A0 http://www.cc= l.net/chemistry/sub_unsub.shtml

Before posting, check wait time at: http://www.ccl.net

Job: http://www.ccl.net/jobs Conferences: http://server.ccl.net/chemistry/announcements/conferences/

Search Messages: http://www.ccl.net/chemistry/sear= chccl/index.shtml
=C2=A0 =C2=A0 =C2=A0http://www.ccl.net/spammers.txt

RTFI: http://www.ccl.net/chemistry/aboutccl/ins= tructions/





--
John McKelvey
545 Legacy Pointe Dr
O'Fallon, MO 63376
636-294-5302
jmmckel a gmail.com
--047d7b86eeaebbdcf2051e4c3414-- From owner-chemistry@ccl.net Thu Aug 27 15:05:01 2015 From: "Francesco Pietra chiendarret-x-gmail.com" To: CCL Subject: CCL:G: Gaussian records during MP2 geometry optimization Message-Id: <-51630-150827144508-21186-Sn+hYpE3sXTOKeEMHncZvg###server.ccl.net> X-Original-From: Francesco Pietra Content-Type: multipart/alternative; boundary=f46d043c802a825e73051e4f5dc4 Date: Thu, 27 Aug 2015 20:45:03 +0200 MIME-Version: 1.0 Sent to CCL by: Francesco Pietra [chiendarret . gmail.com] --f46d043c802a825e73051e4f5dc4 Content-Type: text/plain; charset=UTF-8 Hello: With threaded Gaussian on 16 cores, 120GB mem, unlimited scratch, execution of geometry optimization at MP2/6-31G* is quite slow at the "MP4(R+Q)= 0.84912525D-01" stage. With the 24h limit, I fear that the procedure will be interrupted before writing to chk. In fact, up to the said stage, the chk was overwritten only at the beginning. (the chk came > from previous geom optimization st SCF/6-31G*). Is that seldom writing expected, or is my Gaussian input deficient? ===================== %chk=my.chk %nproc=16 %mem=115GB # MP2/6-31G* MP2=SemiDirect guess=read geom=checkpoint simtype="Geometry optimization" -3 1 ====================== The 120GB were indicated in the PBS to the cluster. thanks francesco pietra --f46d043c802a825e73051e4f5dc4 Content-Type: text/html; charset=UTF-8 Content-Transfer-Encoding: quoted-printable
Hello:

With threaded Gaussian = on 16 cores, 120GB mem, unlimited scratch, execution=C2=A0 of geometry opti= mization at MP2/6-31G* is quite slow at the "MP4(R+Q)=3D=C2=A0 0.84912= 525D-01" stage. With the 24h limit, I fear=C2=A0 that the procedure wi= ll be interrupted before writing to chk. In fact, up to the said stage, the= chk was overwritten only at the beginning. (the chk came from previous geo= m optimization st SCF/6-31G*).

Is that seldom writing expected= , or is my Gaussian input deficient?

=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
%chk=3Dmy.chk
%nproc=3D16
%mem= =3D115GB
# MP2/6-31G* MP2=3DSemiDirect guess=3Dread geom=3Dcheckpoint
<qmtool> simtype=3D"Geometry optimization" </qmtool= >

-3 1
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D

The 120GB were indicated in the PBS to the cluste= r.

thanks

francesco pietra

--f46d043c802a825e73051e4f5dc4--