From owner-chemistry@ccl.net Thu Mar 13 08:50:01 2014 From: "Tobias Schwabe schwabe^^zbh.uni-hamburg.de" To: CCL Subject: CCL: Sensitivity of vertical ionization energy and homo on geometry Message-Id: <-49849-140313051024-7598-J+nfq/vOWC9qpbSXXB8DZA!=!server.ccl.net> X-Original-From: Tobias Schwabe Content-Type: multipart/mixed; boundary="------------020907000102040805060205" Date: Thu, 13 Mar 2014 10:10:34 +0100 MIME-Version: 1.0 Sent to CCL by: Tobias Schwabe [schwabe|zbh.uni-hamburg.de] This is a multi-part message in MIME format. --------------020907000102040805060205 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Dear Scott, > from my understanding you are also concerned about the effect from the level of theory for geometry optimizations on your ionization energy quality. Unfortunately, I have no reference ready to provide, but in general this cannot be ignored depending on the accuracy you seek. Especially critical, in my opinion, are multiple bond length (e.g. C=O bonds) and aromatic systems. The good news here: the standard recommendations for geometry optimizations also apply here. Depending on system size, go for a hybrid functional (with your favorite dispersion correction) plus a triple zeta quality basis set (my preferences here: B3LYP-D/def2-TZVP or PBE0-D/def2-TZVP). With programs like TURBOMOLE or ORCA, those optimizations can be done routinely on quite large systems. If you are interested in more accurate geometries, use SCS-MP2 or double hybrid functionals, but then you also need a larger set of polarization functions (like def2-TZVPP). In case you cannot afford anything of the aforementioned methods, it's getting harder from my experience. Using a (meta-)GGA functional with RI (= density fitting) yields a 10x speed up (or so) but you should be careful if self-interaction problems might occur in your geometry. For example, I suspect that torsion angles for phenyl rings can be a problem (and might bias your ionization energies). I hope this was helpful to you. Otherwise feel free to ask again. Regards, Tobias --------------020907000102040805060205 Content-Type: text/x-vcard; charset=utf-8; name="schwabe.vcf" Content-Transfer-Encoding: 7bit Content-Disposition: attachment; filename="schwabe.vcf" begin:vcard fn:Tobias Schwabe n:Schwabe;Tobias org:University of Hamburg;Center for Bioinformatics adr;quoted-printable;dom:;;Bundesstra=C3=9Fe 43;Hamburg;;20146 email;internet:schwabe#%#zbh.uni-hamburg.de title:Juniorprofessor tel;work:+49 40 42838 7333 url:http://www.zbh.uni-hamburg.de/mitarbeiter/schwabe.html version:2.1 end:vcard --------------020907000102040805060205-- From owner-chemistry@ccl.net Thu Mar 13 16:32:00 2014 From: "Scott McKechnie jsm78---cam.ac.uk" To: CCL Subject: CCL: Sensitivity of vertical ionization energy and homo on geometry Message-Id: <-49850-140313143207-11212-hCEkx9UeJ+2uGKpOf6NoUA:+:server.ccl.net> X-Original-From: Scott McKechnie Content-Type: multipart/alternative; boundary=047d7b10cccf3f559404f4812b12 Date: Thu, 13 Mar 2014 18:32:00 +0000 MIME-Version: 1.0 Sent to CCL by: Scott McKechnie [jsm78-.-cam.ac.uk] --047d7b10cccf3f559404f4812b12 Content-Type: text/plain; charset=ISO-8859-1 Dear Tobias, Many thanks, I realise that my question was a little vague but this is exactly the type of answer I was looking for. The dependence on torsion angle is a nice example and for anyone who is interested he following papers were very helpful: http://dx.doi.org/10.1103/PhysRevB.54.2381 http://dx.doi.org/10.1063/1.449450 I'd welcome any further examples or references but otherwise I'd consider my question solved. In response to the comments about geometric changes that occur during the ionization process, as Dr. Abrash and Dr. Molt correctly pointed out, I am interested in vertical ionization energies where there is no geometric change. Best wishes, Scott On 13 March 2014 09:10, Tobias Schwabe schwabe^^zbh.uni-hamburg.de < owner-chemistry : ccl.net> wrote: > Dear Scott, > > from my understanding you are also concerned about the effect from the >> > level of theory for geometry optimizations on your ionization energy > quality. Unfortunately, I have no reference ready to provide, but in > general this cannot be ignored depending on the accuracy you seek. > > Especially critical, in my opinion, are multiple bond length (e.g. C=O > bonds) and aromatic systems. The good news here: the standard > recommendations for geometry optimizations also apply here. Depending on > system size, go for a hybrid functional (with your favorite dispersion > correction) plus a triple zeta quality basis set (my preferences here: > B3LYP-D/def2-TZVP or PBE0-D/def2-TZVP). With programs like TURBOMOLE or > ORCA, those optimizations can be done routinely on quite large systems. > > If you are interested in more accurate geometries, use SCS-MP2 or double > hybrid functionals, but then you also need a larger set of polarization > functions (like def2-TZVPP). > > In case you cannot afford anything of the aforementioned methods, it's > getting harder from my experience. Using a (meta-)GGA functional with RI (= > density fitting) yields a 10x speed up (or so) but you should be careful if > self-interaction problems might occur in your geometry. For example, I > suspect that torsion angles for phenyl rings can be a problem (and might > bias your ionization energies). > > I hope this was helpful to you. Otherwise feel free to ask again. > > Regards, > Tobias > --047d7b10cccf3f559404f4812b12 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable
Dear Tobias,

Many thanks, I realise tha= t my question was a little vague but this is exactly the type of answer I w= as looking for. The dependence on torsion angle is a nice example and for a= nyone who is interested he following papers were very helpful:


<= /div>
I'd welcome any further examples or references but otherwise = I'd consider my question solved.

<= div>In response to the comments about geometric changes that occur during t= he ionization process, as Dr. Abrash and Dr. Molt correctly pointed out, I = am interested in vertical ionization energies where there is no geometric c= hange.=A0

Best wishes,

Scott
<= br>
<= br>
=A0


=
On 13 March 2014 09:10, Tobias Schwabe schwabe^^= zbh.uni-hamburg.de <owner-= chemistry : ccl.net> wrote:
Dear Scott,

> from my understanding you are also concerned about the effect from the
level of theory for geometry optimizations on your ionization energy qualit= y. Unfortunately, I have no reference ready to provide, but in general this= cannot be ignored depending on the accuracy you seek.

Especially critical, in my opinion, are multiple bond length (e.g. C=3DO bo= nds) and aromatic systems. The good news here: the standard recommendations= for geometry optimizations also apply here. Depending on system size, go f= or a hybrid functional (with your favorite dispersion correction) plus a tr= iple zeta quality basis set (my preferences here: B3LYP-D/def2-TZVP or PBE0= -D/def2-TZVP). With programs like TURBOMOLE or ORCA, those optimizations ca= n be done routinely on quite large systems.

If you are interested in more accurate geometries, use SCS-MP2 or double hy= brid functionals, but then you also need a larger set of polarization funct= ions (like def2-TZVPP).

In case you cannot afford anything of the aforementioned methods, it's = getting harder from my experience. Using a (meta-)GGA functional with RI (= =3D density fitting) yields a 10x speed up (or so) but you should be carefu= l if self-interaction problems might occur in your geometry. For example, I= suspect that torsion angles for phenyl rings can be a problem (and might b= ias your ionization energies).

I hope this was helpful to you. Otherwise feel free to ask again.

Regards,
=A0 =A0 =A0 =A0 Tobias

--047d7b10cccf3f559404f4812b12--