Hello,
Despite how wonderful I think Avogadro is for molecular modeling and its
potential for educational purposes, I hesitate to use Avogadro in the
chemistry laboratory when there seems to be such glaring issues as
predicting the dipole moment for simple inorganic molecules such as carbon
dioxide. When I build carbon dioxide using either 1.0.3 (in Ubuntu) or
1.1.0 (on a PC) and optimize the geometry using MMFF94, the bond lengths
are the same (1.197 angstroms for each C=O bond) and the bond angle is 180
degrees, just as you would predict with VSEPR theory. Yet, the dipole
moment is a robust 9.72 Debye, not 0 Debye as would be predicted by VSEPR
theory. In addition, if you build carbon dioxide in 1.0.3 by selecting
double bonds and turning off adjust hydrogens, the partial charge on the
carbon and oxygen atoms remains zero, and not 0.37 for C and -0.19 for each
O. Unfortunately, Ubuntu has not put the latest version of Avogadro in its
repository.
I cannot explain why such discrepancies exist in the software and so I’m
calling on the users and developers of Avogadro for some help on explaining
or overcoming such discrepancies.
Thank you for your assistance,
Steve
–
Dr. Steven Petrovic
Professor of Chemistry
Southern Oregon University
1250 Siskiyou Blvd.
Ashland, OR 97520
(541) 552-6803
On Sat, Jan 19, 2013 at 12:31 PM, Steven Petrovic petrovis@sou.edu wrote:
Hello,
Despite how wonderful I think Avogadro is for molecular modeling and its
potential for educational purposes, I hesitate to use Avogadro in the
chemistry laboratory when there seems to be such glaring issues as
predicting the dipole moment for simple inorganic molecules such as carbon
dioxide. When I build carbon dioxide using either 1.0.3 (in Ubuntu) or 1.1.0
(on a PC) and optimize the geometry using MMFF94, the bond lengths are the
same (1.197 angstroms for each C=O bond) and the bond angle is 180 degrees,
just as you would predict with VSEPR theory. Yet, the dipole moment is a
robust 9.72 Debye, not 0 Debye as would be predicted by VSEPR theory. In
addition, if you build carbon dioxide in 1.0.3 by selecting double bonds and
turning off adjust hydrogens, the partial charge on the carbon and oxygen
atoms remains zero, and not 0.37 for C and -0.19 for each O. Unfortunately,
Ubuntu has not put the latest version of Avogadro in its repository.
I cannot explain why such discrepancies exist in the software and so I’m
calling on the users and developers of Avogadro for some help on explaining
or overcoming such discrepancies.
Thank you for your assistance,
I think all of these numbers come from Open Babel, and the force
fields we use that were implemented by the Open Babel team. Perhaps
posing this question on the Open Babel list, or looking at other open
source implementations we could integrate would make sense.
I think Geoff and Tim would be more qualified to comment on how much
work it would be to correct some of these issues in Open Babel, we
also integrate with many third party quantum codes and using third
party MD packages might be another viable route to more accurate
numbers.
Marcus
To be clear, are you asking why MMFF94 is so poor at predicting dipole moments?
On 19 January 2013 17:31, Steven Petrovic petrovis@sou.edu wrote:
Hello,
Despite how wonderful I think Avogadro is for molecular modeling and its
potential for educational purposes, I hesitate to use Avogadro in the
chemistry laboratory when there seems to be such glaring issues as
predicting the dipole moment for simple inorganic molecules such as carbon
dioxide. When I build carbon dioxide using either 1.0.3 (in Ubuntu) or 1.1.0
(on a PC) and optimize the geometry using MMFF94, the bond lengths are the
same (1.197 angstroms for each C=O bond) and the bond angle is 180 degrees,
just as you would predict with VSEPR theory. Yet, the dipole moment is a
robust 9.72 Debye, not 0 Debye as would be predicted by VSEPR theory. In
addition, if you build carbon dioxide in 1.0.3 by selecting double bonds and
turning off adjust hydrogens, the partial charge on the carbon and oxygen
atoms remains zero, and not 0.37 for C and -0.19 for each O. Unfortunately,
Ubuntu has not put the latest version of Avogadro in its repository.
I cannot explain why such discrepancies exist in the software and so I’m
calling on the users and developers of Avogadro for some help on explaining
or overcoming such discrepancies.
Thank you for your assistance,
Steve
–
Dr. Steven Petrovic
Professor of Chemistry
Southern Oregon University
1250 Siskiyou Blvd.
Ashland, OR 97520
(541) 552-6803
Master Visual Studio, SharePoint, SQL, ASP.NET, C# 2012, HTML5, CSS,
MVC, Windows 8 Apps, JavaScript and much more. Keep your skills current
with LearnDevNow - 3,200 step-by-step video tutorials by Microsoft
MVPs and experts. SALE $99.99 this month only – learn more at:
http://p.sf.net/sfu/learnmore_122912
Avogadro-Discuss mailing list
Avogadro-Discuss@lists.sourceforge.net
avogadro-discuss List Signup and Options
Noel - do some of the other Open Babel force fields do a better job at this?
Marcus
On Mon, Jan 21, 2013 at 8:59 AM, Noel O’Boyle baoilleach@gmail.com wrote:
To be clear, are you asking why MMFF94 is so poor at predicting dipole moments?
On 19 January 2013 17:31, Steven Petrovic petrovis@sou.edu wrote:
Hello,
Despite how wonderful I think Avogadro is for molecular modeling and its
potential for educational purposes, I hesitate to use Avogadro in the
chemistry laboratory when there seems to be such glaring issues as
predicting the dipole moment for simple inorganic molecules such as carbon
dioxide. When I build carbon dioxide using either 1.0.3 (in Ubuntu) or 1.1.0
(on a PC) and optimize the geometry using MMFF94, the bond lengths are the
same (1.197 angstroms for each C=O bond) and the bond angle is 180 degrees,
just as you would predict with VSEPR theory. Yet, the dipole moment is a
robust 9.72 Debye, not 0 Debye as would be predicted by VSEPR theory. In
addition, if you build carbon dioxide in 1.0.3 by selecting double bonds and
turning off adjust hydrogens, the partial charge on the carbon and oxygen
atoms remains zero, and not 0.37 for C and -0.19 for each O. Unfortunately,
Ubuntu has not put the latest version of Avogadro in its repository.
I cannot explain why such discrepancies exist in the software and so I’m
calling on the users and developers of Avogadro for some help on explaining
or overcoming such discrepancies.
Thank you for your assistance,
Steve
–
Dr. Steven Petrovic
Professor of Chemistry
Southern Oregon University
1250 Siskiyou Blvd.
Ashland, OR 97520
(541) 552-6803
Master Visual Studio, SharePoint, SQL, ASP.NET, C# 2012, HTML5, CSS,
MVC, Windows 8 Apps, JavaScript and much more. Keep your skills current
with LearnDevNow - 3,200 step-by-step video tutorials by Microsoft
MVPs and experts. SALE $99.99 this month only – learn more at:
http://p.sf.net/sfu/learnmore_122912
Avogadro-Discuss mailing list
Avogadro-Discuss@lists.sourceforge.net
avogadro-discuss List Signup and Options
Master Visual Studio, SharePoint, SQL, ASP.NET, C# 2012, HTML5, CSS,
MVC, Windows 8 Apps, JavaScript and much more. Keep your skills current
with LearnDevNow - 3,200 step-by-step video tutorials by Microsoft
MVPs and experts. SALE $99.99 this month only – learn more at:
http://p.sf.net/sfu/learnmore_122412
Avogadro-Discuss mailing list
Avogadro-Discuss@lists.sourceforge.net
avogadro-discuss List Signup and Options
To be clear, are you asking why MMFF94 is so poor at predicting dipole moments?
No, he’s commenting on a bug. O=C=O clearly should have approximately 0.0D dipole moment. Regardless of the charges on the individual atoms, the charges should cancel in a symmetric molecule.
I have no idea the source of the bug, although I can reproduce it. I suspect it’s some sort of “race condition” when the molecule is being modified, the dipole moment is not properly updated.
I’ll take a look later this week. But if you can give me a list of examples for “the dipole moment is totally crazy,” I’d appreciate it.
As for the accuracy of MMFF94 charges vs. other things, both MMFF94 and Gasteiger give fairly reasonable dipole moments for most organic molecules in my testing. It’d be great to have a more universal charge model, but I don’t think one exists at present. It’s on my “wish list” in the next year or two to get a charge model which is halfway decent across most elements.
-Geoff