Sent to CCL by: "Shobe, David" [David.Shobe+/-sud-chemie.com]
The real problem with DFT is as follows. The spin-restricted DFT
(RDFT,
as in Wai-To's response) FORCES pairing of every electron! If you have
a
diradical, RDFT will force the two unpaired electrons into the same
orbital, which may or may not resemble either of the singly occupied
orbitals in the diradical.
And you may be using RDFT without being aware of it, because
quantum-chemical programs in general use RDFT for any singlet species,
unless you specify that UDFT (spin-unrestricted DFT) is to be used.
Several months ago on CCL, there was a discussion on DFT (B3LYP?)
calculations on the twisting of ethylene, and it may be helpful to
review
that discussion. The 90° twisted ethylene is very poorly
described by
DFT, because the two electrons are forced into one of the orthogonal p
orbitals.
Explicitly correlated methods such as CCSD or MP2 may be less
vulnerable
to this effect, because even though they rely on a single-determinant
reference, they at least add in other determinants.
Ideally one would use a multi-determinant reference method for
problems
like this, but they are not easy to use!
--David Shobe
-----Original Message-----
From: owner-chemistry]*[ccl.net [mailto:owner-chemistry]*[ccl.net]
Sent: Wednesday, June 13, 2007 7:45 AM
To: Shobe, David
Subject: CCL:G: Limitations of DFT
Sent to CCL by: Wai-To Chan [chan++curl.gkcl.yorku.ca]
<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
However, one of the reviewers on *this* manuscript
objected to the use of the DFT calculations with the following:
"With regards to using DFT calculations to distinguish between
mechanisms
B
and C, I would put no confidence in such a result. It is well known
that
DFT
calculations greatly overemphasize the stabilization provided by
delocalization,
so they have a built-in bias towards concerted reactions, which is
useful
if the
reaction actually is concerted, but a disaster if it is not."
Is this a true statement? And, if so, where can I find more about this
limitation of DFT theory? I have searched but come up empty so far.
Plus,
I'm
not sure exactly what the reviewer means, since the diradical
intermediate
of
mechanism C would also be subject to delocalization effects...possibly
even
greater than those expected by a "concerted" reaction!
I assume reaction pathway (B) proceeds via a TS characterized
by spin-restricted (R)DFT, (C) a open-shell TS characterized by
unrestricted DFT and the reactant is closed-shell and treated by RDFT.
Did you check the wavefunction stability of the TS for (B)?
Instability of
RHF and/or RDFT
is a telling sign of discontinuity within your PES.
The theoretical foundation underlying the characterization of the kind
of
PES
with stationary points of varying degree of diradical character
using single-reference type of methods is shaky. I however don't see
any
relevance of your
reviewer's remarks to the limiation of DFT. DFT indeed does suffer from
wavefunction stability and
its reliability for the prediction of concerted and stepwise mechanism
involving
diradicals is seldom addressed. But I never heard of any studies which
demonstrate
that DFT is biased towards the concerted pathway. I seem to
remember that DFT has been applied to characterization of the
aromaticity
of transition
state structures in concerted reaction mechanism through NICS
calculations.
I am not familiar with the literature in this area but I doubt very much
one can draw any conclusion from such studies alone on DFT's favouring
of the concerted mehanism in pericyclic reactions.
If I were you I would demand the referee to provide
a reference to back up his/her comments further. Chance is that the
referee's
comments are merely speculations rather than established facts.
It is not just DFT. The entire specturm of commonly employed
single-reference type of methods (HF, MPn, CCSD(T), DFT)
is inherently problematic for a consistently accurate characterization
of a diradical PES. CASSCF in principle could solve the problem arising
from PES discontinuity. In practice CASSCF won't
necessarily give more
accurate structural
and energetic predictions than DFT.
DFT has been applied to the study of concerted and stepwise mechansims
in cycloaddition reaction for years. Occassionaly serious
discrepancies
between
DFT and CASSCF were noted. But no one to my knowledge has suggested
DFT should be avoided in favour of other methods in general. There are
other
promising alternatives to DFT for diradical PES outside GAUSSIAN.
Unless these methods have already gained general acceptance and are
applicable to the
system you are studying you don't have the obligation to 'upgrade'
your DFT calculations in my view. My advice for you is
to come up with a cogent response to the questions of the referee. You
should
in your discussion of your results highlight those limitations of the
theories employed and convince the readers that the computations were
properly executed within the limitations. As long as you lay out the
facts
clearly to allow the
readers to assess your results you should stand a reasonable chance of
getting your work accepted.
Wai-To
Chanhttp://www.ccl.net/cgi-bin/ccl/send_ccl_messagehttp-:-//www.ccl.net/chemistry/sub_unsub.shtmlhttp-:-//www.ccl.net/spammers.txtThis
e-mail message may contain confidential and / or privileged
information.
If you are not an addressee or otherwise authorized to receive this
message, you should not use, copy, disclose or take any action based on
this e-mail or any information contained in the message. If you have
received this material in error, please advise the sender immediately
by
reply e-mail and delete this message.
Thank you.>
--
No virus found in this incoming message.
Checked by AVG Free Edition.
Version: 7.5.472 / Virus Database: 269.8.15/847 - Release Date:
12/06/2007
09:42 p.m.