Los Alamos National Laboratory

Accurate Calculation of Redox Potentials Scheme 1. Born-Haber Cycle.

Using Density Functional Methods

Lindsay E. Roy, Elena Jakubikova, Enrique R. Batista, T-1

D
ensity functional theory (DFT) has provided insight
into the mechanisms of several important catalytic and the Nernst equation then determines the standard one-
cycles in environmental, bioinorganic, and industrial electron redox potentials, E/V
chemistry. Many chemical reactions utilizing transition metal
complexes describe the loss and gain of electrons through the (2)
formation or breaking of chemical bonds. Combined with
electrochemical and spectroscopic evidence of the pertinent where F is the Faraday constant, 23.06 kcal mol1 V1.
intermediates, theoretical chemists have been able to deduce
Although this method has been successful for several organic
favorable thermodynamic and kinetic pathways to these
systems using an implicit solvent model, modeling transition
cycles. The abilities of DFT have also been pushed in hopes
metal complexes has proven to be less reliable [3]. Previously,
of facilitating in the rational design of catalysts with specific
we showed that values of the calculated redox potential for
properties. One such property that has garnered much
a series of small models based on the diiron hydrogenase
attention over the past several years has been the ability of
enzymes reproduced the trends experimentally with an
DFT to accurately predict redox potentials of transition metal
average error of 0.12V using the gradient-corrected DFT, and
complexes. The ability to predict redox potentials a priori
results using the hybrid DFT functional required a systematic
is highly desirable, and one can imagine a combinatorial-
computational approach to finding a redox potential within a shift [4]. There, we realized the importance of referencing
specific range. our results to the calculated absolute half-cell potential for
the Ferrocene couple (i.e., Fc + A Fc++A) in order
There are currently several protocols that exist for the to eliminate additional assumptions such as the solvent
theoretical prediction of the standard redox potentials in surface potential and the electron transfer process at the
solution [1,2]. One of the more popular methods uses the electrode surface necessary when referencing results to the
Born-Haber cycle shown in Scheme 1 where the standard experimental absolute potential of the standard hydrogen
Gibbs free energy of redox half reaction, , consists of electrode (SHE). These results suggest that a fairly high
the free energy change in the gas phase and the solvation free degree of accuracy can be achieved for first-row transition
energies of the oxidized and reduced species. These values elements. To resolve these issues, we set out to provide a
are then used to calculate the overall reaction of the standard rigorous quantum-chemical treatment of systems comprised
Gibbs free energy energy, /kcalmol1 (Eq. 1); of first-, second-, and third-row transition elements. These
compounds contain varying oxidation states and charges and
(1) provide a good benchmark, considering the potentials are not
necessarily reversible on the electrochemical time scale.
Scheme 2 depicts the inorganic and organometallic
compounds chosen for our study. A plot of our results

72 Associate Directorate for Theory, Simulation, and Computation (ADTSC) LA-UR-09-01756

Chemistry and Bioscience

is illustrated in Fig. 1 for a typical generalized gradient Fig. 1. Correlation diagram

approximation (GGA), functional (PBE, [5]), and the hybrid of experimental vs. calculated
generalized gradient (H-GGA) Density Functional, (B3LYP, redox potential, V/Cp2Fe, for
[6,7]),respectively. Firstly, the data are all consistent with one Scheme 2 complexes using
another for a given functional, regardless whether the redox PBE and B3LYP functionals.
occurs at the metal center or on the ligand. The solid line illustrates 1:1
The errors also appear correlation.
Scheme 2. Transition metal to be systematic and
complexes considered in this are not dependent on
study. whether the compound
is a first-, second-, or
third-row element. The
PBE functional provides
excellent agreement with
experiment, and analysis
of the data show that the
quality of fit is 0.97. The
B3LYP functional fared
the worst in terms of
fit and deviation with a
systematic error of 0.83.
In fact, B3LYP results
require a consistent shift
of -0.48 V for all the data
(shown as a dashed line in Fig. 1). When including a baseline For further information contact Enrique R. Batista at
shift for B3LYP, correlation is improved to 0.96. Examination [email protected].
of the data reveals that the only variation occurring with
each calculation is the gas-phase free energy term, [1] J. Li et al., Inorg. Chem. 35, 4694 (1996).
the free energy of solvation, , only differs by ~1 [2] I. Tavernelli, R. Vuilleumier, M. Sprik, Phys. Rev. Lett. 88,
kcal mol-1 for all the functionals that were studied. The 213002 (2002).
better performance of the GGA functional with respect to [3] M.-H. Baik, R.A. Friesner, J. Phys. Chem. A 106, 7407 (2002).
the H-GGA could be the result of an improved gas phase [4] L.E. Roy, E.R., Batista, P.J. Hay, Inorg. Chem. 47, 9228 (2008).
ionization potential. [5] J.P. Perdew, K. Burke, M. Ernzerhot, Phys. Rev. Lett. 77, 3865
(1996).
The average error one can expect using PBE to calculate a [6] A.D. Becke, J. Chem. Phys. 98, 5648 (1993).
redox couple is <0.23 V. The deviation is ~0.16 V for the [7] C. Lee et al., Phys. Rev. B 37, 785 (1988). Funding
redox couple of neutral species, but larger for highly anionic Acknowledgments
or cationic complexes. This protocol is a powerful tool that LANL Directed
will allow theoretical chemists to aid in the design of Research and
redox-active catalysts. Development Program

www.lanl.gov/orgs/adtsc/publications.php
73