Elucidation of the thermochemical properties of triphenyl- or tributyl-substituted Si-, Ge-, and Sn-centered radicals by means of electrochemical approaches and computations

Article abstract of DOI:10.1021/ja044687p

Redox potentials of a number of triphenyl- or tributyl-substituted Si-, Ge-, or Sn-centered radicals, R3M^., have been measured in acetonitrile, tetrahydrofuran, or dimethyl sulfoxide by photomodulated voltammetry or through a study of the oxidation process of the corresponding anions in linear sweep voltammetry. For the results pertaining to the Ph ₃M^. series (including literature data for M = C), the order of reduction potentials follows Sn > Ge > C > Si, while for the two oxidation potentials, it is C > Si. The effect of the R group on the redox properties of R₃Sn^. is pronounced in that the reduction potential is more negative by 490 mV in tetrahydrofuran (390 mV in dimethyl sulfoxide) when R is a butyl rather than a phenyl group. The experimental trends have been substantiated through quantum chemical calculations, and they can be explained qualitatively by considering a combination of effects, such as charge capacity being most pronounced for the heavier elements, resonance stabilization present for the planar Ph ₃C^. and all R₃M⁺, and finally a contribution from solvation. The solvation of R₃M^- is observed to be relatively strong because of a rather localized negative charge in the pyramidal geometry. However, there is no evidence in the calculations to support the existence of covalent interactions between solvent and anions. The solvation of R₃M⁺ is relatively weak, which may be attributed to the planar geometry around the center atom, leading to more spread out charge than that for a pyramidal geometry. Although the calculated solvation energies based on the polarizable continuum model approach exhibit the expected trends, they are not able to reproduce the experimentally derived values on a detailed level for these types of ions. An evaluation of the general performance of the continuum model is provided on the basis of present and previous studies.

Full text of DOI:10.1021/ja044687p

Published on Web 02/03/2005
Elucidation of the Thermochemical Properties of Triphenyl- or
Tributyl-Substituted Si-, Ge-, and Sn-Centered Radicals by
Means of Electrochemical Approaches and Computations
Allan Hjarbæk Holm, Tore Brinck,* and Kim Daasbjerg*,¶
¶
,#
Contribution from the Department of Chemistry, UniVersity of Aarhus, Langelandsgade 140,
DK-8000 Aarhus C, Denmark, and Department of Physical Chemistry, Royal Institute of
Technology, SE-10044 Stockholm, Sweden
Received September 2, 2004; E-mail: kdaa@chem.au.dk (K.D.); tore@physchem.kth.se (T.B.)
Abstract: Redox potentials of a number of triphenyl- or tributyl-substituted Si-, Ge-, or Sn-centered radicals,
•
R
3
M , have been measured in acetonitrile, tetrahydrofuran, or dimethyl sulfoxide by photomodulated
voltammetry or through a study of the oxidation process of the corresponding anions in linear sweep
•
voltammetry. For the results pertaining to the Ph
3
M series (including literature data for M ) C), the order
of reduction potentials follows Sn > Ge > C > Si, while for the two oxidation potentials, it is C > Si. The
•
effect of the R group on the redox properties of R
3
Sn is pronounced in that the reduction potential is more
negative by 490 mV in tetrahydrofuran (390 mV in dimethyl sulfoxide) when R is a butyl rather than a
phenyl group. The experimental trends have been substantiated through quantum chemical calculations,
and they can be explained qualitatively by considering a combination of effects, such as charge capacity
•
being most pronounced for the heavier elements, resonance stabilization present for the planar Ph
3
C and
+
-
3 3
all R M , and finally a contribution from solvation. The solvation of R M is observed to be relatively strong
because of a rather localized negative charge in the pyramidal geometry. However, there is no evidence
in the calculations to support the existence of covalent interactions between solvent and anions. The solvation
+
of R
3
M is relatively weak, which may be attributed to the planar geometry around the center atom, leading
to more spread out charge than that for a pyramidal geometry. Although the calculated solvation energies
based on the polarizable continuum model approach exhibit the expected trends, they are not able to
reproduce the experimentally derived values on a detailed level for these types of ions. An evaluation of
the general performance of the continuum model is provided on the basis of present and previous studies.
Introduction
be followed. In the literature, only few reports have been
concerned with the estimation of the potentials of these types
Chemical compounds containing Si, Ge, or Sn atoms have
attracted considerable attention due to their potential in the
development of new types of materials displaying excellent
pattern transfer and photoconducting properties. Recently,
synthesis of polygermanes and polysilanes by electrochemical
means has been reported,4 making the knowledge of the redox
potentials of the species involved essential for predicting and
ultimately adjusting whether a radical or ionic pathway should
•
of radicals, even if a radical, such as Bu3Sn , has found
7
widespread applications in synthesis. Tanner et al. applied
•
photomodulated voltammetry in the study of Ph3Sn and
1
2,3
•
8
Bu3Sn , but the investigations were difficult because of
experimental limitations.
-6
The primary objective of the present paper is to characterize
the redox properties of Si-, Ge-, and Sn-centered radicals in
order to gain insight into the fundamental thermochemical
properties of group IV elements by evaluating the effect of size
and electron configuration. Ultimately, this may also provide
useful guidelines for controlling polymerization processes
involving these types of species. As chemical objects, we
¶
University of Aarhus.
Royal Institute of Technology.
#
(
(
(
(
1) Hayase, S.; Nakano, Y.; Yoshikawa, S.; Ohta, H.; Sato, Y.; Shiobara, E.;
Miyoshi, S.; Onishi, Y.; Abe, M.; Matsuyama, H.; Ohiwa, Y. Chem. Mater.
2
001, 13, 2186.
2) West, R.; David, L. D.; Djurovich, P. I.; Stearley, K. L.; Srinivasan, K. S.
V.; Yu, H. J. Am. Chem. Soc. 1981, 103, 7352. (b) Miller, R. D. Angew.
Chem., Int. Ed. Engl. 1989, 28, 1733.
•
selected the three radicals, Ph3M (M ) Si, Ge, or Sn), as they
with their comparatively high steric hindrance and delocalization
features should present relatively fewest problems in electro-
chemical experiments with respect to adsorption and sluggish
3) Matsukawa, K.; Tamai, T.; Inoue, H. Appl. Phys. Lett. 2000, 77, 675. (b)
Ostapenko, N. I.; Sekirin, I. V.; Tulchynskaya, D. N.; Suto, S.; Watanabe,
A. Synth. Met. 2002, 129, 19.
4) Ishifune, M.; Kashimura, S.; Kogai, Y.; Fukuhara, Y.; Kato, T.; Bu, H.-
B.; Yamashita, N.; Murai, Y.; Murase, H.; Nishida, R. J. Organomet. Chem.
•
charge transfers. Included in the study was also Bu3Sn to assess
2
000, 611, 26. (b) Kashimura, S.; Ishifune, M.; Yamashita, N.; Bu, H.-B.;
the group effect on the central atom and at the same time
Takebayashi, M.; Kitajima, S.; Yoshihara, D.; Kataoka, Y.; Nishida, R.;
Kawasaki, S.; Murase, H.; Shono, T. J. Org. Chem. 1999, 64, 6615.
5) Okano, M.; Watanabe, K.; Totsuka, S. Electrochemistry 2003, 71, 257.
6) Subramanian, K. J. Macromol. Sci., ReV. Macromol. Chem. Phys. 1998,
C38 (4), 637.
(7) Chatgilialoglu, C. In Radicals in Organic Synthesis; Renaud, P., Sibi, M.
P., Eds.; Wiley: Weinheim, Germany, 2001; p 28.
(8) Tanner, D. D.; Harrison, D. J.; Chen, J.; Kharrat, A.; Wayner, D. D. M.;
Griller, D.; McPhee, D. J. J. Org. Chem. 1990, 55, 3321.
(
(
10.1021/ja044687p CCC: $30.25 © 2005 American Chemical Society
J. AM. CHEM. SOC. 2005, 127, 2677-2685
9
2677

A R T I C L E S
Holm et al.
Scheme 1. Generalized Electrode Processes (M ) Si, Ge, or Sn,
and R ) Ph or Bu)
In the linear sweep experiments, a φ ) 0.68 or 1 mm glassy carbon
Sigradur G, HTW) disk was employed as working electrode. The
(
electrode surface was carefully polished before each experiment with
a 0.25 µm diamond suspension and rinsed in ethanol. The counter
electrode was a platinum coil melted into glass, while a silver wire in
the solvent system separated from the solution by a ceramic disk served
as quasi-reference electrode. The ohmic drop in both the LSV and PMV
experiments was compensated with a positive feedback system
incorporated in the home-built potentiostat.
characterize one of the most commonly used radicals within
the synthetic community. Scheme 1 illustrates the generalized
electrode processes under investigation.
7
The experimental strategy was based on the utilization of two
independent electrochemical techniques, photomodulated vol-
tammetry (PMV) and linear sweep voltammetry (LSV), to
overcome the experimental difficulties encountered and to
establish confidence in the results obtained. At the same time,
quantum chemical calculations were carried out in order to
obtain complementary information about the structure, the extent
of electron delocalization, ionization potentials, electron affini-
ties, and solvation energies. An interesting aspect of the last
set of values is that they may be compared directly with
experimentally derived solvation energies extracted from the
measured potentials. In this manner, we have shown recently
that continuum-based solvation models, in which the solvent is
characterized solely by its dielectric constant, were adequate
for predicting solvation energies of many organic anions,
Procedure. The primary advantage of the PMV technique is that
under optimal conditions, it allows the simultaneous determination of
ox
1/2
red
1/2
the half-wave oxidation and reduction potentials, E and E , of
photolytically generated radicals by recording the current as a function
of the electrode potential.13 Because the photolyzing light is modulated
with a certain frequency, a phase-sensitive detection of the current that
will be modulated with the very same frequency will result in a
substantial increase in the signal/noise ratio, allowing the recording of
voltammograms for reactive radicals even in submicromolar concentra-
•
tions. The most direct approach for generating R M (M ) Si, Ge, or
3
Sn; R ) Ph or Bu) consists of photolyzing the pertinent hexaphenyl or
3 2
hexabutyl compounds, (R Ms) , as shown in eq 1.
hV
•
(R Ms)2
9
8 2 R M
(1)
3
3
9
10,11
12
including carbanions, thiophenoxides,
and arylselenoates.
In most cases, however, this approach for promoting the homolytic
bond cleavage cannot be used as the center σ bond in (Ph Ms) is
quite strong, having bond dissociation energies ranging from 45 (M )
On the other hand, quite severe deviations were seen for
3
2
arylsulfenium10,11 and arylselanylium cations12 as the calculated
values underestimated the absolute solvation energies by as
-
1
14
Sn) to 63 (M ) Ge) and 78 kcal mol (M ) Si). In addition, the
solubility of (Ph Ms) is low in MeCN, which is the preferred solvent
for this kind of measurement because of its relatively low reactivity
toward radicals. In the study of (Ph Sns) , we therefore had to resort
-
1
3
2
much as 20-30 kcal mol . We attributed this discrepancy to
the presence of strong covalent bond formation between the
solvent (i.e., acetonitrile) and the cation center. Indeed, the
observed differences could be accounted for by incorporating
such an adduct in the theoretical calculations using a supermo-
lecular approach. An interesting issue, therefore, is whether the
3
2
to THF. An alternative and indirect approach was also considered, in
which tert-butoxyl radicals were generated by photolyzing di-tert-butyl
peroxide (eq 2), followed by hydrogen abstraction from R
3
MH (eq 3).
+
same sort of interaction would be present for the Ph3M studied
hV
•
(
tert-BuOs)2
9
8 2 tert-BuO
(2)
(3)
herein.
•
•
Experimental Section
R MH + tert-BuO 98 R3M + tert-BuOH
3
Materials. All investigated substrates were commercial and used
in the highest grade available without further purification. Acetonitrile
MeCN) and tetrahydrofuran (THF) were obtained from Lab-Scan.
MeCN was dried through a column of activated alumina, while THF
was distilled over sodium and benzophenone. Dimethyl sulfoxide
The latter reaction is very efficient, proceeding by a rate constant as
8
-1 -1
15
high as 3.5 × 10 M
s
3
in the case of Bu SnH (in benzene).
(
The substrate concentration used was 2-500 mM in deaerated THF
or MeCN containing supporting electrolyte, and in the case of the
second photolytic approach (i.e., eqs 2 and 3) 10 vol % di-tert-butyl
peroxide. Samples were allowed to flow through the cell at a rate of
(
DMSO, sure seal bottle) was purchased from Aldrich and used as
received. The supporting electrolytes, tetrabutylammonium perchlorate
Bu NClO ) and tetrabutylammonium tetrafluoroborate (Bu NBF ),
were prepared and purified by standard procedures.
-
1
2
-3 mL min in order to prevent depletion of substrate and product
(
4
4
4
4
accumulation. The chopping frequency and sweep rate employed
-
1
16
throughout the study were 134 Hz and 0.1 V s , respectively. The
voltammograms were corrected for the background current recorded
in the absence of substrate. All half-wave potentials were measured
Apparatus. The instrumental setup used in the photomodulated
voltammetry technique has been described in detail recently.9,10,12 It
consisted of a 100 W Hg-Xe lamp (L8029, Hamamatsu) for generating
radicals photolytically, a water filter, a chopper (651, EG&G Instru-
ments), a lock-in amplifier (SR810 DSP, Stanford Research Systems),
and a home-built electrochemical flow-cell for detecting the radicals.
As working electrode, a net made of carbon fibers (Grafil 34, Grafil
Inc.) or a gold mini-grid (1000 mesh, Dansk Hollandsk Ædelmetal)
was used. A small platinum rod positioned in the outlet of the
electrochemical cell served as counter electrode. The reference electrode
was a Flex-Ref electrode purchased from World Precision Instruments.
+
relative to the standard potential of the ferrocenium/ferrocene (Fc /
Fc) redox couple and converted to SCE by employing the standard
potentials of 0.410 V versus SCE in MeCN and 0.570 V versus SCE
+
17
in THF determined for Fc /Fc elsewhere.
(
13) Wayner, D. D. M.; Griller, D. J. Am. Chem. Soc. 1985, 107, 7764. (b)
Wayner, D. D. M.; McPhee, D. J.; Griller, D. J. Am. Chem. Soc. 1988,
110, 132.
(14) Saiful, I. S. M.; Ohba, Y.; Mochida, K.; Yamauchi, S. Phys. Chem. Chem.
Phys. 2001, 3, 1011.
(15) Shaw, W. J.; Kandandarachchi, P.; Franz, J. A.; Autrey, T. Organometallics
2004, 23, 2080.
(
9) Brinck, T.; Larsen, A. G.; Madsen, K. M.; Daasbjerg, K. J. Phys. Chem.
B 2000, 104, 9887.
(16) No significant changes were observed to the measured half-wave potentials
upon varying the frequency. For a discussion of the theoretical aspects of
the PMV technique, see: Nagaoka, T.; Griller, D.; Wayner, D. D. M. J.
Phys. Chem. Soc. 1991, 95, 6264.
(17) Daasbjerg, K.; Pedersen, S. U.; Lund, H. In General Aspects in the
Chemistry of Radicals; Alfassi, Z. B., Ed.; Wiley: Chichester, U.K., 1999;
Chapter 12, p 410.
(
(
(
10) Larsen, A. G.; Holm, A. H.; Roberson, M.; Daasbjerg, K. J. Am. Chem.
Soc. 2001, 123, 1723.
11) Brinck, T.; Carlqvist, P.; Holm, A. H.; Daasbjerg, K. J. Phys. Chem. A
2
002, 106, 8827.
12) Holm, A. H.; Yusta, L.; Carlqvist, P.; Brinck, T.; Daasbjerg, K. J. Am.
Chem. Soc. 2003, 125, 2148.
2678 J. AM. CHEM. SOC.
9
VOL. 127, NO. 8, 2005

Triphenyl- and Tributyl-Substituted Si-, Ge-, and Sn-Centered Radicals
A R T I C L E S
In LSV, the oxidation process of R
3
M- was studied in the sweep
-
1
•
-
3 3
rate range of 0.1-20 V s in order to characterize the R M /R M
1
8
couple. The anions were generated in concentrations of 1-2 mM
through a deprotonation of the pertinent hydrides using the potassium
-
+
salt of dimsyl, CH S(O)CH K , as base in 0.1 M Bu NClO /DMSO
3
2
4
4
1
9
(eq 4).
R MH + CH S(O)CH -K+ 98 R3M K + CH S(O)CH (4)
-
+
3
3
2
3
3
In MeCN, the deprotonation could only be accomplished successfully
for Ph SnH. All solutions were carefully deaerated with argon before
use. Digital simulations were carried out using the DigiSim version
3
2
0
3
.03 software. The transfer coefficient, R, was set equal to 0.5, and
-
5
2
-1
the diffusion coefficient, D, was assumed to be 10 cm s for all
species involved. All potentials were converted to SCE by measuring
+
them relative to the standard potential of Fc /Fc (0.430 V versus SCE
in DMSO).
•
Figure 1. Photomodulated voltammograms of (A) Ph3Si generated by
1
7
photolysis of 10 vol % (tert-BuOs)2 + 0.5 M Ph3SiH in 0.1 M Bu4NClO4/
Theoretical Approach. Optimized geometries, harmonic vibrational
MeCN at a gold mini-grid electrode; (B) Ph Sn• generated by photolysis
3
frequencies, and energies for all molecules have been computed using
of 0.025 M (Ph3Sns)2 in 0.3 M Bu4NBF4/THF at a carbon fiber net; (C)
•
21,22
Bu Sn generated by photolysis of 0.02 M (Bu Sns) in 0.3 M
Kohn-Sham density functional theory with the B3LYP
exchange functional in Gaussian 98. The LanL2DZ basis set, which
describes the inner electrons of elements heavier than neon by an
effective core potential, was utilized in the calculations. This basis
set was augmented by d-functions on non-hydrogen atoms and diffuse
correlation-
3
3
2
•
2
3
Bu4NBF4/THF at a gold mini-grid electrode; and (D) Bu3Sn generated by
photolysis of 10 vol % (tert-BuOs)2 + 0.02 M Bu3SnH in 0.1 M
Bu4NClO4/MeCN at a gold mini-grid electrode. All voltammograms were
2
4
-1
recorded at a sweep rate of 0.1 V s and corrected for the background
current. Signal average of forward and backward sweeps is shown by the
dashed curves (- - -).
2
5
p-functions on the central C, Si, Ge, and Sn elements. It is well-
known that diffuse functions are necessary to obtain accurate energetics
for processes involving anions. Therefore, electron affinities were
calculated from single point energies obtained with a basis set that in
addition to d-functions was augmented with diffuse p-functions on all
non-hydrogen atoms. The exponents of these functions were in all cases
1
.3 throughout the study. The larger scale factor generally led to an
-
1
increase in the solvation energy by 2-4 kcal mol for the ionic
compounds. Much smaller changes were observed for neutral radicals.
The solvent parameters, including the dielectric constant (36.64), were
the same as those implemented for MeCN in the program, although it
should be noted that the influence of the exact value of the dielectric
constant was found to be rather modest.
2
5
taken from the work of Check et al.
Solvation energies have been calculated at the B3LYP level using
26
the recent implementation of the polarizable continuum model (PCM)
2
3
in Gaussian 98. The same basis sets were used as in the geometry
optimizations. The solute cavities in the calculations were made up of
Spin densities and atomic charges have been computed using
Mulliken population analysis.29 Since this method has been criticized
for being very basis set-dependent and sometimes leading to unrealistic
overlapping spheres centered at the atomic nuclei. The radii of these
spheres were taken as the van der Waals radii of Bondi,27 as
30
results, we have also computed charges using the CHELPG ap-
proach.31 In this latter method, the charges are fitted to reproduce the
computed electrostatic potential of the molecule. The electrostatic
potential (ESP) is a real physical property, and its calculation is not
dependent on the use of a basis set. However, a problem associated
with ESP charge derivations is that charges of atoms buried inside the
molecule are not always well-defined.32 All quantum chemical calcula-
tions were performed using the Gaussian 98 suite of programs.
implemented in Gaussian 98, and scaled by an appropriate factor. A
scale factor of 1.2 has been found to be suitable for computation of
hydration free energies of organic molecules. This is also the default
scale factor in Gaussian 98. However, because the use of this factor
led to numerical problems in the PCM computations for some of the
molecules studied herein, we choose to employ a larger scale factor of
2
8
+
•
(
(
(
18) Note that the R
3
M /R
3
M couple cannot be studied in a similar manner
+
since the stability of the R
19) Olmstead, W. N.; Margolin, Z.; Bordwell, F. G. J. Org. Chem. 1980, 45,
3
M
precursor would be too low.
Results and Discussions
3
295.
20) Rudolph, M.; Feldberg, S. W. DigiSim, version 3.03; Bioanalytical Systems,
Inc.: West Lafayette, IN.
Photomodulated Voltammetry. Figure 1A shows a typical
•
(
(
21) Becke, A. D. J. Chem. Phys. 1993, 98, 5648.
photomodulated voltammogram recorded for Ph3Si . Both an
22) Stephens, P. J.; Devlin, F. J.; Chablovski, C. F.; Frisch, M. J. J. Phys.
Chem. 1994, 98, 11623.
oxidation and a reduction wave are seen corresponding to the
generalized processes depicted in Scheme 1. It is noteworthy
that there is not complete coincidence of the forward and reverse
sweeps, and that the wave widths, |E3/4 - E1/4|, of 120 and 110
mV, respectively, are larger than the value of 56.4 mV expected
for a nernstian process unaffected by preceding or follow-up
(
23) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M.
A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A.; Stratmann,
E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A. D.; Kudin, K.
N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi, M.; Cammi,
R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.; Ochterski, J.;
Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick, D. K.;
Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.; Ortiz,
J. V.; Baboul, A. G.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.;
Komaromi, I.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham,
M. A.; Peng, C. Y.; Nanayakkara, A.; Gonzalez, C.; Challacombe, M.;
Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Andres, J. L.; Head-
Gordon, M.; Replogle, E. S.; Pople, J. A. Gaussian 98, revision A.7;
Gaussian, Inc.: Pittsburgh, PA, 1998.
3
3
chemistry. Furthermore, it was found in separate experiments
ox
/2
that the half-wave potentials, E1 (-0.41 V versus SCE) and
(29) Mulliken, R. S. J. Chem. Phys. 1955, 23, 1833.
(
24) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 270. (b) Hay, P. J.;
Wadt, W. R. J. Chem. Phys. 1985, 82, 284. (c) Hay, P. J.; Wadt, W. R. J.
Chem. Phys. 1985, 82, 299.
25) Check, C. E.; Faust, T. O.; Bailey, J. M.; Wright, B. J.; Gilbert, T. M.;
Sunderlin, L. S. J. Phys. Chem. A 2001, 105, 8111.
26) Cossi, M.; Barone, V.; Cammi, R.; Tomasi, J. Chem. Phys. Lett. 1996,
(30) Politzer, P.; Harris, R. R. J. Am. Chem. Soc. 1970, 92, 6451. (b) Reed, A.
E.; Weinstock, R. B.; Weinhold, F. J. Chem. Phys. 1985, 83, 735. (c)
Williams, D. E. In ReViews in ComputationaI Chemistry; Lipkowitz, K.
B., Boyd, D. B., Eds.; VCH Publishers: New York, 1991; Vol. 2, p 431.
(31) Breneman, C. M.; Wiberg, K. B. J. Comput. Chem. 1990, 11, 361.
(32) Cornell, W. D.; Cieplak, P.; Bayly, C. I.; Kollman, P. A. J. Am. Chem.
Soc. 1993, 115, 9620.
(
(
2
55, 327.
(
(
27) Bondi, A. J. Phys. Chem. 1964, 68, 441.
28) Tomasi, J.; Persico, M. Chem. ReV. 1994, 94, 2027.
(33) Bard, A. J.; Faulkner, L. R. Electrochemical Methods: Fundamentals and
Applications; Wiley: New York, 2001.
J. AM. CHEM. SOC.
9
VOL. 127, NO. 8, 2005 2679

A R T I C L E S
Holm et al.
red
/2
E1 (-1.39 V versus SCE), determined as an average of the
forward and reverse sweeps, were independent of electrode
material and sweep number, indicating that adsorption is not a
crucial factor. Usually the presence of broad waves is due to
slow charge-transfer kinetics at the electrode surface, that is,
quasi-reversible electrode processes, and consequently, the
ox
/2
red
values of E1 and E cannot a priori be set equal to the
1/2
o
o
33
standard potentials, E
+
•
and E
• -
.
At the same
R M /R M
R M /R M
3
3
3 3
time, there may be an effect on the position of the waves from
+
•
fast homogeneous reactions involving R3M and R3M . While
-
•
R3M has a high stability in aprotic solvents, R3M is expected
to dimerize in a fast process and the strong electrophile, R3M ,
+
would react with any nucleophile present in the solution, be it
residual water or the solvent itself.34 These effects are difficult
to estimate quantitatively, but on account of the similar wave
shapes observed in our PMV studies on arylthiyl, arylselanyl,
and R-hydroxyalkyl radicals,35 and the analysis given therein,
-
Figure 2. Linear sweep voltammograms of the oxidation of 1.2 mM Bu Sn
3
10
12
(- - -), Ph Ge- (-‚‚-), and Ph Sn- (;) in 0.1 M Bu NClO /DMSO at a
glassy carbon electrode. Sweep rate ) 0.5 V s . The current is normalized
3
3
4
4
-1
with respect to the electrode area.
we expect the half-wave potentials to be within 100 mV of the
corresponding standard potentials.
•
For Ph3Ge , none of the two approaches in the PMV method
gave reliable results, presumably because the relatively strong
σ bonds in (Ph3Ges)2 and Ph3GesH made it difficult to
14
generate the radical in sufficiently high concentrations. In that
•
respect, the PMV approach worked much better for Ph3Sn and
•
Bu3Sn , as illustrated in Figure 1B-D. The voltammogram for
•
Ph3Sn is characterized by the appearance of a broad reduction
red
/2
wave with E1 ) -0.51 V versus SCE and |E3/4 - E1/4| ≈ 260
mV, while no oxidation wave is observable. The large width
of the wave seriously questions the thermodynamic significance
•
of such a result. For Bu3Sn , it is possible to observe the
reduction process using both approaches in PMV, that is, direct
photolysis of (Bu3Sns)2 in THF (eq 1) and the indirect hydrogen
abstraction method involving Bu3SnH in MeCN (eqs 2 and 3).
The coincidence of the forward and reverse sweeps is poorer
ox
-
Figure 3. Plots of Ep versus log ν for the oxidation of 1.2 mM Bu3Sn
-
-
for the first approach in THF (Figure 1C), but the average value
(2), Ph3Ge (b), and Ph3Sn (9) in 0.1 M Bu4NClO4/DMSO at a glassy
red
carbon electrode. Sweep rates applied are in the range of 0.1-20 V s-1,
of E1 for the two sweeps is not affected and remains constant,
/2
-1
but only the four points pertaining to 0.1-1 V s are included in the linear
red
independent of the sweep number. The measured E1/2 is -1.00
regression analysis.
V versus SCE for the first approach in THF and -1.12 V versus
SCE for the second approach in MeCN. As to the oxidation
process, it was not possible to obtain anything but poorly
reproducible waves. The explanation probably relies on adsorp-
ox
p
wave appears with the peak potential, E , being in the range
3
6
from -1.0 to -0.65 V versus SCE. To elucidate the mech-
anism, Ep was measured as a function of log ν. Αs seen in
Figure 3, the linear dependence observable at low ν in the range
of 0.1-1 V s is replaced by distinct curvature at higher ν.
The slopes of 20 ( 3 mV decade found for the linear parts
r > 0.95) are in fair agreement with the 19.7 mV decade
expected for a rapid electron transfer (eq 5) followed by a fast
dimerization reaction (eq 6), formally described as an ErC2
ox
8
tion phenomena as also noted by Tanner et al. Experiments
were performed with a variety of changes to the setup, different
electrode materials (gold, platinum, or carbon), supporting
electrolytes (Bu4NClO4 or Bu4NBF4), solvents (THF or MeCN),
and concentrations, but with no success. Thus, only a rough
-
1
-
1
2
-1
(
ox
/2
•
estimate of E1 ) 0.20 V versus SCE for Bu3Sn in MeCN can
be provided.
3
3
mechanism.
Linear Sweep Voltammetry. In LSV, the oxidation process
-
of R3M generated from eq 4 was studied in order to
-
•
R M - e a R M
(5)
(6)
•
-
3
3
characterize the R3M /R3M couples and, in this manner, judge
red
the thermodynamic significance of the E1 values obtained in
kd
/2
2R 98 (R Ms)
M•
3 3 2
the PMV approach. In Figure 2, we have collected voltammo-
-
-
-
grams for Ph3Ge , Ph3Sn , and Bu3Sn recorded at a sweep
-
1
As long as the kinetic control is fully by eq 6, that is, when ν
rate, ν, of 0.5 V s in DMSO; the measurements pertaining to
-
1
ox
-
e 1 V s , the relationship between E and the standard
Ph3Si had to be abandoned because of low reproducibility. In
p
o
37,38
potential E
•
-
is given by eq 7.
R M /R M
all three cases, a characteristic irreversible one-electron oxidation
3
3
+
(
34) In practice, the effect of adding nucleophiles has been found to be small in
the PMV approach; see ref 10.
(36) Note that the dimsyl anion, CH
3
S(O)CH -
2
K , also gives rise to an oxidation
peak in the same potential region. This peak could be “titrated” so that
-
•
(35) Lund, T.; Wayner, D. D. M.; Jonsson, M.; Larsen, A. G.; Daasbjerg, K. J.
Am. Chem. Soc. 2001, 123, 12590.
only the peak pertaining to the R
3 3
M /R M couple was visible.
(37) Sav e´ ant, J.-M.; Vianello, E. Electrochim. Acta 1967, 12, 1545.
2680 J. AM. CHEM. SOC.
9
VOL. 127, NO. 8, 2005

Triphenyl- and Tributyl-Substituted Si-, Ge-, and Sn-Centered Radicals
A R T I C L E S
o
•
potentials obtained in the case of Ph3Sn (within 80 mV) shows
that the solvent effect is small. The same conclusion has been
reached in studies pertaining to both localized and delocalized
2
kdC
ox
p
RT
F
o
RT
3F
2RT
3F
E ) 0.902 + E
•
-
-
ln
(7)
R M /R M
(
)
3
3
ν
9
44
carbanions and radical anions.
In the above expression, 2kd is the dimerization rate constant
On the basis of the measurements of potentials, we may now
state that the electrochemical generation of polysilanes and
polygermanes through an anionic polymerization route relies
on the condition that the applied potentials should be more
negative than approximately -1.4 and -0.6 V versus SCE,
respectively, if the further reduction of the pertinent radicals to
form the anions should be accomplished. For most precursors
employed, this should present no problems.
the many important chain processes based on the active role of
Sn-centered radicals, it is important to ensure that the relevant
potentials of any redox agents present are positioned in the range
o
-
and C is the bulk concentration of R3M . For 2kd, we use the
9
-1 -1
•
value of 3.6 × 10 M
s
determined for Bu3Sn by transient
15
absorption spectroscopy in benzene. Thus, we find that
o
o
E
•
-
) -0.61 V versus SCE (DMSO), E
•
-
)
Ph Ge /Ph Ge
Ph Sn /Ph Sn
3
3
3
3
o
-
0.46 V versus SCE (MeCN), E
SCE (DMSO), and E
•
-
) -0.54 V versus
) -0.93 V versus SCE
DMSO), with an estimated uncertainty of (0.05 V.
The deviating points in the plot of Ep versus log ν at large
sweep rates show that the systems under these conditions no
longer are under full kinetic control by eq 6 but rather that there
is a mixed kinetic control by eqs 5 and 6.40 In this regime, a
rough estimation of the standard heterogeneous rate constant,
k°, for the charge-transfer step in eq 5 can be determined on
the basis of simulations of the voltammograms.20 For all of the
Ph Sn /Ph Sn
3
3
o
Bu Sn /Bu Sn
•
-
3
3
3
9
(
4
-6
Likewise, for
ox
7,45
•
from -1 to 0.2 V versus SCE (in the case of Bu3Sn ) if
interference from reduction or oxidation processes should be
avoided.
Before discussing the effect exerted by the different group
IV elements on the potentials obtained, we will turn to a
computation of geometries, charge distributions, ionization
potentials, and electron affinities to get a better understanding
of the essential features of the systems. These data are gathered
in Tables 1 and 2.
-
1
systems studied, we find a value of k° ≈ 0.02 cm s . Thus,
the effect on k° of substituting butyl with phenyl groups in the
•
-
R3Sn /R3Sn couple is small, suggesting that the extent of
delocalization in the phenyl moieties cannot be extensive.
Table 1 summarizes the electrochemical data obtained for
Ph3Si , Ph3Ge , Ph3Sn , and Bu3Sn . For comparison, we have
included literature data for Ph3C to have almost the complete
series of group IV elements.42 Even if experimental difficulties
because of adsorption prevented some of the measurements, the
4
1
•
•
•
•
Optimized Geometries and Computed Charge Distribu-
•
•
•
tions. Optimized geometries of Ph3Sn , Bu3Sn , and the corre-
sponding ions are depicted in Figure 4. The structures of the
remaining radicals and ions studied are available in the
Supporting Information. Concerning the Ph3M series, the
geometry for the different elements exhibits distinct differences.
red
o
consistency between the two sets of E and E
•
-
values
•
1
/2
R M /R M
3
3
•
•
available for Ph3Sn and Bu3Sn with variations of less than 50
and 190 mV (for different solvents), respectively, is acceptable.
This coherence underlines the thermodynamic significance of
the PMV results, despite the broad waves recorded. In a previous
study, the same two Sn-centered radicals were investigated using
PMV. Although the authors did not place a great deal of
confidence in the measured E values of -0.59 V (THF) for
Ph3Sn and -1.13 V versus SCE (MeCN) and -1.29 V versus
SCE (THF) for Bu3Sn , they are relatively close to the present
results with the largest deviation of 290 mV seen in the latter
•
While Ph3C has a planar geometry around the central C atom,
the heavier radicals are characterized by a pyramidal center.
However, the pyramidal character increases gradually going
down the periodic table; the C-M-C angle is 114.3, 112.0,
and 109.1° for Ph3Si , Ph3Ge , and Ph3Sn , respectively. This
suggests that π-resonance stabilization decreases in the same
order. The gradual decrease can be explained by a decreasing
C-M p-orbital overlap as the M atom gets heavier. The much
8
•
•
•
red
1
/2
•
•
•
stronger resonance interaction in Ph C compared to that of the
3
case. The lower limit of -0.43 V versus SCE estimated for the
other radicals is confirmed by the computed spin densities on
the central atom. While the C atom has a spin density of 0.59,
the heavier atoms all have densities around 0.8. These results
do not support the decrease in resonance interaction going from
Si to Sn suggested by the increasing pyramidality. However,
such a discrepancy may be attributed to the known basis set
•
43
oxidation potential of Bu3Sn on the basis of Marcus theory
ox
is well within the 0.20 V versus SCE determined for E1/2
herein. Because of the experimental challenges, we had to resort
to the use of three different organic solvents, namely, MeCN,
THF, and DMSO. However, the similarity of the reduction
46
dependency problem with the Mulliken population analysis.
(
38) Andrieux, C. P.; Nadjo, L.; Sav e´ ant, J.-M. J. Electroanal. Chem. 1973,
•
A comparison of the computed properties for Bu3Sn and
4
2, 223.
) 3.6 × 10 M-1 s-1 determined in benzene is only
1
0
•
(39) The value of 2k
d
Ph3Sn reveals as expected that the C-Sn bond length as well
1
5
slightly lower than the diffusion-controlled limit. For our experiment
as the spin density is larger for the former radical because of
carried out in acetonitrile, which is characterized by a higher limit of 4 ×
10
-1 -1
1
1
0
M
s
(Grampp, G.; Jaenicke, W. Ber. Bunsen-Ges. Phys. Chem.
the lack of resonance interaction.
991, 95, 904), it is possible that 2k
case, this might underestimate E
d
o
will be comparatively larger. In worst
+
For the Ph3M series, all cations are predicted to be planar.
•
-
by 20 mV. For the three
R M /R M
3
3
At the same time, the C-M bond lengths are shorter than the
experiments carried out in DMSO with a diffusion-controlled limit at 6 ×
9
-1 -1
1
0 M
s , the effect will be small.
(
40) Andrieux, C. P.; Sav e´ ant, J.-M. Electrochemical Reactions. In InVestigations
of Rates and Mechanisms of Reactions, Techniques of Chemistry; Ber-
nasconi, C. F., Ed.; Wiley: New York, 1986; Vol. VI/4E, Part 2, p 305.
41) Although the estimated values of k° imply that the electrode kinetics should
(44) Shalev, H.; Evans, D. H. J. Am. Chem. Soc. 1989, 111, 2667.
(45) Tanaka, H.; Ogawa, H.; Suga, H.; Torii, S.; Jutand, A.; Aziz, S.; Suarez,
A. G.; Amatore, C. J. Org. Chem. 1996, 61, 9402.
(
(46) That there indeed is an ambiguity involved in the partitioning of the electron
density can be understood by comparing the charges obtained by Mulliken
1
6,33
be nernstian at the low sweep rates employed in PMV
, the recorded
waves were found to be relatively broad. However, as discussed in our
previous studies on thiyl and selanyl radicals, there seems to be no simple
population analysis and the CHELPG approach. Both approaches show
•
that Ph
3
C has a significantly more negative central atom than the heavier
1
0,12
•
correlation between |E3/4 - E1/4| and k°.
3
Ph M . However, the CHELPG charges are all negative, while the Mulliken
(
42) (a) Klein, E.; Rojahn, W. Tetrahedron Lett. 1969, 2279. (b) Wasielewski,
M. R.; Breslow, R. J. Am. Chem. Soc. 1976, 98, 4222.
charges are positive. In addition, the CHELPG charges become less negative
going from Si to Sn, while the Mulliken charges show a more erratic
behavior. The CHELPG charges support the expected decrease in resonance-
donation to the center atom when going from Si to Sn.
(
43) Eberson, L. Electron Transfer Reactions in Organic Chemistry; Springer-
Verlag: Heidelberg, Germany, 1987.
J. AM. CHEM. SOC.
9
VOL. 127, NO. 8, 2005 2681

A R T I C L E S
Table 1. Half-Wave Potentials, E1/2 and E , and Standard Potentials, E
Holm et al.
-
, Measured for R3M at Carbon-Based or Gold Electrodes at
red
ox
1/2
o
•
R M /R M
•
3
3
25 °C along with Computed Adiabatic Ionization Potentials (IP) and Electron Affinities (EA)
ox a,b
c
red a,b
c
°
R
a,b
IPd
EAd
E1
|
E
3
/
4
−
E
1
/
4
|
E1
|
E
3
/
4
−
E
1
/
4
|
E
M
•
3
R M
-
/
2
/
2
3
/
Ph3C•
0.27
-0.41
e,f
e
-0.97e,g
5.88
5.45
5.52
5.67
5.71
1.43 (1.56)h
1.46
•
-1.39e
-0.51j
Ph3Si
120
110
•
•
i
Ph3Ge
Ph3Sn
Bu3Sn
-0.61
1.76
2.08
1.52
•
i
e
260
170
-0.54 -0.46
(0.20)e,k
210
-1.12 -1.00
e
j
-0.93
i
a
In V versus SCE. b Uncertainty (0.05 V. In mV. d In eV, computed using B3LYP and extended ECP basis sets. e In 0.1 M BuNClO4/MeCN. f From
c
g
ref 42a, measured by polarography at a dropping mercury electrode. From ref 42b, measured by second harmonic alternating current voltammetry at a gold
electrode. Experimental data from NIST Standard Reference Database. In 0.1 M Bu4NClO4/DMSO. In 0.3 M Bu4NBF4/THF. k Uncertain value because
h
i
j
of experimental difficulties.
-
Table 2. Computed Molecular Properties of the Center Atom (M)
of R3M , R3M , and R3M
The geometrical features of the Ph3C series are similar to
•
+
-
•
-
those of the Ph3M series, in that Ph3C has a planar center
while the heavier anions are pyramidal. On the other hand, the
C-M-C angles are generally smaller and the C-M bond
lengths longer for the anions than for the radicals. These
observations indicate, not surprisingly, that the resonance
Geometrya
Chargec
cation
radical
cation
anion
spin densityb radical
anion
Ph3C•
1.473
20
1.866
14.3
1.966
12.0
2.153
09.1
1.457
120
1.814
120
1.902
120
2.081
120
1.463
0.04
0.18 -0.07
0
0
0
0
0
.59
.82
.80
.77
.83
1
1
1
1
1
120
1.929
102.9
2.053
98.6
2.251
95.3
2.271
95.0
-0.87 -0.62 -1.37
Ph3Si•
Ph3Ge•
Ph3Sn•
Bu3Sn•
0.35
-0.39
0.13
-0.25
0.43
-0.03
0.60
0.08
0.76 -0.06
0.36 -1.34
0.49 -0.19
0.51 -1.22
-
•
interactions in Ph3M are weaker than those in Ph3M . Both
the Mulliken and the CHELPG approaches find small differ-
ences in the charge of the central atom when going down the
periodic table. However, while Mulliken population analysis
predicts the central atom to be near neutral, the CHELPG
approach gives a strongly negative center. The geometry around
0.88
0.93 -1.22
1.00 0.21
1.21 -1.17
0.04
2.188
07.5
2.135
120
-
the center atom as well as the charge is similar for Bu3Sn and
Ph3Sn . This supports the notion that resonance stabilization
a
Computed using B3LYP and extended ECP basis sets; optimized C-M
bond lengths are in angstroms and C-M-C angles in degrees (shown in
-
b
-
italics). Computed M atom spin density using Mulliken population analysis.
in Ph3Sn is relatively small.
c
Computed M atom charges using Mulliken population analysis and the
Ionization Potentials and Electron Affinities. Computed
ionization potentials (IP) and electron affinities (EA) are
included in Table 1. Unfortunately, it is difficult to estimate
the reliability of these values because of the lack of experimental
CHELPG approach (shown in italics).
•
data. The only data we have found for the Ph3M series is the
•
48
EA of Ph3C , which has been measured to be 1.56 ( 0.10 eV.
This is in relatively good agreement with our computed value
of 1.43 eV. As a test of our computational approach, we
•
calculated the IP and EA for the related PhCH2 . The computed
and experimental IP values are 7.09 and 7.24 eV, respectively.
The corresponding EA values are 0.80 and 0.91 eV, respectively.
Thus, our comparisons with experiment indicate that the average
error in the computed IPs and EAs is close to 0.15 eV. This is
also in agreement with the expected accuracy of the B3LYP
4
9
method from testing on the extended G2 set and from our
previous calculations on arylthiyl and arylselanyl radicals.1
It should be noted that the present study contains compounds
with heavier elements than in previous studies, and thus it is
possible that the accuracy in these cases is slightly lower.
•
+
-
Figure 4. Geometries of R3Sn , R3Sn , and R3Sn optimized using B3LYP
and extended ECP basis sets.
1,12
sum of the C and M covalent radii,47 pointing to the presence
of resonance interactions. However, the computed charges
indicate that the resonance-donation to the center decreases
significantly when going down the periodic table, and both
approaches suggest that Ph3Sn has a center charge of around
.9 while for Ph3C it is much less positive. The CHELPG
•
The IP of 5.88 eV obtained for Ph3C is significantly larger
•
than the IP of 5.45 eV for Ph3Si . This may be attributed to the
larger resonance stabilization of the former radical as indicated
+
•
•
by the fact that Ph3C has a planar center, whereas Ph3Si is
pyramidal. There is a relatively small but systematic increase
+
0
approach even finds a large negative charge on the central C
•
•
•
in IP along the series Ph3Si , Ph3Ge , and Ph3Sn , which can be
seen as an effect of a decreasing resonance stabilization of the
corresponding cations for the heavier elements. In general, the
effect of the center atom thus appears to be rather modest on
the IPs. Moreover, the difference between the IPs for Bu3Sn•
+
atom in Ph3C . The difference in geometry and charge distribu-
tion between Bu3Sn+ and Ph3Sn is relatively small, which
confirms that resonance interactions are not particularly sig-
nificant in the latter case.
+
(
47) The bond lengths of C-C, C-Si, C-Ge, and C-Sn, as estimated from
the covalent radii (Cotton, F. A.; Wilkinson, G.; Gaus, P. L. Basic Inorganic
Chemistry; Wiley: New York, 1987), are 1.54, 1.94, 1.99, and 2.17 Å,
respectively.
(48) NIST Standard Reference Database No. 69.
(49) Curtiss, L. A.; Raghavachari, K.; Redfern, P. C.; Pople, J. A. J. Chem.
Phys. 2000, 112, 7374.
2682 J. AM. CHEM. SOC.
9
VOL. 127, NO. 8, 2005

Triphenyl- and Tributyl-Substituted Si-, Ge-, and Sn-Centered Radicals
A R T I C L E S
•
o
Sol
and Ph3Sn is only 0.04 eV, showing that the effect of butyl
Table 3. Differentiated Solvation Energies, ∆∆G (+•)exp and
o
•
and phenyl groups is similar.50
The EA generally increases going from Ph3C to Ph3Sn . This
is most likely an effect of the increasing charge capacity for
the center atom along this series. The charge capacity refers to
the ability of an atom or group to accept or donate charge.
Consideration of charge capacity has allowed the explanation
∆∆G
(-•) , Obtained for R M in MeCN from Experimental
Solution Data and Computed Gas-Phase Data along with
Computed Differentiated Solvation Energies,
Sol
exp
3
•
•
o
Sol
o
Sol
∆∆G (+•)PCM and ∆∆G (-•)PCM
o
sol
a,b
o
sol
a,c
o
sol
a,b
o
sol
a,c
∆
∆G
(+•)
exp
∆∆G
(
+•)
PCM
∆∆G
(−•)
exp
∆∆G
(−•
)
PCM
5
1
Ph3C•
Ph3Si
Ph3Ge
-20
-26
-30.0
-31.7
-31.9
-32.8
g
-54 (-51)d
-44
-40.8
-40.6
-38.6
-38.0
g
•
•
•
-55e
of a number of seemingly anomalous aspects of chemical
behavior,52 such as the low EA of the fluorine atom and that
•
Ph Sn
3
-51
Bu3Sn
(-18)f
-48
the gas-phase acidity of fluoroacetic acid is lower than that of
chloroacetic acid. However, resonance stabilization may also
play a certain role, as the EAs of Ph3C and Ph3Si are almost
identical despite the fact that Si has a significantly higher charge
capacity. As discussed above, resonance stabilization is larger
a
In kcal mol-1. b Uncertainty (5 kcal mol-1. c Computed using B3LYP
and extended ECP basis sets. Computed from the experimental EA. In
DMSO. Uncertain value because of experimental difficulties. Values
could not be obtained because of numerical problems in the PCM
calculations.
•
•
d
e
f
g
-
-
for Ph3C than for Ph3Sn . In a similar manner, resonance
stabilization of the anion might, in principle, explain why the
Solvation Energies. The extraction of solvation energies
based on the experimentally obtained redox potentials can be
•
•
EA of Ph3Sn is 0.56 eV larger than that of Bu3Sn , although
the fact that the phenyl group in contrast to the butyl group is
an inductive electron acceptor should be considered, as well.
In other words, there seems to be a comparatively large group
effect on EA.
Returning now to the measured potentials for Ph3C in Table
, we have a better basis for understanding that the order is Sn
Ge > C > Si with respect to the reduction potentials and C
Si for the two oxidation potentials. From a qualitative point
of view, it would be expected that Ph3C with the central C
atom having an electronegativity of 2.5 should be the radical
relatively hardest to oxidize and easiest to reduce. In comparison,
the electronegativities of Sn, Ge, and Si are all in the range of
.7-1.8. However, the effect from the higher charge capacity
of Ge and Sn also comes through and provides an explanation
for the less negative values of E for Ph3Ge and Ph3Sn
compared to that for Ph3C . Moreover, the relatively strong
delocalization present for the planar Ph3C in contrast to the
other pyramidal radicals will tend to make the redox potentials
of Ph3C more extreme, that is, shift the E and E values in
positive and negative directions, respectively. The change
observed in E1 upon replacing the phenyl groups in R3Sn by
butyl groups constitutes -490 mV in THF and -390 mV in
DMSO. This is mainly due to delocalization and, in particular,
inductive effects for Ph3Sn as discussed for the EA values
above. Note that the group effect amounts to as much as 700
mV17 when the center atom is C due to its smaller charge
capacity. One final important factor influencing the values of
5
4,55
accomplished
through the use of eqs 8 and 9, as described
9
-12
in our previous work.
o
sol
o
sol
o
sol
∆∆G (+•) ≡ ∆G (+) - ∆G (•)
)
exp
exp
exp
o
•
-
IP + FE
+
•
+ C (8)
- C (9)
R M /R M
3
3
1
>
>
o
sol
o
sol
o
sol
∆
∆G (-•) ≡ ∆G (-) - ∆G (•)
)
exp
exp
exp
o
•
EA - FE
•
-
R M /R M
3
3
o
sol
o
sol
o
sol
The parameters, ∆G (+), ∆G (-), and ∆G (•), denote
the solvation energies of the cations, anions, and radicals,
respectively. Without comparison, the major contributions to
1
o
o
o
∆
∆G (+•) and ∆∆G (-•) will originate from the ∆G (+)
sol sol sol
o
red
•
•
1
/2
and ∆G (-) terms because of the strong solvation of char-
sol
•
56
ges. The potentials entered in these expressions are the values
•
ox
/2
o
red
1/2
of E1 and either E
•
-
or E obtained in acetonitrile,
pertaining to DMSO presenting
R M /R M
3
3
o
with the value of E
•
-
ox
red
1/2
Ph Ge /Ph Ge
3
3
•
ox
red
1
/2
the only exception. The use of E1 and E imposes an
/2 1/2
53
additional uncertainty in the extracted numbers because their
thermodynamic significance is not well-defined. Still, the
coherence between the results obtained with the PMV and LSV
techniques justifies the inclusion of these potentials in the
red
/2
•
-
-1
calculations. For the constant C, a number of 109.3 kcal mol
is used, which originates from the value of the absolute potential
57,58
of the standard calomel electrode (-4.74 V).
In Table 3, the experimentally based solvation data,
and ∆∆G (-•) , are collected along with
the corresponding computed values, ∆∆G (+•)PCM and
∆G (-•)PCM, obtained from the PCM approach, as de-
scribed in the Theoretical Approach section. When first the
solvation energies of the anions is considered, it may be noted
that the two sets of ∆∆G (-•) values lie between -55 and
38 kcal mol . This relatively strong solvation is indicative
of a rather localized negative charge, which is consistent with
a pyramidal geometry around the center atom in the anions.
Still, ∆∆G (-•)exp is consistently found to be more negative
red
/2
ox
1/2
E1 and E is the solvation of the species involved. The
o
sol
o
sol
∆∆G (+•)
exp
exp
solvation energies constitute the main difference between the
redox potentials and the corresponding gas-phase parameters,
EA and IP. In the next section, this issue will be our main
concern.
o
sol
o
∆
sol
o
sol
(
50) Although this result cannot be related to oxidation potentials for any of
-
1
-
the present systems, it might be mentioned that quite similar values are
•
ox
/2
obtained for the two carbon-centered radicals, Me
3
C
(E1 ) 0.09 V
versus SCE)1 and Ph
3b
C (E ) 0.27 V versus SCE).
•
ox
1/2
42a
3
(
(
(
51) Huheey, J. E. J. Phys. Chem. 1965, 69, 3284. (b) Huheey, J. E.; Watts, J.
o
C. Inorg. Chem. 1971, 10, 1553. (c) Politzer, P. J. Chem. Phys. 1987, 86,
sol
1
072.
o
than ∆∆G (-•)PCM. Presumably, this should be attributed to
52) Politzer, P.; Huheey, J. E.; Murray, J. S.; Grodzicki, M. J. Mol. Struct.
sol
(
THEOCHEM) 1992, 259, 99. (b) Brinck, T.; Murray, J. S.; Politzer, P.
Inorg. Chem. 1993, 32, 2622.
(54) Matsen, F. A. J. Chem. Phys. 1956, 24, 602.
(55) Peover, M. E. Trans. Faraday Soc. 1962, 58, 1656.
53) For a discussion of radical stabilities, see also: (a) Bordwell, F. G.; Bausch,
M. J. J. Am. Chem. Soc. 1986, 108, 1979. (b) Bordwell, F. G.; Cheng,
J.-P.; Ji, G.-Z.; Satish, A. V.; Zhang, X.-M. J. Am. Chem. Soc. 1991, 113,
(56) The use of IP and EA in these equations is based on the approximation
that they are equal to the corresponding Gibbs energy entities.
(57) Reiss, H.; Heller, A. J. Phys. Chem. 1985, 89, 4207.
(58) Lim, C.; Bashford, D.; Karplus, M. J. Phys. Chem. 1991, 95, 5610.
9
790. (c) Bordwell, F. G.; Zhang, X.-M.; Alnajjar, M. S. J. Am. Chem.
Soc. 1992, 114, 7623.
J. AM. CHEM. SOC.
9
VOL. 127, NO. 8, 2005 2683

A R T I C L E S
Holm et al.
+
the fact that we were forced to use a larger scale factor than
to the more positive charge on the center atom in Ph3Si , as
recommended for the cavity generation in the PCM calculations,
found by both the Mulliken and the CHELPG approaches. In
fact, the PCM method predicts a consistent decrease in
o
leading to an increase in ∆∆G (-•)PCM by around 2-4 kcal
sol
o
+
+
-
1
mol . In addition, the recommended scale factor has been
optimized for neutral molecules, and calculations on anions are
often found to give too positive solvation energies. Furthermore,
the scale factor is optimized for Hartree-Fock wave functions
which generally give more polarized charge distributions and
∆∆G (+•)PCM going from Ph3C to Ph3Sn , in agreement
sol
with the computed change in charge distribution using the
CHELPG method.
Finally, some summarizing comments on the PCM approach
seem appropriate. In a series of articles dealing with different
9
,11,12
thus more negative solvation energies. Considering such effects,
organic cations and anions,
we have analyzed its perfor-
o
it is not unlikely that ∆∆G (-•)PCM is overestimated by 6-8
mance in the computation of solvation energies. First of all,
we find that the accuracy differs greatly for different types of
molecules. The reason for this behavior can be attributed to the
great dependence of the solvation energy on the cavity size
found for PCM and other solvation methods based on dielectric
continuum theory. In general, the accuracy can be increased by
using specially optimized parameters that determine the size
and shape of the cavity based on the hybridization and chemical
environment of the atoms in the investigated compounds.59
However, when studying novel molecules, such optimized
parameters are seldom available or applicable. The present
analysis shows that the average error in the computed solvation
energy may for some sets of compounds be as high as 10 kcal
sol
-
1
kcal mol , which would put the computed values in better
agreement with the experimental ones. Still, some noteworthy
o
sol
-
exceptions remain. For instance, ∆∆G (-•)exp for Ph3C is
-
1
o
1
3 kcal mol more negative than the ∆∆G (-•)PCM value.
The difference is reduced to 10 kcal mol if the experimen-
sol
-
1
tal rather than the computational EA is used to calculate
o
sol
∆
∆G (-•)exp. A plausible explanation for the remaining
difference between the two values could be that there is a
difference between the gas phase and solution geometry. The
-
gas-phase geometry optimization shows that Ph3C in contrast
to the other anions is planar around the center atom. However,
a pyramidal center will give a more localized charge and
therefore a more negative solvation energy. Thus, it is possible
-
1
mol when using standard cavity parameters based on regular
van der Waals radii. On the other hand, in our studies of
parasubstituted thiophenoxides and phenylselenoates, experi-
mentally and theoretically based solvation energies agreed within
-
that a pyramidal geometry for Ph3C is favored in solution. Also
-
for Ph3Sn , there is a relatively large difference between
o
o
∆
∆G (-•)PCM and ∆∆G (-•)exp. Furthermore, it is surpris-
sol sol
o
-
1 11,12
-
-1
2-4 kcal mol .
The average errors on the substituent effects
ing that ∆∆G (-•)exp for Ph3Sn is 7 kcal mol more
negative than that for Ph3Si since both are pyramidal and since
sol
-
on the solvation energies were even lower. These results are
remarkably good considering the expected errors of a few kcal
the larger size of Sn should lead to a less effective solvation.
In this case, we speculate that the accuracy in the computed
-
1
mol in the experimental solvation energies and show that a
relatively high accuracy can be expected of the PCM approach
for analysis of substituent effects. We have also found that when
experimentally determined solvation energies are considerably
more negative than the PCM results, this can often be attributed
to strong specific interactions present between the solvent and
solute. This was demonstrated for the solvation of arylsulfenium
and arylselanylium ions in MeCN, where covalently bonded
adducts were identified by theoretical calculations.1 In such
cases, we have shown that accurate solvation energies can be
obtained by using the PCM method in a supermolecular
approach, where the important solvent molecules are explicitly
considered in the calculations.
•
•
EA of Ph3Sn or Ph3Si may be too low for an accurate
prediction of the solvation energy. The solvation energies of
Ph3Sn- and Bu3Sn- are almost equal as expected on account
of the rather similar charges and geometries computed. For none
of the anions studied, the solvation is strong enough to indicate
the presence of any covalent or charge-transfer interactions with
the solvent. This result is in line with the observation that the
1,12
•
solvent effect on the reduction potentials for Ph3Sn and
•
Bu3Sn constitutes less than 190 mV. If strong interactions
existed between the anions and the solvent, much larger
differences would have been expected.
o
The two sets of ∆∆G (+•) values are in the range from
sol
-
1
-
33 to -20 kcal mol . These values are significantly less
Conclusions
o
sol
negative than the corresponding ∆∆G (-•) numbers, which
Redox potentials of a number of triphenyl- or tributyl-
is expected in the sense that the planar geometry around the
•
substituted Si-, Ge-, or Sn-centered radicals, R3M , have been
+
center atom for Ph3M leads to more spread out charge than
measured in acetonitrile, tetrahydrofuran, or dimethyl sulfoxide
by photomodulated voltammetry or through a study of the
oxidation process of the corresponding anions in linear sweep
voltammetry. The combined results provide the ranges of
potentials in which to expect radical and ion chemistry. This
information is of great importance in the understanding and
control of polymerization processes involving such species. For
the reduction potentials determined in more than one solvent,
the variation is less than 190 mV. As far as the effect exerted
by the different group IV elements on the data obtained for the
for a pyramidal geometry and accordingly to a weaker sol-
vation. At the same time, this also indicates that there is no
strong covalent interaction present with the solvent for these
1
0,11
kinds of cations in contrast to arylsulfenium
and aryl-
1
2
selanylium ions in MeCN. However, it is surprising that
∆G (+•)PCM is more negative than ∆∆G (+•)exp. Because
o
o
∆
sol sol
of the too-large scale factor employed in the PCM calculation,
we would rather have anticipated the opposite order. This seems
to suggest that the discrepancy between experiment and theory,
at least partly, should be attributed to errors in the computed
IP and/or the experimental oxidation potentials. It is interesting
to note that both computations and experiments predict a more
•
Ph3M series is concerned (including literature results for M )
C), the order of reduction potentials follows Sn > Ge > C >
Si, while for the two oxidation potentials, it is C > Si. As to
+
+
negative solvation energy for Ph3Si than for Ph3C , despite
the fact that carbon is the smaller atom. This could be attributed
(59) Barone, V.; Cossi, M.; Tomasi, J. J. Chem. Phys. 1997, 107, 3210.
2684 J. AM. CHEM. SOC.
9
VOL. 127, NO. 8, 2005

Triphenyl- and Tributyl-Substituted Si-, Ge-, and Sn-Centered Radicals
A R T I C L E S
•
+
the substituent effect of the R group in R3Sn , the reduction
planar geometry around the center atom in R3M leads to more
spread out charge and a weaker solvation. Calculated solvation
energies based on the PCM approach can reveal these trends
roughly, but they are not able to reproduce the experimentally
derived values on a detailed level for these types of ions.
potential is more negative by 490 mV in tetrahydrofuran (390
mV in dimethyl sulfoxide) when R is a butyl rather than a phenyl
group. These trends can be explained qualitatively through a
combination of effects, such as charge capacity being most
pronounced for the heavier elements, resonance stabilization
present for the planar Ph3C and all R3M , and finally a
contribution from solvation as supported by quantum-chemical
calculations. The solvation is found to be stronger for R3M
•
+
Acknowledgment. Statens Naturvidenskabelige Forskning-
sråd and Torkil Holms Fond are thanked for financial support.
-
Supporting Information Available: Tables of absolute ener-
+
than for R3M , indicating that no covalent interaction exists
between the cations and the solvent as observed for aryl-
sulfenium and arylselanylium ions in acetonitrile. The relatively
gies and Cartesian coordinates, along with a figure of the
•
+
-
optimized geometries of R3M , R3M , and R3M . This material
is available free of charge via the Internet at http://pubs.acs.org.
-
strong solvation of R3M is attributed to the presence of a rather
localized negative charge in the pyramidal geometry, while the
JA044687P
J. AM. CHEM. SOC.
9
VOL. 127, NO. 8, 2005 2685