Mapping the influence of molecular structure on rates of electron transfer using direct measurements of the electron spin-spin exchange interaction

Article abstract of DOI:10.1021/ja029062a

The spin-spin exchange interaction, 2J, in a radical ion pair produced by a photoinduced electron transfer reaction can provide a direct measure of the electronic coupling matrix element, V, for the subsequent charge recombination reaction. We have develo

Full text of DOI:10.1021/ja029062a

Published on Web 03/06/2003
Mapping the Influence of Molecular Structure on Rates of
Electron Transfer Using Direct Measurements of the Electron
Spin-Spin Exchange Interaction
Aaron S. Lukas, Patrick J. Bushard, Emily A. Weiss, and Michael R. Wasielewski*
Contribution from the Department of Chemistry & Center for Nanofabrication and Molecular
Self-Assembly, Northwestern UniVersity, EVanston, Illinois 60208-3113
Received October 23, 2002; E-mail: wasielew@chem.northwestern.edu
Abstract: The spin-spin exchange interaction, 2J, in a radical ion pair produced by a photoinduced electron
transfer reaction can provide a direct measure of the electronic coupling matrix element, V, for the
subsequent charge recombination reaction. We have developed a series of dyad and triad donor-acceptor
molecules in which 2J is measured directly as a function of incremental changes in their structures. In the
dyads the chromophoric electron donors 4-(N-pyrrolidinyl)- and 4-(N-piperidinyl)naphthalene-1,8-dicarbox-
imide, 5ANI and 6ANI, respectively, and a naphthalene-1,8:4,5-bis(dicarboximide) (NI) acceptor are linked
to the meta positions of a phenyl spacer to yield 5ANI-Ph-NI and 6ANI-Ph-NI. In the triads the same
structure is used, except that the piperidine in 6ANI is replaced by a piperazine in which a para-X-phenyl,
where X ) H, F, Cl, MeO, and Me2N, is attached to the N′ nitrogen to form a para-X-aniline (XAn) donor
to give XAn-6ANI-Ph-NI. Photoexcitation yields the respective 5ANI+-Ph-NI-, 6ANI+-Ph-NI-, and
XAn+-6ANI-Ph-NI- singlet radical ion pair states, which undergo subsequent radical pair intersystem
3
crossing followed by charge recombination to yield *NI. The radical ion pair distances within the dyads
are about 11-12 Å, whereas those in the triads are about ∼16-19 Å. The degree of delocalization of
charge (and spin) density onto the aniline, and therefore the average distance between the radical ion
pairs, is modulated by the para substituent. The 3*NI yields monitored spectroscopically exhibit resonances
as a function of magnetic field, which directly yield 2J for the radical ion pairs. A plot of ln 2J versus rDA
,
the distance between the centroids of the spin distributions of the two radicals that comprise the pair,
yields a slope of -0.5 ( 0.1. Since both 2J and kCR, the rate of radical ion pair recombination, are directly
proportional to V2, the observed distance dependence of 2J shows directly that the recombination rates in
these molecules obey an exponential distance dependence with â ) 0.5 ( 0.1 Å-1. This technique is very
sensitive to small changes in the electronic interaction between the two radicals and can be used to probe
subtle structural differences between radical ion pairs produced from photoinduced electron transfer
reactions.
Introduction
encounter complex, which is followed by diffusive separation
of the radical ions or conformational changes in the linked
The spin dynamics of photogenerated radical ion pairs are
sensitive to external applied magnetic fields, which permit the
manipulation of both the radical ion pair lifetimes and the yields
of products arising from radical ion pair decay.1-3 However,
the presence of a large isotropic spin-spin exchange interaction,
2J, between singlet and triplet radical ion pair states often
inhibits the mixing of these states that is necessary for the
observation of magnetic field effects (MFEs) at easily accessible
magnetic fields. For this reason MFEs are most often observed
in radical ion pairs that are either freely diffusing4 or tethered
via long polymethylene linkers to ensure that 2J is small.5 In
these systems efficient charge separation (CS) occurs in an
systems, both of which increase the distance between the radical
ions, allowing 2J to diminish. Such flexibility results in a wide
distribution of radical ion pair conformations and distances,
thereby complicating analysis of how the magnitude of 2J
correlates with the specific structure of the radical ion pair and
the distance between the two radical ions. Staerk and co-workers
have shown that D-A compounds both with fully flexible
polymethylene linkers and with polymethylene linkers incor-
porating a trans-cyclohexyl group to limit their conformational
freedom sometimes exhibit a resonance in the magnetic field
dependence of the triplet yield following radical ion pair
recombination.5,6
(1) Steiner, U. E.; Ulrich, T. Chem. ReV. 1989, 89, 51-147.
(2) Grissom, C. B. Chem. ReV. 1995, 95, 3-24.
(3) Hoff, A. J.; Gast, P.; van der Vos, R.; Franken, E. M.; Lous, E. J. Z. Phys.
Chem. 1993, 180, 175-192.
The bacterial photosynthetic reaction center is the classic
example of a rigid system that displays large MFEs on the yield
of its triplet charge recombination product. Electron transport
within photosynthetic bacteria is initiated by electron transfer
(4) Schulten, K.; Staerk, H.; Weller, A.; Werner, H.-J.; Nickel, B. Z. Phys.
Chem. 1976, 101, 371-390.
(5) Staerk, H.; Kuhnle, W.; Treichel, R.; Weller, A. Chem. Phys. Lett. 1985,
188, 19-24.
(6) Werner, U.; Kuhnle, W.; Staerk, H. J. Phys. Chem. 1993, 97, 9280-9287.
9
10.1021/ja029062a CCC: $25.00 © 2003 American Chemical Society
J. AM. CHEM. SOC. 2003, 125, 3921-3930
3921

A R T I C L E S
Lukas et al.
Chart 1
Chart 2
Figure 1. Radical ion pair energy levels as a function of magnetic field.
from the lowest excited state of a bacteriochlorophyll dimer
(P) to an adjacent bacteriochlorophyll (B), followed by sequen-
tial thermal electron transfers to bacteriopheophytin (H), and
finally to a quinone (Q),7-11 forming the P+-B-H-Q- radical
ion pair in about 200 ps. Removal or chemical reduction of Q,
followed by photoexcitation produces the P+-B-H- radical ion
pair, which is formed in a pure singlet state. The 1[P+-B-H-]
radical ion pair singlet state undergoes radical pair intersystem
crossing (RP-ISC) by mixing with the three sublevels of the
1
3
triplet state, [P+-B-H-] T [P+-B-H-], driven primarily
by the electron-nuclear hyperfine interactions in each radical
ion of the pair. Subsequent radical ion pair recombination (CR)
of 0.92. This radical ion pair undergoes RP-ISC to yield
3[MeOAn+-6ANI-Me2Ph-NI-], which recombines to give a
high yield of MeOAn-6ANI-Me2Ph-3*NI. The sublevels of
this localized triplet state have unique non-Boltzmann spin
populations, which result in spin-polarized EPR spectra that can
be detected using time-resolved EPR spectroscopy (TREPR),
which closely mimic that observed for photosynthetic reaction
centers.18-21 Subsequently, Gust and co-workers studied a
donor-acceptor triad based on a chromophoric porphyrin donor
coupled to a fullerene electron acceptor and carotenoid second-
ary donor that also displays the TREPR spin polarization pattern
characteristic of RP-ISC.15 They also showed that its CR rate
was decreased by 50% in the presence of a relatively weak
magnetic field (B ) 0.041 T).22 The spin dynamics of radical
ion pair recombination on the picosecond scale has also been
controlled using applied magnetic fields of several tesla in
ferrocene complexes.23 Up until now there have been no
investigations into magnetic field effects on radical ion pairs
and their products within a series of structurally rigid intramo-
lecular D-A compounds.
3
produces *P-B-H. Application of a magnetic field as small
as 0.01 T lowers the yield of 3*P-B-H by about 40%,12 which
is attributed to Zeeman splitting of the triplet sublevels that
diminishes mixing between the singlet and two of the three
triplet radical ion pair states, Figure 1.
Despite many studies of rigid D-A arrays synthesized to
mimic the electron transfer dynamics of photosynthetic bacteria,
spin selective CR to a localized triplet state of the donor or the
acceptor has been observed in only a few cases.13-17 Previous
work from this laboratory demonstrated the first donor-acceptor
molecule, MeOAn-6ANI-Me2Ph-NI, Chart 1, which closely
mimics the spin-dependent CR dynamics within the photosyn-
thetic reaction center.13,14 Transient absorption spectroscopy in
toluene determined that excitation of the 6ANI chromophore
results in stepwise CS: MeOAn-1*6ANI-Me2Ph-NI f
1[MeOAn+-6ANI--Me2Ph-NI]
f
1[MeOAn+-6ANI-
Me2Ph-NI-] to form the final radical pair with a quantum yield
(7) Ogrodnik, A.; Michel-Beyerle, M. E. Z. Naturforsch. 1989, 44a, 763-
764.
We now present data on a family of closely related rigid
donor-acceptor dyads and triads, Chart 2. In the dyads the
chromophoric electron donors are 4-(N-pyrrolidinyl) and 4-(N-
piperidinyl)naphthalene-1,8-dicarboximide, 5ANI and 6ANI,
respectively, in which the imide groups are linked in a meta
(8) DiMagno, T. J.; Norris, J. R. In The Photosynthetic Reaction Center;
Deisenhofer, J., Norris, J. R., Eds.; Academic Press: San Diego, 1993; Vol.
2, pp 105-133.
(9) Sumi, H.; Kakitani, T. Chem. Phys. Lett. 1996, 252, 85-93.
(10) Plato, M.; Mobius, K.; Michel-Beyerle, M. E.; Bixon, M.; Jortner, J. J.
Am. Chem. Soc. 1988, 110, 7279-7285.
(11) Holzapfel, W.; Finkele, U.; Kaiser, W.; Oesterhelt, D.; Scheer, H.; Stilz,
H. U.; Zinth, W. Chem. Phys. Lett. 1989, 160, 1-7.
(12) Blankenship, R. E.; Schaafsma, T. J.; Parson, W. W. Biochim. Biophys.
Acta 1977, 461, 297-305.
(18) Thurnauer, M. C.; Katz, J. J.; Norris, J. R. Proc. Natl. Acad. Sci. U.S.A.
1975, 72, 3270.
(13) Hasharoni, K.; Levanon, H.; Greenfield, S. R.; Gosztola, D. J.; Svec, W.
A.; Wasielewski, M. R. J. Am. Chem. Soc. 1995, 117, 8055-8056.
(14) Hasharoni, K.; Levanon, H.; Greenfield, S. R.; Gosztola, D. J.; Svec, W.
A.; Wasielewski, M. R. J. Am. Chem. Soc. 1996, 118, 10228-10235.
(15) Carbonera, D.; Di Valentin, M.; Corvaja, C.; Agostini, G.; Giacometti, G.;
Liddell, P. A.; Kuciauskas, D.; Moore, A. L.; Moore, T. A.; Gust, D. J.
Am. Chem. Soc. 1998, 120, 4398-4405.
(19) Dutton, P. L.; Leigh, J. S.; Seibert, M. Biochem. Biophys. Res. Commun.
1972, 46, 406.
(20) Regev, A.; Nechushtai, R.; Levanon, H.; Thornber, J. P. J. Phys. Chem.
1989, 93, 2421.
(21) Rutherford, A. W.; Paterson, D. R.; Mullet, J. E. Biochim. Biophys. Acta
1981, 635, 205.
(16) Wiederrecht, G. P.; Svec, W. A.; Wasielewski, M. R. J. Am. Chem. Soc.
1999, 121, 7726-7727.
(17) Wiederrecht, G. P.; Svec, W. A.; Wasielewski, M. R.; Galili, T.; Levanon,
H. J. Am. Chem. Soc. 2000, 122, 9715-9722.
(22) Kuciauskas, D.; Liddell, P. A.; Moore, A. L.; Moore, T. A.; Gust, D. J.
Am. Chem. Soc. 1998, 120, 10880-10886.
(23) Gilch, P.; Pollinger-Dammer, F.; Musewald, C.; Michel-Beyerle, M. E.;
Steiner, U. E. Science 1998, 281, 982-984.
9
3922 J. AM. CHEM. SOC. VOL. 125, NO. 13, 2003

Electron Spin−Spin Exchange Interaction
A R T I C L E S
Chart 3
Information. Characterization was performed with a Gemini 300 MHz,
Varian 400 MHz, or INOVA 500 MHz NMR and a PE BioSystems
MALDI-TOF mass spectrometer. All solvents were spectrophotometric
grade or distilled prior to use.
Cyclic voltammetry measurements were performed in butyronitrile
solution containing 0.1 M tetra-n-butylammonium perchlorate electro-
lyte using a CH Model 622 electrochemical workstation. A 1.0 mm
diameter platinum disk electrode, platinum wire counter electrode, and
Ag/AgxO reference electrode were employed. The ferrocene/ferrocinium
couple (Fc/Fc+, 0.52 vs SCE) was used as an internal reference for all
measurements.
Absorption measurements were made on a Shimadzu (UV-1601)
spectrophotometer. The optical density of all samples was maintained
between 0.7 and 1.0 at 416 nm (ꢀ6ANI,416nm ) 7000 cm-1 M-1).
Femtosecond transient absorption measurements were made using the
420 nm frequency-doubled output from a regeneratively amplified
titanium sapphire laser system operating at 2 kHz as the excitation
pulse.30 Samples were placed in a 2 mm path length quartz cuvette
and stirred using a motorized wire stirrer. Nanosecond transient
absorption measurements were made using the 416 nm, H2-Raman
shifted output from a frequency-tripled 10 Hz Nd:YAG laser (Quan-
taRay DCR-2). Samples were placed in a 10 mm path length quartz
cuvette equipped with a vacuum adapter and subjected to five freeze-
pump-thaw degassing cycles prior to transient absorption measure-
ments. The probe light in the nanosecond experiments was generated
using a xenon flashlamp (EG&G Electro-Optics FX-200) and detected
using a photomultiplier tube with high voltage applied to only 4 dynodes
(Hamamatsu R928). The total instrument response is 7 ns and is
determined primarily by the laser pulse duration. The sample cuvette
was placed between the poles of a Walker Scientific HV-4W
electromagnet powered by a Walker Magnion HS-735 power supply.
The field strength was measured by a Lakeshore 450 gaussmeter with
a Hall effect probe. Both the electromagnet and the gaussmeter were
interfaced with the data collection computer, allowing measurement
and control of the magnetic field to (1 × 10-5 T during data
acquisition.
relationship to a phenyl spacer to which a naphthalene-1,8:4,5-
bis(dicarboximide) (NI) electron acceptor is attached. In the
triads the same structure is used, except that a 4-(N-piperazinyl)-
naphthalene-1,8-dicarboximide serves as a chromophoric elec-
tron acceptor, the primary donor is a para-X-aniline (XAn), in
which X ) H, F, Cl, MeO, and Me2N are attached to the N′
nitrogen of the piperazine, and the secondary acceptor is NI.
The dyads and triads undergo photoinduced electron transfer
to yield the respective 5ANI+-Ph-NI-, 6ANI+-Ph-NI-, and
XAn+-6ANI-Ph-NI- radical ion pair states. These radical
3
ion pairs undergo RP-ISC before recombining to yield *NI.
Application of a 0-1 T external magnetic field prior to
photoexcitation results in Zeeman splitting of the triplet
sublevels, which produces a decrease in the triplet yield if 2J is
small, or a resonance if 2J is somewhat larger, due to enhanced
singlet-triplet mixing for fields at which level crossing occurs,
Figure 1. The structural rigidity of these compounds permits
direct measurement of well-defined 2J resonances. Within the
dyads, resonances are observed at field strengths of several
tenths of a tesla, while resonances within the triads are observed
at much smaller fields. The energy of the 2J resonance correlates
Kinetic traces were recorded over a range of 2 µs. Thirty shots were
averaged at each magnetic field strength with a LeCroy 9384 digital
oscilloscope and sent to a computer, which calculated the ∆A. The
relative triplet yield reported is
∆A(B)
∆A(B ) 0)
T
T0
)
(1)
strongly with the distance separating the radical ion paris, rDA
,
The results presented are an average of three experiments conducted
on separate days with freshly prepared samples.
which relates directly to the electronic coupling matrix element
24-29
for CR, VDA
.
In addition, 2J is sensitive to the geometry
Results
of the bonding network joining the radical ion pairs, thus making
it possible to evaluate the dependence of VDA, and thus electron
transfer rates on the details of the radical ion pair molecular
structure.
Steady State Spectrosopy. The photophysics of the 5ANI
and 6ANI chromophores have been characterized previously
in detail.31,32 Briefly, their ground state optical spectra exhibit
broad charge transfer (CT) absorptions centered at 438 and 397
nm in toluene, respectively. Figure 2 displays the ground state
electronic spectra of 5ANI-Ph-NI, 6ANI, and 6ANI-Ph-NI
in toluene. The 5ANI and 6ANI CT absorption bands are
apparent as well as the vibronic structure (343, 363, and 382
nm) arising from a π-π* transition within the NI acceptor. The
spectra of the triads (not shown) are virtually identical to that
of 6ANI-Ph-NI, because the p-X-aniline (X ) H, F, Cl, MeO,
and Me2N) electron donors do not absorb significantly in the
visible region of the spectrum.
Experimental Section
The syntheses of 6ANI-Ph-NI, 6ANI, HAn-6ANI, and MeOAn-
6ANI have been reported previously,30,31 while those of the remaining
molecules follow similar procedures and are reported in the Supporting
(24) Anderson, P. W. Phys. ReV. 1959, 115, 2-13.
(25) Soos, Z. Annu. ReV. Phys. Chem. 1974, 25, 121-153.
(26) Okamura, M. Y.; Isaacson, R. A.; Feher, G. Biochim. Biophys. Acta 1979,
546, 394.
(27) Nelsen, S. F.; Ismagilov, R. F.; Teki, Y. J. Am. Chem. Soc. 1998, 120,
2200-2201.
(28) Kobori, Y.; Sekiguchi, S.; Akiyama, K.; Tero-Kubota, S. J. Phys. Chem.
A 1999, 103, 5416-5424.
(29) Yago, T.; Kobori, Y.; Akiyama, K.; Tero-Kubota, S. J. Phys. Chem. B
2002, 106, 10074-10081.
(31) Greenfield, S. R.; Svec, W. A.; Gosztola, D.; Wasielewski, M. R. J. Am.
Chem. Soc. 1996, 118, 6767-6777.
(32) Debreczeny, M. P.; Svec, W. A.; Wasielewski, M. R. New J. Chem. 1996,
20, 815-828.
(30) Lukas, A. S.; Miller, S. E.; Wasielewski, M. R. J. Phys. Chem. B 2000,
104, 931-940.
9
J. AM. CHEM. SOC. VOL. 125, NO. 13, 2003 3923

A R T I C L E S
Lukas et al.
Table 2. Redox Potentials of the Indicated Compounds Measured
in Butyronitrile Containing 0.1 M Tetra-n-butylammonium
Perchloratea
compound
XAn+1
ANI+1
ANI-1
NI-1
5ANI-Ph-NI
1.06
1.23
1.25
1.40
1.20
1.22
1.31
-1.30
-1.32
-1.31
-1.30
-1.31
-1.30
-1.29
-0.49
-0.49
-0.50
-0.49
-0.49
-0.50
-0.49
6ANI-Ph-NI
An-6ANI-Ph-NI
ClAn-6ANI-Ph-NI
FAn-6ANI-Ph-NI
MeOAn-6ANI-Ph-NI
Me2NAn-6ANI-Ph-NI
1.02
1.12
1.07
0.82
0.33
a All potentials versus saturated calomel electrode (SCE).
correspond closely to that of naphthalene-1,8-dicarboximide. The
first reduction potential of NI occurs at -0.5 V and is constant
within the series of dyad and triad compounds. The oxidation
potentials of the HAn, FAn, ClAn, MeOAn, and Me2NAn
electron donors correlate strongly with the electron-withdrawing
or -releasing nature of the para substituent attached to the
aniline. Only MeOAn and Me2NAn have oxidation potentials
substantially less positive than that of 6ANI.
Figure 2. Ground electronic spectra of compounds 5ANI-Ph-NI, 6ANI,
and 6ANI-Ph-NI in toluene.
Table 1. Steady State Photophysical Properties of the Indicated
Compounds in Toluene
compound
λABS (nm)
λEM (nm)
ΦF
5ANI
6ANI
438
397
438
400
399
399
399
390
399
399
399
399
390
399
502
497
502
0.92
0.91
0.009
5ANI-Ph-NI
Time-Resolved Charge Separation Dynamics. Transient
absorption spectroscopy permits direct observation of the
intermediates formed via photoinduced electron transfer. The
CS rates were measured after excitation of the 5ANI or 6ANI
chromophores with a 416 nm, 130 fs laser flash. The presence
of NI within 5ANI-Ph-NI and 6ANI-Ph-NI results in rapid
CS to form the 5ANI+-Ph-NI- and 6ANI+-Ph-NI- radical
ion pairs, respectively. These are detected by the presence of
the NI- radical anion, which has distinct absorptions at 480
nm (ꢀ ) 28 300 cm-1 M-1) and 605 nm (ꢀ ) 7000 cm-1 M-1),
as shown in Figure 3A.34 The time constants for CS are listed
in Table 3. Analogously, attachment of electron donors to the
6ANI chromophore within the reference molecules given in
Chart 3 results in electron transfer to form the XAn+-6ANI-
radical pairs, as detected by the presence of the 6ANI- anion
radical, which has distinct absorptions at 420 nm (ꢀ ) 23 500
cm-1 M-1) and 510 nm (ꢀ ) 7000 cm-1 M-1), as well as by
monitoring the formation and decay of stimulated emission
resulting from radiative ion pair recombination. The time
constants for photoinduced electron transfer are listed in Table
3 and show that the reaction XAn-6ANI f XAn+-6ANI- is
very rapid for all substituents X.
Within the triads the reaction XAn-1*6ANI-Ph-NI f
XAn+-6ANI--Ph-NI is much faster than the corresponding
reaction XAn-1*6ANI-Ph-NI f XAn-6ANI+-Ph-NI-, so
that the overall formation of the distal radical ion pair occurs
by the reaction sequence XAn-1*6ANI-Ph-NI f XAn+-
6ANI--Ph-NI f XAn+-6ANI-Ph-NI-. The rate constants
and free energies31 for these reactions are given in Table 3. For
example, within MeOAn-6ANI-Ph-NI the primary CS event
is formation of the MeOAn+-6ANI- -Ph-NI ion pair, as
measured by the appearance of the 510 nm absorption band of
6ANI-, as seen in the transient spectra of MeOAn-6ANI-
Ph-NI at early times within Figure 3B. The 510 nm 6ANI-
peak disappears with a time constant of 120 ps, which is
identical to that for the appearance of the NI- peaks at 480 and
610 nm. This distal radical ion pair is formed with a quantum
yield in excess of 0.99.
6ANI-Ph-NI
498
0.014
An-6ANI
517a
510a
510a
572a
0.48a
FAn-6ANI
0.62a
ClAn-6ANI
0.55a
MeOAn-6ANI
0.035a
<0.001a
0.012a
0.013a
0.014a
0.002a
<0.001a
Me2NAn-6ANI
An-6ANI-Ph-NI
ClAn-6ANI-Ph-NI
FAn-6ANI-Ph-NI
MeOAn-6ANI-Ph-NI
Me2NAn-6ANI-Ph-NI
>600a
517a
510a
510a
570a
>600a
a Fluorescence resulting from charge recombination.
1
1
The *5ANI and *6ANI states can act as either electron
donors or acceptors. Attachment of the NI electron acceptor
via a meta-substituted phenyl bridge to either 5ANI or 6ANI
results in nearly complete quenching of their fluorescence in
toluene, as reflected in the steady state quantum yield data listed
in Table 1 and suggesting that the 5ANI+-Ph-NI- and
6ANI+-Ph-NI- radical ion pairs are formed with quantum
yields exceeding 0.98 (see below). Reference compounds XAn-
6ANI, where X ) H, F, Cl, MeO, and Me2N, have the para-
substituted electron donors attached to 6ANI via a piperazine
bridge as shown in Chart 3. The fluorescence quantum yields
1
from *6ANI in these molecules are all < 0.003, while the
radical ion pair states XAn+-6ANI- all undergo radiative
recombination to ground state, except when X ) N(CH3)2. The
radiative recombination makes it possible to accurately assess
the free energy for the reaction XAn-6ANI f XAn+-6ANI-.31
Redox Potentials. The electrochemical properties of aromatic
imide and diimide chromophores are highly sensitive to the
nature of the imide substituent.33 The 5ANI and 6ANI chro-
mophores undergo reversible oxidation and reduction at modest
potentials, as indicated by cyclic voltammetry. The redox
potentials for all the molecules used in this study are given in
Table 2. The oxidation potentials of 5ANI and 6ANI (1.06 and
1.23 V, respectively) are very similar to the oxidation potentials
of N-methylpyrrolidine and N-methylpiperidine, respectively,
while their reduction potentials are both about -1.3 V and
(33) Viehbeck, A.; Goldberg, M. J.; Kovac, C. A. J. Electrochem. Soc. 1990,
137, 1460-1466.
(34) Wiederrecht, G. P.; Niemczyk, M. P.; Svec, W. A.; Wasielewski, M. R. J.
Am. Chem. Soc. 1996, 118, 81-88.
9
3924 J. AM. CHEM. SOC. VOL. 125, NO. 13, 2003

Electron Spin−Spin Exchange Interaction
A R T I C L E S
population. TREPR measurements on closely related compounds
have shown that the triplet levels of the radical ion pair are
higher in energy than that of the singlet state, which makes 2J
negative within this series of compounds.13 According to the
Heitler-London model, the singlet radical pair is stabilized
while the triplet is destabilized.35 Having noted this fact, for
convenience we will ignore the sign of 2J in our subsequent
discussion. At zero field, the degree of mixing of the singlet
and triplet radical pair states is determined by 2J and the
electron-nuclear hyperfine interactions within each radical.
Application of an external magnetic field splits the triplet
sublevels into T(1 and T0 levels via the Zeeman interaction, as
shown in Figure 1. As the field increases, the T-1 sublevel
comes into resonance with S, which leads to increased S T
T-1 mixing. While absolute changes in the singlet and triplet
sublevel populations cannot be detected using magnetic fields
alone, the increased mixing of the spin states at the level crossing
increases the yield of 3*NI formed from CR of the triplet radical
pair, which is detected by transient absorption. Further increases
in the magnetic field strength increase the S-T-1 energy gap,
which results in a decrease in the triplet yield to a constant value
that is due to S-T0 mixing alone. The overall effect is observed
as an increase in the triplet product of charge recombination
when the S-T-1 resonance condition is satisfied, followed by
a decrease when S can only interact with T0.36,37 Thus by
observing the change in population of the neutral localized triplet
state as a function of magnetic field we have a sensitive direct
gauge of the singlet-triplet energy difference, 2J, within the
radical ion pair.
Figure 3. (A) Transient absorption spectra of 6ANI-Ph-NI at the times
indicated following excitation with a 420 nm, 130 fs laser flash. (B)
Transient absorption spectra of MeOAn-6ANI-Ph-NI at the times
indicated following excitation with a 420 nm, 130 fs laser flash.
3
The production of *NI (formed via RP-ISC) as a function
of external magnetic field strength (0-1 T) within 5ANI-Ph-
NI, 6ANI-Ph-NI, and the triads is shown in Figures 6, 7A,
and 7B. Each of these compounds displays a distinct resonance
at 2J. The values of 2J are determined from these plots and are
listed in Table 4. The CR time constants at zero field, 2J, and
1 T are also given in Table 4.
Charge Recombination in the Absence of an External
Magnetic Field. The transient absorption spectra of MeOAn-
6ANI-Ph-NI following excitation with a 416 nm, 7 ns laser
flash are displayed in Figure 4. At early times (t ) 60 ns), the
spectrum displays sharp features at 480 and 605 nm character-
istic of the MeOAn+-6ANI-Ph-NI- radical ion pair. While
the transient absorption spectra of the singlet and triplet radical
ion pair states are indistinguishable from one another, they
undergo CR via different pathways. The singlet radical ion pair
recombines to ground state, which is the baseline within the
transient absorption experiment, while the triplet radical ion pair
recombines to form the neutral triplet state. The spectrum at
longer times (t ) 600 ns) within Figure 4 arises from
3*[MeOAn-6ANI-Ph-NI]. TREPR has shown that this triplet
state is localized on NI within analogous dyads and triads.13
Therefore the broad absorption centered at 480 nm in the
transient spectrum is due to MeOAn-6ANI-Ph-3*NI. These
CR dynamics give rise to transient kinetics at 480 nm, which
are characterized by an instrument-limited rise followed by
exponential decay to a shelf, as seen in the inset to Figure 4.
The rate constants for CR at B ) 0 T are a sum of the rates for
CR of the singlet and triplet radical ion pairs, labeled within
Figure 5 as kS and kT, respectively. These rate constants most
likely differ considerably due to large differences in the free
energies of reaction.
Discussion
Theory. The formation of triplet states within 5ANI-Ph-
NI, 6ANI-Ph-NI, and XAn-6ANI-Ph-NI occurs via the
RP-ISC mechanism. This mechanism can be described by the
spin-Hamiltonian
HST ) âB (g S + g S ) + a S ‚I +
∑
0
1
1
2
2
1i
1
i
i
a S ‚I - J((1/ ) + 2S ‚S ) (2)
∑
2k
2
k
2
1
2
k
where â is the Bohr magneton, B0 is the applied magnetic field,
g1 and g2 are the electronic g-factors for each radical, S1 and
S2 are electron spin operators for the two radicals within the
radical pair, Ii and Ik are nuclear spin operators, a1i and a2k are
the isotropic hyperfine coupling constants of nucleus i with
radical 1 and nucleus k with radical 2, and J is the scalar spin-
spin exchange interaction constant. Anisotropic exchange
interactions, dipolar hyperfine couplings, and the magnetic
dipole-dipole interaction are neglected because the measure-
Charge Recombination in the Presence of a Magnetic
Field. The amplitude of the long-lived component within the
transient kinetics directly measures the XAn-6ANI-Ph-3*NI
(35) Alexander, M. H.; Salem, L. J. Chem. Phys. 1967, 46, 430.
(36) Weller, A.; Staerk, H.; Treichel, R. Faraday Discuss. Chem. Soc. 1984,
78, 271-278.
(37) Weller, A.; Nolting, F.; Staerk, H. Chem. Phys. Lett. 1983, 96, 24-27.
9
J. AM. CHEM. SOC. VOL. 125, NO. 13, 2003 3925

A R T I C L E S
Lukas et al.
Table 3. Free Energies and Time Constants for Charge Separation and Recombination in Toluene
compound
∆GCS1 (eV)
CS (ps)
1
∆GCR1 (eV)
CR (ns)
1
∆GCS2 (eV)
CS (ps)
2
∆GCR2 (eV)
CR (ns)
2
5ANI-Ph-NI
-0.54
-0.53
-0.08
-0.06
-0.06
-0.32
-0.81
-0.08
-0.06
-0.06
-0.32
-0.81
85
120
18
28
27
7.7
1.2
18
30
25
2.11
2.27
2.72
2.74
2.74
2.48
1.99
2.72
2.74
2.74
2.48
1.99
45
120
5.0
5.2
5.4
6ANI-Ph-NI
An-6ANI
FAn-6ANI
ClAn-6ANI
MeOAn-6ANI
5.3
0.15
Me2NAn-6ANI
An-6ANI-Ph-NI
ClAn-6ANI-Ph-NI
FAn-6ANI-Ph-NI
MeOAn-6ANI-Ph-NI
Me2NAn-6ANI-Ph-NI
-0.19
-0.10
-0.14
-0.48
-0.53
120
120
120
40
-2.53,a -0.48b
-2.55,a -0.50b
-2.60,a -0.55b
-2.00,a 0.05b
-1.46,a 0.59b
120
112
168
73
8.0
1.3
25
100
a Recombination to ground state. b Recombination to NI triplet state.
The small differences in g-factors for organic radicals such as
those studied here contribute to singlet-triplet mixing only at
field strengths of several tesla. There is no experimental evidence
within Figures 6 and 7 for a ∆g contribution because the data
do not show a monotonic increase in triplet product yield as a
function of field strength. Therefore, for the range of field
strengths in the present study, singlet-triplet mixing of the
radical pairs within these compounds is driven largely by the
hyperfine and exchange interactions.
In general, hyperfine interactions (HFI) between the magnetic
moments of the unpaired electrons and those of nearby nuclei
drive singlet-triplet mixing at zero and small magnetic fields.1-3
Weller and co-workers have shown that for a series of organic
radical ion pairs, the magnetic field value at half-saturation due
to HFI can be approximated by eq 4:36,37
Figure 4. Transient absorption spectra of MeOAn-6ANI-Ph-NI at the
times indicated following excitation with a 416 nm, 7 ns laser flash. Inset:
Kinetics monitoring the 480 nm spectral features.
2
BD2 + BA
B1/2 ) 2
(4)
(
)
BD + BA
where BD and BA refer to the energies of the hyperfine
interactions between the nuclear spins and unpaired electron
spins on the donor and acceptor. These may be determined from
the values of the isotropic electron-nuclear hyperfine coupling
constants aik for the nuclear spins Ik on radical i using eq 5:
2 1/2
B ) [ I (I + 1)a ]
(5)
∑
i
k
k
ik
k
The values of aik for NI- and MeOAn+, as well as those for
the radical cations of N-methylpyrrolidine and N-methylpiperi-
dine, were obtained from the literature.38-40 The values of aik
for XAn+ (X ) H, Cl, F) were obtained by scaling their
calculated spin densities given in Table 5 using the calculated
and experimental hyperfine couplings observed for MeOAn+.
Values of B1/2 for 5ANI+-Ph-NI-, 6ANI+-Ph-NI-, and
XAn+-6ANI-Ph-NI- were calculated using these hyperfine
coupling constants and eqs 4 and 5 and are listed in Table 4.
Figure 5. Energy level diagram for relevant donor-acceptor electronic
states.
ments are performed in solution. Also, it is assumed that nuclei
associated structurally with a given radical couple only with
the electron spin within that radical.
Zeeman splitting of the triplet sublevels requires an applied
magnetic field, and thus the first term in eq 2 cannot contribute
to S T T mixing at zero field. As the strength of the magnetic
field increases, contributions from this term will begin to
contribute to S T T mixing due to differences in the isotropic
g-factors of the two electron spins. The frequency for conversion
of the spins from singlet to triplet is given by eq 3:
For compounds that display distinct resonances, B1/2
=
(1/2)∆B, where ∆B is the full width at half-maximum of the
resonance.
The exchange interaction, 2J, is defined as the energy splitting
between the singlet and triplet radical pairs and likely arises
(38) Zhong, C. J.; Kwan, W. S. V.; Miller, L. L. Chem. Mater. 1992, 4, 1423-
∆gâB0
1428.
∆ω )
(3)
(39) Ciminale, F.; Curci, R.; Portacci, M.; Troisi, L. Tetrahedron Lett. 1988,
29, 2463-2466.
(40) Eastland, G. W.; Rao, D. N. R.; Symons, M. C. R. J. Chem. Soc., Perkin
Trans. 2 1984, 1551-1557.
p
where ∆g is the difference in g-factors for the two radicals.
9
3926 J. AM. CHEM. SOC. VOL. 125, NO. 13, 2003

Electron Spin−Spin Exchange Interaction
A R T I C L E S
Table 4. Experimentally Determined Values for 2J, B1/2 and Charge Recombination Rates at B ) 0 G, 2J, and 1 T, and B1/2 Values
Calculated Using Eqs 4 and 5 from HFI Literature Values38-40
B1/2 (mT)
B1/2 (mT)
τCR (ns)
τCR (ns)
τCR (ns)
2J (mT)
exptl
calc
T(0)/RP
B ) 0T
B ) 2J
B ) 1T
5ANI-Ph-NI
305
320
52.5
32.5
60.0
1.5
80
130
72.5
102.5
123
2.2
4.2
4.2
3.9
3.9
4.2
3.8
0.12
0.09
0.24
0.31
0.26
0.25
45
120
120
112
168
73
42
107
96
100
138
84
45
120
153
177
214
68
6ANI-Ph-NI
An-6ANI-Ph-NI
ClAn-6ANI-Ph-NI
FAn-6ANI-Ph-NI
MeOAn-6ANI-Ph-NI
15.0
3.0
Me2NAn-6ANI-Ph-NI
3.9
100
100
Figure 6. Relative triplet yield for the indicated dyads at 0-1 T.
from interactions between the frontier orbitals of the radical
pairs. Figure 1 shows that the Zeeman splitting of the triplet
sublevels causes S T T-1 level crossing when B ) 2J, which
leads to the maximum in the triplet product yields observed in
Figures 6 and 7. At higher field strengths only S T T0 mixing
occurs, at which point the magnetic field effect saturates. Weller
relates the observed resonance at B ) 2J to rDA using
DA-r0)
2J(r) ) 2J0e-R(r
(6)
where 2J(r) is obtained from the peak of the observed resonance,
rDA is the radical ion pair separation, r0 is the van der Waals
contact distance of 3.4 Å, and 2J0 and R are constants. The
preexponential 2J0 includes the dependence of the spin-spin
exhange interaction on the geometry of the radical pair. Equation
6 predicts that the observed resonance should move to lower
fields as the distance between the radical ions increases.
Magnetic Field Effects on Charge Recombination within
the Dyads and Triads. The two dyads, 5ANI-Ph-NI and
6ANI-Ph-NI, which differ only in the number of carbon atoms
in the cyclic amine of the chromophore, display broad 2J
resonances near 0.3 T, Figure 6. Likewise, the calculated B1/2
values are virtually identical for these compounds, Table 4.
However, comparison of the experimental and calculated B1/2
values shows that the observed B1/2 values are 1 to 2 orders of
magnitude larger than those calculated assuming a contribution
from HFI alone. Given that the measured lifetimes of 5ANI+-
Ph-NI- and 6ANI+-Ph-NI- are never less than 42 ns at 0-1
T, Table 4, the line widths of their 2J resonances are not due to
uncertainty broadening (p ) 5.7 × 10-12 T‚s).41 Thus, within
Figure 7. (A) Relative triplet yields for XAn-6ANI-Ph-NI triads at 0-1
T. (B) Relative triplet yield for MeOAn-6ANI-Ph-NI at 0-25 mT.
5ANI+-Ph-NI- and 6ANI+-Ph-NI- the spin Hamiltonian
(eq 2) is dominated by the exchange interaction, 2J, and the
assumption can be made that 2J . EHFI
.
The triads are structurally similar to the dyads, but exhibit
smaller 2J values and much sharper resonances. This is due to
delocalization of charge and spin density away from 6ANI
toward the HAn, FAn, ClAn, and MeOAn electron donors
within the triads and increases the radical ion pair distance
relative to that in 6ANI-Ph-NI depending on the oxidation
potential of XAn. In MeOAn+-6ANI-Ph-NI-, the observed
B1/2 values agree very well with those calculated from the HFI
using eqs 4 and 5. The fact that 2J ≈ EHFI implies that the HFI
plays a significant role in S T T mixing at zero field within
MeOAn+-6ANI-Ph-NI-. In the triads where X ) H, F, and
Cl the observed values of B1/2 are over an order of magnitude
larger than the calculated values, so that 2J . EHFI, as is ob-
(41) Staerk, H.; Treichel, R.; Weller, A. Chem. Phys. Lett. 1983, 96, 28-30.
9
J. AM. CHEM. SOC. VOL. 125, NO. 13, 2003 3927

A R T I C L E S
Lukas et al.
Table 5. Calculated Spin Densities for the Cation Radicals of
5ANI, 6ANI, and XAn-6ANI, where X ) H, F, Cl, MeO, and Me2N,
Determined Using UHF-AM1 Calculationsa
Chart 4
atom
6ANI
5ANI
H
F
Cl
MeO
Me2N
C1
0.400
0.385
0.012
0.009
0.008
0.006
0.002
C2 -0.386 -0.367 -0.012 -0.008 -0.007 -0.006 -0.002
C3 0.429 0.420 0.012 0.009 0.008 0.006 0.002
C4 -0.418 -0.412 -0.013 -0.009 -0.008 -0.006 -0.002
C5 0.386 0.366 0.014 0.009 0.009 0.007 0.002
C6 -0.363 -0.334 -0.013 -0.009 -0.008 -0.007 -0.002
C7 0.617 0.644 0.018 0.013 0.012 0.009 0.003
C8 -0.461 -0.457 -0.016 -0.011 -0.011 -0.008 -0.002
C9 0.545 0.551 0.019 0.014 0.013 0.010 0.003
triplet radical ion pair originated from an excited triplet state.
The same mechanism occurs for MeOAn-6ANI-Ph-NI, in
which repopulation of the radical ion pair results in the longest
CR time constants occurring when S T T-1 mixing is strongest,
i.e., at B ) 2J.
C10 -0.309 -0.261 -0.020 -0.014 -0.014 -0.010 -0.003
C11 -0.079 -0.083 -0.002 -0.001 -0.001 -0.001 -0.000
The 4-dimethylaminoaniline (Me2NAn) donor within
Me2NAn-6ANI-Ph-NI undergoes reversible oxidation at
0.33 V vs SCE, the most negative oxidation potential in this
series of compounds. This results in rapid CS with τCS1 ) 1.3
ps and τCS2 ) 25 ps, forming Me2NAn+-6ANI-Ph-NI- with
unity quantum yield. However, no magnetic field effect is
observed for this compound because the radical pair can
only undergo CR to the singlet ground state since ∆G ) 0.59
C12
0.053
0.051 -0.001 -0.001 -0.001 -0.001 -0.000
O1 -0.089 -0.094 -0.003 -0.002 -0.002 -0.001 -0.000
O2 -0.060 -0.059 -0.002 -0.001 -0.001 -0.001 -0.000
N1
0.543
0.454
0.018
0.000
0.609
0.017
0.000
0.581
0.017
0.000
0.585
0.011
0.000
0.528
0.004
0.000
0.352
N2 -0.017 -0.018
N3
C13
C14
C15
C16
C17
C18
X
-0.280 -0.264 -0.272 -0.151 -0.090
0.468
-0.396 -0.377 -0.389 -0.264
0.531 0.498 0.518 0.386
-0.395 -0.377 -0.389 -0.248
0.469 0.430 0.446
-0.029 -0.060 -0.035
0.429
0.445
0.350
0.010
0.054
0.066
0.055
3
eV for [Me2NAn+-6ANI-Ph-NI-] f [Me2NAn-6ANI-
Ph-3*NI]. While significant S T T mixing is likely to occur
within the radical pair, no field-dependent change is observed
in the CR rate because singlet CR is field independent.
0.295 -0.009
0.096 0.373
a Atoms are labeled according to Chart 4.
Relating 2J Resonances to Radical Ion Pair Structure and
Electronic Coupling. An analysis of the relationship between
the observed 2J resonances and the structures of the radical ion
pairs requires an estimation of the spin densities within the
radical ion pairs and the effective distance between the spin
distributions within the radical ion pairs, rDA. The energy-
minimized structures of the cation and anion radicals within
the dyads and triads were calculated using the unrestricted
Hartree-Fock (UHF) AM1 method.42 The spin density distribu-
tion within the NI- radical anion is symmetric and principally
resides within the naphthalene ring, so that the geometric center
of NI- is the center of its spin distribution. The spin densities
for selected atoms within the cation radicals are listed in Table
5 according to the nomenclature given in Chart 4. Significant
spin density resides on the nitrogen atom (N1) attached to the
4-position of the naphthalene-1,8-dicarboximide ring of 5ANI+
and 6ANI+. In contrast, the largest spin density in each triad
resides on the piperazine nitrogen (N3) that is part of XAn. As
the electron-releasing MeO and Me2N substituents stabilize the
positive charge, the spin distribution within XAn+ shifts away
from N3 and toward these substituents. The calculated spin
densities suggest that the average distance between the spin
density distributions of each radical ion within 6ANI+-Ph-
NI- and 5ANI+-Ph-NI- should be significantly shorter than
the corresponding distances in XAn+-6ANI-Ph-NI-. A
reasonable estimate of the effective radical ion pair distances
in these molecules can be obtained using the spin densities to
weight the distances between the individual atoms in the radical
cation and the center of NI- for each molecule. Since the
through-bond contribution to the overall electronic coupling
between covalently linked radical ion pairs at fixed distances
usually dominates, we measure the distances from the individual
atoms of the radical cation to the center of the phenyl group
attaching it to NI- and then add to it the 7.8 Å distance from
the center of the phenyl to that of NI-. These distances are then
served for the dyads. The degree of S T T mixing at zero field
within the radical pair is best illustrated by the ratio of the local-
3
ized triplet *NI yield to that of the radical ion pair, T(0)/RP,
as indicated in the Figure 4 inset. The absorption ratios directly
reflect the T(0)/RP yield ratio and are listed in Table 4. These
data show that, as expected, the yield of 3*NI following charge
recombination is much higher within the triads than the dyads.
Given the smaller values of 2J for the triads relative to the
dyads, the hyperfine interaction makes a much more significant
contribution to S T T mixing. At low field it is difficult to
separate the contributions of each mechanism in terms of
population and depopulation rates of the mixed states. Neverthe-
less, several observed trends between the CR rates at B ) 0 T,
2J, and 1 T should be mentioned. First, it is clear from
comparing CR time constants at B ) 0 and 1 T that S most
likely interacts with all three triplet sublevels at zero field. As
the field increases, the CR time constants become shorter at B
) 2J due to increased S T T-1 mixing. At higher fields, the
time constants for CR become longer again because S mixes
only with T0. Interestingly, compound MeOAn-6ANI-Ph-
NI is an exception to this trend; its CR time constant is longer
when B ) 2J and shorter at high fields. This is a consequence
of an additional mechanistic complexity for this molecule.
TREPR spectra published earlier show that the emission,
absorption spin polarization pattern of the EPR lines observed
at early times for the closely related MeOAn+-6ANI-
Me2Ph-NI- radical ion pair (Chart 1) is a consequence of its
singlet state precursor, MeOAn-1*6ANI-Me2Ph-NI.13 How-
ever, the energies of MeOAn+-6ANI-Me2Ph-NI- and
MeOAn-6ANI-Me2Ph-3*NI are nearly degenerate, so that
as a function of time an equilibrium between these two states
is established, which leads to a significant population of
3[MeOAn+-6ANI-Me2Ph-NI-] produced from MeOAn-
6ANI-Me2Ph-3*NI. This results in an inversion of the spin
polarization pattern as a function of time, indicating that the
(42) AM1 calculations were performed using HyperChem; Hypercube, Inc., 1115
NW 1114th St., Gainesville, FL 32601.
9
3928 J. AM. CHEM. SOC. VOL. 125, NO. 13, 2003

Electron Spin−Spin Exchange Interaction
A R T I C L E S
weighted by the calculated spin densities of the radical cation
and summed to yield rDA. Generally, rDA increases as the radical
cation is stabilized by substituent effects.
The limited degree of conformational freedom within the
radical ion pairs studied here means that eq 6 should provide a
reasonable description of the dependence of 2J on rDA. However,
comparisons between molecules using eq 6 are valid only if
each radical ion pair in the series has the same free energy of
formation and recombination. This stems from the fact that the
exchange interaction for a radical ion pair is directly related to
the electronic coupling matrix elements, Vn, and the respective
energy gaps at its relaxed geometry, ∆En, between the radical
ion pair and the n states from which it is formed and to which
it decays by a relationship derived from second-order perturba-
tion theory:24-29
Figure 8. Plot of ln(2J(∆GCRT + λ)) vs distance for the indicated dyads
and triads.
2
Vn
2J )
(7)
∑
∆En
n
Given the much smaller energy gap between the triplet radical
ion pair state and the triplet state of NI that results from CR
relative to that between the singlet radical ion pair state and
singlet ground state for each molecule, Table 3, we assume that
the term in eq 7 that couples the triplet radical ion pair state to
the triplet recombination product dominates, so that
2
VCRT
2J ) -
(8)
∆GCRT + λ
where λ is the vertical reorganization energy for CR within the
radical ion pair leading to the local neutral triplet state of NI
and ∆GCRT ) E(triplet) - E(radical ion pair). The total
reorganization energy λ ) λS + λI, where λS is the contribution
from the solvent and λI is the internal structural reorganization
of the donor and acceptor. Noting that the static and high-
frequency dielectric constants of toluene are almost equal, the
Marcus dielectric continuum model43 predicts that λS is negli-
gible. The value of λI typically assigned to aromatic donors and
acceptors is 0.3 eV,44 so that λ ) 0.3 eV.
The term R in eq 6 can be directly related to the exponential
damping factor â for the distance dependence of the rate constant
for nonadiabatic electron transfer.45-47 Since the semiclassical
Marcus-Jortner theory of electron transfer shows that the rate
of electron transfer kET V2, eq 8 implies that kCRT 2J. Thus,
the exponential damping factor R for the distance dependence
of 2J (eq 6) should be equal to â, the comparable factor for the
distance dependence of the charge recombination rate. Plotting
ln(2J(∆GCRT + λ)) vs rDA in Figure 8 results in a best-fit line
with slope R ) 0.5 ( 0.1 Å-1 and an intercept that yields 2J0
) (2.4 ( 0.4) × 104 mT. Our data for the distance dependence
of 2J implies that â ) 0.5 ( 0.1 Å-1, which is consistent with
our previous estimates of â for charge recombination in related
systems.48
Figure 9. Two conformations of MeOAn+-6ANI-Ph-NI- resulting from
chair-boat inversion of the piperazine ring.
One of the most striking features of the MFE data for
MeOAn-6ANI-Ph-NI is the two resonances observed at 1.5
and 15 mT, Figure 7B. The observation of two distinct values
for 2J suggests two distinct values for rDA. The energy-
minimized structure of MeOAn+-6ANI is assumed to cor-
respond to the smaller of the two values of 2J, which is used as
a data point in the plot of ln(2J(∆GCRT + λ)) vs rDA in Figure
8. The distance that corresponds to the observed resonance at
15 mT can be obtained using this value, R ) 0.5 Å-1 and 2J0
) 2.4 × 104 mT obtained from the fit to the data in Figure 8,
r0 ) 3.4 Å, and eq 6 to yield rDA ) 18.2 Å. This distance is 1.1
Å shorter than the 19.3 Å distance obtained for MeOAn+-
6ANI-Ph-NI- from the energy-minimized structures of the
radical cation and anion. This small change in distance is best
explained by assuming that there is a secondary minimum
energy structure.
The most likely source of two conformations within the
radical ion pair is a chair T boat interconversion of the
piperazine ring, Figure 9. The radical ion pair distance is reduced
slightly in the boat conformation (B), resulting in additional
Coulombic stabilization of the radical ion pair. While this small
degree of “harpooning” due to Coulombic attraction of the two
charges may occur,49 our MFE data do not support larger
structural changes due to this mechanism. For example, folding
the MeOAn+ radical cation over 6ANI decreases rDA to values
(43) Marcus, R. A. J. Chem. Phys. 1956, 24, 966.
(44) Closs, G. L.; Miller, J. R. Science (Washington, D.C., 1883-) 1988, 240,
440-447.
(45) Devault, D. Quantum Mechanical Tunneling in Biological Systems;
Cambridge University Press: Cambridge, 1984.
(46) Newton, M. D.; Sutin, N. Annu. ReV. Phys. Chem. 1984, 35, 437-480.
(47) Marcus, R. A.; Sutin, N. Biochim. Biophys. Acta 1985, 811, 265-322.
(48) Lukas, A. S.; Bushard, P. J.; Wasielewski, M. R. J. Phys. Chem. A 2002,
106, 2074-2082.
(49) Lauteslager, X. Y.; van Stokkum, I. H. M.; van Ramesdonk, H. J.; Brouwer,
A. M.; Verhoeven, J. W. J. Phys. Chem. A 1999, 103, 653-659.
9
J. AM. CHEM. SOC. VOL. 125, NO. 13, 2003 3929

A R T I C L E S
Lukas et al.
less than or equal to those within the dyads and would give
rise to a second resonance within the triads at much higher fields,
which is not observed. Earlier work suggests that deviations
from a zigzag arrangement of saturated bonds diminishes the
through-bond coupling for electron transfer reactions.50,51
However, the transannular interaction of the nitrogen lone pair
orbitals depicted in Figure 9, structure B, most likely results in
slightly enhanced electronic coupling consistent with the modest
15 mT value of 2J.
The XAn-6ANI-Ph-NI triads, where X ) H, F, and Cl,
are structurally similar to MeOAn-6ANI-Ph-NI, differing
only in the substituent on XAn. However, they have substantially
larger values for both 2J and B1/2. The spin density calculations
show that the spin is more strongly localized on the nitrogen
atom of XAn (N3) within these triads, which results in a
decrease in rDA. Due to their structural similarity, it is likely
that the radical pairs within all of the triads have both A and B
conformers. However, the greater line widths within the triads
where X ) H, F, and Cl prohibit resolving the two resonance
peaks.
The possibility of two conformations, however, does not
ensure the observation of two distinct 2J resonances. To observe
distinct resonances, the time constant for interconversion must
be slow relative to the energy difference between the resonances.
The difference in 2J values within MeOAn+-6ANI-Ph-NI-
is 13.5 mT, which corresponds to a transform-limited time of
approximately 0.4 ns. Thus, interconversion of the A and B
conformers, if it does occur, must occur at times slower than
0.4 ns to observe distinct resonances. The energy barrier for
ring inversion in N,N′-dimethylpiperazine measured by NMR
is 13 kcal/mol.52,53 Assuming Arrhenius behavior and a prefactor
of about 1013 s-1, this corresponds to an inversion rate of only
about 4 × 103 s-1. However, in a radical ion pair the Coulombic
attraction of the charges (harpooning) can increase the chair f
boat inversion rate substantially. For example, Lauteslager et
al.54 have observed that harpooning occurs with rate constants
of about 108-109 s-1 in photogenerated radical ion pair
compounds having an N-phenylpiperidine donor attached to a
cyanonaphthalene acceptor. These rate constants are consistent
with our data for MeOAn+-6ANI-Ph-NI-. From the ampli-
tude of the two resonances, there appears to be a preference for
the A conformation with the longer 19.3 Å distance. However,
it must be remembered that for such small 2J values HFIs play
a significant role in driving RP-ISC. While both resonances
within MeOAn+-6ANI-Ph-NI- have contributions from
hyperfine and exchange interactions, the relative magnitude of
HFI is significantly greater at the lower field resonance.
Therefore, the amplitudes of the two resonances do not
necessarily reflect the relative populations of the two radical
ion pair conformations.
Conclusions
We have shown that the effect of applied magnetic fields on
the RP-ISC yield of triplet products within a series of structurally
rigid compounds with fixed distances allows for evaluation of
the spin-spin exchange interaction, 2J, as a function of rDA
,
the donor-acceptor distance. For those compounds that exhibit
exchange interactions on the order of several tenths of a tesla,
the singlet-triplet mixing within the radical ion pair is
dominated by the exchange interaction, and 2J . EHFI
.
However, at larger rDA (smaller 2J), EHFI can contribute to the
observed resonance significantly, and experimentally observed
B1/2 within the triad with the smallest value of 2J agrees very
well with the calculated B1/2 value. Changes in both the
magnitude and number of 2J resonances can be directly
correlated with radical ion pair structure. The relationship
between 2J and the electronic coupling matrix element for
electron transfer, VDA, makes it possible to probe the dependence
of VDA on subtle changes in molecular structure at an unprec-
edented level of detail, thereby providing insights into how to
optimize structures for efficient electron transfer. This type of
detailed understanding of how spin-spin exchange interactions
map molecular structure and thereby dictate charge separation
lifetimes should prove very useful in the design of systems for
photochemical conversion and storage of solar energy.
Acknowledgment. This work was supported by the Division
of Chemical Sciences, Office of Basic Energy Sciences, U.S.
Department of Energy (Grant No. DE-FG02-99ER14999).
(50) Oliver, A. M.; Craig, D. C.; Paddon-Row, M. N.; Kroon, J.; Verhoeven, J.
W. Chem. Phys. Lett. 1988, 150, 366-373.
Supporting Information Available: Details regarding the
synthesis and characterization of the molecules used in this
study. This material is available free of charge via the Internet
at http://pubs.acs.org.
(51) Wasielewski, M. R.; Niemczyk, M. P.; Johnson, D. G.; Svec, W. A.; Minsek,
D. W. Tetrahedron 1989, 45, 4785-4806.
(52) Harris, R. K.; Spragg, R. A. Chem. Commun. 1966, 314-316.
(53) Abraham, R. J.; Macdonald, D. B. Chem. Commun. 1966, 188-189.
(54) Lauteslager, X. Y.; Bartels, M. J.; Piet, J. J.; Warman, J. M.; Verhoeven,
J. W.; Brouwer, A. M. Eur. J. Org. Chem. 1998, 2467-2481.
JA029062A
9
3930 J. AM. CHEM. SOC. VOL. 125, NO. 13, 2003