Carbon-carbon bond fragmentation in aminoalcohol radical cations. Kinetics, thermodynamic correlations, and mechanism

Article abstract of DOI:10.1021/ja960378q

A detailed study of the kinetics, thermodynamics and mechansim of carbon-carbon bond fragmentation in a series of aminoalcohol radical cations is presented. The compounds that provide the basis for this investigation are derived from the parent structure,

Full text of DOI:10.1021/ja960378q

J. Am. Chem. Soc. 1996, 118, 5655-5664
5655
Carbon-Carbon Bond Fragmentation in Aminoalcohol Radical
Cations. Kinetics, Thermodynamic Correlations, and
Mechanism
Richard D. Burton, Michael D. Bartberger, Yin Zhang, John R. Eyler, and
Kirk S. Schanze*
Contribution from the Department of Chemistry, UniVersity of Florida, P.O. Box 117200,
GainesVille, Florida 32611
ReceiVed February 5, 1996X
Abstract: A detailed study of the kinetics, thermodynamics and mechansim of carbon-carbon bond fragmentation
in a series of aminoalcohol radical cations is presented. The compounds that provide the basis for this investigation
are derived from the parent structure, erythro-2-(phenylamino)-1,2-diphenylethanol, by substitution at the para position
of the N-phenyl with methoxy, methyl, (hydrogen), chloro, and cyano groups (compounds 1a-e, respectively). The
rates for C-C bond fragmentation for radical cations 1a-e•+ in CH3CN solution were determined by laser flash
photolysis and vary from 3.9 × 104 (1a) to 7.4 × 106 s-1 (1e). The activation parameters for bond fragmentation
in 1c-e•+ are characterized by low activation enthalpies and relatively large, negative activation entropies. The
bond fragmentation rates increase with the peak potential for anodic oxidation of the neutral aminoalcohols, Ep(1).
Correlation of the free energy of activation for bond fragmentation (∆GBFq) with FEp(1) (F is the Faraday constant)
implies that the dependence of ∆Gq on ∆G°BF is relatively weak, consistent with bond fragmentation in 1a-e•+
BF
being weakly endothermic or exothermic. The transient absorption spectra of the reactive intermediates produced
by fragmentation of 1a-e•+ are consistent with a mechansim involving heterolytic fragmentation of the C1-C2
bond with concomitant loss of the hydroxyl proton. By contrast, FT-ICR studies of 1a-e•+ indicate that in the gas
phase homolytic fragmentation of the C1-C2 bond predominates. Semiempirical calculations using the AM1
Hamiltonian demonstrate that in the gas phase homolysis is the thermodynamically preferred pathway, consistent
with the FT-ICR results.
Introduction
understanding the thermodynamic6,14,15 and kinetic origins of
the significantly enhanced reactivity of these reactive intermedi-
ates.
Radical ions are the primary reactive intermediates produced
by photoinduced electron transfer.16,17 The reactivity of these
high-energy intermediates affords the means to apply photo-
(7) (a) Dinnocenzo, J. P.; Banach, T. E. J. Am. Chem. Soc. 1989, 111,
8646. (b) Dinnocenzo, J. P.; Conlon, J. P. Tetrahedron Lett. 1995, 36, 7415.
(c) Dinnocenzo, J. P.; Karki, S. B.; Jones, J. P. J. Am. Chem. Soc. 1993,
115, 7111. (d) Karki, S. B.; Dinnocenzo, J. P.; Jones, J. P.; Korzekwa, K.
R. J. Am. Chem. Soc. 1995, 117, 3657.
(8) (a) Mirafzal, G. A.; Kim, T.; Liu, J.; Bauld, N. L. J. Am. Chem. Soc.
1992, 114, 10968. (b) Yueh, W.; Bauld, N. L. J. Am. Chem. Soc. 1995,
117, 5671. (c) Yueh, W.; Bauld, N. L. J. Chem. Soc., Perkins Trans. 2
1995, 871. (d) Bellville, D. J.; Pabon, R. A.; Bauld, N. L. J. Am. Chem.
Soc. 1985, 107, 4978.
(9) (a) Roth, H. D.; Herbertz, T. J. Am. Chem. Soc. 1993, 115, 9801. (b)
Ishiguro, K.; Turro, N. J.; Roth, H. D. J. Am. Chem. Soc. 1994, 116, 6933.
(c) Weng, X.; Du, X.-M.; Roth, H. D. J. Am. Chem. Soc. 1995, 117, 135.
(d) Weng, X.; Sheik, Q.; Roth, H. D. J. Am. Chem. Soc. 1995, 117, 10655.
(10) (a) Schlesener, C. J.; Amatore, C.; Kochi, J. K. J. Am. Chem. Soc.
1984, 106, 7472. (b) Masnovi, J. M.; Sankararaman, S.; Kochi, J. K. J.
Am. Chem. Soc. 1989, 111, 2263. (c) Sankararaman, S.; Perrier, S.; Kochi,
J. K. J. Am. Chem. Soc. 1989, 111, 6448. (d) Perrier, S.; Sankararaman, S.;
Kochi, J. K. J. Chem. Soc., Perkin Trans. 2 1993, 825.
(11) (a) Reitsto¨en, B.; Parker, V. D. J. Am. Chem. Soc. 1990, 112, 4968.
(b) Parker, V. D.; Chao, Y.; Reitsto¨en, B. J. Am. Chem. Soc. 1991, 113,
2336. (c) Parker, V. D.; Tilset, M. J. Am. Chem. Soc. 1991, 113, 8778.
(12) Horner, J. H.; Martinez, F. N.; Musa, O. M.; Newcomb, M.; Shahin,
H. E. J. Am. Chem. Soc. 1995, 117, 11124.
Radical ions are important reactive intermediates in single
electron transfer (SET) chemistry.1 Given the importance of
SET chemistry to chemical synthesis and mechanism, there has
been a recent upsurge of research activity focused on under-
standing the unique reactivity of organic radical cations and
radical anions.2-13 These studies have led to new synthetic
methodologies3e,f,4 and provided a unique opportunity to un-
derstand bonding in odd-electron species.6,9,10 Furthermore,
research on radical ions is beginning to provide a basis for
X Abstract published in AdVance ACS Abstracts, June 1, 1996.
(1) For recent reviews in the area of SET chemistry see: AdVances in
Electron Transfer Chemistry; Mariano, P. S., Ed.; JAI Press: Greenwich;
1991-1994; Vols. 1-4.
(2) Chanon, M.; Rajzmann, M.; Chanon, F. Tetrahedron 1990, 46, 6193.
(3) (a) Arnold, D. R.; Maroulis, A. J. J. Am. Chem. Soc. 1976, 98, 5931.
(b) Popielarz, R.; Arnold, D. R. J. Am. Chem. Soc. 1990, 112, 3068. (c)
Nicholas, A. M.; Arnold, D. R. Can. J. Chem. 1982, 60, 2165. (d) Nicholas,
A. M.; Boyd, R. J.; Arnold, D. R. Can. J. Chem. 1982, 60, 3011. (e)
McMahon, K.; Arnold, D. R. Can. J. Chem. 1993, 71, 450. (f) Arnold, D.
R.; Du, X. Can. J. Chem. 1994, 72, 403.
(4) (a) Eitsu, H.; Xu, W.; Mariano, P. S. J. Am. Chem. Soc. 1988, 110,
8099. (b) Jeon, Y. T.; Lee, C.-P.; Mariano, P. S. J. Am. Chem. Soc. 1991,
113, 8847. (c) Xu, W.; Zhang, X.-M.; Mariano, P. S. J. Am. Chem. Soc.
1991, 113, 8863. (d) Yoon, U. C.; Mariano, P. S. Acc. Chem. Res. 1992,
25, 233. (e) Zhang, X.; Yeh, S.-R.; Hong, S.; Freccero, M.; Albini, A.;
Falvey, D. E.; Mariano, P. S. J. Am. Chem. Soc. 1994, 116, 4211.
(5) (a) Workentin, M. S.; Johnston, L. J.; Parker, V. D. J. Am. Chem.
Soc. 1994, 116, 8279. (b) Workentin, M. S.; Schepp, N. P.; Johnston, L. J.
J. Am. Chem. Soc. 1994, 116, 1141. (c) Johnston, L. J.; Kwong, P.;
Shelemay, A. J. Am. Chem. Soc. 1993, 115, 1664.
(6) (a) Maslak, P. In Photoinduced Electron Transfer V. Top. Curr. Chem.
1993, 168, 1. (b) Maslak, P. 1990, 29, 283. (c) Maslak, P.; Chapman, W.
H., Jr.; Vallombroso, T. M.; Watson, B. M. J. Am. Chem. Soc. 1995, 117,
12380. (d) Maslak, P.; Vallombroso, T. M.; Chapman, W. H., Jr.; Narvaez,
J. N. Angew. Chem., Int. Ed. Engl. 1994, 33, 73.
(13) Baciocchi, E.; Del Giacco, T.; Elisei, F. J. Am. Chem. Soc. 1993,
115, 12290.
(14) (a) Zhang, X.-M.; Bordwell, F. G.; Satish, A. V. Pure Appl. Chem.
1995, 67, 735. (b) Bordwell, F. G.; Zhang, X.-M.; Cheng, J.-P. J. Org.
Chem. 1993, 58, 6410. (c) Bordwell, F. G.; Satish, A. V. J. Am. Chem.
Soc. 1992, 114, 10173. (d) Bordwell, F. G.; Harrelson, J. A., Jr.; Satish, A.
V. J. Org. Chem. 1989, 54, 3101. (e) Bordwell, F. G.; Cheng, J.-P. J. Am.
Chem. Soc. 1989, 111, 1792.
(15) Wayner, D. D. M.; Parker, V. D. Acc. Chem. Res. 1993, 26, 287.
S0002-7863(96)00378-2 CCC: $12.00 © 1996 American Chemical Society

5656 J. Am. Chem. Soc., Vol. 118, No. 24, 1996
Burton et al.
chemistry, and more specifically photoinduced electron transfer,
to drive energetically uphill processes in chemical synthesis and
energy conversion.16,17 Furthermore, knowledge of the rates
of specific radical ion reactions such as deprotonation,7a,10a,11,13
carbon-carbon6,10c,d and carbon-heteroatom4,5c bond scission,
rearrangments,9 and addition to unsaturated bonds8a,12 provides
a method to probe the dynamics of geminate ion-radical pairs
born in the primary event of photoinduced electron transfer.
Amine radical cations have been the focus of much work in
radical ion chemistry. Early photochemical and electrochemical
work suggested that the acidity of protons on a carbon that is
R to an amine radical cation is greatly enhanced relative to the
corresponding neutral amines.18-20 More recent studies by
Lewis,21 von Sonntag,22 Dinnocenzo,7 Parker,11c and Mariano4e
provide a quantiative thermodynamic and kinetic description
of this effect. Studies also reveal that the departure of a variety
of electrofugal groups is significantly enhanced by an R-amine
radical cation center.4
the relative stability of the radical cation for the series of
structurally related 2-aminoalcohols, 1a-e. The peak potentials
for irreversible oxidation of the neutral aminoalcohols were
determined by electrochemistry, while the rates and activation
parameters for C-C bond fragmentation of the radical cations
1a-e•+ were determined by laser flash photolysis. The results
of this study indicate that there is a strong correlation between
the free energy of activation for C-C bond fragmentation and
the peak potential for anodic oxidation of the aminoalcohol.
Furthermore, comparison of the reaction in CH3CN solution and
in the gas phase reveals interesting features concerning the
mechanism of the C-C bond fragmentation process.
Our interest has centered on the reactivity of radical cations
derived from 2-aminoalcohols and 1,2-diamines.23 These radical
cations undergo facile C-C bond fragmentation in a manner
consistent with the expectation that the amine radical cation
activates the adjacent C-C σ bond, e.g.,
Experimental Section
Materials. Solvents and chemicals used for synthesis were of
reagent grade and used without purification unless noted. Silica gel
(Merck, 230-400 mesh) was used for chromotography. NMR spectra
were obtained on a GE QE 300-MHz spectrometer. Metal complex
fac-[(bpy)ReI(CO)3(4-benzylpyridine)+][PF6-] (3) was synthesized and
characterized as described previously.25a
erythro-2-[(p-Methoxyphenyl)amino]-1,2-diphenylethanol (1a). p-
Anisidine (1.25 g, 10.2 mmol) was added to a Schlenk tube which
was subsequently placed under argon. Then 50 mL of dry CH2Cl2
was added to the Schlenk tube via a cannula. The Schlenk tube was
cooled to 0 °C on an ice-water bath whereupon 5.7 mL of a 2.0 M
solution of Al(CH3)3 in hexane (11.4 mM) was added via a syringe.
The solution was stirred for 1 h after which time 2.0 g of trans-stilbene
oxide (10.2 mmol) dissolved in 5 mL of CH2Cl2 was slowly added.
The reaction was monitored by TLC (silica gel, 35% hexanes/CH2Cl2);
a spot at Rf ) 0.35 was observed after 45 min. After 2 h, the trans-
stilbene oxide spot at Rf ≈ 0.9 had disappeared and the reaction was
quenched by addition of 5% aqueous NaOH. The solution was
extracted with two 25-mL portions of CH2Cl2. The combined organic
fractions were dried over MgSO4, and the solvent was then removed
under reduced pressure to yield 1 g of a brownish oily solid. The crude
product was purified by flash chromotography (silica gel, 25% ethyl
acetate/hexanes). Pure 1a was obtained as a white crystalline solid,
0.5 g (yield 15%). NMR analysis confirmed that the sample contains
less than 5% of the threo isomer.
Whitten’s work along with our own indicated that the dynamics
of C-C bond fragmentation depend on a variety of factors
including the structure of the electrofugal group, the stereo-
chemistry, and the relative stability of the radical cation as
reflected by the oxidation potential of the netural amine.23,24
The objective of the present study was to determine the
relationship between the dynamics of C-C bond scission and
(16) For an excellent series of reviews of photoinduced electron transfer
and the reactivity of radical ions derived therefrom see: (a) Photoinduced
Electron Transfer, Fox, M. A., Chanon, M. D., Eds.; Elsevier: Amsterdam,
1988; Parts A-D. (b) Photoinduced Electron Transfer I. In Top. Curr.
Chem. 1990, 156, 1-125. (c) Photoinduced Electron Transfer III. In Top.
Curr. Chem. 1991, 159, 1-259. (d) Photoinduced Electron Transfer IV. In
Top. Curr. Chem. 1992, 163, 1-245. (e) Photoinduced Electron Transfer
V. In Top. Curr. Chem. 1993, 168, 1-270.
(17) (a) Mattes, S. L.; Farid, S. Science 1984, 226, 917. (b) Gould, I.
R.; Ege, D.; Moser, J. E.; Farid, S. J. Am. Chem. Soc. 1990, 112, 4290. (c)
Gould, I. R.; Moser, J. E.; Armitage, B.; Farid, S. Res. Chem. Intermed.
1995, 21, 793.
1
Spectral data: TLC (silica, 35% hexanes/CH2Cl2) Rf ) 0.35; H
NMR (300 MHz, CDCl3) δ 3.69 (s, 3H), 4.62 (d, J ) 4.6 Hz, 1H),
5.07 (d, J ) 4.6 Hz, 1H), 7.00-7.31 (m, 14H); 13C NMR (75 MHz,
CDCl3) δ 55.7, 64.7, 77.1, 114.8, 115.5, 126.6, 127.3, 127.9, 128.1,
128.2, 138.8, 140.1, 140.9, 152.5. Anal. Calcd for C21H21NO2: C,
78.95; H, 6.64; N, 4.38. Found: C, 78.76; H, 6.63; N, 4.30.
erythro-2-[(p-Methylphenyl)amino]-1,2-diphenylethanol (1b). This
compound was synthesized following the procedure outlined for 1a,
except that 1.64 g of p-toluidine (15.3 mmol) was used in place of
p-anisidine. The amounts of the other reagents were scaled-up
accordingly.
(18) Cohen, S. G.; Parola, A.; Parsons, G. H. Chem. ReV. 1973, 73, 141.
(19) Peters, K. S.; Schaeffer, C. G. J. Am. Chem. Soc. 1980, 102, 7566.
(20) Mann, C. K.; Barnes, K. K. Electrochemical Reactions in Non-
Aqueous Systems; Marcel Dekker: New York, 1970.
(21) (a) Lewis, F. D. Acc. Chem. Res. 1986, 19, 401. (b) Lewis, F. D.
AdV. Photochem. 1986, 13, 165. (c) Lewis, F. D.; Bassani, D. M.; Burch,
E. L.; Cohen, B. E.; Engleman, J. A.; Reddy, G. D.; Schneider, S.; Jaeger,
W.; Gedeck, P.; Gahr, M. J. Am. Chem. Soc. 1995, 117, 660.
(22) Das, S.; von Sonntag, C. Z. Naturforsch. 1986, 416, 505.
(23) (a) Wang, Y.; Hauser, B. T.; Rooney, M. M.; Burton, R. D.; Schanze,
K. S. J. Am. Chem. Soc. 1993, 115, 5675. (b) Lucia, L. A.; Burton, R. D.;
Schanze, K. S. J. Phys. Chem. 1993, 97, 9078. (c) Wang, Y.; Lucia, L. A.;
Schanze, K. S. J. Phys. Chem. 1995, 99, 1961. (d) Lucia, L. A.; Wang, Y.;
Nafisi, K.; Netzel, T. L.; Schanze, K. S. J. Phys. Chem. 1995, 99, 11801.
(e) Wang, Y.; Schanze, K. S. J. Phys. Chem., in press.
(24) (a) Ci, X.; Lee, L. Y. C.; Whitten, D. G. J. Am. Chem. Soc. 1987,
109, 2536. (b) Kellett, M. A.; Whitten, D. G. J. Am. Chem. Soc. 1989, 111,
2314. (c) Ci, X.; Whitten, D. G. J. Am. Chem. Soc. 1989, 111, 3459. (d)
Bergmark, W. R.; Whitten, D. G. J. Am. Chem. Soc. 1990, 112, 4042. (e)
Ci, X.; Kellett, M. A.; Whitten, D. G. J. Am. Chem. Soc. 1991, 113, 3893.
(f) Leon, J. W.; Whitten, D. G. J. Am. Chem. Soc. 1993, 115, 8038. (g)
Gan, H.; Leinhos, U.; Gould, I. R.; Whitten, D. G. J. Am. Chem. Soc. 1995,
99, 3566.
After extraction, 1.0 g of crude 1b was obtained as a white solid.
The crude material was purified by recrystallization from EtOH/H2O
(1:1 v/v). The purified product was obtained as white crystals, yield
750 mg (16%). NMR analysis confirmed that the product contained
less than 5% of the threo isomer.
1
Spectral data: TLC (silica, 35% hexanes/CH2Cl2) Rf ) 0.35; H
NMR (300 MHz, CDCl3) δ 2.25 (s, 3H), 2.48 (s, 1H), 4.67 (d, J ) 4.5
Hz, 1H), 5.09 (d, J ) 4.5 Hz, 1H), 7.15-7.30 (m, 14H); 13C NMR (75
(25) (a) Wang, Y.; Schanze, K. S. Inorg. Chem. 1994, 33, 1354. (b)
Wang, Y.; Schanze, K. S. Chem. Phys. 1993, 176, 305.

C-C Bond Fragmentation in Aminoalcohol Radical Cations
J. Am. Chem. Soc., Vol. 118, No. 24, 1996 5657
MHz, CDCl3) δ 20.3, 64.1, 77.1, 114.2, 126.6, 127.2, 127.5, 127.9,
128.2, 129.6, 138.7, 140.1, 144.5. Anal. Calcd for C21H21NO: C,
83.12; H, 6.99; N, 4.62. Found: C, 83.17; H, 7.01; N, 4.55.
Methods and Equipment for Photophysics and Electrochemistry.
Stern-Volmer luminescence quenching experiments were carried out
by monitoring the emission lifetime of 3 as a function of the
concentration of 1a-e. Luminescence decays were determined by time-
correlated single photon counting on an instrument manufactured by
Photochemical Research Associates (model FLI).
Electrochemistry was carried out on a cyclic voltammograph
(Bioanalytical Systems, CV-27) using a glassy carbon disk working
electrode, a Pt wire auxilliary electrode, and an SCE reference electrode.
Cyclic voltammetry of aminoalcohols 1a-e (c ≈ 10 mM) was carried
out in CH3CN solution with 0.1 M tetrabutylammonium hexafluoro-
phosphate (TBAH) as a supporting electrolyte.
erythro-2-(Phenylamino)-1,2-diphenylethanol (1c). This com-
pound was synthesized following the procedure outlined for 1a, except
that 1.64 g of aniline (18.0 mmol) was used in place of p-anisidine.
The amounts of the other reagents were scaled-up accordingly.
After extraction, 1.5 g of crude 1c was obtained as a yellow solid.
The crude material was purified by recrystallization from EtOH/H2O
(1:1 v/v). The purified product was obtained as white crystals, yield
1.05 g (20%). NMR analysis confirmed that the product contained
less than 5% of the threo isomer.
Transient absorption studies were carried out on an apparatus that
has been described previously.25b Excitation was effected by using the
3rd harmonic output of a Nd:YAG laser (Spectra Physics, GCR-14S),
the excitation dose was 6 mJ‚pulse-1 (flux of 18 mJ‚cm-2‚pulse-1) and
the pulse width was 8 ns. For transient absorption experiments the
concentration of 3 was approximately 0.1 mM (absorbance at 355 nm
was 0.5) and the concentrations of 1a-e were adjusted so that the
MLCT excited state of 3 was quenched with ≈65% efficiency.
Transient absorption studies were carried out with samples contained
in a flow cell to minimize the effect of sample decomposition on the
kinetics. Typically a 100-mL volume of sample solution was used for
each data acquisition. Temperature-dependent flash photolysis studies
were carried out by using a jacketed flow cell that was maintained at
constant temperature by a recirculating bath operating with a water/
glycol mixture as the coolant. Activation parameters were derived from
kinetic data sets comprising at least 10 rates measured at temperatures
ranging from -15 to +40 °C. All Eyring plots were nicely linear.
Fourier Transform Ion Cyclotron Resonance (FT-ICR) Studies.
FT-ICR experiments were carried out using an Ionspec data workstation
equipped with a 3 T superconducting magnet. The instrument was
operated in broadband mode with detection over a range of 100-400
m/z. The aminoalcohols were introduced via a heated solids probe to
a constant background pressure of 5 × 10-8 Torr in the analyzer cell
region. Ionization was effected by electron impact using low energies
(8-16 eV) and the ions were trapped in the FT-ICR analyzer cell using
a potential of +1.0 V. Short reaction delays were utilized to the limits
of the instrumentation (ca. 10 ms). The excitation detection employed
a standard frequency-chirp excitation. To maximize S/N rations, 10
scans (8K data points) were obtained and co-added prior to Fourier
transformations.
1
Spectral data: TLC (silica, 35% hexanes/CH2Cl2) Rf ) 0.25; H
NMR (300 MHz, CDCl3) δ 2.48 (s, 1H, amino), 4.70 (d, J ) 4.5 Hz,
1H), 5.11 (d, J ) 4.5 Hz, 1H), 7.08-7.31 (m, 15H); 13C NMR (75
MHz, CDCl3) δ 63.8, 77.2, 114.0, 118.0, 126.5, 127.6, 128.0, 128.2,
129.1, 138.5, 140.1, 146.8. Anal. Calcd for C20H19NO: C, 83.01; H,
6.62; N, 4.84. Found: C, 82.91; H, 6.66; N, 4.74.
erythro-2-[(p-Chlorophenyl)amino]-1,2-diphenylethanol (1d). p-
Chloroaniline (3.2 g, 25 mmol) and trans-stilbene oxide (1.0 g, 5.1
mmol) were ground up together by using a mortar and pestle. The
chloroaniline/stilbene oxide mixture and a magnetic stir bar were placed
into a heavy-walled glass tube that was fitted with a Teflon high-vacuum
stopcock. The tube was evacuated under high vacuum and then sealed
by closing the stopper. The tube was heated to 150 °C in a sand bath
while the mixture was stirred by using a magnetic stirring motor. (This
temperature was sufficient to melt the mixture.) The tube was cooled
at intervals of 24 h and the reaction mixture was analyzed by TLC
(silica gel, 35% hexanes/CH2Cl2). After 4 days the trans-stilbene oxide
spot at Rf ) 0.9 had disappeared and at this point the reaction was
deemed complete. The crude reaction product was dissolved in CH2Cl2
and filtered and then the solvent was removed under reduced pressure.
The resulting crude product was a yellowish-brown solid which was
purified by recrystallization two times from ethyl acetate/hexanes (4:1
v/v). The recrystallized product was obtained as a powdery white solid,
yield 700 mg (43% based on trans-stilbene oxide). NMR analysis
confirmed that the product contained less than 1% of the threo isomer.
Spectral data: 1H NMR (300 MHz, CDCl3) δ 2.35 (s, 1H, amino),
4.70 (d, J ) 4.5 Hz, 1H, methine adjacent to hydroxyl), 5.18 (d, J )
4.5 Hz, 1H, methine adjacent to amino), 7.02-7.38 (m, 14H, phenyls);
13C NMR (75 MHz, CDCl3) δ 63.9 (adjacent to hydroxyl), 77.2
(adjacent to amino), 115.0, 126.5, 127.7, 127.9, 128.1, 128.3, 128.9,
138.1, 139.9, 145.4 (all aromatic) Anal. Calcd for C20H18ClNO: C,
74.17; H, 5.61; N, 4.33. Found: C, 74.34; H, 5.67; N, 4.25.
Computational Methods. Semiempirical calculations were carried
out using the AM1 parameter set26a as implemented within the MOPAC
6.0 program system. Geometries of all species were optimized using
the Restricted Hartree-Fock (RHF) method. In the case of open-shell
systems, this leads to the “Half-Electron Configuration Interaction” (HE-
CI) treatment of Dewar.26b Although the RHF/HE-CI method was
exceedingly slow, it was selected over the method described by
Camaioni27 (UHF geometry optimization followed by a single point
RHF/HE-CI calculation) because the self-consistent field (SCF) calcula-
erythro-2-[(p-Cyanophenyl)amino]-1,2-diphenylethanol (1e). p-
Aminobenzonitrile (2.41 g, 20.4mmol) and trans-stilbene oxide (2.0
g, 10.2 mmol) were ground up together by using a mortar and pestle.
The aminobenzonitrile/stilbene oxide mixture and a magnetic stir bar
were placed into a heavy-walled glass tube that was fitted with a Teflon
high-vacuum stopcock. The tube was evacuated under high vacuum
and then sealed by closing the stopper. The tube was heated to 150
°C in a sand bath while the mixture was stirred by using a magnetic
stirring motor. (This temperature was sufficient to melt the mixture.)
The reaction was monitored by TLC (25% ethyl acetate/hexanes); a
new spot at Rf ) 0.20 was observed after 2 h. The reaction was allowed
to proceed for 14 h until it appeared that the trans-stilbene oxide spot
at Rf ) 0.9 had disappeared. The brownish mixture was cooled to
room temperature after which time the resulting solid was dissolved in
CH2Cl2. The solution was filtered and dried over MgSO4 and the
solvent was removed under reduced pressure. The resulting crude
product was purified by recrystallization two times from ethyl acetate/
hexanes (4:1 v/v). The recrystallized product was obtained as a
powdery white solid, yield 750 mg (23% based on trans-stilbene oxide).
NMR analysis confirmed that the product contained less than 1% of
the threo isomer.
tions did not converge under UHF with 1c•+
. A similar SCF
convergence problem was also encountered with 1c•+ using GAMESS
with UHF and the AM1 parameter set.28 For all species, optimized
geometries were determined by using a variety of different starting
geometries in order to minimize the possibility for locating structures
that correspond to a local, but not a global energy minimum.
Calculations were carried out on a Sun SPARCcenter 2000 compute
server and geometries were visualized by using Chem 3D on a
Macintosh desktop computer.
Results
General Objectives and Methodology. The principal aim
of the experimental studies described herein was the follow-
Spectral data: TLC (silica, 25% hexanes/CH2Cl2) Rf ) 0.20. 1H
NMR (300 MHz, CDCl3) δ 2.49 (s, 1H), 4.68 (d, J ) 4.5 Hz, 1H),
5.15 (d, J ) 4.5 Hz, 1H), 7.05-7.45 (m, 14H); 13C NMR (75 MHz,
CDCl3) δ 62.5, 75.4, 120.9, 96.4, 113.2, 127.2, 127.4, 127.8, 128.0,
128.8, 133.5, 140.2, 143.4, 151.5. Anal. Calcd for C21H18N2O: C,
80.22; H, 5.78; N, 8.91. Found: C, 80.48; H, 5.89; N, 8.83.
(26) (a) Dewar, M. J. S.; Zoebisch, E. G.; Healy, E. F.; Stewart, J. J. P.
J. Am. Chem. Soc. 1985, 107, 3902. (b) Dewar, M. J. S.; Hashmall, J. A.;
Venier, C. G. J. Am. Chem. Soc. 1968, 90, 1953.
(27) Camaioni, D. M. J. Am. Chem. Soc. 1990, 112, 9475.
(28) Schmidt, M. W.; Baldridge, K. K.; Boatz, J. A.; Jensen, J. H.;
Koseki, S.; Gordon, M. S.; Nguyen, K. A.; Windus, T. L.; Elbert, S. T.
QCPE Bull. 1990, 10, 52.

5658 J. Am. Chem. Soc., Vol. 118, No. 24, 1996
Burton et al.
Table 1. Electrochemistry of Aminoalcohols and N,N ′-
Dimethylanilinea
dEp/d(log V)
substituent
Ep(1) Vb
(mV‚decade-1
)
E1/2(2/2•+) (V)c
-OCH3
-CH3
-H
0.73
0.83
1.02
1.04
1.31
178
73
148
121
156
0.57
0.73
0.88
0.91
1.17
-Cl
-CN
a All data for CH3CN solutions. Potentials are relative to SCE
reference electrode. b Peak potential of irreversible anodic wave for first
sweep at V ) 100 mV/s. c Data from ref 11c.
ing: (1) to determine the kinetics of fragmentation for the series
of aminoalcohol radical cations 1a-e•+ and (2) to correlate the
kinetic data with the relative thermodynamic stability of the
radical cations, as reflected by the thermodynamic oxidation
potentials of the aminoalcohols.
Figure 1. Plot of Ep(1) as a function of E1/2(2/2•+). Data are from
Table 2 and the slope of the solid line is 1.0.
Kochi and co-workers have carefully studied the electrochemical
properties of compounds that exhibit irreversible anodic waves
due to an EC mechanism.32 Their work demonstrates that under
conditions where anodic oxidation of a series of structurally
related substrates is “totally irreversible”, Ep values for the series
correlate well with the thermodynamic potentials (E°) for the
oxidations.32a Since we were interested in establishing the
relationship between Ep(1) and the thermodynamic potential,
E°(1/1•+), experiments were carried out to demonstrate that
anodic oxidation of compounds 1a-e is totally irreversible.
One of the principal criteria for establishing that an anodic
process is totally irreversible is the observation of a sweep rate
dependence for Ep in excess of 30 mV/decade.32a,33 Therefore,
the dependence of Ep(1) on the sweep rate (V) for 1a-e was
examined. Plots of Ep(1) vs log V for 1a-e are linear, with
slopes in most cases in excess of 100 mV/decade (Table 1), as
expected if the oxidations are totally irreversible. Thus, given
these relatively large dEp/dV values, it is reasonable to assume
that anodic oxidation of compounds 1a-e is totally irreversible.
It follows then that Ep(1) should correlate linearly with E°(1/
1•+) for the series.32a Further support for this premise comes
from correlation of Ep(1) values with the half-wave potentials
for reVersible oxidation of the series of para-substituted N,N-
dimethylanilines (2a-e), E1/2(2/2•+).11c Figure 1 illustrates a
plot of Ep(1) vs E1/2(2/2•+); in this plot a solid line with a slope
of 1.0 is drawn through the data points. Quite remarkably, this
correlation indicates that the substituent effect on Ep(1) is exactly
parallel to the substituent effect on the thermodynamic potential
for the 2/2•+ couple. This observation is strong evidence in
support of our assertion that there is a 1:1 correlation between
the substituent effects on Ep(1) and E°(1/1•+).
Laser flash photolysis was used to determine the fragmenta-
tion kinetics of the radical cations. Thus, the transition metal
complex sensitizer, [(2,2′-bipyridine)ReI(CO)3(4-benzylpyridine)+]-
[PF6-] (3),29 was photochemically excited with 355-nm pulses
from a Nd:YAG laser. The excited state metal complex then
reacts with aminoalcohols 1a-e by electron transfer to produce
the corresponding aminoalcohol radical cations 1a-e•+, which
subsequently decay by C-C bond fragmentation. In order to
facilitate the flash photolysis studies the aniline chromophore
was incorporated into the structure of the aminoalcohols because
the aniline radical cation has a pronounced mid-visible absorp-
tion that can be easily detected by transient absorption
spectroscopy.4e,23b,30 It was not known at the outset of this study
that R-amino radicals 4a-e (Scheme 1), which are produced
by fragmentation, exhibit pronounced absorption in the near-
UV region. This serendipitous feature greatly facilitated the
analysis of the transient absorption kinetics.
There are several reasons why metal complex sensitizer 3
was selected for this study. First, 3 has a long-lived metal-to-
ligand charge transfer (MLCT) excited state. The MLCT state
is a comparatively strong oxidant, with a reduction potential of
approximately +1.22 V vs SCE.29 Furthermore, previous work
indicated that this sensitizer produces high yields of free radical
ions as a result of excited state electron transfer reactions with
amines.31
Electrochemistry of Aminoalcohols. Cyclic voltammo-
grams of aminoalcohols 1a-e feature an irreversible anodic
wave with peak potentials, Ep(1), ranging from +0.7 to +1.3
V (Table 1). Ep(1) increases systematically with the electron
withdrawing ability of the aniline substituent, consistent with
assignment of the anodic wave to production of radical cations
1a-e•+ via oxidation of the aniline unit. The electrochemical
oxidation process is irreversible owing to the rapid chemical
reaction of 1a-e•+; in other words, the anodic oxidation is an
EC process,
Stern-Volmer Luminescence Quenching. Stern-Volmer
quenching experiments were carried out to determine the
efficiency by which aminoalcohols 1a-e quench the MLCT
excited state of 3. Metal complex 3 exhibits a moderately
intense MLCT emission in CH3CN (λmax ) 595 nm, Φem
)
air
0.045, τem ) 210 ns, and τem ) 113 ns).25a Stern-Volmer
quenching experiments were carried out by using time-correlated
single photon counting to determine the MLCT emission lifetime
of an air-saturated CH3CN solution of 3 (τemair) as a function
of the concentration of 1a-e. Table 2 lists the second-order
rate constants for MLCT emission quenching (kq) calculated
from the slopes of the Stern-Volmer plots. The Stern-Volmer
studies demonstrate that each aminoalcohol quenches the MLCT
emission of 3 efficiently, presumably via electron transfer. This
is expected, since given the excited state reduction potential of
(29) Schanze, K. S.; MacQueen, D. B.; Perkins, T. A.; Cabana, L. A.
Coord. Chem. ReV. 1993, 122, 63.
(30) Shida, T. Electronic Absorption Spectra of Radical Ions; Elsevier:
Amsterdam, 1988.
(32) (a) Klingler, R. J.; Kochi, J. K. J. Am. Chem. Soc. 1980, 102, 4790.
(b) Klingler, R. J.; Kochi, J. K. J. Am. Chem. Soc. 1982, 104, 4186.
(33) Nicholson, R. S.; Shain, I. Anal. Chem. 1964, 36, 706.
(31) Lucia, L. A.; Schanze, K. S. Inorg. Chim. Acta 1994, 225, 41.

C-C Bond Fragmentation in Aminoalcohol Radical Cations
J. Am. Chem. Soc., Vol. 118, No. 24, 1996 5659
Table 2. Kinetic Properties of Aminoalcoholsa
b
q
q
d
kq
k
BF(298 K)c
(s-1
∆GBFq(298 K)
(kcal/mol)
∆HBF
∆SBF
kbase
(M-1 s-1
)
comp
substituent
(M-1 s-1
)
)
(kcal)
(cal/(mol‚K))
1a
1b
1c
1d
1e
-OCH3
-CH3
-H
7.6 × 109
5.6 × 109
2.3 × 109
4.2 × 109
8.0 × 108
3.9 × 104
2.6 × 105
3.7 × 105
5.5 × 105
7.4 × 106
11.2
10.1
9.8
9.6
8.1
<105
8.0 × 106
1.8 × 108
4.3 × 108
7.5 × 108
7.6 ( 1
3.0 ( 1
4.1 ( 1
-9 ( 4
-20 ( 4
-12 ( 4
-Cl
-CN
a All data for CH3CN solutions at 25 °C. b Rate constant for Stern-Volmer quenching of luminescence of complex 3 in degassed solution. c Rate
constant for C-C bond fragmentation of radical cations 1a-e•+. Estimated error is (10%. d Rate constant for pyridine base catalyzed C-C bond
fragmentation of cation radicals, 1•+
.
Scheme 1
the MLCT state of 3 (+1.22 V), electron transfer quenching is
exothermic for most of the aminoalcohol donors.34 Further
evidence in support of the electron transfer quenching mecha-
nism comes from the fact that the kq values for 1a-e increase
as Ep(1) decreases.
mechanism outlined in Scheme 1. Thus, excitation of 3
produces the metal to ligand charge transfer (MLCT) excited
state, 3*, which is then quenched by electron transfer from 1c.
Quenching produces reduced metal complex 3 (red) and
aminoalcohol radical cation 1c•+. Radical cation 1c•+ subse-
quently undergoes heterolytic C-C bond fragmentation to
produce R-amino radical 4c and benzaldehyde (Vide infra).
Under air-saturated conditions 4c reacts rapidly with O2 to
produce imine 5c.23c Imines of type 5c hydrolyze under the
conditions of reversed-phase HPLC analysis to produce a second
equivalent of benzaldehyde and aniline (6c).23a Although
detailed steady-state photochemical studies were not carried out
on the other aminoalcohols, a consistent pattern was observed
in the transient spectroscopy of each compound (Vide infra)
which strongly implies that Scheme 1 applies to all five
aminoalcohols (1a-e).
Laser Flash Photolysis Studies. Kinetics of C-C Bond
Fragmentation of the Radical Cations. Degassed CH3CN
solutions containing metal complex 3 and aminoalcohols 1a-e
were subjected to laser flash photolysis. Excitation of solutions
containing 3 and one of the aminoalcohol quenchers gave rise
to moderate transient absorption throughout the near-UV and
visible which persisted for greater than 100 µs. Figure 2
illustrates the time evolution of the transient absorption of
solutions containing 3 and 1c or 1d at times ranging from 0 to
approximately 10 µs after excitation. Time-resolved absorption
Steady-State Photochemical Studies. Steady-state photo-
chemical studies were carried out to confirm that irradiation of
mixtures of metal complex 3 and aminoalcohol 1c produces
products consistent with photoinduced electron transfer and
subsequent C-C bond fragmentation of radical cation 1c•+.
Thus, 3.0 mL of an air-saturated CH3CN solution of metal
complex 3 (c ) 1 mM) and 1c (c ) 2.3 mM) was irradiated at
366 nm (I ) 6.6 × 10-9 einsteins/s) for 10 min.35 The resulting
solution was analyzed by HPLC using a C18 reversed phase
column and a UV detector (λ ) 254 nm), and the analysis
indicated that the reaction products were benzaldehyde and
aniline. Quantum yield studies carried out under the conditions
listed above indicated that the reaction is highly efficient;
formation of benzaldehyde and disappearance of 1c occur with
quantum yields of 1.0 and 0.5, respectively.
The products and stoichiometry observed for steady-state
irradiation of the mixture of 3 and 1c are consistent with the
(34) The free energy for electron transfer quenching of 3* is given
approximately by ∆GET ) Ep(1) - 1.22 V, where 1.22 V is the reduction
potential of 3*.29 For 1a-e, ∆GET ranges from -0.49 (1a) to +0.1 eV
(1e).
(35) Only metal complex 3 absorbs light at 366 nm.

5660 J. Am. Chem. Soc., Vol. 118, No. 24, 1996
Burton et al.
the kinetics of the transient absorption spectral evolution were
simulated very well over the entire spectral range from 330 to
600 nm by using a first-order kinetic model.39 The first-order
rate constants derived from the global analysis are summarized
in Table 2. Inspection of these data reveals that the bond
fragmentation rate increases systematically with Ep(1), which
indicates that kBF increases as the stability of the radical cation
decreases. The implications of the correlation between kBF and
Ep(1) are discussed in detail below.
Activation parameters for bond fragmentation (∆Hq and
BF
∆SqBF) in 1c-e•+ were determined from Eyring analysis of kBF
over the temperature range from -10 to +35 °C (Table 2). The
temperature-dependent kinetic data reveal that bond fragmenta-
tion in the radical cations is characterized by comparatively low
activation enthalpies and negative activation entropies.
Base-Catalyzed C-C Bond Fragmentation. Previous stud-
ies with 3 and 1c revealed that addition of pyridine or substituted
pyridines accelerates the rate of C-C bond fragmentation of
1c•+.23b Kinetic studies of 1c as a function of pyridine
concentration demonstrated that the rate of fragmentation in the
presence of pyridine (kobs) followed a linear relationship of the
form
Figure 2. Time-resolved transient absorption difference spectra for
CH3CN solutions containing 3 (c ) 1.2 × 10-4 M) and either 1c or
1d: (a) aminoalcohol 1d (c ) 5 × 10-3 M); (b) aminoalcohol 1c (c )
5 × 10-3 M). Delay times range from 0 to 4 µs following excitation
with a pulse from a Nd:YAG laser (355 nm, 6 mJ/pulse, 8 ns fwhm).
The solid line marked with polygons is the spectrum immediately after
laser pulse and arrows indicate the direction of change in absorbance
with increasing delay time.
kobs ) kBF + kbase[pyridine]
(7)
where kbase is the second-order rate constant for the base-
catalyzed reaction pathway. Similarly, kbase was determined for
1c•+ with a series of substituted pyridines,40 and a correlation
of kbase with pKa for the series of pyridines yielded a Brønsted
coefficient, â ) 0.62.23b Finally, kbase for reaction of 1c-OD
with pyridine yielded a deuterium isotope effect, kHbase/kbDase
)
spectra of 3 with the other aminoalcohols were similar. For
each aminoalcohol, at early delay times the difference spectra
display absorption throughout the near-UV and visible region;
however, an absorption band in the 450-500-nm region is
clearly discernible. This mid-visible absorption is clearly
delineated for aminoalcohol 1d (Figure 2a) where band maxima
appear at 465 and 495 nm. This mid-visible absorption is
attributed to the aniline radical cation chromophore (1a-e•+,
Scheme 1), which is produced by photoinduced electron
transfer.4e,23b,30 Figure 2 also reveals that the difference spectra
evolve during the 0-10 µs time domain and the evolution is
characterized by the following: (1) a decrease in the absorption
of the aniline radical cation chromophore in the 450-500-nm
region and (2) appearance of a strong absorption band at 360-
370 nm in the near-UV. The spectral evolution is attributed to
heterolytic C-C bond fragmentation of radical cations 1a-e•+
to produce benzaldehyde and R-amino radicals, 4a-e (eq 5,
Scheme 1). The absorption band that appears at 360-370 nm
concomitant to disappearance of the aniline radical cation
absorption is attributed to radicals 4a-e. The large absorptivity
and low energy of this absorption band is consistent with the
highly delocalized electronic structure 4a-e, and the spectral
assignment is also supported by the observation of similar
absorption features in structurally similar radicals.36
2.0.
In the present study, the effect of pyridine on the rates of
C-C bond fragmentation in 1a•+, 1b•+, 1d•+, and 1e•+ was
determined. The rates of fragmentation were measured as a
function of [pyridine], and kbase was determined from linear
correlations of kobs vs [pyridine] according to eq 7. As noted
for 1c•+, pyridine accelerates the rate of fragmentation of 1b•+,
1d•+, and 1e•+, and kbase values for these radical cations are
listed in Table 2. Rate acceleration was not observed for
fragmentation of 1a•+ for [pyridine] e 0.1 M, which indicates
that kbase e 105 M-1 s-1 for this radical cation. Inspection of
the data in Table 2 reveals that kbase increases with Ep(1) and
the significance of this correlation is discussed below.
Gas-Phase Bond Fragmentation Studies. Fourier Trans-
form Ion Cyclotron Resonance (FT-ICR) studies were carried
out to determine the fragmentation pathways for radical cations
1a-e•+ in the gas phase. It was also hoped that it might be
possible to monitor the kinetics of the gas-phase fragmentation
reactions. Aminoalcohols 1a-e were introduced into the FT-
ICR cell via a heated solids probe and the radical cations were
generated by electron impact ionization. The lowest possible
electron energy was used in order to increase the probability of
observing the parent radical cations (8-16 eV was typically
required). Despite this effort, for each aminoalcohol only
fragment ions were detected, even at the earliest accessible delay
times. As a result, it was not possible to monitor the gas-phase
fragmentation kinetics. For each aminoalcohol, the most
prominent ion peak in the mass spectrum corresponded to
iminium ion 7, which implies that the dominant fragmentation
Global kinetic analysis38 was applied to the multiwavelength
transient absorption data for aminoalcohols 1a-e to determine
the rates of C-C bond fragmentation for the corresponding
radical cations (kBF, eq 5, Scheme 1). For each aminoalcohol,
(36) The (N-methylanilino)methyl radical (e.g., the conjugate base of
2c•+) has a strong absorption band at 330 nm.37 Also, R-amino radicals
produced by C-C bond fragmentation of 1,2-diaminoethane radical cations
generally exhibit strong absorption in the 340-370-nm region.23
(37) Holcman, J.; Sehested, K. J. Phys. Chem. 1977, 81, 1963.
(38) (a) SPECFIT (version 2.10), Spectrum Software Associates: Chapel
Hill, NC, 1996. (b) Stultz, L. K.; Binstead, R. A.; Reynolds, M. S.; Meyer,
T. J. J. Am. Chem. Soc. 1995, 117, 2520.
(39) The kinetics of the spectral evolution that is attributed to C-C bond
fragmentation of cation radicals 1a-e•+ is independent of the laser excitation
power. This is consistent with the fact that C-C bond fragmentation is a
first-order process.
(40) The series of pyridines included 4-chloropyridine, pyridine, 4-meth-
ylpyridine, and 3,4-dimethylpyridine.23b

C-C Bond Fragmentation in Aminoalcohol Radical Cations
Table 3. AM1 Calculated Heats of Formation
J. Am. Chem. Soc., Vol. 118, No. 24, 1996 5661
a
b
c
∆Hf
∆Hhom
∆Hhet
structure
acronym
conformation
(kcal/mol)
(kcal/mol)
(kcal/mol)
H
N
1c
C1 and C2 phenyls gauched
31.6
41
Ph
OH
Ph
Ph
1c•+
4c
C1 and C2 phenyls gauched
phenyls anti
204.4
73.9
10
29
H
N
Ph
Ph
OH
Ph
Ph
H
N
Ph
H
H
7c
phenyls trans
215.4
Ph N+
Ph
H
Ph
8
9
phenyl and H syn
phenyl and H syn
-1.4
H
H
O
H
Ph
159.7
O
H
+
a Calculated values obtained by use of AM1 (MOPAC 6.0). Even electron species computed by using RHF and odd electron species calculated
by using RHF/HE-CI. b AM1 calculated enthalpy for C-C bond homolysis. c AM1 calculated enthalpy for C-C bond heterolysis. d See Figure 3
for structures.
the C1 and C2 phenyl groups are anti (Figure 3). Thus, the
calculations suggest that the steric interaction between the
N-phenyl and C1 phenyl is larger than that between the C1 and
C2 phenyls. This observation is significant, because it has been
previously suggested that in the preferred conformation of the
transition state for C-C bond fragmentation in 1c•+, the aniline
nitrogen and hydroxy oxygen are anti and the C1 and C2 phenyl
groups are anti.24b,e Therefore, if the calculations are correct,
they imply that part of the activation energy for bond fragmen-
tation of 1c•+ may be due to the requirement for endothermic
conformational rearrangement to the anti conformer.
The ∆Hf data can be used to calculate enthalpies for
homolysis of the C1-C2 bond in 1c (∆Hhom), and for homolysis
and heterolysis (∆Hhet) of the C1-C2 bond in 1c•+ (Table 3).
First, ∆Hhom calculated for 1c (41 kcal/mol) is in reasonable
agreement with AM1 calculated27 and experimental6c ∆Hhom
values for other 1,2-substituted diphenylethanes. Second, for
radical cation 1c•+ two bond scission modes can be considered
(eqs 8 and 9, Scheme 2). The AM1 calculations imply that in
the gas phase the thermodynamically preferred mode of
fragmentation is bond homolysis, which has ∆Hhom ) 10 kcal/
mol; further, this pathway is preferred over heterolysis by
approximately 20 kcal/mol (∆Hhet ) 29 kcal/mol).
Figure 3. Vector plots of RHF or RHF/HE-CI energy minimized struc-
tures: (a) 1c, lowest energy conformation, C1 and C2 phenyls gauche;
(b) 1c, local energy minimum, C1 and C2 phenyls anti; (c) 1c•+, lowest
energy conformation, C1 and C2 phenyls gauche; and (d) 1c•+, local
energy minimum, C1 and C2 phenyls anti.
pathway for radical cations 1a-e•+ in the gas phase is homolytic
C-C bond fragmentation (see eq 9 in Scheme 2).
Discussion
Semiempirical Calculations. Semiempirical calculations
were carried out using the AM1 parameter set within MOPAC
6.0 to provide insight concerning C-C bond fragmentation in
1c•+. Computational methods are described in the Experimental
Section. Table 3 contains a listing of RHF or RHF/HE-CI
calcuated ∆Hf values for 1c, the corresponding radical cation
1c•+, and several reactive intermediates that are produced by
either heterolytic or homolytic bond scission of the aminoalcohol
neutral or radical cation. Where ever comparisons can be made,
our calculated energies compare favorably with results previ-
ously reported by Camaioni.27 An interesting feature that
emerges from the AM1 calculations is that in the energy-
minimized structures of 1c and 1c•+ (Figure 3), the aniline
nitrogen and hydroxy oxygen are gauche and the C1 and C2
phenyl groups are gauche. For 1c and 1c•+ the energies of these
conformers lie approximately 4 kcal/mol lower than the energies
of conformations in which the nitrogen and oxygen are anti and
Mechanism and Thermodynamics of C-C Bond Frag-
mentation. As implied above, carbon-carbon bond fragmenta-
tion in 1a-e•+ must occur either via heterolytic or homolytic
scission of the C-C σ bond (eqs 8 and 9, respectively, Scheme
2). These two pathways are distinguishable based on the
reactive intermediates that are formed by the fragmentation:
R-amino radicals 4a-e are produced by bond heterolysis while
the R-hydroxy benzyl radical (8) and iminium ions 7a-e are
produced by bond homolysis.
The experimental data indicate that in CH3CN solution 1a-
e
•+ fragment via heterolysis. Transient absorption spectroscopy
reveals that a species with a strong near-UV absorption (λmax
≈ 360 nm) is produced by fragmentation; this species cannot
be ketyl radical 8, which has its most prominent absorption band
below 300 nm.41 Furthermore, observation of base catalysis
(41) Go¨rner, H.; Kuhn, H. J. J. Phys. Chem. 1986, 90, 5946.

5662 J. Am. Chem. Soc., Vol. 118, No. 24, 1996
Burton et al.
Scheme 2
for fragmentation points to involvement of proton loss in the
rate determining step. Both of these lines of evidence support
the premise that heterolytic fragmentation occurs in solution.
By contrast, the FT-ICR data clearly indicate that in the gas
phase, radical cations 1a-e•+ fragment homolytically. This is
indicated by the observation of iminium ions 7a-e as the
predominant charged species produced when 1a-e are ionized
in the gas phase.42
Consideration of the thermodynamics for C-C bond frag-
mentation in 1a-e•+ provides a rationale for the difference in
bond fragmentation modes observed in solution and the gas
phase. Figure 4 outlines three thermodynamic cycles that can
be applied to calculate the free energy for C-C bond fragmen-
tation in 1a-e•+ (∆G4), and Table 4 provides a tabular summary
of the necessary equations and the calculated thermodynamic
quantities.15 Three modes of bond fragmentation are considered
in this diagram: (a) homolytic fission (∆G4a); (b) heterolytic
fission without loss of a proton from the benzaldehyde oxygen
(∆G4b); and (c) heterolytic fission with concomitant loss of a
proton from the benzaldehyde oxygen (∆G4c). Three free
energies are required for estimation of ∆G4 (Table 4): the free
energy for C-C bond homolysis in the neutrals 1a-e (∆G1);
the free energy for electrochemical reduction of 1a-e•+ relative
to the SCE (∆G2); and the free energy for electrochemical
reduction of one of the products of bond fission relative to the
SCE (∆G3a, ∆G3b or ∆G3c).
Figure 4. Thermodynamic cycles for calculation of free energy for
C-C bond fragmentation in aminoalcohol radical cations.
Sufficient experimental data are available in the literature or
from the present study to evaluate each of the required free
energies. Table 4 provides a tabular summary of how each
thermodynamic parameter was obtained or estimated. First, an
estimate for the free energy of homolysis of 1c (∆G1) was
arrived at by taking the literature bond enthalpy of 2-amino-
ethanol (∆H ) 80 kcal/mol), subtracting the steric and electronic
effect typical of 1,2-diphenyl substitution on the C-C bond
enthalpy of ethane (∆H ≈ 22 kcal/mol), and using an entropy
value which is typical for bond homolysis of substituted ethanes
(∆S ≈ 36 cal/(mol‚K)).43 These values lead to an estimate of
∆G1 ) 47 kcal/mol, which is in accord with experimentally
determined free energies for bond homolysis of substituted 1,2-
diphenylethanes in which the C-C bond is weakened by steric
and electronic effects (30-43 kcal/mol).6c Second, ∆G2 was
estimated based on Ep(1c), which provides a value which is
intermediate for the series 1a-e. Finally, ∆G3a, ∆G3b, and ∆G3c
were derived from electrochemical data as follows: ∆G3a from
the reduction potential of iminium ion 7;44a ∆G3b from the
reduction potential of benzaldehyde (Ep ) -1.55 V),44b the pKa
Table 4. Thermodynamics of Bond Fragmentation Estimated by
Thermochemical Cyclesa
value
b
parameter (kcal‚mol-1
)
method of calculation
∆G1
∆G2
∆G3a
+47
-23
+19
estimated as described in text
∆G2 ) -FEp(1c)
∆G3a ) -FE1/2(7/4),
where E1/2(7/4) ) -0.84 V44a
∆G3b ) -FE1/2(9/8), where E1/2(9/8) )
-66 V calculated described in text
∆G3c calculated as described in text
∆G4a ) ∆G1 + ∆G2 - ∆G3a
∆G4b ) ∆G1 + ∆G2 - ∆G3b
∆G4c ) ∆G1 + ∆G2 - ∆G3c
∆G3b
+15
∆G3c
∆G4a
∆G4b
∆G4c
+25
+5
+9
-1
a Refer to Figure 4 for definition of thermodynamic quantities. F is
the Faraday constant (23.06 kcal‚mol-1‚eV-1). b Estimated errors in
∆G1-∆G3 are (10%.
of the R-hydroxybenzyl radical (structure 8, pKa ) 8.1),44b and
the pKa of the R-hydroxybenzyl cation (structure 9, pKa ) -7);45
and ∆G3c from the reduction potential of benzaldehyde and the
pKa of the R-hydroxybenzyl radical.46
Some very interesting features are apparent from consider-
ation of the ∆G4 values (Table 4) which are calculated for the
(42) DeJohgh, D. C.; Liu, D. C. K.; Leclair-Lanteigne, P.; Gravel, D.
Can. J. Chem. 1975, 53, 3175.
(43) McMillin, D. F.; Golden, D. M. Annu. ReV. Phys. Chem. 1982, 33,
493.
(44) (a) Andrieux, C. P.; Saveant, J.-M. J. Electroanal. Chem. 1970,
26, 223. (b) Andrieux, C. P.; Grzeszczuk, M.; Saveant, J.-M. J. Am. Chem.
Soc. 1991, 113, 8811.
(45) March, J. AdVanced Organic Chemistry, 4th ed.; Wiley-Inter-
science: New York, 1992; p 250.

C-C Bond Fragmentation in Aminoalcohol Radical Cations
J. Am. Chem. Soc., Vol. 118, No. 24, 1996 5663
pathways in Figures 4a-c. First, all of the ∆G4’s are <10 kcal/
mol, indicating that regardless of mechanism, C-C bond
scission in 1a-e•+ is weakly endothermic at most. Thus, single
electron oxidation of the aminoalcohols significantly activates
the C-C bond toward scission. Even more interesting is the
relationship between the ∆G4’s that are calculated for the three
fragmentation pathways. The pathways in Figures 4a and 4b
involve bond scission without concomitant loss of the proton
from the carbonyl oxygen. Under these circumstances, ho-
molysis (pathway a, ∆G4a) is favored by ca. 4 kcal/mol over
heterolysis (pathway b, ∆G4b). However, the calculations also
show that when heterolytic C-C bond scission occurs with
concomitant loss of a proton (pathway c, ∆G4c), this pathway
is thermodynamically preferred over homolysis (pathway a) by
more than 5 kcal/mol.
Figure 5. Linear free energy correlation for bond fragmentation of
radical cations 1a-e•+
.
Specifically, the observed difference in the thermodynamics of
the gas-phase and solution-phase reactions can be explained if
the solvation energy of the R-hydroxybenzyl cation (9) is greater
than that of the iminium ion (7).
With this thermodynamic information at hand, it is now
possible to rationalize why C-C bond scission in 1a-e•+ occurs
via homolysis in the gas phase and via heterolysis in solution.
In the gas phase, there are no readily available proton acceptors.
Under this circumstance iminium ions 7a-e are the best
electrofugal groups and bond homolysis predominates. By
contrast, in solution the solvent and/or added base facilitates
heterolytic fragmentation by accepting the proton from the
incipient carbonyl oxygen. Under this circumstance the benz-
aldehyde fragment is the best electrofugal group and bond
heterolysis predominates. This analysis provides a good indica-
tion that even in the absence of a basic catalyst (such as
pyridine), in solution C-C bond scission in 1a-e•+ occurs with
concomitant loss of the proton from the incipient carbonyl
oxygen.
Kinetics of Radical Cation Bond Fragmentation. Cor-
relations and Mechanistic Implications. As noted above, the
rates of C-C bond fragmentation in 1a-e•+ increase with Ep(1).
The strong correlation between kBF and Ep(1) is highlighted by
a plot of the free energy of activation for bond fragmentation
(∆GBFq) vs FEp(1), where F is the Faraday constant (Figure 5).
The solid line in the correlation was generated by linear least-
squares regression (r ) 0.9722, slope ) -0.21). Since both
parameters used in the free energy correlation are expressed in
units of kilocalories per mole, the slope of the line provides a
direct measure of the effect of radical cation stability (as
reflected by Ep(1)) on ∆GBFq. Thus, the relatively small slope
of the correlation indicates that the effects of the para substituent
on the energies of the reactant (radical cation) and transition
state are similar. This is consistent with the reaction having an
early transition state, as expected if fragmentation is weakly
endothermic or exothermic.47
Previous work indicates that remote substituents have little
effect on radical stability.48,49 On this basis, we suggest that
the para substituents have little effect on the energy of the
R-amino radicals (4a-e) produced by heterolytic bond frag-
mentation of 1a-e•+. Thus, the substituent effect on the rate
of bond scission is caused by the effect of the substituent on
the energy of the radical cation reactant and the transition state.
Furthermore, if the substituents have little influence on the
energy of 4a-e, then the correlation of ∆GBFq with FEp(1) can
be viewed as a free energy correlation, where the x-coordinate
is directly proportional to the free energy of bond fragmentation,
∆G°BF (e.g., FEp(1) ) ∆G°BF + constant).
We now briefly compare the results of the thermodynamic
cycle calculations, which are based upon experimental data, with
those of the AM1 calculations. The RHF calculated bond
homolysis enthalpy for neutral aminoalcohol 1c (∆Hhom ) 41
kcal/mol) is lower than expected based on the estimate of ∆G1
(47 kcal/mol), which was arrived at by the thermodynamic cycle
calculation. Camaioni has noted that AM1 consistently under-
estimates ∆Hf values for carbon-based radicals; furthermore,
his results indicate that the degree of underestimation is larger
when the carbon radical bears an R heteroatom substituent.27
Since the AM1-calculated ∆Hhom for 1c relies on ∆Hf values
for benzyl radicals which feature an R heteroatom (e.g., 4c and
8), it is likely that our AM1-calculated ∆Hhom is lower than the
true value because of underestimation of the ∆Hf values for
these two radicals. Thus, we believe that the free energy for
bond homolysis which is based on the thermodynamic cycle
calculation (e.g., ∆G1) is probably a better estimate for the bond
energy of aminoalcohol 1c.
Maslak has demonstrated that the free energy dependence
for the rate of bond scission in a series of benzylic radical ions
correlates with Marcus theory.6d If the data for the aminoal-
cohols is considered within the framework of a similar analysis,
Another interesting feature that emerges from the calculations
is that AM1 correctly predicts that bond homolysis is the
preferred pathway for the gas-phase fragmentation of 1c•+.
However, a disparity between the AM1 calculations and those
based on thermodynamic cycles is that AM1 predicts that the
pathway in Figure 4a is favored over that in Figure 4b by over
20 kcal/mol, while the thermodynamic cycle suggests only a 4
kcal/mol difference. It is important to note that the AM1
calculations refer to gas-phase processes while the thermody-
namic cycle calculations are based primarily on free energies
for the solution-phase counterparts. Thus, it is possible that
the large difference in the enthalpies for the two homolysis
pathways (e.g., Figures 4a and 4b) in the gas phase is offset
partly in the solution-phase counterparts by solvation effects.
q
then the observed slope of the correlation of ∆GBF vs ∆G°BF
(Figure 5, slope ) -0.21) implies that bond fragmentation of
1a-e•+ is weakly exothermic. This conclusion follows from
Marcus theory, which indicates that the slope of a plot of ∆Gq
vs ∆G° for a reaction with ∆G° < 0 is less than 0.5.50
The bond fragmentation kinetics of 1c-e•+ are characterized
by comparatively low activation enthalpies and moderately large
negative activation entropies. Similar activation parameters
were observed by Maslak and co-workers in their recent studies
of C-C bond fragmentation in sterically congested 1,2-
(47) Hammond, G. S. J. Am. Chem. Soc. 1955, 77, 334.
(48) Dust, J. M.; Arnold, D. R. J. Am. Chem. Soc. 1983, 105, 1221.
(49) Bordwell, F. G.; Bausch, M. J. J. Am. Chem. Soc. 1986, 108, 2473.
(50) For examples of the application of this concept see: (a) Fukuzumi,
S.; Wong, C. L.; Kochi, J. K. J. Am. Chem. Soc. 1980, 102, 2928. (b) Bock,
C. R.; Connor, J. A.; Guitierrez, A. R.; Meyer, T. J.; Whitten, D. G.;
Sullivan, B. P.; Nagle, J. K. J. Am. Chem. Soc. 1979, 101, 4815.
(46) In the redox process which defines ∆G3c, PhCHO + H+ + e-
f
PhCHOH•, “H+” refers to H+ in the standard state (e.g., aqueous H+, pH
0).

5664 J. Am. Chem. Soc., Vol. 118, No. 24, 1996
Burton et al.
As noted above, if deprotonation and fragmentation are
concerted, then kbase corresponds to the rate of deprotonation
of the radical cation. Therefore, if the reaction is concerted,
variation of kbase with Ep(1) reveals the effect of radical cation
stability on the thermodynamic acidity (as reflected by the rate
of deprotonation). Since deprotonation of the aminoalcohol
radical cations presumably occurs at the hydroxy group, we
expect the remote para substituent to have only a modest effect
on the thermodynamic and kinetic acidity. This is consistent
with the experimental observations for radical cations 1c-e•+
(e.g., for these radical cations the substituent has a comparatively
weak effect on kbase). On this basis we suggest that base-
catalyzed fragmentation in 1c-e•+ occurs via an E2-like
concerted mechanism. By contrast, kbase decreases sharply with
Ep(1) for the radical cations derived from 1a and 1b. While
the underlying basis for this effect is unclear, it is possible that
for these radical cations base catalysis occurs via an E1cB
mechanism, or by an entirely different mechanism involving
deprotonation at a different position.
Figure 6. Plot of log(kbase) for aminoalcohol radical cations as a
function of FEp(1) (F is faraday constant, 23.06 kcal‚V-1). The line is
simply a smooth curve drawn through the data points.
diphenylethane radical cations.6c Importantly, the consistency
of our findings with those of Maslak clearly indicates that an
unfavorable activation entropy is typical for bond scission of
radical cations in moderately polar solvents. The large negative
∆Sq values that are observed for the radical cation fragmenta-
tions likely have their origin in the interactions of solvent
molecules with the radical cation reactant and the transition state
for fragmentation. Further, it is possible that part of the
unfavorable activation entropy component may be related to
stereoelectronic demands of the transition state.8d,27
Summary and Conclusions
The kinetics and mechanism of bond fragmentation in radical
cations derived from aminoalcohols 1a-e have been examined.
The rates of bond fragmentation in the radical cations in CH3CN
solution were determined by laser flash photolysis. A correla-
tion of the observed rates with the peak potentials for anodic
oxidation of the neutral aminoalcohols indicates that the rate
of C-C bond fragmentation increases, albeit weakly, with
decreasing stability of the radical cations. Prominent features
in the transient absorption spectra of the reactive intermediates
produced by the reaction indicate that (in solution) heterolytic
bond fragmentation occurs. By contrast, FT-ICR studies of the
products arising from fragmentation of 1a-e•+ indicate that, in
the gas phase, homolytic C-C bond fragmentation predomi-
nates. AM1-calculated enthalpies for bond homolysis and
heterolysis of 1c•+ reveal that C-C bond homolysis is ther-
moydnamically preferred over heterolysis by 20 kcal/mol,
consistent with the gas-phase experimental results. By contrast,
free energies for C-C bond homolysis and heterolysis estimated
from (solution phase) thermodynamic data indicate that in
solution, heterolytic bond fragmentation with concurrent depro-
tonation is the thermodynamically preferred reaction pathway,
again consistent with the experimental data.
Mechanism of Base-Catalyzed Fragmentation. The ob-
servation of base catalysis for bond fragmentation in 1a-e•+
indicates that proton loss is involved in the rate determining
step for fragmentation, at least in the presence of an added base.
Two mechanisms can be considered for the base-catalyzed
fragmentation reaction. The first, which corresponds to an E2
elimination, involves concerted deprotonation at the hydroxyl
group and C-C bond fragmentation. In this case, proton
transfer is rate determining and the observed second-order rate
(kbase) corresponds to the deprotonation rate. The second
mechanism corresponds to an E1cB elimination and involves
initial deprotonation at the hydroxyl group to form the conjugate
base of the aminoalcohol radical cation. Deprotonation is then
followed by C-C bond fragmentation in the conjugate base of
the radical cation. In the E1cB mechanism, the rate for
disappearance of the radical cation is first order in base, but
the relationship of the observed rate to the microscopic rate
constants depends upon the relative rates of the individual steps.
Fragmentation of 1b-e•+ is catalyzed by pyridine base and
the observed second-order rate constants for fragmentation (kbase
)
increase with Ep(1). Figure 6 illustrates a plot of log(kbase) as
a function of FEp(1). This plot indicates that kbase varies only
weakly with Ep(1) for 1c-d•+, but becomes significantly slower
for 1b•+ and immeasurably slow for 1a•+. If the mechanism
for the base-catalyzed fragmentation reaction is the same among
Acknowledgment. We gratefully acknowledge the National
Science Foundation (Grant No. CHE-9401620) for support of
this work.
the entire series of radical cations, the correlation of log(kbase
)
JA960378Q
vs Ep(1) is expected to be linear;51 thus, the nonlinearity of the
correlation suggests that the mechanism for base catalysis may
change as a function of Ep(1).
(51) (a) Bunting, J. W.; Stefanidis, D. J. Am. Chem. Soc. 1988, 110,
4008. (b) Bunting, J. W.; Kanter, J. P. J. Am. Chem. Soc. 1991, 113, 6950.