Concerted proton-electron transfer in pyridylphenols: The importance of the hydrogen bond

Article abstract of DOI:10.1002/anie.200702486

(Chemical Equation Presented) Slowed down by a break up: One-electron oxidation of intramolecularly hydrogen-bonded phenol-pyridine compounds 1 and 2 involves proton transfer coupled with electron transfer (see scheme; B = base). At the same driving force, 2 reacts substantially slower than 1. The methylene group between the phenol and pyridine groups in 2 breaks the conjugation between the rings, which has a marked effect on the hydrogen bond and is likely the origin of the difference in rate constants.

Full text of DOI:10.1002/anie.200702486

Communications
DOI: 10.1002/anie.200702486
Concerted Proton–Electron Transfer
Concerted Proton–Electron Transfer in Pyridylphenols: The
Importance of the Hydrogen Bond**
Todd F. Markle and James M. Mayer*
Proton-coupled electron transfer (PCET) reactions are of
much current interest because of their role in many chemical
and biological processes.[1] The oxidation of tyrosine residues
to tyrosyl radicals,for example,is important in the function of
a number of proteins.[2] The Kok S-state cycle of photo-
system II involves oxidations of tyrosine-Z through long-
range electron transfer to P680+ whereby the phenolic proton is
likely transferred to a hydrogen-bonded imidazole (H190) in a
single kinetic step,[3] called separated CPET (concerted
proton–electron transfer).[4] This conclusion has been gener-
ally supported by studies of phenol model systems in which
electron transfer (ET) is coupled to proton transfer (PT) to a
different species.[4–9] Concerted transfer of eÀ and H+ is often
preferred because DGꢀCPET is more favorable than DGEꢀ T for
Scheme 1. Pyridylphenols and their one-electron oxidation.
initial electron transfer.[4–9] When the driving forces are equal,
A+ =oxidant.
however,separated-CPET reactions typically occur with
lower rate constants than related ET reactions. The origin of
the slower rates has been discussed in theoretical treat-
ments[1,10] and the importance of hydrogen bonding has been
noted by Hammarström and co-workers.[6b] Herein,we report
that the nature of the hydrogen bond has a large influence on
the facility of CPET.
We recently described chemical and electrochemical
oxidations of phenol base compounds HOAr-B (B = base)
that proceed by intermolecular ET concerted with intra-
molecular PT to the hydrogen-bonded base (Scheme 1).[9]
CPET reactions of pyridylphenol 1 are around 102 times
imidazole analogue of 1 has a much longer d(O···N) than 2 (by
0.06 Š) yet is oxidized substantially faster.[9] The key feature
seems to be that the faster phenol–pyridine and phenol–
imidazole compounds contain resonance-assisted hydrogen
bonds (RAHBs) owing to conjugation between the proton
donor and acceptor. This conjugation leads to stronger
hydrogen bonds and influences the NMR and IR spectra
(see below).[9,12]
To test the apparent correlation between RAHBs and
increased CPET rates,the nonconjugated phenol–pyridine
4,6-di-tert-butyl-2-(2’-pyridinylmethyl)phenol (3) was pre-
pared. 2-Pyridylmagnesium chloride (2-pyMgCl) was added
to benzyl ether protected 3,5-di-tert-butylsalicylaldehyde,
followed by acylation of the resulting alcohol and catalytic
hydrogenation. Compound 3 has been fully characterized,
including by X-ray crystallography.[13] The presence of the
inserted CH2 group causes the pyridine ring to twist out of the
plane of the phenol ring and makes the hydrogen bond longer
and more linear than that in 1 [dO···N = 2.6914(13) vs.
2.567(3) Š, a(OHN) = 167.7(15)8 vs. 154(2)8].[9] The chem-
ical shift of the phenolic proton in nonconjugated 3 is 3.7 ppm
faster than those of amine analogue 2 with the same type of
[9]
oxidant and for the same value of DGCꢀ PET
.
This result was
unexpected because free pyridine is a weaker base than a
[11]
primary amine (pKBH+ = 12 and 18,respectively,in MeCN).
The difference between 1 and 2 is unlikely to be due to
different outer-sphere (solvent) reorganization because these
molecules are of comparable size and should have similar
changes in solvation upon CPET. The average crystallo-
graphic proton donor–acceptor distance d(O···N) is slightly
shorter in 1 than in 2 (by 0.01 Š).[9] However,this distance
does not appear to be the dominant factor as a phenol–
upfield of that in 1 (dOH = 11.15 vs. 14.83 ppm in CD3CN). IR
À1
spectra show that the nOH peaks of 3 (3094,2772 cm
in
=
MeCN) are blue-shifted relative to that of 1 (2650 cmÀ1, nOD
[*] T. F. Markle, Prof. Dr. J. M. Mayer
Department of Chemistry
2000 cmÀ1,Figure 1). The low-field 1H NMR resonance and
low-frequency nOH peak for 1 are characteristic of shorter,
stronger RAHB systems.[12]
University of Washington
Seattle, WA 98195-1700 (USA)
Fax: (+1)206-685-8665
E-mail: mayer@chem.washington.edu
Cyclic voltammograms of 3 are chemically reversible and
0/+
1
give a value of E = = 0.44 V vs. [Cp2Fe] in MeCN (Cp =
2
[**] This work was supported by the US National Institutes of Health
(R01 GM50422). We thank Dr. A. DiPasquale for X-ray crystallog-
raphy.
cyclopentadienyl). The peak separation DEp is large relative
to that of 1 (220 mV vs. 100 mV at n = 0.2 VsÀ1),which
suggests slower electrochemical kinetics.[9,13,14] Slow ET is also
evident in the kinetics of 3 with a number of one-electron
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
738
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 738 –740

Angewandte
Chemie
Figure 1. IR spectra of solutions of a) 3 and b) 1 in MeCN. The nOH
peaks are the very broad absorptions between 2000 and 3500 cmÀ1
The peak at around 2400 cmÀ1 in (a) is due to incomplete solvent
subtraction.
.
oxidants in anaerobic MeCN (Table 1).[13] A primary kinetic
isotope effect (kH/kD = 5.9 Æ 0.8 at 248C) is observed for the
oxidation of [H/D]-3 with [Fe(Me2bpy)3]3+,which is indicative
of a CPET mechanism. This kH/kD ratio is large relative to
those measured for other HOAr-B oxidations (1.6–2.8).[9]
Arrhenius analysis of the temperature dependence of kH
and kD (Figure 2a,Table 2) gives values of log( AH/AD) =
Figure 2. a) Plot of k versus T for the reactions of [Fe(Me2bpy)3]3+ with
[9]
&
*
*
[H/D]-3 ( =kH; =kD) and 1 ( ). Fits are to the Arrhenius
equation: k=Aexp[ÀEa/RT]. b) Plot of logk versus Erxn for reaction of
[9]
[Fe(N–N)3]3+ with 3 ( , solid line), 1 ( , long dashes), and 2 (
,
!
&
*
À1.0 Æ 0.8 and Ea ÀEaH = 2.3 Æ 1.2 kcalmolÀ1 (estimated 2s
D
short dashes).[9] The lines are fits to k=ZexpÀ[(lÀErxn)2/4lkB T] with
values of log(Z/mÀ1 sÀ1)=9.9 (for 1), 10.4 (for 2), and 9.5 (for 3).[9,13,15]
errors).
Table 1: Kinetic data for oxidation of 3 at 296Æ2 K.[a]
Table 2: Arrhenius parameters for the reactions of 1 and [H/D]-3 with
k [mÀ1 sÀ1
]
[Fe(Me2bpy)3]3+
.
Oxidant
Erxn [V]
[Fe(3,4,7,8-Me4phen)3]3+
[Fe(4,7-Me2phen)3]3+
[Fe(5,5’-Me2bpy)3]3+
[Fe(bpy)3]3+
0.02
0.09
0.14
(2.3Æ0.2)”104
(2.5Æ0.3)”105
(1.7Æ0.2)”105
(1.8Æ0.2)”106
(8Æ2)”102
Phenol DG8
krt [mÀ1 sÀ1 [b]
]
log(A [mÀ1 sÀ1]) Ea
[kcalmolÀ1
]
[kcalmolÀ1 [a]
]
1
0Æ0.7
À3.2Æ0.7
À3.2Æ0.7
(6.6Æ1.0)”105
(1.7Æ0.2)”105
9.6Æ0.7
9.2Æ0.5
5.3Æ1.0
5.4Æ0.8
7.7Æ0.9
0.26
[H]-3
[D]-3
[N(p-C6H4OMe)3]+C
[N(tol)3]+C
À0.28
À0.06
0.04
(2.9Æ0.2)”104 10.2Æ0.6
(1.2Æ0.1)”105
(5.3Æ0.5)”105
[N(p-C6H4OMe)(p-C6H4Br)2]+C
[a] Calculated from cyclic voltammetry data (DG8=ÀFErxn). [b] Data
collected at 298 K for 1 and 297 K for [H/D]-3.
[a] bpy=2,2’-bipyridine; phen=1,10-phenanthroline; tol=tolyl.
tion. However,some insight can be gained from the data
above and some initial calculations. For the oxidations of 1
and 3 with [Fe(Me2bpy)3]3+,both the Arrhenius A and Ea
values (Table 2) are the same within error,even though the
reaction of 3 is energetically more downhill by 3.2 kcalmolÀ1.
The rate constants for CPET are plotted versus the
reaction driving force (Erxn = EoxidantÀEHOAr-B) in Figure 2b
and S13.[13] The lines to guide the eye assume the typical
parabolic dependence of logk on Erxn.
[15] At the same driving
force,values of k for 3 are similar to those for 2,and 25–150
times slower than those found for 1. Even though the proton
acceptor in 3 is a pyridine ring,the reactivity of this
nonconjugated phenol is much closer to that of the non-
conjugated phenolamine 2 than that of the conjugated
phenol–pyridine 1.
What is the origin of the substantial differences in rate
constants for 1 and 3? The complete theoretical analysis
needed to address this question—a non-adiabatic treatment
with a fully quantum-mechanical transferring proton and
rates averaged over the range of proton donor–acceptor
(OH···N) distances—is beyond the scope of this communica-
The AH/AD and Ea ÀEa values for 3 + [Fe(Me2bpy)3]3+
appear to be outside the semiclassical limits,[16] implying
that tunneling plays a significant role and indicating that
crossing from reactants to products could occur in different
regions of the potential-energy surface for H versus D.[17]
To probe the differences in the potential-energy surfaces
for oxidations of 1 and 3,DFT calculations were performed
using the “four-point” method developed by Nelsen et al. for
ET reactions.[18] For a self-exchange reaction such as HOAr-
B + COAr-BH+,this method gives the energy to take the
reactants to the geometry of the products without allowing
the electron to transfer—that is,the vertical reorganization
D
H
Angew. Chem. Int. Ed. 2008, 47, 738 –740
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
739

Communications
energy l. With the assumption of parabolic surfaces,the
shifts and IR stretching frequencies,as well as on the rates of
CPET oxidation. These results may have implications in the
understanding of biological and other CPET reactions;
further experimental and computational studies are in
progress.
crossing point between the reactant and product surfaces in
1
this symmetrical case is = l. This method implicitly defines
4
the problem as an electron transfer with proton transfer
contributing to the relaxation energies,and does not consider
proton tunneling and the quantum nature of the proton.
Therefore,any conclusions from this approach must be
considered tentative pending a more complete analysis.
Nelsenꢀs method estimates l from calculations of four
points of a cycle: 1) the neutral HOAr-B at its optimized
geometry (Eng0),2) vertical ionization to give the radical
cation at the optimized geometry of the neutral compound
(Eng+),3) relaxation of the radical cation to its optimized,
Received: June 8,2007
Revised: October 10,2007
Published online: December 13,2007
Keywords: electron transfer · hydrogen bonds · oxidation ·
.
proton transfer
proton-transferred,geometry
(
Ecg+),and finally,4) the
[1] a) R. I. Cukier,D. G. Nocera, Annu. Rev. Phys. Chem. 1998, 49,
337 – 369; b) J. M. Mayer, Annu. Rev. Phys. Chem. 2004, 55,363 –
390; c) S. Hammes-Schiffer, Acc. Chem. Res. 2001, 34,273 – 281.
[2] a) J. Stubbe,W. A. van der Donk, Chem. Rev. 1998, 98,705 – 762;
b) R. P. Pesavento,W. A. van der Donk, Adv. Protein Chem.
2001, 58,317.
closed-shell neutral compound at the optimized cation
geometry (Ecg0). l is the sum of the relaxation energies for
+
the neutral compound and the cation [(Eng ÀEcg+) +
0
(Ecg ÀEng0)].[13,18] The calculations indicate that the value of
l for 1 is significantly smaller (by 8–11 kcalmolÀ1) than those
[3] a) G. Renger, Biochim. Biophys. Acta Bioenerg. 2004, 1655,195 –
204; b) H. Kuhne,G. W. Brudvig, J. Phys. Chem. B 2002, 106,
8189 – 8196; c) V. Petrouleas,D. Koulougliotis,N. Ionnidis,
Biochemistry 2005, 44,6723 – 6728.
0
for 2 and 3 (Table 3). This difference arises mostly from Ecg
,
Table 3: Calculations of l.[a]
[4] C. Costentin,M. Robert,J.-M. SavØant, J. Am. Chem. Soc. 2006,
128,8726 – 8727.
B
Eng
Ecg
Ecg
l[b]
0
+
Eng
+
0
[5] a) L. Biczók,N. Gupta,H. Linschitz, J. Am. Chem. Soc. 1997,
119,12601; b) N. Gupta,H. Linschitz, J. Am. Chem. Soc. 1997,
119,6384.
py
NH2
CH2py
(1)
(2)
(3)
0
0
0
134
135
136
120
118
117
7
12
13
21
29
32
[6] a) M. Sjödin,S. Styring,H. Wolpher,Y. Xu,L. Sun,L.
Hammarström, J. Am. Chem. Soc. 2005, 127,3855 – 3863; b) M.
Sjödin,T. Irebo,J. E. Utlas,J. Lind,G. MerØnyi,B. Škermark,L.
Hammarström, J. Am. Chem. Soc. 2006, 128,13076 – 13083.
[7] S. Y. Reece,D. F. Nocera, J. Am. Chem. Soc. 2005, 127,9448 –
9458.
[8] I. J. Rhile,J. M. Mayer, J. Am. Chem. Soc. 2004, 126,12718 –
12719.
[9] I. J. Rhile,T. F. Markle,H. Nagao,A. G. DiPasquale,O. P. Lam,
M. A. Lockwood,K. Rotter,J. M. Mayer, J. Am. Chem. Soc.
2006, 128,6075 – 6088.
[a] B3LYP/6-311+G**//B3LYP/6-31G* energies (in kcalmolÀ1 relative to
Eng for each system) with solvent (MeCN) modeled using PCM.[19]
0
+
0
0 [18,19]
[b] l=Eng ÀEcg++Ecg ÀEng
.
the energy of the neutral at the cation geometry,which is a
zwitterionic structure in which the proton is nitrogen-
bound.[13] This structure is stabilized by resonance in 1
[Eq. (1)] but not in 2 or 3 because of the saturated a carbon
atom. Interestingly,the root-mean square displacements of
[10] J. H. Skone,A. V. Soudackov,S. Hammes-Schiffer, J. Am. Chem.
Soc. 2006, 128,16655 – 16663,and references therein.
[11] K. Izutsu, Acid-Base Dissociation Constants in Dipolar Aprotic
Solvents,Blackwell Scientific,Boston,MA, 1990,pp. 17 – 35.
[12] a) P. Gilli,V. Bertolasi,V. Ferretti,G. Gilli, J. Am. Chem. Soc.
2000, 122,10405 – 10417; b) Z. Rozwadowski,E. Majeewski,T.
Dziembowska,P. Hansen, J. Chem. Soc. Perkin Trans. 2 1999,
2809 – 2817.
[13] See the Supporting Information for details.
[14] C. Costentin,M. Robert,J.-M. Saveant, J. Am. Chem. Soc. 2007,
129,9953 – 9963.
[15] a) See references [1,6,10,14,18]; b) for reactions with [Fe(N–
N)3]3+ (N–N = bidentate N-donor ligand), Erxn was corrected for
electrostatic work as in reference [9].
the non-hydrogen atoms between neutral and cation ground
states is almost identical in 1 and 3 (within 0.001 Š),so this is
not the origin of the difference in l.[13] Thus,the computations
are consistent with the suggestion that the faster reactions of 1
are due to the conjugation between the proton donor and
acceptor. Conjugated 1 has the shorter,stronger hydrogen
bond,as indicated by the structural and spectroscopic data,
which makes the proton-transfer potential flatter and results
in more facile CPET.
In conclusion,CPET reactions of the nonconjugated
phenol–pyridine 3 are much slower than those of conjugated
analogue 1 and are comparable to those of the nonconjugated
aminophenol 2. The nature of the base (NH2 vs. py) is not
critical. We suggest that the key feature in this case is the
presence or absence of conjugation between the phenol and
the basic nitrogen atom. This conjugation has a marked effect
on the hydrogen bond,as indicated by 1H NMR chemical
[16] R. P. Bell, The Tunnel Effect in Chemistry,Chapman and Hall,
New York, 1980,pp. 77 – 105.
[17] a) R. A. Marcus, Philos. Trans. R. Soc. London Ser. B 2006, 361,
1445 – 1455; b) R. A. Marcus, J. Phys. Chem. B 2007, 111,6643 –
6654; c) for a somewhat different interpretation: J. P. Klinman,
Biochim. Biophys. Acta Bioenerg. 2006, 1757,981 – 987.
[18] a) S. F. Nelsen,S. C. Blackstock,Y. Kim, J. Am. Chem. Soc. 1987,
109,677; b) S. F. Nelsen,M. N. Weaver,J. R. Pladziewicz,L. K.
Ausman,T. L. Jentzsch,J. J. OꢀKonek, J. Phys. Chem. A 2006,
110,11665 – 11676.
[13]
[19] M. J. Frisch,et al. Gaussian03,Revision D.02,2004.
740
www.angewandte.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 738 –740