Symmetry of O-H-O and N-H-N hydrogen bonds in 6-hydroxy-2-formylfulvene and 6-aminofulvene-2-aldimines

Article abstract of DOI:10.1006/bioo.2001.1222

The symmetry of the hydrogen bonds in 6-hydroxy-2-formylfulvene and two N,N′-diaryl-6-aminofulvene-2-aldimines is probed by the NMR technique of isotopic perturbation. Observed deuterium-induced ¹³C NMR isotope shifts at several positions can be attributed to a combination of an intrinsic shift and the perturbation of a tautomeric equilibrium. The most dramatic are at the aldehydic or aldiminic carbon signals, where the observed isotope shift for the unlabeled carbon is +376 or +223 ppb. This large downfield shift is contrary to the small upfield shift expected for a four-bond intrinsic shift and can be attributed only to a perturbation shift. Therefore these intramolecular hydrogen bonds are asymmetric, the proton resides in a double-minimum potential surface, and each molecule exists as a pair of rapidly interconverting tautomers, regardless of solvent. The symmetry of the hydrogen bond is not governed only by the O-O or N-N distance. It is proposed that symmetric hydrogen bonds can be observed in crystalline phases but not as yet in solution because the disorder of the solvation environment induces an asymmetry of the hydrogen bond, whereas a crystal can guarantee a symmetric environment. These results provide no insight into the source of the stabilization attributed to low-barrier hydrogen bonds if they lack the special feature of symmetry.

Full text of DOI:10.1006/bioo.2001.1222

Bioorganic Chemistry 30, 3–15 (2002)
doi:10.1006/bioo.2001.1222, available online at http://www.idealibrary.com on
Symmetry of O-H-O and N-H-N Hydrogen Bonds in
6-Hydroxy-2-formylfulvene and 6-Aminofulvene-2-aldimines1
Charles L. Perrin2 and Brian K. Ohta
Department of Chemistry and Biochemistry, University of California at San Diego,
La Jolla, California 92093-0358
Received June 5, 2001
The symmetry of the hydrogen bonds in 6-hydroxy-2-formylfulvene and two N,NЈ-diaryl-6-
aminofulvene-2-aldimines is probed by the NMR technique of isotopic perturbation. Observed
deuterium-induced 13C NMR isotope shifts at several positions can be attributed to a combination
of an intrinsic shift and the perturbation of a tautomeric equilibrium. The most dramatic are at
the aldehydic or aldiminic carbon signals, where the observed isotope shift for the unlabeled
carbon is ϩ376 or ϩ223 ppb. This large downfield shift is contrary to the small upfield shift
expected for a four-bond intrinsic shift and can be attributed only to a perturbation shift. Therefore
these intramolecular hydrogen bonds are asymmetric, the proton resides in a double-minimum
potential surface, and each molecule exists as a pair of rapidly interconverting tautomers,
regardless of solvent. The symmetry of the hydrogen bond is not governed only by the O-O or
N-N distance. It is proposed that symmetric hydrogen bonds can be observed in crystalline
phases but not as yet in solution because the disorder of the solvation environment induces an
asymmetry of the hydrogen bond, whereas a crystal can guarantee a symmetric environment.
These results provide no insight into the source of the stabilization attributed to low-barrier
hydrogen bonds if they lack the special feature of symmetry.
᭧ 2002 Elsevier Science (USA)
Key Words: symmetry; hydrogen bonds; NMR; isotopic perturbation; LBHBs.
INTRODUCTION
Hydrogen bonds are a key feature of molecular structure and the focus of an
enormous number of theoretical and experimental studies (1). Usually the motion of the
hydrogen-bonded hydrogen is described by a double-well potential, but in symmetric
hydrogen bonds it is a single well. Such hydrogen bonds require donor atoms of
˚
equal basicity and an unusually short O-O distance of Յ2.5 A (2). These hydrogen
bonds are unusually strong, perhaps because resonance energy is maximized when
two resonance forms are of equal energy, or because of covalent character (3). These
have been called short, strong hydrogen bonds or low-barrier hydrogen bonds (LBHBs)
or centered or symmetric hydrogen bonds, depending on the observational criterion.
1 This paper is dedicated to F. H. Westheimer, on the occasion of his 90th birthday.
2 To whom correspondence and reprint requests should be addressed. Fax: (858) 822-0386. E-mail:
cperrin@ucsd.edu.
3
0045-2068/02 $35.00
᭧ 2002 Elsevier Science (USA)
All rights reserved.

4
PERRIN AND OHTA
Recently LBHBs have been proposed to play a role in enzymatic catalysis (4). An
extra stabilization is claimed to reduce the activation barrier for the formation of
high-energy intermediates or transition states. Yet hydrogen-bond strengths are often
considerably lower than the 10–20 kcal/mol suggested (5). Consequently this proposal
has generated considerable controversy (6), and there are some enzymic reactions
where LBHBs are not required (7). Authoritative recent reviews are available (8).
The potential significance of symmetric hydrogen bonds has prompted us to search
for examples in solution.
The question is whether the proton in the hydrogen bond resides in a single- or a
double-minimum potential surface. Alternatively, the crucial distinction is between a
single symmetric structure and a pair of rapidly interconverting asymmetric tautomers.
Our previous studies addressed the intramolecular O-H-O hydrogen bonds in
3-hydroxy-2-phenylpropenal (9) and in dicarboxylate monoanions (10), as well as
the N-H-N hydrogen bonds in protonated 1,8-bis(dimethylamino)naphthalenes (11).
We now extend those studies to the O-H-O and N-H-N hydrogen bonds of neutrals
6-hydroxy-2-formylfulvene (1), N,NЈ-diphenyl-6-aminofulvene-2-aldimine (2), and
N,NЈ-bis(3,5-dimethylphenyl)-6-aminofulvene-2-aldimine (3).
These molecules are good candidates for symmetric hydrogen bonding. The short
˚
O-O and N-N distances, 2.51 and 2.79 A, respectively (12), favor transformation of
a double-well potential into a single-well one. The donor and acceptor basicities are
necessarily matched. An unusually strong bond is indicated by 15N-15N and 15N иии
H scalar couplings across the N-H-N hydrogen bond in a desymmetrized analog
(13). Moreover, aromaticity in the cyclopentadienyl ring may provide additional
stabilization to the symmetric structure.
There is considerable evidence against symmetric hydrogen bonds in these mole-
cules, from X-ray and neutron-diffraction data (12), quadrupole coupling constants
(14), UV-visible absorption (15), X-ray photoelectron spectroscopy (16), and micro-
wave studies (17). A 6-31G**/MP2 calculation indicates that the symmetric structure
of 1 is 1.8 kcal/mol less stable than the asymmetric one, but the stabilities reverse,
by 0.5 kcal/mol, when zero-point energy is included (18). Although N-H-N hydrogen
bonds are generally longer and thus weaker than O-H-O, the O-O distance in a fully
planar 1,2-dicarboxylate monoanion (a seven-membered ring, including H) would be
shorter than optimum (19), so that the N-N distance in 2 or 3 may be more favorable
for a strong, symmetric hydrogen bond.
This study also addresses the influence of solvation on the symmetry of the hydrogen
bond. Because solvation stabilizes localized charges more than delocalized ones (20),
the hydrogen bonds in the ions studied previously may suffer from a bias toward
asymmetry that is not operative with neutrals. Although previous work showed the

SYMMETRY OF O-H-O AND N-H-N HYDROGEN BONDS
5
O-H-O hydrogen bond of uncharged 3-hydroxy-2-phenylpropenal to be asymmetric
(9), this may simply be a consequence of too long an O-O distance. Besides, oxygens
possess additional lone pairs that can hydrogen bond to solvent. We now study
N-H-N systems that are neutral and that lack lone pairs or NH that can hydrogen
bond to solvent, as well as a comparison O-H-O system.
METHODOLOGY OF ISOTOPE SHIFTS
Isotopic perturbation of equilibrium is a powerful and widely applicable NMR
technique for distinguishing a single symmetric structure from a pair of asymmetric
tautomers (21). It succeeds even when interconversion is so rapid that separate signals
n
from individual tautomers are not seen. It relies on measuring the isotope shift ⌬obs
,
the difference (Eq. (1)) between the 13C chemical shifts in molecules with and without
n
deuterium (22). This includes an intrinsic contribution ⌬0, which is usually Ͻ0
(upfield) and falls off rapidly with n, the number of bonds between the 13C and the D.
n
⌬
ϭ ␦C(D) Ϫ ␦C(H)
(1)
obs
The isotopolog of 1, 2, or 3 with exactly one deuterium at C6,7 can be synthesized
by known procedures. If the hydrogen bond is asymmetric, there is an additional
contribution to ⌬obs, arising from perturbation of the equilibrium between the two
tautomers A and B (X ϭ O, NAr). Because an enol or enamine has a higher C-H
stretching frequency than an aldehyde or imine, tautomer A has a higher zero-point
energy. Therefore the equilibrium favors the other tautomer B, and the time-averaged
chemical shifts are displaced from those in d0, where the two tautomers are of identical
stability. This is seen as a perturbation (of equilibrium) isotope shift given by Eq.
(2), where D ϭ ␦XH Ϫ ␦ϭX, the chemical-shift difference between exchange-related
carbons proximal and distal to the OH or NH in a static tautomer.
K Ϫ 1
2(K ϩ 1)
n⌬e ϭ
D
(2)
The perturbation shifts can be estimated. By analogy to 3-hydroxy-2-phenylpro-
4
penal, ⌬e at the aldehydic carbons of 1 is expected to be ca. ϩ0.76 ppm (9). For 2
or 3 at 25ЊC the equilibrium constant K is ca. 1.09, from the difference of 18.3 cmϪ1
in CH vs CD zero-point energies between enamine (v ϭ 2990 cmϪ1) and aldimine
(v ϭ 2865 cmϪ1) (23). The chemical shifts can be estimated (␦ ϭ 130, 160 ppm)
from these same models (24). Then, according to Eq. (2), the aldiminic CH is expected
4
4
to exhibit a ⌬e of ϩ0.66 ppm, and the CD a ⌬e of Ϫ0.66 ppm, but shifted further
upfield by 1⌬0 and observable only with 2H decoupling. The perturbation shift 4⌬e at
C4 is automatically zero by symmetry. The chemical shifts for the fulvene carbons
of a frozen tautomer of 1 or 2 can be estimated from the acetyl and trimethylsilyl

6
PERRIN AND OHTA
derivatives, and the chemical shifts for phenyl carbons in a frozen tautomer of 2 or
3 can be estimated by using N-benzylaniline and N-benzylideneaniline as models
(25). Table 1 lists the estimated pertubation isotope shifts, all but one of which are
positive (downfield for the carbon nearer H).
Isotope shifts are now used to determine the shape of the potential surface of the
O-H-O hydrogen bond of 1 and the N-H-N hydrogen bonds of 2 and 3. If the hydrogen
bond is symmetric, the labeled compound will exhibit only intrinsic shifts. Instead,
a combination of an intrinsic isotope shift and a shift due to perturbation of a tautomeric
equilibrium is seen at several carbons. Thus each of these species exists as a pair of
rapidly interconverting tautomers, and the hydrogen bond is characterized by a double-
well potential.
MATERIALS AND METHODS
6-Dimethylamino-2-(N,N-dimethylformiminium)fulvene perchlorate (26). This pre-
cursor was prepared from 6-(dimethylamino)fulvene (27) and dimethylformamide-
POCl3 complex (28) in THF, precipitated with NaClO4 in methanol, and recrystallized:
17% yield, mp. 229–233ЊC (lit. 235–237ЊC). 13C NMR ([2H6]DMSO) ␦ 48.5, 118.4,
125.3, 128.5, 154.3.
6-Hydroxy-2-formylfulvene (1) (26). The perchlorate salt above was stirred with
chloroform and aqueous NaOH. The organic material was collected, washed with
water, and evaporated. The residue was stirred with aqueous NaOH at 50ЊC under
N2, then acidified with aqueous HCl, extracted with CH2Cl2, and evaporated carefully:
64% yield.
N,NЈ-Diphenyl-6-aminofulvene-2-aldimine (2) (15). A solution of perchlorate salt
in ethanol was refluxed with aniline and cooled, and the resulting precipitate was
recrystallized: 53% yield, mp 98–100ЊC (lit. 100ЊC).
N,NЈ-Bis(3,5-dimethylphenyl)-6-aminofulvene-2-aldimine (3). Perchlorate salt was
refluxed in ethanol with 3,5-dimethylaniline and cooled, and the resulting precipitate
1
was recrystallized: 54% yield, mp. 155–158ЊC. H NMR (CDCl3) ␦ 2.36 (s, 12H),
6.46 (t, J ϭ 3.75 Hz, 1H), 6.85 (s, 2H), 6.94 (s, 4H), 7.06 (d, J ϭ 3.5 Hz, 2H),
8.30 (s, 2H). 13C NMR (CDCl3) ␦ 21.3, 117.1, 120.8, 122.2, 126.8, 134.3, 139.4,
145.3, 150.5.
Preparation of deuterated compounds. Labeled compounds were synthesized using
TABLE 1
Estimated Perturbation Isotope Shifts at CHX, Fulvene, and Phenyl Carbons of 1 or 2
Carbon
⌬e1, ppm
⌬e2, ppm
C6,7
C1,2
C3,5
Ipso
0.76
0.47
0.89
—
0.66
0.07
0.12
0.09
Ortho
Meta
Para
—
—
—
0.18
Ϫ0.01
0.19

SYMMETRY OF O-H-O AND N-H-N HYDROGEN BONDS
7
TABLE 2
1H-Coupled 13C NMR Data of 2 in [2H4]Methanol
Carbon
␦, ppm
Multiplicity
J, Hz
Ortho
4
1,2
Para
Meta
3,5
Ipso
6,7
120.1
121.9
124.0
126.0
130.9
136.7
147.1
151.8
ddd
dt
m
157, 7.3, 6.0
164, 4.5
dt
157, 8.4
152, 8.3, 2.3
163
ddd
dm
m
dd
160, 2.3
[2H7]dimethylformamide-POCl3 complex as the source of the second formyl group.
This results in exactly one deuterium and one protium on the two aldehydic or
1
aldiminic carbons. The multiplicities and integration of the H NMR spectra were in
agreement with the expected product.
NMR Spectra and sample preparation. Spectra were recorded on a Varian Unity
1
500 spectrometer operating at a 13C resonance frequency of 125 MHz and a H
resonance frequency of 500 MHz. Spectra are referenced to the solvent. Samples for
measurement of isotope shifts were prepared by mixing equal weights of d1 and d0
compounds, except for 6-hydroxy-2-formylfulvene, which was synthesized as a mix-
ture of d1 and d0. Decoupling of both 1H and 2H was achieved by sending an additional
76.85-MHz signal through the lock channel.
RESULTS
NMR signal assignments. Partially assigned NMR spectra of 2 had been reported
1
(14). The full H spectrum could be assigned on the basis of signal integrations and
1
splitting patterns. The full 13C spectrum was assigned on the basis of H-couplings
1
and a 2D HMQC spectrum that correlates H and 13C chemical shifts. Table 2 lists
the carbon multiplicities and C-H coupling constants. Assignments for 3 were made
by analogy to 2.
6-Hydroxy-2-formylfulvene. The 13C NMR data for the mixture of isotopologs of
1 in CDCl3 are listed in Table 3. The second and third columns list n and nЈ, the
number of bonds between the label and the nearer or farther, respectively, of each
TABLE 3
13C NMR Data from Mixture of 1 and 6-[2]1 in CDCl3
Carbon
n
nЈ
␦0, ppm
⌬down, ppb
⌬up, ppb
6,7
1,2
3,5
4
1
2
3
4
4
3
4
4
176.0
126.5
141.2
125.2
ϩ376
ϩ200
ϩ151
Ͻ5
—
Ϫ279
Ϫ195
Ͻ5

8
PERRIN AND OHTA
set of carbons. The fourth column lists the NMR shifts for unlabeled 1. The fifth and
sixth columns list the observed downfield and upfield isotope shifts of 6-[2H]1 relative
to 1.
2
Without H decoupling the only signal observed from C6,7 of 6-[2H]1 is the CH,
since the CD is split into a triplet and lacks a nuclear Overhauser enhancement. The
CH shows a large downfield shift of 376 ppb. For both C1,2 and C3,5 one signal of
6-[2H]1 is shifted upfield and the other downfield. The signal for C4 is a singlet,
4
since it is too far from the label to show a resolvable ⌬0 and since it is located
symmetrically relative to the deuterated and undeuterated carbons and therefore cannot
show a ⌬e.
N,NЈ-Diphenyl-6-aminofulvene-2-aldimine. Figure 1 shows individual NMR sig-
nals of cyclopentadienyl and CHN carbons from a 1:1 mixture of 2 and 6-[2H]2. The
spectra are aligned along the signal of unlabeled 2 and are plotted on the same scale.
The signal for C4 is a singlet. Three signals are observed from some carbons, owing
to isotope shifts. Because of the stoichiometry the two smaller signals can be assigned
to 6-[2H]2. For C1,2 and C3,5 of 6-[2H]1 one signal is shifted upfield and the other
downfield, by unequal amounts. For ipso and ortho carbons (not shown) the signals
are shifted by equal or nearly equal amounts. For meta carbons, the signals are
singlets, with unresolvable isotope shifts. The para carbon shows a high-field shoulder,
corresponding to a small intrinsic isotope shift.
The 13C NMR chemical shifts and isotope shifts of a mixture of 2 and 6-[2H]2 in
[2H6]DMSO are listed in Table 4. The results in CDCl3 in Table 5 are quite similar.
Table 6 lists the data in [2H4]methanol. The assignment of C6,7 was facilitated by
2
simultaneously decoupling not only 1H but also H, since otherwise the CD signal is
invisible, as demonstrated by comparing Fig. 1a with Fig. 2, which shows C6,7 in
1
2
the H- and H-decoupled 13C NMR spectrum of 2.
N,NЈ-Bis(3,5-dimethylphenyl)-6-aminofulvene-2-aldimine. The assignment of C6,7
1
signals in 6-[2H]3 was made from a H-coupled 13C NMR spectrum. The special
feature of this derivative is that it permits the assignment of the ipso carbon signals
1
of 6-[2H]3, and by extension that of 6-[2H]2. The H-coupled spectrum is simplified
by 3,5-dimethyl substitution, which eliminates the 3JCH between ipso carbon and meta
CH of 2. Consequently the ipso carbon of 6-[2H]3 shows a well-resolved downfield
3
6.6-Hz doublet, due to JCH to the aldiminic CH, which must be distal. In contrast,
3
the other ipso carbon is an upfield singlet, broadened by an unresolvable JCD, so it
must be proximal to the label. The 13C NMR data for the mixture of isotopologs of
3 in [2H6]DMSO are listed in Table 7.
DISCUSSION
6-Hydroxy-2-formylfulvene. The isotope shifts in Table 3 are not purely intrinsic.
2
The signal from C6,7 of 6-[2H]1 that is readily observed, without H decoupling, is
the CH. Its large downfield shift of 376 ppb, relative to 1, is entirely contrary to the
4
small upfield shift expected from a ⌬0 (22). This must be a perturbation shift.
Likewise, the isotope shifts for both C1,2 and C3,5 in 6-[2H]1 are too large to be
intrinsic. The inequality of upfield and downfield isotope shifts for each of these
carbons means that intrinsic isotope shifts are not negligible. The combination of
intrinsic and perturbation shifts could not be separated, but the data are consistent

SYMMETRY OF O-H-O AND N-H-N HYDROGEN BONDS
9
FIG. 1. 13C NMR spectra of 2 and 6-[2H]2 in [2H6]DMSO expanded about (a) C6,7, (b) C1,2,
(c) C3,5, and (d) C4.
with perturbation shifts of Ϯ200 and Ϯ151 ppb at C1,2 and C3,5, respectively, and
2
3
intrinsic shifts ⌬0 and ⌬0 of Ϫ79 and Ϫ44 ppb.
N,NЈ-Diphenyl-6-aminofulvene-2-aldimine. At C4 and at the meta carbons of 2 no
isotope shifts can be resolved, neither intrinsic nor perturbation. The para carbon
shows a very small 6⌬0 of Ϫ16 ppb in DMSO and Ϫ24 ppb in methanol. Nevertheless,
as for 1, the isotope shifts for all the other carbons in Tables 4–6 are not purely
intrinsic. A downfield shift of 223 ppb is observed from C6,7. This must be the CH

10
PERRIN AND OHTA
TABLE 4
13C NMR Data from Mixture of 2 and 2-d in [2H6]DMSO
Carbon
n
nЈ
␦0, ppm
⌬down, ppb
⌬up, ppb
6,7
1,2
3,5
4
Ipso
Ortho
Meta
Para
1
2
3
4
3
4
5
6
4
3
4
4
6
7
8
9
151.5
122.2
135.6
120.8
145.1
119.2
130.0
125.1
ϩ223
ϩ178
ϩ87
Ͻ5
ϩ194
ϩ66
Ͻ5
Ϫ542a
Ϫ252
Ϫ147
Ͻ5
Ϫ211
Ϫ66
Ͻ5
Ͻ5
Ϫ16
a Upfield carbon is proximal to deuterium.
2
of 6-[2H]2, since it is seen without H decoupling. This is much too large and even
of the wrong sign to be a four-bond intrinsic shift, so it must be a perturbation shift.
With 2H decoupling the CD signal of 6-[2H]2 becomes visible 765 ppb upfield of the
CH. This represents an isotope shift of Ϫ542 ppb, which equals the sum of Ϫ⌬e and
4
1⌬0. If ⌬0 is negligible at the CH, as at C4, the isotope shifts can be separated into
perturbation shifts of Ϯ223 ppb and an intrinsic shift of Ϫ319 ppb for the CD, which
1
is consistent with other ⌬0 (22).
The data in Tables 4–6 and the spectra in Fig. 1 show isotope shifts that decrease
as the distance to the label (number of intervening bonds) increases. The difference
between the upfield and downfield shift at C1,2 or C3,5 again implies substantial
intrinsic isotope shifts. However, each observed isotope shift is the sum of an intrinsic
shift and a perturbation shift, which could not be separated. The data in [2H6]DMSO
are consistent with perturbation shifts of Ϯ178 and Ϯ87 ppb at C1,2 and C3,5,
respectively. The similar magnitudes of the upfield and downfield shifts at phenyl
carbons indicate that intrinsic shifts are small, as might be expected from the large
number of intervening bonds. Instead the large downfield isotope shifts at ipso and
ortho carbons do represent perturbation shifts.
N,NЈ-Bis(3,5-dimethylphenyl)-6-aminofulvene-2-aldimine. As with 2 the 13C NMR
TABLE 5
13C NMR Data from Mixture of 2 and 6-[2H]2 in CDCl3
Carbon
n
nЈ
␦0, ppm
⌬down, ppb
⌬up, ppb
6,7
1,2
3,5
4
Ipso
Ortho
Meta
Para
1
2
3
4
3
4
5
6
4
3
4
4
6
7
8
9
150.9
122.3
134.9
121.2
145.5
119.3
130.8
125.0
ϩ218
ϩ176
ϩ92
Ͻ5
182
ϩ59
Ͻ5
—
Ϫ249
Ϫ157
Ͻ5
Ϫ189
Ϫ57
Ͻ5
Ͻ5
Ϫ19

SYMMETRY OF O-H-O AND N-H-N HYDROGEN BONDS
11
TABLE 6
13C NMR Data from Mixture of 2 and 6-[2H]2 in [2H4]Methanol
Carbon
n
nЈ
␦0, ppm
⌬down, ppb
⌬up, ppb
6,7
1,2
3,5
4
Ipso
Ortho
Meta
Para
1
2
3
4
3
4
5
6
4
3
4
4
6
7
8
9
152.0
124.1
136.8
122.1
147.2
120.3
131.0
126.1
ϩ267
ϩ194
ϩ97
Ͻ5
ϩ213
ϩ79
Ͻ5
—
Ϫ273
Ϫ170
Ͻ5
Ϫ239
Ϫ85
Ͻ5
Ͻ5
Ϫ24
spectrum of 3 exhibits isotope shifts inconsistent with purely intrinsic shifts. The
3
spectra are quite similar, except for the detectability of a JCH between the aldiminic
proton and the ipso carbon. Since it is the downfield ipso carbon of 6-[2H]3 that
1
appears as a doublet in a H-coupled spectrum, this must be proximal to the H and
3
distal to the D. Therefore we can conclude that ⌬e ϭ ϩ193 ppb, not Ϫ205 ppb.
This agrees with the sign predicted in Table 1 and supports those predictions. By
3
inference the downfield ipso carbon of 6-[2H]2 is also proximal to H, so that ⌬e is
ϩ194 ppb in DMSO and ϩ213 ppb in methanol, not Ϫ211 or Ϫ239 ppb.
Environmental effects. The 13C NMR spectrum of 2 does not change significantly
FIG. 2. 1H- and 2H-decoupled 13C NMR spectrum of 2 and 6-[2H]2 in DMSO-d6 expanded about C6,7.

12
PERRIN AND OHTA
TABLE 7
13C NMR Data from Mixture of 3 and 6-[2H]3 in [2H6]DMSO
Carbon
n
nЈ
␦0, ppm
⌬down, ppb
⌬up, ppb
6,7
1,2
3,5
4
Ipso
Ortho
Meta
Para
CH3
1
2
3
4
3
4
5
6
6
4
3
4
4
6
7
8
9
9
150.9
122.1
126.6
120.4
144.7
116.8
139.1
134.3
20.9
214
178
85
Ͻ5
193
67
Ͻ5
Ͻ5
Ͻ5
—
Ϫ254
Ϫ145
Ͻ5
Ϫ205a
Ϫ65
Ͻ5
Ͻ5
Ͻ5
a Upfield carbon is proximal to the deuterium.
in different solvents (Tables 4–6). The small differences in the observed isotope shifts
are probably due to changes in the perturbation shifts, through unpredictable variation
of the chemical shift difference D in Eq. (2). The fact that observed shifts in [2H4]metha-
nol are of larger magnitude is likely to be due to a greater K in Eq. (2), arising from
the different vibrational frequencies with N-D-N hydrogen bonds. Such an increase
was detected in O-[2H]phthalate monoanion (10).
A purpose of this study was to probe the influence of solvation on the symmetry
of hydrogen bonds. Previous examples of asymmetric hydrogen bonds involved ions,
for which solvation favors charge localization (20). Here we study the hydrogen bonds
in some nonionic species, to see if they might become symmetric in an appropriate
solvent. Nevertheless, the observation of perturbation isotope shifts in 1, 2, and 3
indicates that their hydrogen bonds are asymmetric too.
Although the net charge in 1, 2, or 3 is zero, these molecules may have some local
ionic character. There are substantial contributions from zwitterionic resonance forms
with a cyclopentadienide anion and an oxenium or iminium cation. That cationic
character may make these molecules sensitive to solvation, just as ions seem to be,
and therefore unfavorable for symmetric hydrogen bonds.
Magnitude and sign of isotope shifts. The intrinsic shifts are uniformly negative,
as is nearly universal (22), and they fall off with distance from the isotope. The
1
observed (upfield) shift of Ϫ542 ppb for the CD in 1 is dominated by a ⌬0 of ca.
Ϫ319 ppb. The intrinsic shifts at C1,2, C3,5, and ipso carbons of 2 are also negative.
The assignment of the upfield ipso carbon of 6-[2H]3 as proximal to CD shows that
6
3⌬0 ϩ ⌬0 must be negative. By inference this conclusion also holds for 6-[2H]2.
The data show that perturbation isotope shifts also generally decrease with increas-
ing number of intervening bonds from the label, but the decrease is not uniform. The
largest is at the aldehydic or aldiminic carbon, and none is detectable at meta and
para carbons. The perturbation shifts in the phenyl seem to be somewhat larger than
4
in the fulvene. No ⌬e is seen at C4, since this must be zero by symmetry.
According to Eqs. (1–2) or Table 1, the estimated perturbation shift at the aldehydic
or aldiminic CH of 1 or 2 is 760 or 660 ppb. The observed shifts in Tables 3 and 4

SYMMETRY OF O-H-O AND N-H-N HYDROGEN BONDS
13
are only 376 and 223 ppb. Such a reduction was also seen for 3-hydroxy-2-phenylpro-
penal and was attributed to imperfections of the models used to estimate perturbation
shifts (9). The key result is that the isotope shift is positive for the unlabeled carbon,
as predicted. The labeled carbon in 2 has a perturbation shift of Ϫ223 ppb, as well
1
as a ⌬0 of Ϫ319 ppb, summing to the observed shift of Ϫ542 ppb.
The other carbon for which absolute isotope shifts could be unequivocally assigned
3
is the ipso carbon of 3. Since the upfield carbon is proximal to CD, requiring ⌬0 ϩ
3
6⌬0 to be negative, we can conclude that ⌬e of the other ipso carbon is ϩ193 ppb.
This is larger than the estimate in Table 1, but of the correct sign.
The magnitude of the isotope shifts at the aromatic carbons of 2 and 3 falls off
with distance. None could be detected at the meta carbons, as expected from the
estimates in Table 1. However, none was detected at para either, despite the estimates.
Again this is likely to be due to an imperfection of the models.
The O-H-O hydrogen bond of 1 is qualitatively similar to the N-H-N hydrogen
bond of 2 or 3. The magnitudes of the observed isotope shifts in the former are
slightly larger at each carbon. It is likely that the differences are due to a greater K
and/or D (Eq. (2)) for 1, but this is minor compared to the similarity that demonstrates
that all of these have asymmetric hydrogen bonds.
CONCLUSIONS
Observed isotope shifts of labeled 1, 2, and 3 can be attributed to a combination
of intrinsic and perturbation shifts. The key result is that the perturbation isotope shifts
establish the existence of a tautomeric equilibrium. Therefore these intramolecular
hydrogen bonds are asymmetric, the proton resides in a double-minimum potential
surface, and each molecule exists as a pair of rapidly interconverting tautomers.
Changing the solvent does not qualitatively change the observed spectra or the isotope
shifts for these uncharged species.
A significant conclusion from these studies is that the symmetry of the hydrogen
bond is not determined simply by the O-O or N-N distance. Even though that distance
is short here and could have been shorter in these seven-membered (including H)
rings, it does not lead to a symmetric hydrogen bond. Yet crucial questions persist.
If LBHBs are responsible for stabilization in some enzymatic catalysis, where does
that stabilization come from if the hydrogen bonds do not have the special feature
of symmetry? Why are symmetric hydrogen bonds observed in crystalline phases but
not as yet in solution? Perhaps the disorder of the solvation environment induces an
asymmetry of the hydrogen bond, whereas a crystal can guarantee a symmetric
environment. If so, the unusual strength of such hydrogen bonds cannot be due to a
maximum resonance stabilization associated with symmetry, since solvation is suffi-
cient to disrupt the symmetry. It has been suggested that relief of strain may be
responsible for the apparent strength of some hydrogen bonds (18).
ACKNOWLEDGMENTS
This research was supported by NSF Grants CHE94-20739 and CHE99-82103 and NIH Grant GM44527.
Purchase of the 500-MHz NMR spectrometer was made possible by grants from NIH and NSF.

14
PERRIN AND OHTA
REFERENCES
1. Pimentel, G. C., and McClellan, A. L. (1960) The Hydrogen Bond, Freeman, San Francisco. Schuster,
P., Zundel, G., and Sandorfy C. (Eds.) (1976) The Hydrogen Bond: Recent Developments in Theory
and Experiments, North-Holland, Amsterdam. Jeffrey, G. A., and Saenger, W. (1991) Hydrogen
Bonding in Biological Structures, Springer, Berlin. Scheiner, S. (1997) Hydrogen Bonding: A Theoreti-
cal Perspective, Oxford, New York.
2. Emsley, J. (1980) Chem. Soc. Revs. 9, 91. Hibbert, F., and Emsley, J. (1991) Adv. Phys. Org. Chem.
26, 255.
3. Gilli, P., Bertolasi, V., Ferretti, V., and Gilli, G. (1994) J. Am. Chem. Soc. 116, 909.
4. Gerlt, J. A., Kreevoy, M. M., Cleland, W. W., and Frey, P. A. (1997) Chem. Biol. 4, 259. Cleland,
W. W., and Kreevoy, M. M. (1994) Science 264, 1887. Gerlt, J. A., and Gassman, P. G. (1993) J.
Am. Chem. Soc. 115, 11552. Frey, P. A., Whitt, S. A., and Tobin, J. B. (1994) Science 264, 1927.
Frey, P. A. (1995) Science 269, 104. Cleland, W. W., and Kreevoy, M. M. (1995) Science 269, 104.
Hur, O., Leja, C., and Dunn, M. F. (1996) Biochemistry 35, 7378. Ash, E. L., Sudmeier, J. L., De
Fabo, E. C., and Bachovchin, W. W. (1997) Science 278, 1128. Cleland, W. W., Frey, P. A., and
Gerlt, J. A. (1998) J. Biol. Chem. 273, 25529. Harris, T. K., and Mildvan, A. S. (1999) Proteins:
Struct. Funct. Genet. 35, 275.
5. Mock, W. L., and Chua, D. C. Y. (1995) J. Chem. Soc., Perkin Trans. 2 2069. Scheiner, S., and Kar,
T. (1995) J. Am. Chem. Soc. 117, 6970. Schwartz, B., and Drueckhammer, D. G. (1995) J. Am.
Chem. Soc. 117, 11902. Kato, Y., Toledo, L. M., and Rebek, J., Jr. (1996) J. Am. Chem. Soc. 118,
8575. Warshel, A., and Papazyan, A. (1996) Proc. Natl. Acad. Sci. 93, 13665. Shan, S., Loh, S., and
Herschlag, D. (1996) Science 272, 97.
6. Guthrie, J. P., and Kluger, R. (1993) J. Am. Chem. Soc. 115, 11569. Warshel, A., Papazyan, A., and
Kollman, P. A. (1995) Science 269, 102. Alagona, G., Ghio, C., and Kollman, P. A. (1995) J. Am.
Chem. Soc. 117, 9855. Guthrie, J. P. (1996) Chem. Biol. 3, 163.
7. Mulholland, A. J., Lyne, P. D., and Karplus, M. (2000) J. Am. Chem. Soc. 122, 534. Mohr, M., and
Zipse, H. (1999) Chem. Eur. J. 5, 3046.
8. Perrin, C. L., and Nielson, J. B. (1997) Annu. Rev. Phys. Chem. 48, 511. Frey, P. A., and Cleland,
W. W. (1998) Bioorg. Chem. 26, 175. Cleland, W. W. (2000) Arch. Biochem. Biophys. 382, 1.
9. Perrin, C. L., and Kim, Y. J. (1998) J. Am. Chem. Soc. 120, 12641. Perrin, C. L., Nielson, J. B., and
Kim, Y.-J. (1998) Ber. Bunsen-Ges. Phys. Chem., 102, 403.
10. Perrin, C. L., and Thoburn, J. D. (1989) J. Am. Chem. Soc. 111, 8010. Perrin, C. L., and Thoburn,
J. D. (1992) J. Am. Chem. Soc. 114, 8559. Perrin, C. L. (1994) Science 266, 1665. Perrin, C. L., and
Nielson, J. B. (1997) J. Am. Chem. Soc. 119, 12734.
11. Perrin, C. L., and Ohta, B. K., (2001) J. Am. Chem. Soc. 123, 6520.
12. Ammon, H. L., and Mueller-Westerhoff, U. (1974) Tetrahedron 30, 1437. Fuess, H., and Linder, H.
J. (1975) Chem. Ber. 108, 3096.
13. Claramunt, R. M., Sanz, D., Alarco´n, S. H., Torralba, M. P., Elguero, J., Foces-Foces, C., Pietrzak,
M., Langer, U., and Limbach, H.-H. (2001) Angew. Chem. Int. Ed. 40, 420.
14. Jackman, L. M., Trewella, J. C., and Haddon, R. C. (1980) J. Am. Chem. Soc. 102, 2519.
15. Mueller-Westerhoff, U. (1970) J. Am. Chem. Soc. 92, 4849.
16. Brown, R. S., Tse, A., Nakashima, T., and Haddon, R. C. (1979) J. Am. Chem. Soc. 101, 3157.
17. Pickett, H. M. (1973) J. Am. Chem. Soc. 95, 1770.
18. Morpurgo, S., Bossa, M., and Morpurgo, G. O. (1998) J. Mol. Struct. (THEOCHEM) 429, 71.
19. McCoy, L. L. (1967) J. Am. Chem. Soc. 89, 1673.
20. Pan, Y. P., and McAllister, M. A. (1997) J. Am. Chem. Soc. 119, 7561. Chen, J. G., McAllister, M.
A., Lee, J. K., and Houk, K. N. (1998) J. Org. Chem. 63, 4611. Pan, Y. P., and McAllister, M. A.
(1998) J. Am. Chem. Soc. 120, 166. Garcia-Viloca, M., Gonzalez-Lafont, A., and Lluch, J. M. (1999)
J. Am. Chem. Soc. 121, 9198.
21. Saunders, M., Telkowski, L., and Kates, M. R. (1977) J. Am. Chem. Soc. 99, 8070. Siehl, H.-U.
(1987) Adv. Phys. Org. Chem. 23, 63. Forsyth, D. A. (1987) in Isotopes in Organic Chemistry,
(Buncel, E., and Lee, C. C., Eds.), Vol. 6, Chap. 1, Elsevier, Amsterdam.

SYMMETRY OF O-H-O AND N-H-N HYDROGEN BONDS
15
22. Batiz-Hernandez, H., and Bernheim, R. A. (1967) Prog. Nucl. Magn. Reson. Spectrosc. 3, 63. Hansen,
P. E. (1983) Annu. Rep. NMR Spectr. 15, 105. Jameson, C. J., and Osten, H. J. (1986) Ann. Rep.
NMR Spectrosc. 17, 1. Hansen, P. E. (1988) Progr. Nucl. Magn. Reson. Spectrosc. 20, 207. Hansen,
P. E. (2000) Magn. Reson. Chem. 38, 1.
23. Muller, K., Previdoli, F., and Desilvestro, H. (1981) Helv. Chim. Acta 64, 2497. Lewis, J. W., and
Sandorfy, C. (1982) Can. J. Chem. 60, 1720.
24. Bureau, R., Mortier, J., and Joucla, M. (1993) Bull. Soc. Chim. Fr. 584. Fraser, R. R., and Chuaqui-
Offermanns, N. (1981) Can. J. Chem. 3007.
25. Hartke, K., Kohl, A., and Kaempchen, T. (1983) Chem. Ber. 116, 2653. Sadtler Standard Spectra
(1969) spectra 7807C and 2016C, Sadtler Research Laboratories, Philadelphia.
26. Hafner, K., Vopel, K. H., Ploss, G., and Konig, C. (1963) Justus. Liebigs. Ann. Chem. 661, 52.
27. Hafner, K., Vopel, K. H., Ploss, G., and Konig, C. (1967) Org. Syn. 47, 52.
28. Bredereck, H., Gompper, R., Klemm, K., and Rempfer, H. (1959) Chem. Ber. 92, 837.