Reaction pathways of proton transfer in hydrogen-bonded phenol-carboxylate complexes explored by combined UV-Vis and NMR spectroscopy

Article abstract of DOI:10.1021/ja201113a

Combined low-temperature NMR/UV-vis spectroscopy (UVNMR), where optical and NMR spectra are measured in the NMR spectrometer under the same conditions, has been set up and applied to the study of H-bonded anions A· middot;H middot; middot;X^- (AH = 1-¹³C-2-chloro-4-nitrophenol, X^- = 15 carboxylic acid anions, 5 phenolates, Cl^-, Br ^-, I^-, and BF₄^-). In this series, H is shifted from A to X, modeling the proton-transfer pathway. The ¹H and ¹³C chemical shifts and the H/D isotope effects on the latter provide information about averaged H-bond geometries. At the same time, red shifts of the π-π* UV-vis absorption bands are observed which correlate with the averaged H-bond geometries. However, on the UV-vis time scale, different tautomeric states and solvent configurations are in slow exchange. The combined data sets indicate that the proton transfer starts with a H-bond compression and a displacement of the proton toward the H-bond center, involving single-well configurations A-H...X^-. In the strong H-bond regime, coexisting tautomers A · ·H...X^- and A^-...H ··X are observed by UV. Their geometries and statistical weights change continuously when the basicity of X^- is increased. Finally, again a series of single-well structures of the type A^-...H-X is observed. Interestingly, the UV-vis absorption bands are broadened inhomogeneously because of a distribution of H-bond geometries arising from different solvent configurations.

Full text of DOI:10.1021/ja201113a

ARTICLE
pubs.acs.org/JACS
Reaction Pathways of Proton Transfer in Hydrogen-Bonded
PhenolꢀCarboxylate Complexes Explored by Combined UVꢀVis
and NMR Spectroscopy
Benjamin Koeppe,† Peter M. Tolstoy,*,†,‡,§ and Hans-Heinrich Limbach†
†Institut f€ur Chemie und Biochemie, Freie Universit€at Berlin, Takustrasse 3, D-14195 Berlin, Germany
‡V.A. Fock Institute of Physics, St. Petersburg State University, Uljanovskaja 1, 198504 St. Petersburg, Russia
S Supporting Information
b
ABSTRACT: Combined low-temperature NMR/UVꢀvis
spectroscopy (UVNMR), where optical and NMR spectra are
measured in the NMR spectrometer under the same conditions,
has been set up and applied to the study of H-bonded anions
A
H
Xꢀ (AH = 1-13C-2-chloro-4-nitrophenol, Xꢀ = 15
3 3 3 3
carboxylic acid anions, 5 phenolates, Clꢀ, Brꢀ, Iꢀ, and BF4ꢀ).
In this series, H is shifted from A to X, modeling the proton-
transfer pathway. The 1H and 13C chemical shifts and the H/D
isotope effects on the latter provide information about averaged
H-bond geometries. At the same time, red shifts of the πꢀπ* UVꢀvis absorption bands are observed which correlate with the
averaged H-bond geometries. However, on the UVꢀvis time scale, different tautomeric states and solvent configurations are in slow
exchange. The combined data sets indicate that the proton transfer starts with a H-bond compression and a displacement of the
proton toward the H-bond center, involving single-well configurations AꢀH Xꢀ. In the strong H-bond regime, coexisting
3 3 3
tautomers A
H
Xꢀ and Aꢀ
H
X are observed by UV. Their geometries and statistical weights change continuously when
3 3 ꢀ3 3 3
3 3 3 3 3
the basicity of X is increased. Finally, again a series of single-well structures of the type Aꢀ HꢀX is observed. Interestingly, the
3 3 3
UVꢀvis absorption bands are broadened inhomogeneously because of a distribution of H-bond geometries arising from different
solvent configurations.
’ INTRODUCTION
change is the chemical shift of the hydrogen-bonded proton,
which senses the H-bond compression by a low-field shift,9ꢀ11
whereas signals of heavy atoms such as 15N and 19F probe the
proton position,12,13 as well as H/D isotope effects on chemical
shifts.10,14ꢀ16 In cases of hydrogen bonds formed by carboxylic
acids, the NMR parameters of carboxylic carbon atoms have been
exploited.10,15ꢀ17
These NMR correlations have been applied to liquid solutions
in order to establish averaged H-bond geometries.18 However,
the proton-transfer pathway of Figure 1a, corresponding to a
sequence of averaged H-bond geometries, constitutes a crude
model which does not explicitly take into account the possibility
of single- or double-well potentials, nor distributions of H-bond
geometries arising from soluteꢀsolvent interactions.19 The latter
have also been called “solvatomers”.20
Hydrogen-bond-mediated intermolecular acidꢀbase interac-
tions are of prime importance for the function of biomolecules.1,2
They depend strongly on the environment. In aqueous solution,
acidꢀbase interactions are characterized by pKa values, which
allow one to calculate the equilibrium constant of proton transfer
between the solvated acid and the solvated base. However, this
concept breaks down when AH and B form a direct hydrogen
bond, AHB, for example in environments such as the organic
solid state, aprotic solvents, or the interior of proteins. The
location of H may depend not only on the acidity of AH and the
basicity of B but also on the local electrostatics. The pathways of
proton transfer between A and B have been explored by neutron
crystallography3 and solid-state NMR4 of series of hydrogen-
bonded complexes, where each complex is assumed to represent
a snapshot of the proton-transfer pathway. These studies showed
that when the acidity of AH is increased, H transfer to B occurs
Using the NMR method of isotopic perturbation, Perrin et al.
obtained evidence in the case of several intramolecular hydrogen-
bonded systems for the absence of quasi-symmetric structures
via a series of geometric states, A
H
B via A H B to
3 3 3 3 3 3 3
3 3 3 3
A
H B and the presence of tautomeric equilibria, as illu-
3 3 3 3
A
H B, where a compression of the intermolecular
3 3 3 3 3 3 3
strated in Figure 1b, characterized by an equilibrium constant
hydrogen bond occurs when H is shifted toward the H-bond
center. Similar effects were observed for isolated complexes in
the gas phase exposed to strong electric fields.5 Solid- and liquid-
state NMR H-bond correlations indicated typical changes of
NMR parameters during this process.6ꢀ8 The most important
K.21 An increase of the acidity of AH then leads to an increase of
Received: February 4, 2011
Published: May 02, 2011
r
2011 American Chemical Society
7897
dx.doi.org/10.1021/ja201113a J. Am. Chem. Soc. 2011, 133, 7897–7908
|

Journal of the American Chemical Society
ARTICLE
K and hence to H transfer from A to B. However, it has not yet
been possible to determine the geometries of the tautomers or to
solve the question of how far the protons are transferred, whether
the jump distances are smaller than proton displacements caused
by the solvent cages as has been discussed previously,20 and
hence whether tautomers are clearly separated from solvatomers.
In contrast to NMR, which is not able to follow H transfers in
strong hydrogen bonds in solution in real time, optical spectros-
copy is able to distinguish an acid and its deprotonated base.
Thus, using UVꢀvis spectroscopy, tautomeric equilibria have
been previously detected for complexes of phenols with proton
acceptors.22ꢀ26 Unfortunately, the spectra of solutions of AH
and B are often influenced by a number of different hydrogen-
bonded species such as homoconjugated complexes AHAꢀ and
BHBþ, as well as higher associates of the type (AH)nBm, which
makes the analysis often difficult. Moreover, potential informa-
tion about H-bond geometries from UVꢀvis spectra has not
been exploited to our knowledge, so again it is difficult to obtain
information about the potential energy surface of the proton
motion by UVꢀvis spectroscopy.
However, if links between UVꢀvis bands and H-bond geo-
metries could be established, this not only would be of impor-
tance for understanding the pathways of proton transfer in
solution but also may have important applications in biophysics.
For example, photoactive proteins such as the photoactive yellow
protein (PYP) contain a phenolic cofactor hydrogen-bonded to a
carboxylate,27 which after photoexcitation forms a series of
structural intermediates exhibiting characteristic absorption
frequencies.28 However, the associated changes in the H-bond
geometries are largely unknown to date.
In order to establish the desired correlation between H-bond
geometries and UVꢀvis parameters, we have proposed the use of
combined low-temperature UVꢀvis and NMR spectroscopy
(UVNMR), where both types of spectra are measured at the
same time for a single sample under the same conditions. The
setup for this technique has been described recently.29 The NMR
spectrum allows one to determine the structure of the complexes
present in the sample, i.e., removes the existing uncertainties of
UVꢀvis spectroscopy, and allows one to obtain information
about the averaged H-bond geometries from NMR chemical
shifts. Thus, the high potential of UVꢀvis spectroscopy to detect
fast dynamics, averaged out in NMR spectra, can be exploited,
and UVꢀvis bands can be correlated with H-bond geometries.
To model interactions in photoactive proteins, we have
chosen to perform low-temperature UVNMR studies of anionic
AHXꢀ hydrogen-bonded complexes depicted in Figure 2, where
AH = 2-chloro-4-nitrophenol (1) and Xꢀ = 2-chloro-4-nitro-
phenolate 2 (Aꢀ), 15 carboxylate anions, 6 phenolates, and the
halogen anions Clꢀ, Brꢀ, Iꢀ, and BF4ꢀ. In addition, we have
studied the neutral AH (1) and its anion Aꢀ (2) (see Figure 2).
Figure 1. H-transfer pathways from an acid A to a base B in a hydrogen
bond AHB upon an increase of the acidity of AH. (a) Continuous change
of the H-bond geometry from the initial to the final state, assisted by
H-bond compression when H approaches the H-bond center. (b)
Traditional proton-transfer equilibrium between A and B.
Figure 2. Structures of studied species: 2-chloro-4-nitrophenol (1), its anion 2, and its hydrogen-bonded conjugated complexes with the given acids and
phenols, 3ꢀ27. Positions of bridging protons in complexes vary as discussed in the text.
7898
dx.doi.org/10.1021/ja201113a |J. Am. Chem. Soc. 2011, 133, 7897–7908

Journal of the American Chemical Society
ARTICLE
In all cases, tetraethylammonium (TEA) was used as a counter-
cation. As solvent we used CD2Cl2 down to its freezing point at
ꢀ97 °C. We chose 1 as chromophore because it is a rather acidic
phenol and can mimic the acidity increase of weaker acids upon
photoexcitation, allowing one to detect a wide range of proton-
ation states in complexes with anions of carboxylic acids and
other phenols. In this paper we will refer to the 2-chloro-4-
nitrophenolate moiety as the “chromophore” in all of its forms,
be it neutral AH, anionic Aꢀ, or part of an AHXꢀ complex.
This paper is structured as follows. After the Experimental
Section, the results of our UVNMR measurements are presented
and discussed. A correlation between UV-band frequencies and
H-bond geometries is proposed, and the range of geometries is
specified in which a tautomeric equilibrium is present. Based
thereupon, a qualitative description of the proton-transfer path-
way is presented which may, in the future, be extended to a
quantitative description.
danger of too high losses of sample substance needs to be considered.
Finally, the sample is redissolved in CD2Cl2.
Specific Techniques for the Preparation of UVꢀVis and
NMR Samples. To obtain all the data on the species HA, Aꢀ, and
AHXꢀ described in this work, two types of experiments on two
independent sample types were performed for each species. On the
one sample (type A), combined low-temperature 1H NMR and UVꢀvis
(UVNMR) measurements were performed at concentrations of the
chromophore moiety Aꢀ of ∼1 mM. These kinds of samples were
prepared in UVNMR cuvettes as described below. The other sample
type (type B) served for 13C NMR measurements and contained
∼3 mM 1-13C-enriched Aꢀ (and the corresponding amounts of other
components) to increase the signal-to-noise ratios in 13C NMR spectra.
Samples of type B were prepared in NMR tubes of medium wall size
equipped with J. Young valves (Norell). For each sample of each species
and type of experiment, relative sample compositions determined by
1H NMR are listed in the Supporting Information (SI) Table S1 and
presented graphically in Figure S1. Procedural details will follow
immediately. To ensure consistent the results, low-temperature
1H NMR spectra of the samples of types A and B were compared, and
it was verified that the 1H peak positions were the same (0.06 ppm mean
square deviation; see Figure S9). In other words, the 3-fold difference in
the absolute concentration of the samples of types A and B did not affect
the NMR results, apart from the improved sensitivity in the case of type
B samples. In all samples, CD2Cl2 (99.9% or 99.7%, Eurisotop; 99.8%,
Deutero, Armar) dried over 4 Å molecular sieves was used as the solvent;
the overall sample volumes varied between 300 and 450 μL.
’ EXPERIMENTAL SECTION
Synthesis of 1-13C-2-Chloro-4-nitrophenol (1). 1-13C-4-Ni-
trophenol was obtained by cyclization30 of 2-13C-acetone with sodium
nitromalonaldehyde.31 The latter was then converted into 1-13C-2-
chloro-4-nitrophenol (1) using a well-established procedure32 which
had, however, to be adapted to a small scale as described in the following.
First, 0.10 g of 1-13C-4-nitrophenol was dissolved in 10 mL of 25%
aqueous HCl with gentle heating. The solution was cooled in an ice bath
under vigorous stirring, upon which it formed a gel. Under continued
stirring, 0.35 mL of sodium hypochlorite solution (150 g/L) was added
dropwise within 2 min. After 1 h of continued stirring, the reaction
mixture was extracted with diethyl ether. The organic phase was
separated and dried. The solvent was evaporated, yielding 0.11 g of
the crude product. Removal of a yellow impurity was achieved by
preparative thin-layer chromatography (TLC, silica; dichloromethane/
methanol 20:1). Residual starting material, if present, was removed by
recrystallization from 25% aqueous HCl.
General Techniques for the Preparation of UVNMR and
NMR Samples. To facilitate the sample preparation, certain prelimin-
ary and general actions were taken, which are described below. The
specific details for each of the samples are given in the following
subsections.
Stock Solutions. Stock solutions, 5ꢀ100 mM (0.5 mL), of the
different acids and phenols in dry CD2Cl2, CH2Cl2, or methanol were
prepared. For that purpose, substances were weighted and dissolved in
2 mL glass vials with PTFE lined screw caps (Wheaton). Solvents and
solutions were handled with microliter pipets (Eppendorf). Mixing of
the components for the sample preparation was based on these stock
solutions.
Preparation of Tetraethylammonium Salts of Proton Donors HX. A
certain amount (∼1ꢀ20 μmol) of HX (as stock solution) was mixed in a
10 mL flask with 0.7ꢀ1.0 equiv of TEA hydroxide in methanol solution.
The solvents were removed on a rotary evaporator (40 °C water bath,
<10 mbar). After pressure equilibration with inert gas, ∼1 mL of CH2Cl2
(dried over 4 Å molecular sieves) was added and evaporated in vacuum
to remove residual methanol and water. The latter procedure was
repeated twice, leaving the dry salt.
Deuteration of NMR Samples in Mobile Proton Sites. Whenever
deuteration of the sample was needed, the NMR tube equipped with a J.
Young valve and containing the already prepared sample (see below)
was attached to a high-vacuum line, and the solvent was evaporated.
About 0.2 mL of methanol-OD (99.5%) was added to the sample and
subsequently evaporated. In samples with volatile components, a trade-
off between the desirable complete evaporation of methanol and the
Samples of 2-Chloro-4-nitrophenol (1). A 1 mM solution of 1 and a
3 mM solution of 1-13C-enriched 1 were used for UVNMR (typeA) and
13C NMR (type B), respectively.
Samples of TEA 2-Chloro-4-nitrophenolate (2). For UVNMR (type
A), 20 μmol of TEA pivalate (prepared from the acid as described above)
acting as proton scavenger were dissolved in 300 μL of CD2Cl2 and
transferred to a UVNMR cuvette. Next, 0.3 μmol of 1 as a 10 mM
solution was added. For 13C NMR (type B), 1 μmol of 2 was prepared
from 1-13C enriched 1 and TEA hydroxide as described in the previous
subsection, dissolved in 350 μL of solvent, and transferred to a sample
tube.
Samples for Complexes 4ꢀ7. The appropriate commercially avail-
able salt TEAþXꢀ was dried by repeated addition and evaporation of dry
CH2Cl2 in a rotary evaporator, taken up in 350 μL of dry CD2Cl2, and
transferred to a sample tube. To this were added ∼0.4 μmol of HA for
UVNMR samples (type A) and ∼1.2 μmol HA for 13C NMR samples
(type B), readily yielding complexes 4ꢀ7 as the exclusive species
containing chromophore moiety formed in CD2Cl2 solution at low
temperature. Type B samples were deuterated in the mobile proton sites
as described in the general procedure above.
Samples for Complexes 8ꢀ27. The appropriate salt TEAþXꢀ
(freshly prepared as described above) was dissolved in 350 μL of dry
CD2Cl2 and transferred to a sample tube. To this were added ∼0.4 μmol
of HA for UVNMR samples (type A) and ∼1.2 μmol of HA for 13C
NMR samples (type B). A solution of the acid HA was then added
stepwise under 1H NMR monitoring in order to shift equilibria in the
direction of the desired heteroconjugated complexes AHXꢀ. For type A
(UVNMR) samples, the composition was carefully adjusted such that
AHXꢀ became in good approximation the exclusive species containing
the chromophore moiety at measurement conditions. This procedure is
described in more detail in ref 29. In complexes 26 and 27, exclusiveness
in the optical spectra could not be obtained due to a rapid irradiation-
induced blue-drift of absorption during measurement;33 spectra of
AHXꢀ complexes 26 and 27 were determined by measurements in
the presence of chromophore anion Aꢀ (2) and subsequent deconvolu-
tion. The UVꢀvis spectrum of 24 cannot be obtained easily under the
given regime due to overlaps in absorption of the 2-chloro-4-nitrophenol
7899
dx.doi.org/10.1021/ja201113a |J. Am. Chem. Soc. 2011, 133, 7897–7908

Journal of the American Chemical Society
ARTICLE
Figure 3. Experimental setup used for UVNMR experiments.
and 4-nitrophenol moieties. In all cases, type B samples were deuterated
at the mobile proton sites, as described above (except for the sample of
complex 20, in which case acetic acid was found to be too volatile for the
given procedure), as soon as a satisfactory fraction of AHXꢀ (ADXꢀ)
was found. As a result, in type B sample, the AHXꢀ complex is not the
exclusive form of the chromophore moiety.
Note that sample solutions of complexes, especially those containing
weak proton donors, decompose on storage and irradiation due to
nucleophilic attack of phenolates on dichloromethane;33 various small
amounts of coupling products were found in corresponding samples.
Samples of Hydrogen Di-2-chloro-4-nitrophenolate (3). A titration
approach analogous to that used for the samples of heteroconjugated
complexes 8ꢀ27 was employed. Sample solutions of appropriate
amounts of 2-chloro-4-nitrophenolate (2) obtained from 1 and TEA
hydroxide were prepared as described above, followed by a stepwise
optimization by addition of a solution of 1.
1
Figure 4. Partial experimental H and 13C NMR spectra and super-
imposed experimental and calculated UVꢀvis spectra of mixtures of
2-chloro-4-nitrophenol (1) with TEA carboxylates in CD2Cl2 around
175 K. The numbers on the peaks correspond to the species depicted in
Figure 2. The dashed bands in the center column were calculated as
described in the text.
where a = ν~0 ꢀ H{F/(F2 ꢀ 1)}, b = H{F/(F2 ꢀ 1)} exp c2, and
c = F/(2 ln 2)1/2. The meanings of the four fitting parameters are as
follows: A is the amplitude factor, ν~0 is the wavenumber at the band’s
maximum, H is the full width at half-height, and F is the skewness.
For 1, 2, and complexes 4ꢀ9, 22ꢀ27, simply fitting with a single log-
normal function gave good results (see Figure 4 and SI Figures S5 and
S6). For all single log-normal fits, a skewness parameter of 1.5 was found
to be satisfactory. For complexes 3, 10ꢀ21, the bands in the UVꢀvis
spectra were fitted assuming the presence of two bands with asymmetry
parameter of 1.5 and a bandwidth chosen as described in the Results
section. The physical meaning of the bands and their widths will be
Acquisition and Processing of Combined 1H NMR/UVꢀVis
and 13C NMR Spectra. UVNMR Measurements. The experimental
setup for a Bruker 500 MHz spectrometer shown in Figure 3 was similar
to that described in ref 29, except for the following improvements.
Samples were measured in “NMR cuvettes” custom-made by Hellma
(M€ullheim, Germany): the round bottoms of 5 mm quartz NMR sample
tubes (0.8 mm wall size) were cut off and replaced by flat, optically
polished quartz disks (0.4 mm thick). Inside the NMR cuvettes were
placed PTFE inserts with 0.2 mm spacers, providing a homogeneous 0.4
mm optical path length for absorption measurements in reflection.
Technical drawings of an NMR cuvette and a PTFE insert are given in SI
Figure S2. Sample insertion depth in the NMR spinner was chosen such
that samples rested on the optical reflection probe during measure-
ments. Blank UVꢀvis measurements were performed prior to the actual
sample measurements for each combination of NMR cuvette and PTFE
insert using 50 μL of CD2Cl2 solvent at 200 K. 1H NMR spectra were
calibrated to CHDCl2 (5.32 ppm). In all figures, UVꢀvis spectra were
normalized to equal maxima in absorbance.
1
addressed in the Discussion section. From the H NMR spectra, we
estimate that if there are other H-bonded complexes in solution which
include 1 or 2 as one of the partners, the amount of such species does not
exceed 5% of the main 1:1 complex. We estimate that this converts into
the possible 5% error in the intensity of the fitted bands and in the error
of the band position of no more than 5 nm.
Nomenclature. Throughout the paper, the chemical shifts of com-
plexes 3ꢀ27 are labeled as δ(COHX), in which CO refers to the 1-C of
1 or 2 and its phenolic oxygen atom, and X to the acceptor atom in the
partner (mostly oxygen). The observed nucleus is underlined. It should
be clear from the context to which complex the chemical shift refers. For
compounds 1 and 2, the chemical shifts are denoted as δ(COH) and
δ(COꢀ), correspondingly. The H/D isotope effects on the chemical
shift of the 1-C carbon of the 2-chloro-4-nitrophenyl moiety are defined
as δ(CODX) ꢀ δ(COHX).
13C NMR Measurements. For 13C NMR measurements, a standard
(not inversed) 1H/13C NMR probe was used. 13C NMR spectra were
calibrated to 13CD2Cl2 (53.5 ppm).
’ RESULTS
UVꢀVis Spectra Deconvolution. For reasons that will become
apparent in the Results and Discussion sections, UVꢀvis spectra
were analyzed by least-squares fitting using log-normal band shapes in
the wavenumber dimension.34 For convenience, following the nomen-
clature of ref 34, the fitting function I was written as
UVNMR Spectroscopy. Selected spectral results obtained in
this study for samples containing 1:1 complexes of 2-chloro-4-
nitrophenolate (2) with carboxylic acids (HX) in CD2Cl2 are
depicted in Figure 4. The signals are labeled by bold numbers
which correspond to the structures of heteroconjugated com-
plexes AHXꢀ and the homoconjugated complex AHAꢀ depicted
in Figure 2. Roughly, the acidity of the acid component HX
decreases from the top to the bottom. The stars and double stars
characterize the hydrogen dicarboxylates (XHXꢀ) and
0
@
1
A
(
)
2
ꢀ
ꢁ
~
Ab
1
c2
ν ꢀ a
2
~
~
Iðν, A, ν0, H, FÞ ¼
expðꢀc Þ exp
ꢀ
ln
ð1Þ
~
ν ꢀ b
b
7900
dx.doi.org/10.1021/ja201113a |J. Am. Chem. Soc. 2011, 133, 7897–7908

Journal of the American Chemical Society
ARTICLE
phenol moiety are visible, as only this position was enriched with
13C. The spectra in Figure 4a and h feature signals at 155.9 and
173.9 ppm, respectively, stemming from neutral 1 and the anion
2. The 13C NMR spectra in Figure 4bꢀg have several signals,
assigned to different complexes, as indicated by the numbers.
Several species involving 2-chloro-4-nitrophenol might be pre-
sent in solution, because, unlike for UVNMR measurements, the
composition of the samples for 13C NMR spectra was not
optimized, such that the desired complex is the only species
involving 2-chloro-4-nitrophenol formed at low temperatures.
Additionally, after partial deuteration, most of the signals split
into two, assigned to the nondeuterated (“H”) and deuterated
(“D”) isotopologs of the complex, δ(COHX) and δ(CODX),
respectively. The values δ(COHX) of the conjugated complexes
vary in the range 157ꢀ171 ppm, having the chemical shifts
δ(COH) and δ(COꢀ) as limits. The increasing values of
δ(COHX) were used as a criterion for ordering all spectra,
leading to the trends described above. There is an inversion in the
sign of the secondary isotope effect on the 1-13C chemical shifts,
δ(CODX) ꢀ δ(COHX), within the series.
Figure 5. Chemical structures of (a) hydrogen dicarboxylate and
(b) dihydrogen tricarboxylate. In this work, tetraethylammonim served
as a countercation. The signals of hydrogen dicarboxylate and dihydro-
gen tricarboxylate are marked with * and **, respectively, in Figures 5, S2,
and S3.
The complete set of NMR and UVꢀvis data for all complexes
studied, including the phenol 1 and its anion 2, is assembled in
Table 1.
dihydrogen tricarboxylates (XHXHX)ꢀ whose structures are
depicted in Figure 5. These structures have been characterized
1
recently by H and 13C NMR in the cases of acetic acid and
UVꢀVis Band Shape Analysis. An examination of the shapes
of the UVꢀvis bands reveals the following features. At high and
low energies single bands are observed, as illustrated by the
spectra depicted in Figure 4a,b,g, and h. By contrast, dual bands
are observed for complexes 11, 13, 15, and 19 at intermediate
energies (Figure 4cꢀf). In Figure 6, the experimental full widths
at half-heights, Δν~, are plotted versus the center of gravity of the
band, ν~COG. Starting on the high-energy side, we first observe for
the single bands of 1 and 4ꢀ10 a monotonic decrease of the
bandwidths with decreasing band energy. However, in the dual-
band region, starting from complex 11 there is a sharp increase of
Δν~ that culminates for complex 15, followed by a sharp decrease
until complex 21. Finally, a monotonic decrease is observed again
for complexes 22ꢀ27 reaching a minimum for the anion 2. The
monotonic decrease of Δν~ in both single-band regions is well
described by the dashed line. The latter was used to represent the
intrinsic bandwidths in the dual-band region. These bandwidths
were then treated as fixed parameters of the log-normal function
eq 1, used to simulate the blue and red components of the dual
UV bands of Figure 4; only the band intensities and positions
were varied until an optimal fit of the experimental spectra was
obtained, leading to the superimposed experimental and calcu-
lated spectra in the center column of Figure 4. For clarity, the
calculated individual bands are also included as dashed curves.
Furthermore, we have represented each band by a vertical bar in
Figure 4, located at the center of gravity, where the relative
heights represent the band fractions given by
chloroacetic acid.10 For reference we have included the spectra of
free 2 as well as of 2-chloro-4-nitrophenol (1). The spectra were
measured around 170ꢀ180 K. The full set of spectra is included
in SI Figures S3ꢀS6.
In the left column of Figure 4, the low-field parts of the
1H NMR spectra are depicted. When the acidity of the added acid
HX is decreased, the 1H signal of the bridging proton of AHXꢀ is
shifted to low field in the case of 4ꢀ14; a maximum occurs
around 19 ppm for 15, and then the signal shifts to higher field
again in the case of 16ꢀ27. As the phenolic proton of 1 resonates
1
around 7 ppm at 175 K,29 the low-field H NMR region of
Figure 4a does not contain any signal.
The center column of Figure 4 depicts the corresponding
UVꢀvis spectra between 270 and 500 nm of the chromophore
moiety,
2-chloro-4-nitrophenol/2-chloro-4-nitrophenolate.
These spectra were taken in the NMR spectrometer at the same
1
time as the H NMR spectra shown in the left column. As
described in the Experimental Section, the samples had been
prepared under NMR control in such a way that the UVꢀvis
bands stem exclusively from single species involving the chro-
mophore moiety. We note that XHXꢀ and (XHXHX)ꢀ, which
give rise to low-field 1H NMR lines, do not absorb between 300
and 500 nm. The UVꢀvis absorption bands shown in Figure 4
and in SI Figures S5 and S6 arise from the electronic πꢀπ*
transitions of 1, 2, and the heteroconjugated 1:1 complexes
4ꢀ25. A small contribution of 2 was taken into account in order
to obtain the spectra of 26 and 27 as depicted in Figure S6. When
the acidity of the acid component HX is decreased, a continuous
red shift of the πꢀπ* transition is observed. More on the shape of
the UVꢀvis bands is given in the following subsection and in the
Discussion.
Ablue
xblue
¼
,
xred ¼ 1 ꢀ xblue
ð2Þ
Ablue þ Ared
The right column of Figure 4 contains the 13C NMR signals of
the 2-chloro-4-nitrophenolate/phenol moiety labeled with 13C in
the 1-C position, characterized by the chemical shift δ(COHX).
13C NMR signals at natural abundance are not shown. The
full set of 13C NMR spectra is included in SI Figure S4. In the
low-field part of 13C NMR spectra (Figure 4, right), only the
signals of the 1-C carbons of the 2-chloro-4-nitrophenolate/
Here, Ablue and Ared are the integrated band intensities. In the
case of the dual bands, the blue band dominates when a transition
occurs at high energies, and the red band dominates when the
transition occurs at low energies. The examples shown in Figure 4
hold for the full set of UVꢀvis spectra as depicted in SI Figures
S5 and S6. All parameters of the UV band shape analyses are
included in Table 1.
7901
dx.doi.org/10.1021/ja201113a |J. Am. Chem. Soc. 2011, 133, 7897–7908

Journal of the American Chemical Society
ARTICLE
Table 1. Experimental 1H and 13C NMR Chemical Shifts and Parameters of the UVꢀVis Absorption Bands of 1ꢀ27 in CD2Cl2 at
175 Ka
b
c
d
e
HX
δ(COHX) δ(COHX) δ(CODX) ν~COG Δν~ xblue ν~blue Δν~blue ν~red Δν~red
q1
q2
rOH ΔrOH
1
2
3
ꢀ
ꢀ
6.70
155.9
172.9
165.5
33220 6000
23540 2660
1
0
ꢀ0.540 2.977 0.948 0.05
ꢀ
2-chloro-4-nitrophenol
18.54
165.40
27360 5410 0.28 30580 5310 26300 3850 ꢀ0.143 2.455 1.084 0.10
0.143 2.455 1.370 0.15
4
5
6
7
8
9
hydrogen tetrafluoroborate
hydrogen iodide
8.96
9.81
157.45
157.91
158.60
159.22
159.93
160.77
161.48
162.06
157.32
157.76
158.43
159.02
159.60
160.28
160.76
161.15
32230 5720
31730 5650
31570 5590
31160 5460
30690 5340
30200 5210
29880 5040
1
1
1
1
1
1
1
32230
31730
31570
31160
30690
30200
29880
ꢀ0.440 2.805 0.962 0.07
ꢀ0.410 2.757 0.968 0.08
ꢀ0.371 2.697 0.977 0.08
ꢀ0.333 2.643 0.989 0.09
ꢀ0.266 2.561 1.014 0.10
ꢀ0.236 2.528 1.028 0.11
ꢀ0.204 2.499 1.045 0.11
hydrogen bromide
hydrogen chloride
trifluoroacetic acid
dichloroacetic acid
11.04
12.29
14.65
15.70
16.74
17.35
10 2,3,5-trichlorobenzoic acid
11 chloroacetic acid
29290 5330 0.81 29880 5110 27190 4190 ꢀ0.206 2.500 1.044 0.12
0.025 2.416 1.233 0.14
12 3,5-dichlorobenzoic acid
13 formic acid
18.26
18.70
19.02
19.18
18.76
19.11
18.68
18.43
18.16
17.98
162.84
163.84
164.61
165.74
166.06
166.37
166.68
167.74
168.15
168.43
161.79
162.90
163.84
165.64
166.23
166.66
167.20
168.64
28890 5530 0.76 29720 5070 26650 4050 ꢀ0.193 2.490 1.052 0.12
0.076 2.426 1.289 0.14
28270 6040 0.65 29490 4990 26420 3900 ꢀ0.173 2.474 1.064 0.12
0.100 2.434 1.317 0.15
14 4-chlorobenzoic acid
15 benzoic acid
27930 6240 0.58 29430 4940 26200 3820 ꢀ0.167 2.470 1.068 0.12
0.124 2.445 1.346 0.15
27530 6440 0.49 29370 4950 26050 3750 ꢀ0.162 2.467 1.071 0.12
0.141 2.454 1.368 0.15
16 4-chlorophenylacetic acid
17 4-tert-butylbenzoic acid
18 phenylacetic acid
19 3-phenylpropionic acid
20 acetic acid
27260 6200 0.44 29300 4920 25930 3700 ꢀ0.156 2.463 1.075 0.12
0.154 2.462 1.385 0.15
27030 5880 0.42 29020 4830 25810 3660 ꢀ0.132 2.449 1.092 0.13
0.168 2.471 1.403 0.15
26890 5750 0.40 28970 4820 25730 3620 ꢀ0.129 2.447 1.095 0.13
0.179 2.479 1.418 0.15
26390 4810 0.29 28900 4800 25550 3550 ꢀ0.123 2.444 1.099 0.13
0.202 2.497 1.450 0.16
26310 4600 0.26 29040 4840 25500 3530 ꢀ0.134 2.450 1.091 0.13
0.208 2.502 1.459 0.16
21 4-phenylbutyric acid
169.42
25910 4080 0.18 28730 4740 25390 3480 ꢀ0.108 2.437 1.110 0.13
0.224 2.517 1.482 0.16
22 pivalic acid
17.47
16.53
16.49
15.40
13.70
12.86
169.05
169.36
169.48
170.08
170.73
171.20
169.93
170.04
170.16
170.48
170.92
171.41
25450 3750
25590 3590
0
0
25450
25590
0.181 2.480 1.421 0.20
0.211 2.505 1.463 0.16
0.212 2.506 1.465
23 2,4,5-trichlorophenol
24 4-nitrophenol
25 2,4-dichlorophenol
26 4-chlorophenol
27 4-methylphenol
25321 3470
24911 3110
24836 3190
0
0
0
25321
24911
24836
0.244 2.536 1.512 0.16
0.293 2.592 1.589 0.12
0.317 2.622 1.628 0.15
a Chemical shifts δ in ppm. Distances q1 and rOH in Å. Wave numbers in cmꢀ1. ν~COG, center of gravity of single or dual bands; ν~blue and ν~red, centers of
gravity of the dual bands; Δν~COG, experimental bandwidths; Δν~blue and Δν~red, corresponding bandwidths calculated using the log normal function of
eq 1. b Single values of q1 were calculated from δ(COHX) using eq 4. Pairs of q1 values: the upper (lower) values were obtained by fitting ν~blue (ν~red) to
the dashed correlation curve in Figure 9b. c Values corresponding to the dotted curve in Figure 9a. d Calculated from the dotted curve in Figure 9a.
e Width of the distribution of O H distances, estimated according to Figure 9c.
3 3 3
’ DISCUSSION
C represents the carbon position C-1 of AH. The experiments
were performed using CD2Cl2 as solvent around 175 K. Indeed,
we find typical changes in the NMR as well as in the UV spectra,
which together provide novel information about the proton-
transfer pathways, as will be discussed in the following. In
particular, in the case of complexes AHXꢀ exhibiting strong
hydrogen bonds, we observe dual UVꢀvis bands (Figure 4)
indicating the presence of two tautomeric forms, whereas the
NMR spectra provide NMR parameters averaged over both forms.
The purpose of the present study was to elucidate how the UV
absorption frequencies and important NMR parameters of
hydrogen-bonded 1:1 complexes AHXꢀ of the chromophore
AH = 2-chloro-4-nitrophenol (1) with various anions
Xꢀ (Figure 2) behave when H is shifted away from the phenol
1
oxygen to the anion. In particular, we have studied the H and
13C chemical shifts as well as the H/D isotope effects on the 13
C
chemical shifts of the hydrogen bridges COHX of AHXꢀ, where
7902
dx.doi.org/10.1021/ja201113a |J. Am. Chem. Soc. 2011, 133, 7897–7908

Journal of the American Chemical Society
ARTICLE
Figure 6. Experimental full width at half-height of πꢀπ* absorption
band of neutral 1, its anion 2, and conjugate complexes of 3ꢀ27,
measured in CD2Cl2 solution at 170ꢀ180 K, plotted versus the center of
gravity of the whole band. The dashed line represents a second-order
polynomial fit to the data of species 1, 2, 4ꢀ10, 22ꢀ27 that follow a
monotonic decrease in bandwidth against decreasing transition energy.
NMR Parameter Correlations. Let us first discuss the NMR
parameters of AHXꢀ. In Figure 7a the chemical shifts δ(COHX)
of the hydrogen-bonded protons are plotted as a function of
the 13C chemical shift δ(COHX) of the phenolic carbon C-1.
δ(COHX) provides a convenient measure of the proton posi-
tion. The data points of the complexes with HX = benzoic acids
are symbolized by triangles, those with HX = phenols by circles,
and those with the remaining saturated carboxylic and inorganic
acids as well as neutral 1 by squares. These symbols will be used
also in the following figures.
Figure 7. Experimental NMR parameters of 2-chloro-4-nitrophenol
(1) and its conjugated anions 3ꢀ27 (see Table 1) plotted as a function
of the 13C chemical shift δ(COHX) of C-1 of 1. (a) 1H chemical shift of
bridging proton δ(COHX). (b) H/D isotope effect δ(CODX) ꢀ
δ(COHX). Triangles refer to complexes where the partner of 1 is a
benzoate, circles where the partner is a phenolate. In other cases the
symbol is a square. The vertical dashed line corresponds to the anion 2.
The vertical dashed line represents the free anion 2, providing
the limiting 13C chemical shift in the absence of a mobile proton.
An excellent bell-shaped correlation is observed. Neutral 1
changing sign when the bell-shaped correlation curve of
Figure 7a exhibits its maximum. These findings indicate that
δ(COHX) increases monotonically when H is shifted away from
the phenolic oxygen. Whereas different chemical substitution
does not seem to play a major role, we note that it is important to
compare only data observed using the same solvent and similar
temperatures. Finally, we note that similar dispersion-shaped
curves have been observed previously in the case of H/D isotope
effects on heavy-atom chemical shifts of various hydrogen-
bonded complexes of carboxylic acids.16
An additional correlation is observed between the values of
δ(COHX) and the chemical shifts of the protons H-6 of the
2-chloro-4-nitrophenol moiety. This signal had been used in the
adjustment of sample composition as described in the Experi-
mental Section. For further details we refer the reader to SI
Figure S7 and the accompanying text.
UVꢀVis/NMR Correlation. In Figure 8a we have plotted the
UVꢀvis band frequencies of the πꢀπ* transitions of the systems
studied as a function of the corresponding 13C chemical shifts,
δ(COHX). The filled symbols, ν~COG (or λCOG), represent the
experimental centers of gravity of the UVꢀvis bands. We observe
a linear correlation between ν~COG and the chemical shift of the
phenolic carbon δ(COHX). Thus, the values of ν~COG represent
measures for the H-bond geometries as the chemical shifts, where
the averaging is performed in both cases over all possible
1
exhibits the lowest H chemical shift of 7 ppm, corresponding
to the average value of weakly hydrogen-bonded self-associates.
We expect the free 1 to resonate at higher fields in view of the
finding that, for nonsubstituted phenol in the gas phase at 422 K,
a value of 3.77 ppm with respect to gaseous TMS has been
reported.35 Dilution experiments revealed chemical shifts of 4.29
ppm for monomeric phenol and 5.16 ppm for 4-nitrophenol in
carbon tetrachloride.36 For given values of δ(COHX), the
proton chemical shifts δ(COHX) seem to be slightly larger for
complexes with benzoates as partners than for complexes with
carboxylates. We further note that the value of δ(COHX) of the
homoconjugated complex 3 is 18.5 ppm, slightly smaller than
those of several heteroconjugated complexes; the maximum
chemical shift of 19.2 ppm is obtained for 15. These values are,
in turn, smaller than the ones of ∼20ꢀ21 ppm obtained for the
strongest OHO hydrogen bonds9 and indicate the absence of
extremely strong symmetric hydrogen bonds. Below it will be
shown that these findings are consistent with the occurrence of
double bands in UVꢀvis.
In Figure 7b, the secondary H/D isotope effect on carbon
chemical shift, δ(CODX) ꢀ δ(COHX), is plotted as a function
of δ(COHX). A dispersion-shaped correlation is observed,
7903
dx.doi.org/10.1021/ja201113a |J. Am. Chem. Soc. 2011, 133, 7897–7908

Journal of the American Chemical Society
ARTICLE
Figure 8. NMR and UVꢀvis parameters of the πꢀπ* transition of
2-chloro-4-nitrophenol (1), its anion 2, and its conjugated anions 3ꢀ27
(see Table 1). (a) Center of gravity of the experimental absorption band
(solid symbol) and centers of gravity of log-normal fit functions (open
symbols) obtained as explained in the text versus the 13C chemical shift
of C-1. (b) Fraction xred of integral area of the “red band” fit function
(for more details see text).
tautomers, potential wells, or solvent configurations. However,
whereas the averaging is dynamic in the case of NMR, the
UVꢀvis bands contain information about the distribution of
H-bond geometries.
Figure 9. (a) Hydrogen bond correlation q2 vs q1 of OHO hydrogen
bonds in the solid state, adapted from ref 9. The solid line refers to
equilibrium geometries. The dotted line represents an empirical correc-
tion for anharmonic zero-point vibrations. (b) Center of gravity of the
experimental absorption band plotted versus the H-bond coordinate q1
estimated using eq 5. The data points corresponding to the UVꢀvis
spectra for which fitting with a single log-normal band (solid symbols)
and two log-normal bands (open symbols) was done are plotted. The
meaning of the shadowed areas is explained in the text. (c) Center of
gravity of the experimental absorption band plotted versus the rOH
distance in a similar way as in (b). Error bars illustrate the width of
inhomogeneous distribution of rOH estimated from the experimental
UVꢀvis bandwidths.
When H is close to A or close to X, single UVꢀvis bands are
observed, depending on the relative basicities of Aꢀ and of Xꢀ.
Thus, in these cases only single wells of the type AꢀH Xꢀ or
3 3 3
Aꢀ HꢀX are observable. By contrast, at intermediate basicities,
3 3 3
strong and short hydrogen bonds are formed, as indicated by the
1
low-field H NMR shifts; in this region, dual UVꢀvis bands are
observed. The centers of gravity of each sub-band, ~νblue of the blue
component and~νred oftheredcomponent, aredepictedinFigure8a
by open symbols. We note that in the dual-band region~νblue and~νred
are not constant but decrease slightly when δ(COHX) is increased.
The fraction xred = 1 ꢀ xblue of the red components obtained
by band shape analysis is depicted in Figure 8b as a function of
δ(COHX). When H is close to A, xred is zero but increases
monotonically to a value of 1 in the dual-band region. Later, we
will interpret these findings in terms of an equilibrium between
two tautomeric forms, AꢀH Xꢀ and Aꢀ HꢀX.
the data of the heteroconjugated complexes. On the other hand,
the center of gravity of the experimental band, ν~COG, represented
by the solid circle located centrally in Figure 8a does not differ
substantially from the linear correlation with δ(COHX).
As illustrated in SI Figure S8 (see also Table S2), using the
frequencies of the absorption maxima instead of centers of
gravity would lead to similar results besides a constant red-shift
3 3 3
3 3 3
The only data point that deviates somewhat from the correla-
tions in Figure 8 arises from the homoconjugated anion 3, in
which the red band component dominates, and where the blue
band appears at a higher energy than expected by extrapolation of
of about 500 cmꢀ1
.
7904
dx.doi.org/10.1021/ja201113a |J. Am. Chem. Soc. 2011, 133, 7897–7908

Journal of the American Chemical Society
ARTICLE
UVꢀVis/Hydrogen Bond Correlation. In order to interpret
the spectroscopic changes of AHXꢀ upon increasing the basicity
of Xꢀ in terms of the proton-transfer pathway from A to X, we
use H-bond correlations established previously using neutron
and NMR crystallography as well as computational methods.3ꢀ9
The main result of these correlations is that the two distances,
rAH and rHX, of a hydrogen bridge or their combinations are
correlated with each other.
band positions are less affected by changes of the proton position.
The other features of Figure 9c are similar to those of Figure 9b,
besides the vertical and horizontal “distribution bars” which we
have included for some of the data points arising from the dual-
band region. These bars will be discussed below in more detail.
The region of the dual bands which most likely indicate the
presence of two tautomers was analyzed as follows. The “blue”
(“red”) band arises from the tautomer with the hydrogen-bonded
proton closer to (farther away from) the phenol chromophore.
We used then eq 5 and the equations of ref 9 to calculate the
values of q1 and of rOH of the tautomer pairs from the values of
ν~blue and ν~red. The resulting data points are shown in Figure 9b,c
by open circles, located exactly on the dashed correlation curve.
All data are included in Table 1. The tautomer pairs are identified
in a couple of cases in Figure 9b,c by broken lines.
q1 ¼ ðrAH ꢀ rHXÞ=2, q2 ¼ rAH þ rHX
ð3Þ
In the case of linear hydrogen bonds, the proton coordinate q1
(in Å) represents the distance of H from the H-bond center and
q2 (in Å) the heavy-atom distance. The correlation is depicted in
Figure 9a for the case of OHO hydrogen bonds.9 When H is
transferred through the H-bond center, i.e., q1 is changed from
negative to positive values, q2 is decreased and increased again,
exhibiting a minimum when H is located in the H-bond center at
q1 = 0. The solid line refers to equilibrium structures where H is
considered as a point in space. The dotted line refers to a
correction for the shape of the proton. Due to anharmonic
zero-point vibrations, the volume of the proton is not zero, and
hence the shortest heavy -atom distance cannot be achieved. For
details of the mathematical relations between q1 and q2, the
reader is referred to ref 9.
We note that dual bands, i.e., tautomerism, occur in the range
ꢀ0.2 < q1 < þ0.2 Å or 1.05 < rOH< 1.50 Å, as symbolized in
Figure 9b,c by the diagonally hatched areas. On the other hand, a
region where H-bond geometries are excluded is illustrated by
the horizontally hatched areas ꢀ0.1< q1 < þ0.05 Å or 1.1 < rOH
<
1.25 Å.
UV Bandwidth and Distribution of Hydrogen Bond Geo-
metries. Before we discuss the implications of these findings, let
us first discuss how a distribution of H-bond geometries will
influence the UVꢀvis bands. A distribution arises from a number
of different solvent configurations, as has been suggested pre-
viously for a series of neutral complexes with OHN hydrogen
bonds.19 The latter have also been called “solvatomers”.20 As the
UV band frequencies depend on the H-bond geometries, a
distribution of the latter should lead to inhomogeneously
broadened UVꢀvis bands if the interconversion of the different
configurations is slow on the UV time scale. Indeed, the observed
decrease of the bandwidths Δν~ depicted in Figure 6 can be
consistently explained in this way.
For that purpose we have introduced in Figure 9c vertical and
horizontal “distribution bars” for some of the data points arising
from the dual-band region. In order to derive these bars we
proceeded as follows. We noticed that the experimental band-
width Δν~(2) of the anion 2 is particularly small, as illustrated in
Figure 4h, which could arise from the fact that there is no
hydrogen-bonded proton close to the chromophore. Hence,
Δν~(2)is not affected by H-bond geometries. Thus, we assume
that the contribution Δν~d(i) for complex i from the H-bond
distribution can be estimated from the difference
Using solid-state NMR, a simple relation between 1H chemical
shifts and q1 of OHO hydrogen bonds has been proposed which
we use to estimate average H-bond geometries of complexes
studied here,9
2
δðCO H XÞ ¼ δo þ Δ expð ꢀ 6:2q1 Þ
ð4Þ
where the maximum shift is δo þ Δ = 20.5 ppm, independent of
the chemical environment. δo refers to the free proton donor
whose chemical shift depends, on the other hand, on the
chemical environment. Thus, for water and aliphatic hydroxyl
groups, δo ≈ 0.7 ppm, whereas 6 ppm was calculated for free
carboxylic acids. In the cases of the phenols studied here, we
estimate δo = 4 ppm, derived from the value of the free phenol
monomer.35,36
Using the experimental 1H chemical shifts, we have calculated
in a first stage the values of q1 for the systems exhibiting a single
UVꢀvis band. The data points are included in Table 1, together
with those of q2 and rOH. In Figure 9b, we have plotted the
UVꢀvis band positions as a function of the values of the proton
coordinate q1, leading to the data points symbolized by the solid
circles. They indicate a correlation which could be expressed by
the dashed sigmoidal curve corresponding to the empirical
equation
~
~
~
ΔνdðiÞ ¼ ΔνðiÞ ꢀ Δνð2Þ
ð6Þ
where Δν~(i) represents the observed bandwidth of complex i,
assembled in Table 1. Thus, all other broadening mechanisms,
including nonresolved vibronic transitions and phonon side
bands as well as their interconversion by spectral diffusion, are
then included in Δν~(2). Further discussion of this quantity is
beyond the scope of this study. Δν~d(i) values are collected in SI
Table S2.
The values of Δ~νd(i) were introduced into Figure 9c as vertical
distribution bars. The horizontal distribution bars ΔrOH could
then be calculated from the dashed correlation curve. The values
are included in Table 1. The graph shows that a distribution
width ΔrOH leads to a small contribution to the inhomogeneous
broadening of the UV bandwidth when rOH is large, but to a
substantial broadening when rOH is small. From the graph it
follows that the ΔrOH increases slightly when rOH is increased,
although the bandwidth decreases.
ꢀ
ꢀ
ꢁ
ꢁ
q1 þ 0:11
0:22
~
ν ¼ 23500 þ 10500= exp
þ 1
ð5Þ
where ~v is in cmꢀ1 and q1 is in Å. We have included the data
points of neutral 1 and of the anion at infinite negative and
positive values of q1 from which the two wave numbers in eq 5
were estimated. When the proton is far away from the H-bond
center, the band positions do not change with q1, but they change
strongly when the proton goes through the H-bond center. In
Figure 9c we have transformed the abscissa to the distance rOH
.
When H is located near the chromophore, i.e., if the latter is
neutral and rOH small, changes of rOH induce large changes of the
UV band frequencies. By contrast, as expected, when the
chromophore carries the negative charge and rOH is large, the
7905
dx.doi.org/10.1021/ja201113a |J. Am. Chem. Soc. 2011, 133, 7897–7908

Journal of the American Chemical Society
ARTICLE
distances between 2.5 and 2.4 Å, and i.e. to the shortest and
strongest OHO hydrogen bonds. Moreover, the geometric
region without any tautomer is confined to values of q1 between
only about ꢀ0.10 and 0.05 Å, as illustrated by the horizontally
hatched area in Figure 9b. These values are comparable with the
width of the square of the vibrational wave function of the
proton. Neutron structures of strong hydrogen bonds indicate an
even larger “width” of at least 0.5 Å.37 Thus, the tautomerism is
not a normal one as expected for a simple reaction, as was
illustrated in Figure 1b. Usually, the equilibrium between the two
forms is characterized by an equilibrium constant of tautomer-
ism. However, that requires a diabatic potential for the proton
motion, as illustrated schematically in Figure 11a. In this case,
both tautomers exhibit individually different electronic transi-
tions, different H-bond geometries, and well distinct partition
functions. On the other hand, an adiabatic potential may also be
realized. One possible configuration is illustrated in Figure 11b. A
series of adiabatic potentials along the proton-transfer pathway
has been proposed by Hynes et al.38 In this case, because of the
small barrier, the partition functions of both tautomers are
entangled, and the interpretation of the dual-absorption bands
in terms of two different tautomers may break down.
Figure 10. Schematic representation of the proton-transfer pathway
combining geometrical shifts and population distribution in a two-state
approximation found in this work for conjugate anions of 2-chloro-4-
nitrophenol. Negative charges have been omitted.
In all cases, solvent reorientation will play an important role.
Thus, it is conceivable that a given tautomer is associated with a
range of given solvent configurations, as illustrated schematically
in Figure 11c. Around the moiety exhibiting the negative charge,
e.g., Xꢀ, the countercation Kþ will be located and the solvent
dipoles will exhibit a higher ordering than around the neutral
chromophore AH. The opposite will be true if the chromophore
carries the negative charge. In the wording proposed by Perrin
et al.,20 each tautomer will exhibit a given subset of “solvatomers”,
leading to a diabatic potential for the tautomerism.
The interpretation of our findings in terms of a two-state
tautomerism is not in disagreement with H-bond correlations
described by some of us previously, which have been applied not
only to solids but also to liquids.7ꢀ9 In the latter case, the
distances used in the correlations represent averages over all
solvent configurations and possible tautomers. On the other
hand, our findings support the finding of Perrin et al.,20 who
showed using isotope NMR methods that the symmetry of
potentially symmetric hydrogen bonds is lifted in the liquid state.
This leads us to the problem of the homoconjugate anion 3
(AHAꢀ). From the 1H chemical shift correlation of eq 4, we obtain
for the proton chemical shift of 18.5 ppm the values q1 = ( 0.143 Å,
which indicates the presence of a degenerate tautomerism between
two forms or a symmetric double well. This is confirmed by the
observation of a dual band. Whereas ~νred is located near the dashed
correlation curve in Figure 9b, the value of ~νblue is substantially larger
than predicted. This means that the two chromophores interact
strongly with each other, despite their significant nonequivalence.
Further theoretical studies will be necessary to describe the com-
bined UVꢀvis/NMR results of this system in the future.
Figure 11. (a) Diabatic potential for the tautomerism in a strong
hydrogen bond. (b) Adiabatic double-well potential. (c) Schematic
illustration of the different organization of the countercation Kþ and
the solvent dipoles around a negatively charged hydrogen-bonded anion
exhibiting a two-state tautomerism.
In conclusion, we can understand qualitatively the monotonic
increase of the bandwidth in Figure 4 when the basicity of the
accepting anion is decreased. Moreover, neutral 1 exhibits a
particularly large bandwidth. This can then be explained in terms
of a distribution of hydrogen-bonded associates containing a
varying number of molecules, resulting in slightly different
distances rOH
.
Proton-Transfer Pathways Elucidated by UVNMR. The
results described above are summarized schematically in Fig-
ure 10. When H is located near A, i.e., the acidity of AH or the
basicity of X small, a single tautomer is formed (Figure 10, top).
When the acidity of AH or the basicity of X is increased, H is
shifted toward the H-bond center and the hydrogen bond is
compressed, i.e., the A X distance decreases. At a certain
3 3 3
point, a second tautomer appears (Figure 10, center) which
becomes more and more populated. The bridging proton posi-
tions in both tautomers shift monotonically toward the acceptor
’ CONCLUSIONS
X. At the same time, the A X distance decreases in the
3 3 3
tautomer AꢀH Xꢀ and increases in Aꢀ HꢀX. Finally,
The main result of this study is the finding that the UVꢀvis
band frequencies of a phenolic chromophore forming OHO
hydrogen bonds with proton acceptors are correlated with
3 3 3
3 3 3
the second tautomer survives alone, but its structure continues to
change in a monotonic way (Figure 10, bottom).
1
averaged NMR parameters such as H and 13C chemical shifts
The geometry range in which the tautomerism occurs is
surprisingly small and confined to values of q1 between about
ꢀ0.2 and 0.2 Å, as illustrated by the diagonally hatched area in
and H/D isotope effects on the latter. As in previous work,7,9,16 in
which some of us had established correlations of NMR para-
meters with H-bond geometries, the new results indicate a
Figure 9b. This corresponds to values of q2, i.e. to A
X
3 3 3
7906
dx.doi.org/10.1021/ja201113a |J. Am. Chem. Soc. 2011, 133, 7897–7908

Journal of the American Chemical Society
ARTICLE
correlation of the UVꢀvis band frequencies with the O
H
(3) (a) Steiner, Th.; Saenger, W. Acta Crystallogr. 1994, B50, 348–
357. (b) Steiner, Th. J. Chem. Soc., Chem. Commun. 1995, 1331–1332.
(c) Gilli, P.; Bertolasi, V.; Ferretti, V.; Gilli, G. J. Am. Chem. Soc. 1994,
116, 909–915. (d) Steiner, Th. J. Phys. Chem. A 1998, 102, 7041–7052.
(4) (a) Benedict, H.; Limbach, H. H.; Wehlan, M.; Fehlhammer,
W. P.; Golubev, N. S.; Janoschek, R. J. Am. Chem. Soc. 1998, 120, 2939–
2950. (b) Lorente, P.; Shenderovich, I. G.; Golubev, N. S.; Denisov,
G. S.; Buntkowsky, G.; Limbach, H. H. Magn. Reson. Chem. 2001,
39, S18–S29.
(5) Ramos, M.; Alkorta, I.; Elguero, J.; Golubev, N. S.; Denisov,
G. S.; Benedict, H.; Limbach, H. H. J. Phys. Chem. A 1997, 101, 9791–
9800.
(6) Smirnov, S. N.; Golubev, N. S.; Denisov, G. S.; Benedict, H.;
Schah-Mohammedi, P.; Limbach, H. H. J. Am. Chem. Soc. 1996,
118, 4094–4101.
3 3 3
distances, a result which had not previously been recognized. The
advantage of UVꢀvis spectroscopy is that different H-bond
configurations are in slow exchange. Hence, once the samples
have been characterized by NMR, UVꢀvis provides information
about whether the proton motion involves single or double wells
and how H-bond configurations are influenced by the solvent. In
the case of weaker hydrogen bonds, we observed single UV bands
indicating the occurrence of single wells for the proton motion.
By contrast, dual bands were observed for strong and short
hydrogen bonds, indicating the presence of two configurations, i.
e., tautomeric states. As a consequence, the inhomogeneous part
of the UVꢀvis bands provides evidence about the distribution of
H-bond geometries. These findings may be important in the
interpretation of UVꢀvis shifts after irradiation of phenolic
chromophores in proteins such as the photoactive yellow
protein28 with light. A quantitative treatment of the correlations
of the above-mentioned spectroscopic parameters with H-bond
geometry, based on existing quantitative relationships, is pending
and will be published elsewhere. In any case, the experimental
results obtained in this study may require further theoretical studies.
(7) Limbach, H. H.; Denisov, G. S.; Golubev, N. S. In Isotope Effects
in Chemistry and Biology; Kohen, A., Limbach, H. H., Eds.; Taylor
& Francis: Boca Raton, FL, 2006; pp 193ꢀ230.
(8) Tolstoy, P. M.; Schah-Mohammedi, P.; Smirnov, S. N.; Golubev,
N. S.; Denisov, G. S.; Limbach, H. H. J. Am. Chem. Soc. 2004, 126,
5621–5634.
(9) Limbach, H. H.; Tolstoy, P. M.; Pꢀerez-Hernꢀandez, N.; Guo, J.;
Shenderovich, I. G.; Denisov, G. S. Isr. J. Chem. 2009, 49, 199–216.
(10) (a) Sternberg, U.; Brunner, E. J. Magn. Reson. A 1994, 108,
142–150. (b) Harris, T. K.; Zhao, Q.; Mildvan, A. S. J. Mol. Struct. 2000,
552, 97–109.
’ ASSOCIATED CONTENT
(11) Jeffrey, G. A.; Yeon, Y. Acta Crystallogr. B 1986, B42, 410–413.
(12) Shenderovich, I. G.; Limbach, H. H.; Smirnov, S. N.; Tolstoy,
P. M.; Denisov, G. S.; Golubev, N. S. Phys. Chem. Chem. Phys. 2002,
4, 5488–5497.
(13) Shenderovich, I. G.; Tolstoy, P. M.; Golubev, N. S.; Smirnov,
S. N.; Denisov, G. S.; Limbach, H. H. J. Am. Chem. Soc. 2003, 125,
11710–11720.
(14) Detering, C.; Tolstoy, P. M.; Golubev, N. S.; Denisov, G. S.;
Limbach, H. H. Dokl. Phys. Chem. 2001, 379, 1–4.
(15) Schah-Mohammedi, P.; Shenderovich, I. G.; Detering, C.;
Limbach, H. H.; Tolstoy, P. M.; Smirnov, S. N.; Denisov, G. S.; Golubev,
N. S. J. Am. Chem. Soc. 2000, 122, 12878–12879.
(16) Tolstoy, P.; Guo, J.; Koeppe, B.; Golubev, N.; Denisov, G.;
Smirnov, S.; Limbach, H. H. J. Phys. Chem. A 2010, 114, 10775–10782.
(17) Gu, Z.; Zambrano, R.; McDermott, A. J. Am. Chem. Soc. 1994,
116, 6368–6372.
(18) (a) Sharif, S.; Denisov, G. S.; Toney, M. D.; Limbach, H. H. J.
Am. Chem. Soc. 2007, 129, 6313–6327. (b) Sharif, S.; Denisov, G. S.;
Toney, M. D.; Limbach, H. H. J. Am. Chem. Soc. 2006, 128, 3375–3387.
(c) Sharif, S.; Fogle, E.; Toney, M. D.; Denisov, G. S.; Shenderovich,
I. G.; Tolstoy, P. M.; Chan-Huot, M.; Buntkowsky, G.; Limbach, H. H. J.
Am. Chem. Soc. 2007, 129, 9558–9559.
S
Supporting Information. Sample composition (types A
b
and B) as determined by 1H NMR and ratios n(HX)/n(Xꢀ) for
type A samples; schematics of UVNMR sample tubes; complete
set of low-field parts of 1H NMR, 13C NMR, and UVꢀvis spectra
for complexes 1ꢀ27; wavelengths of the UVꢀvis bands' maxima,
1
centers of gravity, and bandwidths as well as the H NMR
chemical shifts of the ortho protons of the chromophore for
1
complexes 1ꢀ27; and comparison of the position of H NMR
signals of complexes 3ꢀ22 and 24ꢀ27 for type A and type B
samples. This material is available free of charge via the Internet
at http://pubs.acs.org.
’ AUTHOR INFORMATION
Corresponding Author
tolstoy@chemie.fu-berlin.de
Present Addresses
§Max Born Institut f€ur Nichtlineare Optik and Kurzzeitspektroskopie,
Max-Born-Strasse 2A, D-12489 Berlin, Germany
(19) Golubev, N. S.; Denisov, G. S.; Smirnov, S. N.; Shchepkin,
D. N.; Limbach, H. H. Z. Phys. Chem. 1996, 196, 73–84.
(20) Perrin, C. L.; Lau, J. S. J. Am. Chem. Soc. 2006, 128, 11820–
11824.
’ ACKNOWLEDGMENT
We thank Dr. E. T. J. Nibbering, Max-Born Institute, Berlin,
and Prof. G. S. Denisov, St. Petersburg State University, as well as
Prof. Maurice Kreevoy, Minneapolis, for helpful discussions. This
work has been supported by the Deutsche Forschungsge-
meinschaft, Bonn, the Fonds der Chemischen Industrie, Frank-
furt, the Russian Foundation of Basic Research, and the
European Research Council under the European Union’s Se-
venth Framework Program (FP7/2007-2013/ERC grant agree-
ment no. 247051 T.E.).
(21) Perrin, C. L. Acc. Chem. Res. 2010, 43, 1550–1557.
(22) (a) Nagakura, S.; Baba, H. J. Am. Chem. Soc. 1952, 74,
5693–5698. (b) Baba, H.; Suzuki, S. J. Chem. Phys. 1961, 35, 1118–
1127. (c) Baba, H.; Suzuki, S. J. Chem. Phys. 1963, 38, 349–353.
(23) (a) Bell, C. L.; Barrow, G. M. J. Chem. Phys. 1959, 31, 1158–
1163. (b) Baba, H.; Matsuyama, A.; Kokubun, H. J. Chem. Phys. 1964,
41, 895–896. (c) Baba, H.; Matsuyama, A.; Kokubun, H. Spectrochim.
Acta 1969, 25A, 1709–1722. (d) Hudson, R. A.; Scott, R. M.; Vino-
gradov, S. N. J. Phys. Chem. 1972, 76, 1989–1993. (e) Romanowski, H.;
Sobczyk, L. J. Phys. Chem. 1975, 79, 2535–2542.
’ REFERENCES
(24) Kreevoy, M. M.; Liang, T. M. J. Am. Chem. Soc. 1980, 102,
3315–3322.
(1) Isotope Effects in Chemistry and Biology; Kohen, A., Limbach,
H. H., Eds.; Taylor & Francis: Boca Raton, FL, 2006.
(2) Hydrogen Transfer Reactions; Hynes, J. T., Klinman, J.,
Limbach, H. H., Schowen, R. L., Eds.; Wiley-VCH: Weinheim, 2007;
Vols. 1ꢀ4.
(25) (a) Magonski, J. J. Solution Chem. 1990, 19, 597–607.
(b) Magonski, J. J. Phys. Org. Chem. 2002, 15, 204–210.
(26) (a) Pawlak, Z.; Magonksi, J.; Jasinski, T. J. Mol. Struct. 1978,
47, 329–338. (b) Pawlak, Z.; Magonksi, J. J. Mol. Struct. 1980,
7907
dx.doi.org/10.1021/ja201113a |J. Am. Chem. Soc. 2011, 133, 7897–7908

Journal of the American Chemical Society
ARTICLE
60, 179–185. (c) Magonksi, J.; Pawlak, Z. J. Mol. Struct. 1982,
80, 243–250.
(27) Anderson, S.; Crosson, S.; Moffat, K. Acta Crystallogr. D 2004,
60, 1008–1016.
(28) Joshi, C. P.; Otto, H.; Hoersch, D.; Meyer, T. E.; Cusanovich,
M. A.; Heyn, M. P. Biochemistry 2009, 39, 9980–9993.
(29) Tolstoy, P. M.; Koeppe, B.; Denisov, G. S.; Limbach, H. H.
Angew. Chem., Int. Ed. 2009, 48, 5745–5747.
(30) Drehmann, U.; P€urschel, U.; Wauschkun, H. J. Prakt. Chem.
1961, 14, 122–126.
(31) Fanta, P. Org. Synth. 1952, 32, 95–96.
(32) Ermakov, A. A.; Ermakov, S. A.; Polomarchuk, N. I. Russian
Patent RU 268877, 2006.
(33) According to Hopf et al., phenolates can substitute both
chlorines of dichloromethane, leading to di(aryloxy)methanes and
chloride. Here, the reactionmay occur according to 2AHXꢀ þ CD2Cl2 T
CD2A2 þ 2HX þ 2Clꢀ, AHXꢀ þ HX T AHXꢀHX, leading to higher
accociates depending on concentrations and temperature. Hopf, H.;
Utermoehlen, R.; Jones, P. G.; Desvergne, J. P.; Bouas-Laurent, H. J. Org.
Chem. 1992, 57, 5509–5517.
(34) Siano, D. B.; Metzler, D. E. J. Chem. Phys. 1969, 51, 1856–1861.
(35) Chauvel, J. P., Jr.; True, N. S. Chem. Phys. 1985, 95, 435–441.
(36) Guillaume, F.; Sꢀeguin, J. P.; Nadio, L.; Uzan, R. Spectrochim.
Acta A 1982, 38, 661–664.
(37) Steiner, T.; Majerz, I.; Wilson, C. C. Angew. Chem., Int. Ed.
2001, 40, 2651–2654.
(38) (a) Kiefer, P. M.; Hynes, J. T. J. Phys. Chem. A 2002,
106, 1834–1849. (b) Kiefer, P. M.; Hynes, J. T. J. Phys. Chem. A 2002,
106, 1850–1861. (c) Kiefer, P. M.; Hynes, J. T. J. Phys. Chem. A 2003,
107, 9022–9039.
7908
dx.doi.org/10.1021/ja201113a |J. Am. Chem. Soc. 2011, 133, 7897–7908