Low-temperature NMR spectra of fluoride-acetic acid hydrogen-bonded complexes in aprotic polar environment

Article abstract of DOI:10.1016/j.molstruc.2004.01.024

Using low-temperature NMR (¹H, ¹⁹F) technique in the slow exchange regime, the solutions containing tetrabutylammonium (TBA) acetate and HF have been studied in an aprotic freon mixture, CDF₃/CDF ₂Cl, exhibiting a dielectric permittivity, which increases strongly by lowering the temperature. Two different hydrogen bonded anionic clusters, a 1:1 cluster of the type AcO^δ-?H?F ^-1+δ+ ([AcOHF]^-) and a 2:1 cluster of the type AcOH?F^-?HOAc ([(AcOH)₂F]^-) have been detected in equilibrium with each other, both forming ion pairs with the TBA countercation. [AcOHF]^- exhibits an extremely strong hydrogen bond, with a proton shared between partially negatively charged oxygen and fluorine atoms. The NMR chemical shifts and scalar spin-spin coupling constants, ¹J(FH), have been measured in the temperature range between 110 and 160 K, where separate NMR signals are observed for both species. In addition, H/D isotope effects on the ¹⁹F NMR chemical shielding have been measured for both clusters. In contrast to the related complexes [(FH) _nF]^- (n=1-4) studied previously, the NMR parameters of [AcOHF]^- and of [(AcOH)₂F]^- depend strongly on temperature. This effect is associated with the increasing polarity of the solvent with decreasing temperatures, established earlier, which displaces the proton from fluorine to oxygen. As a motive power of this conversion, preferential solvation of the compact fluoride ion as compared to acetate is proposed.

Full text of DOI:10.1016/j.molstruc.2004.01.024

Journal of Molecular Structure 700 (2004) 3–12
www.elsevier.com/locate/molstruc
Low-temperature NMR spectra of fluoride–acetic acid hydrogen-bonded
complexes in aprotic polar environment
Nikolai S. Golubeva, Peter M. Tolstoya,b, Sergei N. Smirnova, Gleb S. Denisova,
Hans-Heinrich Limbachb,
*
aV.A. Fock Institute of Physics, St Petersburg University, Ulyanovskaja 1, Peterhof, Russian Federation
bInstitut fuer Chemie der Freien Universitaet Berlin, Takustrasse 3, D14195 Berlin, Germany
Received 20 January 2004; revised 20 January 2004; accepted 21 January 2004
Abstract
Using low-temperature NMR (1H, 19F) technique in the slow exchange regime, the solutions containing tetrabutylammonium (TBA)
acetate and HF have been studied in an aprotic freon mixture, CDF3/CDF2Cl, exhibiting a dielectric permittivity, which increases strongly by
lowering the temperature. Two different hydrogen bonded anionic clusters, a 1:1 cluster of the type AcOd2· · ·H· · ·F21 þ dþ ([AcOHF]2) and
a 2:1 cluster of the type AcOH· · ·F2· · ·HOAc ([(AcOH)2F]2) have been detected in equilibrium with each other, both forming ion pairs with
the TBA countercation. [AcOHF]2 exhibits an extremely strong hydrogen bond, with a proton shared between partially negatively charged
oxygen and fluorine atoms. The NMR chemical shifts and scalar spin–spin coupling constants, 1JðFHÞ; have been measured in the
temperature range between 110 and 160 K, where separate NMR signals are observed for both species. In addition, H/D isotope effects on the
19F NMR chemical shielding have been measured for both clusters.
In contrast to the related complexes [(FH)nF]2 ðn ¼ 1–4Þ studied previously, the NMR parameters of [AcOHF]2 and of [(AcOH)2F]2
depend strongly on temperature. This effect is associated with the increasing polarity of the solvent with decreasing temperatures, established
earlier, which displaces the proton from fluorine to oxygen. As a motive power of this conversion, preferential solvation of the compact
fluoride ion as compared to acetate is proposed.
q 2004 Elsevier B.V. All rights reserved.
Keywords: Low-temperature NMR; Hydrogen bond; Isotope effect; Acetic acid; Fluoride anion
1. Introduction
experimental and theoretical results are in favour of solvent
assisted hydrogen bond weakening, lengthening and
asymmetrization, which makes the participation of such
bonds in enzymic processes doubtful [12]. On the other
hand, in some recent studies [13–15] it was shown that
hydrogen bonds in homoconjugated anions are strengthened
and contracted by embedding into a polar medium. This
finding is in agreement, in particular, with the increase of
the scalar 2hJðFHFÞ spin coupling constant between fluorine
nuclei across the hydrogen bonds with increasing polarity of
the solvent for the dihydrogen trifluoride ion, which exhibits
two coupled short hydrogen bonds of the type
[F–H· · ·F· · ·H–F]2 [16,17]. A simple explanation of this
effect was proposed, i.e. that the electrostatic repulsion
between negatively charged fluorine atoms is reduced by a
medium of high dielectric constant 1 [15].
Strong and short (‘low-barrier’) hydrogen bonds in
charged (anionic or cationic) complexes (clusters) have
attracted recently the attention of many biochemists due to
presumed role of these bonds in enzymic catalysis [1–11].
One of the main questions when invoking the concept of
low-barrier hydrogen bonding to biochemical reactions is
how these bonds are affected by the surrounding water
molecules or other polar groups. In fact, the problem of how
solvent polarity influences the length, strength and sym-
metry of a short hydrogen bond in charged complex
has discussed in the literature for a long time. Some
*
Corresponding author. Tel.: þ49-30-8385-5375; fax: þ49-30-8385-
5310.
E-mail addresses: limbach@chemie.fu-berlin.de (H.-H. Limbach),
nick.golubev@pobox.spbu.ru (N.S. Golubev), tolstoy@chemie.fu-berlin.
de (P.M. Tolstoy).
Taking in mind, that a solvent may act on H-bonds of
different type in different ways, in this communication we
0022-2860/$ - see front matter q 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.molstruc.2004.01.024

4
N.S. Golubev et al. / Journal of Molecular Structure 700 (2004) 3–12
describe the influence of the solvent polarity on the NMR
parameters of ionic clusters with strongly asymmetric
hydrogen bonds, formed by hydrogen fluoride with the
acetate-anion, in ion pairs with the ‘inert’ tetrabutylam-
monium (TBA) counter cation. As described in our
previous papers [18–20], we took advantage of a very
strong dependence of 1 of the low-freezing solvent CDF3/
CDF2Cl on temperature. According to Ref. [18], 1 of this
mixture rises rapidly from about 25 to 45 when
temperature is decreased from 170 to 100 K. This makes
it possible to study the dependence of the NMR
parameters as a function of 1 without changing the
solvent. At temperatures as low as 100–170 K, the slow
hydrogen bond exchange regime is usually reached where
NMR signals of different acid–base complexes being in
equilibrium with each other are well resolved, a
circumstance which increases considerably the informa-
tive power of liquid state NMR. As a result of this work,
we succeeded in finding and assigning NMR features of
several hydrogen-bonded species being present in the
solutions in chemical equilibrium, in the slow hydrogen
bond exchange regime. In this communication, the spectra
of two of them, namely, AcOd2· · ·H· · ·F21 þ dþ
([AcOHF]2) and AcOH· · ·F2· · ·HOAc [(AcOH)2F]2 are
described and characterized.
chemical shifts were measured using fluoroform, CHF3
(CDF3), as internal standard, and were converted to the
conventional TMS (tetramethylsilane) scale. The total
deuterium fraction was determined by comparison of the
integrated signal intensity of the residual mobile proton site
with those of the immobile CH-protons. It should be noted
that, as a rule, the resolution in our low-temperature spectra
is much worse than in usual (ambient temperature) liquid
state spectra. It is explained by a considerable viscosity of
freon mixtures, and is markedly revealed in spectra of nuclei
with high magnetic shielding anisotropy, like 19F.
3. Hydrogen bond correlations
In order to link the NMR parameters of the hydrogen
bonded FHO and OHFHO systems studied to hydrogen
bond geometries we will make use of several hydrogen bond
correlations, which have been established previously by
various authors.
To a given hydrogen bond of the type A–H· · ·B one
can normally associate two distances, r1 for the diatomic
unit AH and r2 for the diatomic unit H· · ·B. According to
Pauling [21], these distances correspond to the valence
bond orders
p1 ¼ exp{ 2 ðr1 2 r10Þ=b1};
p2 ¼ exp{ 2 ðr2 2 r20Þ=b2}; with p1 þ p2 ¼ 1:
ð1Þ
2. Experimental
For H, the total valence has to be unity. The correlation
of r1 with r2 also implies a correlation between the
hydrogen bond coordinates q1 ¼ 1=2ðr1 2 r2Þ and q2 ¼
r1 þ r2: For a linear hydrogen bond, q1 represents the
distance of H from the hydrogen bond center and q2 the
distance between atoms A and B. When H is transferred
from one heavy atom to the other, q1 increases from
negative values to positive values. q2 goes through a
minimum which is located at q1 ¼ 0 for AHA and near 0
for AHB systems. The parameters in Eq. (1) have been
derived empirically from various experiments such as
neutron diffraction or solid state NMR and the results of
quantum-mechanical calculations. For the FH and
OH bonds, the following parameters have been proposed
[17,19], i.e.
2.1. Materials
A measured amount of 40% hydrofluoric acid was added
to a solution of TBA acetate (Aldrich) in dichloromethane
(Aldrich). Then water was removed by repeated azeotropic
distillation with dry dichloromethane. The H/D replacement
of the mobile (OH, FH) proton was performed by dissolving
a salt in methanol–OD: proton/deuterium exchange
between acetic acid–fluoride anion and excess of methanol
leads to the partial substitution of H for D in the salt.
Methanol was then removed under reduced pressure,
leaving in the deuterated salt in the flask. The oily product
was placed into a hermetic thick-walled NMR sample
tube equipped with PTFE (polytetrafluoroethylene) valves
(Wilmad, Bueno), connected to a high vacuum line and
filled with the solvent via vacuum transfer. The solvent,
CDF3/CDF2Cl, with freezing point below 100 K was added
to the samples by vacuum transfer. The approximate overall
concentration of the samples, estimated by weighing
chemicals and measuring the volume of the solution at
low temperatures (around 120 K), was about 0.02 M.
0
ꢀ
ꢀ
r1 ¼ 0:897 A and b1 ¼ 0:360 A for HF and
0
ꢀ
ꢀ
r2 ¼ 0:942 A; b2 ¼ 0:371 A for OH:
ð2Þ
r10 and r20 refer to the values of the hypothetical diatomic
molecules. b1 and b2 describe the decrease of the bond
orders with increasing distances.
For the correlation of HF coupling constants in FHB
systems, some of us have proposed the relation [20]
2.2. NMR spectra
0
1JðFHBÞ ¼ 1JðFHÞ pFH 2 8DJðFHBÞpFHp2HB
;
The Bruker AMX-500 NMR spectrometer used was
equipped with a low-temperature probe-head which allowed
us to perform experiments down to 100 K. 1H and 19F NMR
0
1JðFHÞ ¼ 600 Hz;
ð3Þ

N.S. Golubev et al. / Journal of Molecular Structure 700 (2004) 3–12
5
with DJðFHBÞ ¼ 162:5 Hz for B ¼ F and DJðFHBÞ ¼ 190
Hz for B ¼ N. We will assume later that the average value is
valid for B ¼ O, DJðFHOÞ ¼ 176 Hz:
The parameters A and B are empirical and will depend on
the type of the hydrogen bond studied. We will use here the
values A ¼ 220 and B ¼ 5 for the FHF and the FHO
hydrogen bonded systems considered in this work. For a
given set of values pFH and pOH one can calculate using Eq.
(6) the corresponding values of pFD and pOD and hence the
fluorine chemical shifts of the deuterated systems. For this
purpose, an appropriate simple computer program was
written.
1
For the analysis of the H chemical shifts we use the
relation
0
0
dðFHÞ ¼ dðFHÞ pFH þdðHBÞ pHB þ4DdðFHBÞpFHpHB ð4Þ
with DdðFHBÞ ¼ 15:2ppm: For B ¼ F, we assume a value
0
0
of dðFHÞ ¼ dðHFÞ ¼ 0 ppm; and for B ¼ O that dðH
0
OÞ ¼ 6ppm; referring to free HF and acetic acid molecules.
For the fluorine chemical shifts we assume that
4. Results
0
dF ¼ dðFHÞ pFH þ dF1pHB þ 8DdðFHBÞpFHp2HB
ð5Þ
4.1. NMR spectra at 120 K
Here, we set for B ¼ O DdðFHOÞ ¼ 100 ppm; dðFÞ1
¼
1
0
In Fig. 1 are depicted the low-field H NMR signals of
2100 ppm and dðFHÞ ¼ 2210 ppm:
solutions TBAþ acetate (AcO2) with hydrogen fluoride,
with acetate/HF ratios of 3:1 (top) and 3:2 (bottom) and the
deuterium fractions xD ¼ 0 (top) and xD ¼ 0:7 (bottom) in
the hydrogen bonded sites.
We note that these parameters are not ‘hard’ parameters
but may change, when further experimental and theoretical
results will be available. Therefore, one has to keep in mind
that geometries derived using these equations from NMR
parameters might be subject to systematic errors. However,
they are sufficient for qualitative comparison of different H-
bonded systems.
The upper spectrum exhibits two intense singlets at
19.3 and 16.5 ppm, which are typical for the acetic acid–
acetate homoconjugated anions labeled here as [OHO]2
and [OHOCOHO]2. These anions have been character-
ized recently; they are subject to a fast acetate–acetic
acid interconversion as revealed by 13C NMR [19]. In
addition, we find two doublets, exhibiting chemical shifts
of 18.33 and 15.58 ppm and spin–spin coupling con-
stants 1JðFHÞ near 60 Hz at 120 K. These signals were
assigned to the anions [AcOHF]2 ; [FHO]2 and
[(AcOH)2F]2 ; [OHFHO]2 as indicated in Fig. 1. As
shown in the lower spectrum, two new signals appear at
Finally, we assume the following relations between the
valence bond orders of the deuterated and the protonated
hydrogen bonds
Dp ¼pFH 2 pFD 2 ðpOH 2 pODÞ ¼
B
AðpFHpOHÞ ðpFH 2 pOHÞ;
pFH þ pOH ¼ pFD þ pOD ¼ 1:
ð6Þ
Fig. 1. Low-field part of 1H NMR spectra of solutions of tetrabutylammonium acetate and hydrogen fluoride in CDF3/CDF2Cl at 120 K. Top: molar ratio 3:1,
deuterium fraction in the hydrogen bond sites xD ¼ 0: Bottom: molar ratio 3:2, xD ¼ 0:7:

6
N.S. Golubev et al. / Journal of Molecular Structure 700 (2004) 3–12
Fig. 2. 19F NMR signals of solutions of tetrabutylammonium acetate and hydrogen fluoride (molar ratio 3:2) in CDF3/CDF2Cl at 120 K. Left: signals of
[OHFHO]2, L ¼ H, D at xD ¼ 0:73: Right: signals of [FHO]2 at xD ¼ 0:53:
16.8 and 15.84 ppm after partial deuteration, assigned to
the complexes [OHOCODO]2 and [OHFDO]2.
The assignment of [OHO]2 and [OHOCOHO]2 was
based on the following observations: (i) a decrease of the
acetate/HF ratio reduces the intensity of the low-field signal,
and that of the high-field one is increased; (ii) partial H/D
replacement does not alter the type of the low-field signal,
but produces a new peak for [OHOCODO]2 as mentioned
above. This isotope shift has been labeled as a geometric
‘vicinal’ isotope effect [19].
4.2. Variable temperature NMR spectra
In order to elucidate the effect of temperature on the
NMR parameters of the [FHO]2 and [OHFHO]2 clusters
we have performed variable temperature measurements on
samples of TBA acetate and HF mixture in CDF3/CDF2Cl
1
which are depicted in Figs. 3 and 4. The H doublet of
[FHO]2 shifts downfield and the one of [OHFHO]2 upfield
1
as temperature is decreased; at the same time the JðFHÞ
splitting decreases. Both signals exhibit differential line
widths of the doublet signal components, which are more
pronounced at low temperatures. Generally, asymmetric
doublets are caused by an interference between dipolar and
chemical shift anisotropy relaxation mechanisms and an
increase of the rotational correlation times t: For HF
doublets, the low-field component is broader than the high-
field component if the coupling constant 1JðFHÞ is positive
and the high-field component is broader than the low-field
The assignment of the fluorine containing species was
done in a similar way. When the acetate/HF ratio is
decreased, the intensity of the low-field signal decreases,
thus this signal has to be assigned to [FHO]2 and the high
field signal to [OHFHO]2. Again, partial H/D substitution
reduces the intensity of the [FHO]2 signal, and introduces a
new 1H signal for [OHFDO]2. The situation is analogous to
the corresponding homoconjugate acetate–acetic acid
anions.
Some typical 19F signals of samples partially deuter-
ated in the hydrogen bond sites are depicted in Fig. 2.
Signals of 1:1 complexes [FHO]2 and [FDO]2 are
assigned by varying the deuterium fraction. The signals
are broad and a scalar coupling with H is not resolved in
the case of [FHO]2. The origin of this effect will be
discussed later.
Table 1
Experimental NMR parameters of the 1:1 and 2:1 complexes of acetic acid
with fluoride in CDF3/CDF2Cl at 120 K
NMR parameter
[OHF]2
[OHFHO]2
dðFHOÞ ^ 0:005 (ppm)
1JðFHOÞ ^ 0:5 (Hz)
18.327
The 19F signals of [OHFHO]2, [OHFDO]2 and
[ODFDO]2 are sharper, and appear at lower field as
indicated in Fig. 2. Only [OHFHO]2 and [OHFDO]2
exhibit a triplet and a doublet splitting with 1JðFHÞ coupling
61.0
dðFHOÞ ^ 0:01 (ppm)
dðFDOÞ ^ 0:01 (ppm)
dðOHFHOÞ ^ 0:01 (ppm)
dðOHFDOÞ ^ 0:01 (ppm)
1JðOHFHOÞ ^ 0:5 (Hz)
1JðOHFDOÞ ^ 0:5 (Hz)
dðOHFHOÞ ^ 0:01 (ppm)
dðOHFDOÞ ^ 0:01 (ppm)
dðODFDOÞ ^ 0:01 (ppm)
2132.21
2130.73
15.58
15.84
1
constants of about 60 Hz as expected from the H spectra.
260.5
260.5
Deuteration leads to a successive high-field shift. Scalar
couplings to D which should be of the order of 9 Hz are not
resolved.
2108.64
2109.26
2109.90
All NMR parameters are collected in Table 1.

N.S. Golubev et al. / Journal of Molecular Structure 700 (2004) 3–12
7
1
Fig. 3. Variable temperature H NMR signals of [FHO]2 (left) and [OHFHO]2 (right) dissolved in CDF3/CDF2Cl. For further explanation see text.
1
component if JðFHÞ is negative, as discussed previously
[16,20]. Thus, we conclude that 1JðFHOÞ . 0 and 1JðOHF
HOÞ , 0: At high temperatures, the line widths of the
[OHFHO]2 species are larger than those of [FHO]2, a
phenomenon which has to be assigned to the onset of
hydrogen bond and proton exchange. At lower tempera-
tures, the situation is inversed. Thus, [FHO]2 seems to be
involved in a so far unresolved dynamic process.
This process is better resolved in the 19F signals of this
species, depicted in Fig. 4. The hydrogen bond exchange
between the [FHO]2 and for [FDO]2 species is slow in the
NMR time scale at the highest and lowest temperature
accessible, and sharp lines are observed in both limits. At
170 K, doublet [FHO]2 is observed arising from scalar
1JðFHÞ coupling. The splitting decreases with decreasing
temperatures. Around 150 K, the lines are broadened and
are shifted to low-field. A maximum broadening is observed
around 130 K, and then again a sharpening of the lines
below this temperature.
It is unlikely that these spectral features arise from
transverse relaxation phenomena, for example from an
increase of the fluorine magnetic shielding anisotropy,
effective especially at 130 K. By contrast, these spectral
featuresaretypicalforthepresenceofanequilibriumbetween
two species A and B, where A dominates at high and B at low
temperatures, where the equilibrium constant is given by K ¼
xB=xA ¼ kAB=kBA: xA and xB are the mole fractions of both
Fig. 4. Variable temperature 19F NMR signals of a mixture of [OHF]2 and
[ODF]2 in CDF3/CDF2Cl. For further explanation see text.
species and kAB and kBA are the first-order or pseudo-first-
order rate constants of the interconversion. At 130 K, both

8
N.S. Golubev et al. / Journal of Molecular Structure 700 (2004) 3–12
Fig. 5. Dependencies of the measured NMR parameters (chemical shifts and scalar spin–spin coupling constants) for two clusters, [FHO]2 (left) and
[OHFHO]2 (right), on temperature in the region, 110–160 K.
species exhibit similar mole fractions, and in addition, the
exchange rate constants kAB < kBA are by chance of the order
of the chemical shift difference in Hz of the exchanging
species, pðnA 2 nBÞ: As the values of ni are temperature
dependent, we did not try to perform a quantitative line shape
Data points for all variable temperature experiments are
collected in Tables 2 and 3.
5. Discussion
analysis. However, for 130 K we can estimate kAB < kBA
<
pðnA 2 nBÞ roughly to be about 300 s21. We note that this
interconversion does neither lead to a collapse of spin–spin
splitting nor of the isotope splitting. This means, that the
processdoesnotinvolveproton norhydrogenbondexchange.
On theother hand, wenote that theequilibriumconstantK has
to change strongly with temperature in order to produce a
complete transformation of form B to form A within
20 K. This indicates a very large negative entropy and a
very large negative enthalpy of the transition A ! B:
The above results show that acetic acid and fluoride
anion can form 1:1 and 2:1 clusters in polar solvent such
as CDF3/CDF2Cl. In this section, we will first link the
NMR parameters of these clusters with the corresponding
hydrogen bond geometries. We will then discuss the
effects of H/D substitution as well as temperature
dependent solvation on these geometries, in comparison
with the clusters of HF with fluoride, which have been
studied previously [16,17]. Finally, the nature of the
dynamic process in which the 1:1 complex is involved is
discussed.
Finally, in Fig. 5 we have plotted the 1H chemical shifts
and JHF coupling constants of [FHO]2, [OHFHO]2 and
1
[OHFDO]2 as a function of temperature.
Table 2
1H NMR parameters of the 1:1 and 2:1 complexes of acetic acid with fluoride, [FHO]2, [OHFHO]2 and [OHFDO]2, in CDF3/CDF2Cl as a function of
temperature
T
[FHO]2
[OHFHO]2
[OHFDO]2
(K)
dðFHOÞ
1JðFHOÞ
W1
W2
dðFHOÞ
1JðFHOÞ
W1
W2
dðFHOÞ
1JðFHOÞ
W1
W2
(ppm)
(Hz)
(Hz)
(Hz)
(ppm)
(Hz)
(Hz)
(Hz)
(ppm)
(Hz)
(Hz)
(Hz)
110
120
130
140
150
160
18.322
18.327
18.318
18.296
18.251
18.222
51.5
61.0
67.0
72.5
75.5
81.5
80
43
23
27
28
25
38
35
18
23
24
23
15.55
15.58
15.61
15.63
15.71
15.76
260.5
259.0
258.0
254.5
252.5
247.5
24
16
18
42
70
70
40
21
21
46
70
70
15.80
15.84
15.87
61.0
59.0
57.0
25
16
27
50
26
31
dðFHOÞ; chemical shift of the H-bond proton; W1; signal line width in Hz of the low-field; W2; of the high-field component.

N.S. Golubev et al. / Journal of Molecular Structure 700 (2004) 3–12
9
Table 3
19F NMR parameters of the 1:1 acetate–HF complex [FHO]2 in
those published previously for various hydrogen fluoride–
fluoride anions, [(FH)nF]2, where n ¼ 1–4:
CDF3/CDF2Cl as a function of temperature
In Fig. 6a and f are depicted the correlations of the
hydrogen bond coordinate, q1 ¼ 1=2ðr1 2 r2Þ with q2 ¼
r1 þ r2: As has been shown previously [16,17], the values
obtained by ab initio calculations for [(FH)nF]2 are well
located on the correlation curve, represented by the solid
line in Fig. 6a. The geometric data in Fig. 6f were obtained
empirically as described below.
T
H form
D form
dðFDOÞ 2 dðFHOÞ
(K)
(ppm)
dðFHOÞ
W
dðFDOÞ
W
(ppm)
(Hz)
(ppm)
(Hz)
110
120
130
140
150
160
170
2131.60
2132.21
2133.47
2134.82
2135.80
2136.60
2137.53
78
199
428
344
145
50
2129.86
2130.73
2132.21
2133.81
2135.02
2136.01
2137.19
54
196
566
213
93
1.74
1.49
1.26
1.01
0.78
0.59
0.34
In Fig. 6b and g are depicted the values of the coupling
constants 1JðFHBÞ; B ¼ F, O as a function of q1: As
described previously both for FHF and FHN systems, a
64
1
change of the sign of JðFHBÞ is observed when H moved
35
46
away from F just after crossing the H-bond center. After
reaching a minimum, the coupling constants go to 0,
because the FH distance becomes larger (obviously, the
coupling constant is 0 at infinite distance). In the case of the
FHN systems we observed recently that the minimum was
deeper as compared to FHF [20]. The same is true in the
case of [FHO]2 and [OHFHO]2, we observe a similar sign
change: the values of 1JðFHOÞ are positive for [FHO]2 and
negative for [OHFHO]2. The minimum value is again
larger than the one for FHF. Assuming the validity of
W; signal line width.
5.1. Hydrogen bond geometries of [FHO]2 and [OHFHO]2
and their deuterated isotopologs
In Fig. 6 we have depicted various geometric and NMR
hydrogen bond correlations of the [FHO]2 and [OHFHO]2
clusters studied here. For comparison, we have included
Fig. 6. Hydrogen bond correlations of clusters of [(FH)nF]2 (left), adapted from Ref. [17] and of [FHO]2 and [OHFHO]2 (right). The solid lines correspond to
the equations listed in the hydrogen bond correlations section. For further explanation see text.

10
N.S. Golubev et al. / Journal of Molecular Structure 700 (2004) 3–12
Eq. (3), we obtain the solid line of Fig. 6g. Now, we could
adjust the values of q1 in such a way that the experimental
Table 4
Hydrogen bond distances of [FHO]2 and [OHFHO]2 obtained by NMR
1
values of JðFHOÞ are located on the correlation curve.
˚
rFH (A)
˚
q2 (A)
˚
q1 (A)
˚
rOH (A)
Species
T (K)
Fixing the values of q1 determined also those of q2 in
Fig. 6f.
[OHF]2
110
120
130
140
150
160
170
110
120
130
140
150
160
110
120
130
140
150
160
1.212
1.208
1.200
1.192
1.186
1.181
1.176
1.206
1.201
1.191
1.181
1.174
1.169
1.444
1.436
1.432
1.418
1.410
1.395
2.234
2.353
2.351
2.350
2.349
2.348
2.347
2.353
2.351
2.349
2.348
2.347
2.346
2.477
2.472
2.469
2.460
2.455
2.444
0.035
1.142
1.145
1.151
1.158
1.162
1.167
1.171
1.147
1.151
1.158
1.167
1.173
1.178
1.034
1.036
1.037
1.041
1.044
1.049
1
In Fig. 6c and h are depicted the H chemical shifts of
[OHF]2
0.031
0.024
0.017
0.012
0.007
0.002
0.030
0.025
0.016
0.007
0.001
20.004
0.205
0.200
0.198
0.188
0.183
0.173
[OHF]2
both types of systems, and the agreement with Eq. (4) is
satisfactory. In Fig. 6d and i we have included the
corresponding fluorine chemical shifts. As these shifts are
much more sensitive to the hydrogen bond geometries, they
were used to fine tune the q1 values. The H/D isotope effects
on the fluorine chemical shifts, DdðFHBÞ ¼ dðFDBÞ 2 dðF
HBÞ are included in Fig. 6e and j. They are well reproduced
by the solid lines calculated using Eq. (6). Again, the data
for the FHF system were taken from Ref. [17].
[OHF]2
[OHF]2
[OHF]2
[OHF]2
[ODF]2
[ODF]2
[ODF]2
[ODF]2
[ODF]2
[ODF]2
[OHFHO]2
[OHFHO]2
[OHFHO]2
[OHFHO]2
[OHFHO]2
[OHFHO]2
We note that the DdðFHFÞ values are negative for H
close to F, and even for [FHF]2 a negative value of
20.32 ppm was found. In our case, the values for [FHO]2
are positive, a result which can be taken as indication that H
is a little bit closer to O than to F, although Fig. 6 indicates
that [FHO]2 is very similar to the symmetric [FHF]2. By
increasing temperature, the hydron in [FHO]2 is shifted
towards F, and at 160–170 K, a symmetric proton shared
geometry is reached. Note, that there are no unusual spectral
changes in this temperature region, which is not surprizing
while for the asymmetric hydrogen bonds q1 ¼ 0 does not
correspond to any special physical situation. By contrast, the
hydrons in [OHFHO]2 are closer to O than to the central
fluoride, in a similar way as in the corresponding cluster
[FHFHF]2. Again, increasing temperature shifts both H
towards F, but the centered H-bond is not reached at
160 K. As a motive power of this displacement, preferential
solvation of compact fluoride anion as compared to the
acetate is proposed. All hydrogen bond distances estimated
from the analysis of Fig. 6 are included in Table 4. These
distances are affected by systematic errors, in view of many
assumptions both concerning the forms of Eqs. (1)–(6) as
well as the parameters used. However, we think that relative
changes are well reproduced, in particular the observed sign
change of the coupling constants.
dðAHXDBÞ 2 dðAHXHBÞ of a coupled hydrogen bonded
system AHXHB is non-zero. This means that the chemical
shift of H in AHX depends on whether the neighbouring
bond XHB is protonated or deuterated. For [OHOCOHO]2
a value of þ 0.29 ppm was found (see Fig. 1 and Ref. [19]),
a value which is very close to the one of 0.26 observed for
[OHFHO]2 (Fig. 1 and Table 1). For comparison, the value
of [FHFHF]2 is 0.155 ppm. These values indicate a mutual
influence of the two coupled hydrogen bonds in all the three
systems which exhibit an anti-cooperativity. Thus, deutera-
tion of XHB leads to an increase of the X· · ·B distance, but a
decrease of the A· · ·X distance. In the case of anions, this
feature is understandable and in terms of a higher charge on
X as compared to A and B. When the donor HB moves away
from X2, the latter becomes more ‘basic’ and attracts the
proton from AH, leading then to a distance increase of AH
and a shortening of the A· · ·X distance. Thus, from vicinal
isotope effects qualitative conclusions about the hydrogen
bond structure can be obtained.
The reasons for the temperature shifts have been
discussed in previous papers [18–20]. As temperature is
lowered, the local polarity increases because of solvent
ordering around the solutes, and structures with larger
dipole or quadrupole moments are favoured. In non-charged
‘neutral’ hydrogen bonds, the zwitterionic structures are
reached by shifting the proton towards or across the
hydrogen bond center. In the case of ions, a non-
symmetrical charge distribution will lead to similar effects.
The geometric changes of [FHO]2 with temperature as
expressed in Fig. 6 and Table 4 are then the consequence of
this phenomenon.
5.2. Structure and dynamics of [FHO]2 at low-temperature
So far, we have only considered the geometry of the
hydrogen bond in [FHO]2 but did not yet discuss the whole
structure of this complex, which is characterized also by an
orientational degree of freedom of the acetate moiety with
respect to the HF moiety. In fact, using different techniques,
rotational isomers have been detected previously for formic
acid monomer in gas phase and for the complex of formic
acid with hexamethylphosphorotriamide in polar solvents
[22,23].
Finally, we come to the discussion of the vicinal isotope
effects [19,20] found previously for [FHFHF]2 and the
[OHOCOHO]2 systems, and here for [OHFHO]2. This
phenomenon implies that the chemical shift difference
The rotational isomerism may be characterized by
different COF angles b as depicted in Scheme 1 for two

N.S. Golubev et al. / Journal of Molecular Structure 700 (2004) 3–12
11
Scheme 1.
arbitrary values of b leading to two isomers C and D.
Naturally, one could equally well conceive a broad
distribution of angles b: The different configurations will
exhibit different electrical properties, and therefore, by the
interaction with the surrounding solvent molecules and the
countercation isomers with larger dipole moments will be
favoured at lower temperatures.
against this assignment is that one expects very large rate
constants of the interchange between forms C and D, in
contrast to what is found for A and B experimentally.
However, it may be that the dynamics of the interchange
are determined by an unknown aggregation process
involving not only complexes [FHO]2 but also solvent
molecules and the countercation. Depending on the
aggregation state, the angle b may adopt a preferential
value.
Another degree of freedom arises from the possibility of
a hydrogen bond flip of the HF moiety between the two
oxygen atoms of acetate as indicated in Scheme 1.
Especially form C will be subject to such a fast hydrogen
bond flip, whereas the corresponding flip in form D will be
slower. The flip may involve a transition state exhibiting a
symmetric configuration with bifurcated hydrogen bond;
however, such a symmetric structure cannot be excluded as
main structure of the complex. Some of us have previously
found evidence for similar very fast hydrogen bond flips in
low-temperature solutions of hydrogen bonded
complexes [24].
Authors would like to thank the referee for pointing
out another mechanism, which could be responsible for
the spectral changes shown in Fig. 4. In this mechanism
A is CH3COOH· · ·F2 and B is CH3COO2· · ·HF, which
differ only by the average proton position, while the
dynamic behaviour is attributed to the slow solvent
reorganization that is needed to convert one of these
forms to the other. This attribution is supported by the
continuous change of average q1 values (see Table 4).
Slow (in the NMR time scale) proton transfer, CH3-
COOH· · ·F2 $ CH3COO2· · ·HF, can really be a reason
of the observed NMR effects, but solvent reorganization
as a process retarding the proton transfer is more than
doubtful. Indeed, dielectric relaxation in non-viscous
liquids takes usually the time as low as 10210–10212 s,
and low viscosity of the used freon mixture at 120–
140 K is proved by comparatively sharp NMR lines of
the solvent. More probable seems to be the process of
slowing down, shown in Scheme 2.
The question arises, whether the observed equilibrium
between the high-temperature form A and the low-
temperature form B can be explained in terms of the
reaction network of Scheme 1. A dominates at higher
and B at lower temperatures which means that B should
exhibit a larger dipole moment than A. At 130 K, the
interconversion rate is about 300 s21. As mentioned
above, the interconversion is not related to hydrogen
bond breaking processes which would lead to a
coalescence of the signals of [FHO]2 and [FDO]2, in
contrast to the experimental findings, nor to proton
Such interconversion involves not only proton transfer
but also H-bond reorganization, which requires some
activation energy for breaking down one of the hydrogen
bonds. This hypothetical mechanism can be confirmed
or rejected by a calculation of the whole reaction
pathway.
1
exchange, which would lead to a collapse of the JðFH
OÞ spin–spin splitting pattern. On the other hand, from
the graphs of Fig. 6 there is no evidence for major
changes of the hydrogen bond geometry. Therefore, the
process is also not related to proton jumps between two
sites in the hydrogen bond, one closer to F and the other
closer to O. These constraints as well as the simple
molecular structure of [FHO]2 limit strongly the
possibilities.
Thus, as a preliminary working hypothesis, we
tentatively assign A to C and B to D. An argument
Scheme 2.

12
N.S. Golubev et al. / Journal of Molecular Structure 700 (2004) 3–12
[5] C. Viragh, T.K. Harris, P.M. Reddy, M.A. Massiah, A.S. Mildvan,
6. Conclusions
I.M. Kovach, Biochemistry 39 (2000) 16200.
[6] Y. Kim, K.H. Ahn, Theor. Chem. Acc. 106 (2001) 171.
[7] D.B. Northrop, Acc. Chem. Res. 34 (2001) 790.
[8] W.W. Cleland, Arch. Biochem. Biophys. 382 (2000) 1.
[9] K.B. Schowen, H.H. Limbach, G.S. Denisov, R.L. Schowen,
Biochim. Biophys. Acta 1458 (2000) 43.
We have shown that using low-temperature NMR it is
possible to detect non-trivial properties of simple hydrogen
bonded clusters of fluoride with carboxylic acid. As fluoride
and hydroxide have many similarities, this work may also be
interesting for modeling complex hydrogen bonded net-
works in functional biomolecules. Using the valence bond
order correlations, it is possible to link the worlds of NMR
parameters and hydrogen bond geometries, and obtain
information about the effects of a polar solvent on the latter.
We hope that such results will stimulate further theoretical
work in this area of research.
[10] B.A. Katz, J.R. Spencer, K. Elrod, C. Luong, R.L. Mackman, M. Rice,
P.A. Sprengeler, D. Allen, J. Janc, J. Am. Chem. Soc. 124 (2002)
11657.
[11] P.A. Frey, J. Mol. Struct. 615 (2002) 153.
[12] (a) A. Warshel, A. Papazyan, P.A. Kollman, Science 269 (1995) 102.
(b) A. Warshel, A. Papazyan, Proc. Natl Acad. Sci. USA 93 (1996)
13665. (c) A. Warshel, J. Biol. Chem. 273 (1998) 27035. (d) C.L.
Perrin, J.B. Nielson, Annu. Rev. Phys. Chem. 48 (1997) 511.
[13] C.J. Smallwood, M.A. McAllister, J. Am. Chem. Soc. 119 (1997)
11277.
[14] G.A. Kumar, Y. Pan, C.J. Smallwood, M.A. McAllister, J. Comput.
Chem. 19 (1998) 1345.
Acknowledgements
[15] N.S. Golubev, I.G. Shenderovich, P.M. Tolstoy, D.N. Shchepkin,
J. Mol. Struct. (2004) (in press).
This work is financially supported by the Russian
Foundation for Basic Research, Grants 03-03-32272 and
03-03-04009, the Deutsche Forschungsgemeinschaft, Bad
Godesberg, and the Fonds der Chemischen Industrie,
[16] I.G. Shenderovich, S.N. Smirnov, G.S. Denisov, V.A. Gindin, N.S.
Golubev, A. Dunger, R. Reibke, S. Kirpekar, O.L. Malkina, H.-H.
Limbach, Ber. Bunsen. Phys. Chem. 102 (1998) 422.
[17] I.G. Shenderovich, H.-H. Limbach, S.N. Smirnov, P.M. Tolstoy,
G.S. Denisov, N.S. Golubev, Phys. Chem. Chem. Phys. 4 (2002)
5488.
´
´
Frankfurt. P.M.T. thanks the Fonds for a Kekule scholar-
ship. Authors are grateful to the referee for valuable
comments and suggestions.
[18] I.G. Shenderovich, A.P. Burtsev, G.S. Denisov, N.S. Golubev, H.-H.
Limbach, Magn. Res. Chem. 39 (2001) S1/91.
[19] P.M. Tolstoy, P. Schah-Mohammedi, S.N. Smirnov, N.S. Golubev,
G.S. Denisov, H.-H. Limbach, J. Am. Chem. Soc., submitted for
publication.
References
[20] I.G. Shenderovich, P.M. Tolstoy, N.S. Golubev, S.N. Smirnov, G.S.
Denisov, H.-H. Limbach, J. Am. Chem. Soc. 125 (2003) 11710.
[21] (a) L. Pauling, J. Am. Chem. Soc. 69 (1947) 542. (b) I.D. Brown, Acta
Crystallogr. B48 (1992) 553.
[1] W.W. Cleland, M.M. Kreevoy, Science 264 (1994) 1887.
[2] P.A. Frey, S.A. Whitt, J.B. Tobin, Science 264 (1994) 1927.
[3] N.S. Golubev, G.S. Denisov, V.A. Gindin, S.S. Ligay, H.-H. Limbach,
S.N. Smirnov, J. Mol. Struct. 322 (1994) 83.
[22] E. Bjarnov, W.H. Hocking, Z. Naturforsch. 33A (1978) 610.
[23] N.S. Golubev, E.G. Pushkareva, Vestnik LGU 11 (1985) 87.
[4] A.S. Mildvan, M.A. Massiah, T.K. Harris, G.T. Marks, D.H.T.
Harrison, C. Viragh, P.M. Reddy, I.M. Kovach, J. Mol. Struct. 615
(2002) 163.
¨
[24] F. Mannle, H.-H. Limbach, G.S. Denisov, J. Am. Chem. Soc.,
submitted for publication.