Self-association of water and water-solute associations in chloroform studied by NMR shift titrations

Article abstract of DOI:10.1021/jp952596w

Association constants and the corresponding free energies ΔG were obtained by nonlinear least-square fitting of the water ¹H NMR shift changes observed at different concentrations. The only until now available earlier value (Shaw, J.-H. L.; Wan

Full text of DOI:10.1021/jp952596w

J. Phys. Chem. 1996, 100, 5533-5537
5533
Self-Association of Water and Water-Solute Associations in Chloroform Studied by NMR
Shift Titrations†
Frank Eblinger and Hans-Jo1rg Schneider*
FR Organische Chemie der UniVersita¨t des Saarlandes, D 66041 Saarbru¨cken, Germany
ReceiVed: September 6, 1995; In Final Form: December 14, 1995X
Association constants and the corresponding free energies ∆G were obtained by nonlinear least-square fitting
of the water 1H NMR shift changes observed at different concentrations. The only until now available earlier
value (Shaw, J.-H. L.; Wang, S. M.; Li, N. C. J. Phys. Chem. 1973, 77, 236) is shown to be based on erraneous
evaluations and on inaccurate data. Analysis for the self-association of water in chloroform based on a 1:1
dimer model yields ∆G ) -3.5 kJ/mol; based on a trimer model one obtains an almost equally good fit with
∆G ) -15.4 kJ/mol. The ∆G value in acetonitrile as solvent is with +9 kJ/mol positive (vide infra) as are
all ∆G values observed for water-solute associations in chloroform with DMF, DMSO, diphenyl sulfoxide,
and benzene, with the exception of HMPT (hexamethylphosphotriamide) and a special diamide designed to
take up water by 2 hydrogen bonds. The complexation induced shifts, determined simultaneously from the
NMR titrations, are between 3 and 4.5 ppm. They are similar to those observed with hydroxy compounds
and show that the simulations are based on realistic models. No correlation is found between the CIS and
the corresponding ∆G values. The only upfield shift with -4.4 ppm is found for the benzene-water complex,
in agreement with theoretically derived structures.
Introduction
results of this early investigation, which suffers not only from
necessarily larger technical problems but also, in our view, from
misinterpretation of the reported shielding variations. If we
evaluate these (given in Table 1 of ref 7a) with the usual
nonlinear least-square fit procedures (see experimental and
computational details) we obtain for the assumed7 1:1 associa-
tion an equilibrium constant of, e.g., K ) 8.6 M-1 (at 306 K),
which is about 20 times higher than that reported.7a Noticeably,
the fit for these data is at least as good with our reevaluation
(Figure 1a) as in the original paper.
Self-association of water in lipophilic solvents has received
until now little attention, although it can play an often decisive
role for hydrogen-bonded supramolecular complexes. The same
holds for corresponding interactions of water with solutes; the
few available investigations2 already show that water can interfer
substantially with host-guest complex formation. Even the
presence of only 3% methanol in CCl4 lowers association
constants of diamide-type structures3 from K ) 104 to below
10.
The underlying assumption of dominating water dimers might
not be correct, even though the data fit perfectly to the model.
Various structures have been proposed for water associations
in lipophilic solvents. Earlier workers believed that water
dissolves in aliphatic and aromatic hydrocarbons and CCl4
primarily as monomer,8 whereas in slightly polar solvents such
as partially chlorinated hydrocarbons water was described as
somewhat polymerized (reported formation constant of the
trimer in tetrachloroethane:9 2-5 L2 mol-2). On the other hand,
dimerization constants7a,10 were reported to be around 0.5 mol/
L. In contrast, a multitude of theoretical and spectroscopic
investigations for which we cite only leading references11 are
predicting cyclic trimers to be particularly stable in the gas state.
These studies allowed detailed insight into the structures not
only of associations between water11 but also of water with
several simple acceptor molecules.12 On the other hand, there
is a scarcity on thermodynamic data for these equilibria from
many of these spectroscopic studies. NMR measurements
should allow convenient access to such missing affinity data.
In contrast to complexes with organic molecules, NMR studies
give in the present case little insight into the structure or even
stochiometry of these simple but in comparison to larger organic
or biological rather featureless complexes.
NMR titrations, based on established techniques,4 offer a
convenient way to obtain new and practically useful numbers
on association constants or free energies ∆G . We found them
to lead to results which are partially in significant contrast to
published data or predictions. An advantage of the NMR
method using instruments with higher sensitivity is that one can
measure association constants as low as 10-4 M-1 or as high
as 104 M-1, limited only by solubilities and the necessary NMR
spectrum accumulation time. The measurements were es-
sentially restricted to chloroform as solvent in view of the
necessary minimum concentrations of water for the observation
of its NMR signals. This prohibited measurements in more inert
solvent such as carbon tetrachloride, in which the majority of
many earlier, mostly IR-based investigations were performed.5,6
Even the large compilations of Abraham5 and Raevsky6 et al.
contain relatively few data on associations with water. The
present study was initiated also with the intend to improve the
experimental basis for the prediction of corresponding equilib-
rium constants.
Results
Self-Association of Water. The self-association of water
in CDCl3 has been investigated by NMR already over 30 years
ago.7 A major purpose of the present paper is to correct the
For the water dimer association in CDCl3, we obtain on the
basis of a simple 1:1 model an equilibrium constant of 4.1 L/mol
corresponding to ∆G ) -3.5 kJ/mol (Table 1), with a fit without
systematical deviations (Figure 1b). Care was taken that any
phosgene content in the chloroform solutions will not by
† Supramolecular Chemistry, part 57; for part 56 see: Rammo, J.; Hettich,
R.; Roigk, A.; Schneider, H.-J. J. Chem. Soc., Chem. Commun., in press.
X Abstract published in AdVance ACS Abstracts, March 1, 1996.
0022-3654/96/20100-5533$12.00/0 © 1996 American Chemical Society

5534 J. Phys. Chem., Vol. 100, No. 13, 1996
Eblinger and Schneider
TABLE 1: Self-Association of Water in Chloroform: Free
Energies of Association ∆G and Association-Induced NMR
Shifts CIS (from NMR Titrations at 298 K; for Structures
See Scheme 1)
∆G [kJ/mol]
CIS [ppm]
∆G [kJ/mol]
1. in CD3CNa
+9.14
-3.53
-15.4
+3.8
+3.9
+4.4
a
2. in CDCl3
b
3. in CDCl3
+0.66 calcc
+0.66 calcd
+0.62 calce
-1.95 exph
+6.58 expi
c
3. in CCl4
4. in CCl4
5. in CCl4
7. in CCl4
c
e
h
8. in gas phasei
a Fitted to a dimer model. b Fitted to a trimer model. c From factor
values: see ref 5. d From factor values: see ref 6. e From factor values:
private communicaton from C. Laurence. f From LFER: Coetze, J. F.
J. Solut. Chem. 1982, 11, 395. g IR results: private communication from
C. Laurence. h IR result: Magnusson, L. B. J. Phys. Chem. 1970, 74,
4221. i Gas-phase result: Curtiss, L. A.; Blander, M. Chem. ReV. 1988,
88, 827.
are realistic is also visible in the complexation-induced NMR
shifts (CIS values for 100% complexation); the observed
deshielding as consequence of the hydrogen-bond formation is
close to CIS values observed with other hydroxylic functions.15
As pointed out by a referee, the CIS values are in fact
unexpectedly large in view their being the average of bound
and “free” water protons. However, if one proton is involved
in a hydrogen bond, the other one is likely deshielded also as
consequence of the charge depletion at the oxygen brought about
by the primary hydrogen-bond interaction. In addition we note
that the CIS values are susceptible to larger errors than the
equilibrium constants K, not only due to their sensitivity against
acid traces15 but also due to the fact that solubility problems
limit the titrations to points far from saturation (see figures).
The free energies of complexation ∆G for the dimer and the
trimer model are not as far apart as it might seem: if the dimer
contains one hydrogen bond and the trimer three, as suggested
by theoretical studies,11 one would arrive at a value of
approximately 4 kJ/mol for one hydrogen bond for both
structures. The higher value for the trimer model, which would
oppose larger entropy disadvantages, can be due to a higher
acidity and basicity of water molecules already involved in
hydrogen bonding.
The ∆G for self-association drops drastically from CDCl3 to
acetonitrile as solvent (Table 1), which is in accord with the
known hydrogen-bond acceptor factor values for this solvent .
A value of ∆G ) 12.9 kJ/mol can be calculated for the free
energy of the water/acetonitrile association from published factor
values in CCl4;5 in comparison one calculates from factor values
for water/CHCl3 6.3 kJ/mol. The available factor values allow
for water/water interactions only the prediction with CCl4 as
solvent, which for solubility reasons is not accessible to NMR
titrations. The resulting ∆G values are partially at substantial
variance with the numbers derived from NMR titrations (Table
1).
Association of Water with Simple Solutes. The observed
affinities of water (presumably in the form of a trimer, see
above) to simple hydrogen-bond acceptor molecules (Scheme
1) show little agreement with published data for CCl4 as solvent
(Table 2) . For all solutes a satisfactory fit was obtained for a
simple water/solute (1:1) model (Figure 2), which led us to
refrain from simulations based on other, more extended models.
Again, the CIS values obtained from the 1:1 simulations support
that the model was not unrealistic. The deshielding is in a
similar range as that observed with water self-association (Table
1) and with complexes involving hydroxy compounds and other,
Figure 1. Experimental H NMR shifts of water protons and nonlinear
least-square fit for self-association of water: (1a) reevaluation of
literature data;7a (1b) with new data, fit based on dimer model; (1c)
with new data, fit based on trimer model.
liberation of free HCl interfere with the measurements (see
Experimental Section).
For the theoretically preferred11 trimer we obtain an associa-
tion free energy ∆G ) -15.4 kJ/mol (Table 1). That the
simulations on the basis of both the dimer and the trimer model

Water-Solute Associations in Chloroform
J. Phys. Chem., Vol. 100, No. 13, 1996 5535
SCHEME 1: Water Dimer (1) and Trimer (2) Models
from Literature,17 Water-Benzene Complex (3)17 and
(4-9) Structures of Other Acceptors
lower than the largest values (around 5 ppm) observed and
calculated for a single proton being as close as possible above
the center of aryl groups in cyclophanes.16 This suggests that
water protons are located in tight contact to the π-moiety without
significant contributions of conformers in which water oxygen
atoms would serve as acceptor for the slightly polarized benzene
protons. Such a structure (see Scheme 1) is is in good
agreement with theoretical17 and experimental18 results. The
CIS value, which is unusually high also in view of the fact that
it reflects the time averaged shift of all water protons, suggests
that in addition to the ring current anisotropy effect there is
disruption of water hydrogen bonds in dimers or trimers, or to
the solvent, by the benzene acceptor, which enhances the total
CIS value significantly.
Experimental and Computational Details
The solutes and solvents were commercially available except
the diamide 5 (see below); they were purified and dried
according to literature procedures.17 Water and solute concen-
trations were measured/checked by NMR integration, using
internal reference compounds (1,1,1-trichloroethane and 1,4-
dioxane) of known concentrations.
That phosgene in chloroform, which was used as purchased
from Aldrich without stabilizer, did not interfer in the titrations
was secured by measuring acidity from chloroform after
extraction with water: mixtures of chloroform and water (5:1
v/v) were for removal of CO2 degassed by bubbling through
nitrogen and during 10 h deliberately exposed to daylight which
is known to lead to phosgene traces. The pH of the water
solution then was 5.0 (corresponding to 1.0 × 10-5 M H+); in
comparison, water treated the same way but not with chloroform
showed a pH of 6.2 (6.7 × 10-7 M H+). Thus, the maximum
proton concentration in the solutions, which contained on the
average 1 mM water, was well below 1% of the proton
concentration stemming from water.
Pyridine-2,6-bis(N,N-dibenzyl)diamide (5): 2 g (9.8mmol)
of pyridine 2,6-dicarboxylic acid and 2.2 g (10.8 mmol) of PCl5
were refluxed for 12 h with 100 mL of freshly distilled thionyl
chloride. The excess thionyl chloride was evaporated, and the
remaining oil was flushed twice with 50 mL of dry toluene in
order to remove unreacted thionyl chloride. At 0 °C the acid
chloride was added dropwise to a mixture of 3.87 g (19.6 mmol)
of dibenzylamine and 4.5 mL of triethylamine in 100 mL of
dry methylene chloride. After 5 h of stirring at room temper-
ature, the organic phase was washed three times with 100 mL
of water. The methylene chloride solution was dried over
K2CO3 and evaporated. The resulting brown oil was purified
by column chromatography (SiO2/ethyl acetate). The diamide
cristallized on addition of dry ether. The Rf value on silica gel
TLC with EtOAc is 0.64; white powder; mp 182-184 °C.
TABLE 2: Associationsa with Water as Donorj with several
Solutes (from NMR Titrations at 298 K; for Structures See
Scheme 1)
in CDCl3
in CCl4
∆G [kJ/mol]
∆G [kJ/mol] CIS[ppm]
(Me2N)3PO
pyr-diamide
DMF
DMSO
Ph2SO
-2.06
-0.27
+0.55
+0.71
+1.1
+3.9
+3.1
+3.9
+3.7
+3.7
-5.8
-8.3c
-8.4d
-9.4g
-3.8c
-5.2c
-3.5d
-5.2d
-3.2g
-5.2g
-3.53c
C6H6
+10.4
+4.05c +5.19d
3
1H NMR (CDCl3) δ (ppm) 7.88 (1H, t, J ) 7.72 Hz, Pyr-
a Footnotes a-g as in Table 1. j With [H2O], (3-6) × 10-3 M.
3
4-H); 7.76 (2H, d, J ) 7.79, Pyr-h-3,5); 7.31-7.08 (m, 20H,
partially ionic receptors.13c As observed earlier13c there is again
no correlation between the CIS values and the strength of a
hydrogen bond as far the latter is depicted by the corresponding
∆G values. Noticeably, only hexamethylphosphortriamide
(HMPT, Scheme 1), which is known to be an excellent hydrogen
bond acceptor, gives a negative ∆G value (corresponding to
>50% complexation under standard condition in 1 M solution)
except the special pyridinediamide 5 (see Scheme 1), which
was designed to complex a single water molecule by strainfree
2-fold hydrogen bonding.
Ar-H); 4.63, 4.45 (2s, 2 × 4H, benzyl-CH2 cis + trans isomer
around amide bond). 13C NMR (CDCl3) δ (ppm) 168.6 (CdO);
153.0 (Pyr C2, C6); 138.2 (Pyr C4); 136.59, 136.40 (Phenyl-
C1, cis + trans); 124.8 (Pyr-C3, C5); 128.6§ (phenyl-C2); 128.3§
(phenyl-C3); 127.4 (phenyl-C4); 51.1, 47.65 (benzyl-CH2, cis
+ trans) [§: assignments interchangeable].
Anal. Calcd for C35H31N3O2: C, 79.97; H, 5.94; N, 7.99.
Found: C, 79.30; H, 5.75; N, 7.97.
NMR measurements were carried out with Bruker AM400
and DRX 500 systems at 298 K.
Benzene is the only solute producing with CIS ) -4.4 ppm
a substantial shielding on the water proton which is only slightly
NMR Shift Titrations. (a) For the dimer model: Calculation
by nonliner least-squares fit using the program Sigmaplot from

5536 J. Phys. Chem., Vol. 100, No. 13, 1996
Eblinger and Schneider
Figure 2. Experimental H NMR shifts of water protons and nonlinear least-squares fit for associations of water with several solutes: (2b) with
DMF 4; (2b) with pyridinediamide 5; (2c) with DMSO 6; (2d) with benzene 9A (for structures see Scheme 1).
Jandel (Marquardt-Levenberg algorithm), based on equations
Acknowledgment. Our work is supported by the Deutsche
Forschungsgemeinschaft, Bonn, and the Fonds der Chemischen
Industrie, Frankfurt. We appreciate discussions with Drs. M.
Abraham, K. Connors, C. Laurence, and O. Raevsky, within
the framework of a NATO CRG. Dr. V. P. Solov’ev from
IPAC, Chernogolovka near Moscow, is thanked for help with
his program Chem Equi.
for self-association:
Cd ) (1 + 4C0Ka)/8Ka - b
2
b ) (((1 + 4C0Ka)/8Ka)2 - C0 /4)1/2
d ) (C0 - 2Cd)dm/C0 + Cddd/C0
References and Notes
where Cd ) equilibrium concentration of dimer; dm ) shift of
monomer; dd ) shift of dimer; Ka ) association constant; C0
) analytical concentration; d ) observed shift.
(1) Adrian, J. C.; Wilcox, C. S. J. Am. Chem. Soc. 1992, 114, 1398
and references therein.
(2) Simova, S.; Juneja, R.; Schneider, H.-J., unpublished results. See
also: Schneider, H.-J.; Juneja, R. K.; Simova, S. Chem. Ber. 1989, 112,
1211.
(b) For Water-Solute Interactions:3
d ) dS + Dd/2S0[Kd + R0 + S0 - {(Kd + R0 + S0)2 -
4R0S0}1/2]
(3) Wilcox, C. S. In: Schneider, H.-J., Du¨rr, H., Eds.; Frontiers in
Supramolecular Organic Chemistry and Photochemistry; VCH: Weinheim,
1991; pp 123-143.
(4) Schneider, H.-J.; Kramer, R.; Simova, S; Schneider, U. J. Am.
Chem. Soc. 1988, 110, 6442.
where d ) observed shift; dS ) shift of observed compound
before complexation; S0 ) concentration of observed compound;
R0 ) concentration of added compound; Dd ) complexation
induced shift (CIS); Kd ) dissociation constant ) 1/Ka.
(c) For the Trimer Model: The formation constant for the
water trimer was fitted by Monte Carlo optimization using the
program Chem Equi from Dr. Vitaly P. Solo’vev, IPAC,
Chernogolovka, Russia.
(5) Raevsky, O. A.; Grogor'ev, V. Y.; Kireev, D. B.; Zefirov, N. S.
Quantum Struct.sAct. Relat. 1992, 49.
(6) (a) Abraham, M. H. Chem. Soc. ReV. 1993, 73. (b) Abraham, M.
H.; et al. J. Chem. Soc. Perkin Trans. 2 1994, 1777.
(7) (a) Shaw, Y.-H. L.; Wang, S. M.; Li, N. C. J. Phys. Chem. 1973,
77, 236. (b) For a review see: Christian, S. D.; et al. Q. ReV. 1970, 24, 20.
(8) (a) Gordon, M.; Hope, C. S.; Loan, L. D.; Roe, K.-J. Proc. R. Soc.
1960, 258, 215. (b) Ackermann, T. Z. Phys. Chem. 1964, 42, 119. (c)
Risbourg, A.; Liebaert, R. Compt. Rend. 1967, 264, 237.

Water-Solute Associations in Chloroform
J. Phys. Chem., Vol. 100, No. 13, 1996 5537
(9) Lin, T. F.; Christian, S. D.; Affsprung, H. E. J. Phys. Chem. 1965,
69, 2980.
(10) Masterson, W. L.; Gendrano, M. C. J. Phys. Chem. 1966a, 70, 2895.
(11) (a) Wales, D. J. J. Am. Chem. Soc. 1993, 115, 11180. (b) Fowler,
J. E.; Schaefer III, H. F. Ibid. 1995, 117, 446 and references therein.
(12) With a lactam: Held, A.; Pratt, D. W. J. Am. Chem. Soc. 1993,
115, 9708.
(14) Schneider, H.-J.; Ru¨diger, V.; Cuber, U. J. Org. Chem. 1995, 60,
996.
(15) Alhambra, C.; Luque, F. J.; Orozco, M. J. Phys. Chem. 1995, 99,
3084.
(16) Blake, G. A.; Suzuki, S.; Green, P. G.; Bumgarner, R. E.; Dasgupta,
S.; Goddard, W. A. Science 1992, 257, 942.
(13) See e.g.: (a) Sanders Pearce, C. M.; Sanders J. K. M. J. Chem.
Soc., Perkin Trans 2 1994, 1119. (b) Bonar-Law, R. P.; Sanders, J. K. M.
J. Am. Chem. Soc. 1995, 117, 259. (c) Coteron, J. M.; Hacket, F.; Schneider,
H.-J. J. Org. Chem., in press.
(17) Tietze, L. F.; Eicher, Th. Reaktionen und Synthesen; Thieme
Verlag: Stuttgart, 1981.
JP952596W