The role of water in the oligomerization equilibria involving Bis(pentafluorophenyl)borinic acid in dichloromethane solution

Article abstract of DOI:10.1021/om0494519

The association equilibria of water involving bis(pentafluorophenyl)borinic acid in dichloromethane solution was investigated. The amount of water present in solution affected the rate of attainment of the 1_m/1_t equilibrium, the oligomerization being exceedingly slow in anhydrous conditions. The catalytic role of water can be attributed to the increased nucleophilicity of the BOH group upon water coordination, which allows alternative aggregation pathways. Semiempirical computations, at the PM3 level, provided a picture of the pligomerization in the presence and in the absence of water that well agrees with the experimental findings.

Full text of DOI:10.1021/om0494519

Organometallics 2004, 23, 5493-5502
5493
The Role of Water in the Oligomerization Equilibria
Involving Bis(pentafluorophenyl)borinic Acid in
Dichloromethane Solution
Tiziana Beringhelli, Giuseppe D’Alfonso,* Daniela Donghi, and Daniela Maggioni
Dipartimento di Chimica Inorganica, Metallorganica e Analitica, Facolta` di Farmacia,
Universita` di Milano, Via Venezian 21, 20133 Milano, Italy
Pierluigi Mercandelli* and Angelo Sironi
Dipartimento di Chimica Strutturale e Stereochimica Inorganica, CNR-ISTM,
Universita` di Milano, Via Venezian 21, 20133 Milano, Italy
Received July 21, 2004
1
The H, 19F, and 11B NMR data indicated that in CD2Cl2 solution monomeric bis(penta-
fluorophenyl)borinic acid, (C6F5)2BOH (1m), is in equilibrium with the cyclic trimer (1t)
observed in the solid state. The position of the association equilibrium shifted to the right
on increasing the concentration, on decreasing the temperature, and on decreasing solvent
polarity, in the series CD2Cl2, CDCl3, CCl4, in agreement with the higher polarity of the
monomer (2.38 D for 1m and 0.65 D for 1t, according to PM3 computations). At temperatures
1
lower than 210 K the H and 19F NMR spectra revealed the simultaneous (reversible)
formation of two novel compounds, which have been formulated as the (C6F5)2BOB(C6F5)2
anhydride (2) and the trimeric species [(C6F5)2BOH]3‚OH2 (3), of C2 symmetry, with a water
molecule formally inserted into a B-O(H)-B bridge of 1t, to give a very strong BO(H)‚‚‚HO(H)B
hydrogen bond (δ 18.6). 1H and 19F EXSY experiments at 184 K revealed exchange between
3 and 1m, and not 1t. The data showed that the formation of 3, observed at temperatures
where the monomer-trimer equilibrium is frozen, occurs by aggregation of monomeric units
and not by cycle opening from 1t. The stabilization of the water molecule in 3 is strong
enough to promote the dehydration of 1 to give the anhydride 2; for entropic reasons, the
reaction occurs only at very low temperatures and is reversed on raising the temperature.
At higher temperatures, the position of the monomer-trimer equilibrium is affected by the
amount of water, which stabilizes the trimeric form, owing to the formation of a hydrogen-
bond adduct 4 containing exocyclic water. At low temperatures, in the presence of the
monomer, this species progressively dehydrated, due to the formation of 3. The amount of
water present in solution also affected the rate of attainment of the 1m/1t equilibrium, the
oligomerization being exceedingly slow in anhydrous conditions. The catalytic role of water
can be attributed to the increased nucleophilicity of the BOH group upon water coordination,
which allows alternative aggregation pathways. Semiempirical computations, at the PM3
level, provided a picture of the oligomerization in the presence and in the absence of water
that well agrees with the experimental findings.
Introduction
via strong BO(H)H‚‚‚OH2 hydrogen bonds. More re-
cently we have shown that bis(pentafluorophenyl)-
borinic acid (1)3 in the solid state exists as a self-
associated trimeric species, which in toluene solution
dissociates to give the (C6F5)2BOH monomer (hereafter
Ar2BOH, Ar ) C6F5).4 This molecule, owing to presence
of a variety of reactive sites (Lewis acid and Lewis base,
Brønsted acid, hydrogen-bond donor and acceptors) can
be involved in a multiplicity of intra- or intermolecular
interactions and association equilibria. An in-depth
investigation of the forms in which the molecule is really
present in solution is then crucial, in the light of the
high current interest, both academic and industrial, for
We are currently investigating the association equi-
libria involving Lewis acids such as perfluoroaryl-
boranes. It has been previously reported1,2 that tris-
(pentafluorophenyl)borane in toluene solution is able to
bind quantitatively and in a stepwise manner up to
three water molecules, the first one through a Lewis
acid-base (C6F5)3B-OH2 interaction and the other two
(1) Beringhelli, T.; Maggioni, D.; D’Alfonso, G. Organometallics
2001, 20, 4927.
(2) (a) Bergquist, C.; Bridgewater, B. M.; Harlan, C. J.; Norton, J.
R.; Friesner, R. A.; Parkin, G. J. Am. Chem. Soc. 2000, 122, 10581. (b)
Doerrer, L. H.; Green, M. L. H. J. Chem. Soc., Dalton Trans. 1999,
4325. (c) Danopoulos, A. A.; Galsworthy, J. R.; Green, M. L. H.;
Cafferkey, S.; Doerrer, L. H.; Hursthouse, M. B. Chem. Commun. 1998,
2529. (d) Siedle, A. R.; Lamanna, W. M. U.S. Patent 5,296,433, 1994.
(e) Siedle, A. R.; Lamanna, W. M.; Newmark, R. A. Macromol. Symp.
1993, 66, 215.
(3) Chambers, R. D.; Chivers, T. J. Chem. Soc. 1965, 3933.
(4) Beringhelli, T.; D’Alfonso, G.; Donghi, D.; Maggioni, D.; Mer-
candelli, P.; Sironi, A. Organometallics 2003, 22, 1588.
10.1021/om0494519 CCC: $27.50 © 2004 American Chemical Society
Publication on Web 10/07/2004

5494 Organometallics, Vol. 23, No. 23, 2004
Beringhelli et al.
Chart 1
1) quite close to those measured in toluene solution. The
attribution of the minor set B to the trimeric cyclic form
1t (Chart 1) is based on the points below.
(i) The 11B chemical shift value (δ 8.4) is typical of
tetracoordinated boron atoms (while that of 1m falls in
the range of tricoordinated boron derivatives, δ 43.8).8,9
(ii) The values of the longitudinal relaxation times T1,
much shorter for resonances B than A (Table 1), indicate
that the species responsible for set B is a molecule larger
than that responsible for signals A. (iii) The relative
intensity R of signals B with respect to A is higher in
spectra recorded immediately after dissolution of 1 (in
solid 1 exists in the trimeric form),4 then slowly de-
creases to equilibrium values. (iv) The equilibrium
values of R increase on increasing the overall concen-
tration (Figure 2a) and (v) on decreasing the tempera-
ture (Figure 2b).
Figure 1. Typical 19F NMR spectrum of a CD2Cl2 solution
of 1 (283 K, 0.18 M). Signals A are due to 1m and B to 1t.
this compound.5,6 In particular, it has been recently
shown that, in combination with aluminum alkyls, it is
an efficient cocatalyst for olefin polymerization.7
We have already shown that the pπ-pπ OfB dona-
tion imparts a partial double-bond character to the
B-OH interaction, responsible for the restricted rota-
tion observed by low-temperature NMR in toluene
solution.4
The latter behavior is typical of association processes,
which are entropically unfavorable. All of the above
evidence indicates that in the present case this process
is a self-oligomerization reaction.10 The occurrence of
such processes is a well-known phenomenon for R2BY
derivatives in which Y contains one or more lone pairs
(Y ) halogens, OR, NR2).9g,11 The trimeric nature of the
[Ar2BOH]n oligomer has been ascertained by measuring
the slope of a suitable plot of the intensity ratios R at
different overall concentrations C (as detailed in the
Supporting Information, Figure S1).
It has been therefore definitely confirmed that signals
B are due to the cyclic trimer 1t, in equilibrium with
the monomer 1m (eq 1). According to the NMR data (one
protonic resonance, one set only of 19F signals), in
solution 1t has a (dynamic) symmetry higher than that
observed in the solid state (C2).4
We have now investigated the behavior of 1 in
dichloromethane and found a much more complex
picture than in toluene. In this solvent sizable amounts
of the trimeric form of Ar2BOH are present, even at
room temperature. Adventitious water, which acts as a
Lewis base and as hydrogen-bond acceptor/donor, affects
both the position and the rate of the monomer-trimer
equilibrium, owing to the formation of a number of
monomeric or oligomeric adducts. The nature of these
adducts and their interconversion pathways have been
investigated, also with the use of semiempirical com-
putations. In this context, a quite unusual spontaneous
dehydration reaction has been discovered, driven by the
formation of a very strong hydrogen bond within an
oligomeric adduct containing “endo-cyclic” water.
Results and Discussion
3 Ar2BOH a [Ar2BOH]3
(1)
1
Monomer-Trimer Equilibrium. The H and 19F
spectra of bis(pentafluorophenyl)borinic acid (1) in
CD2Cl2 solution usually show two sets of resonances,
labeled as A and B in Figure 1. The major set A is
immediately attributed to the monomeric form (1m) of
Ar2BOH, on the basis of the chemical shift values (Table
Equilibrium 1 is slow, and its rate is affected by the
presence of traces of water (as will be discussed in the
(8) Kidd, R. J. In NMR of Newly Accessible Nuclei; Laszlo, P., Ed.;
Academic Press: New York, 1983; Vol. 2.
(9) See for instance: (a) Dagorne, S.; Guzei, I. A.; Coles, M. P.;
Jordan, R. F. J. Am. Chem. Soc. 2000, 122, 274. (b) Fraenk, W.;
Klapo¨tke, T. M.; Krumm B.; Mayer, P. Chem. Commun. 2000, 667. (c)
Kehr, G.; Fro¨hlich, R.; Wibbeling, B.; Erker, G. Chem. Eur. J. 2000, 6,
258. (d) Jacobsen, H.; Berke, H.; Do¨ring, S.; Kehr, G.; Erker, G.;
Fro¨hlich, R.; Meyer, O. Organometallics 1999, 18, 1724. (e) Vagedes,
D.; Fro¨hlich, R.; Erker, G. Angew. Chem., Int. Ed. 1999, 38, 3362. (f)
Parks, D. J.; Piers, W. E.; Parvez, M.; Atencio, R.; Zaworotko, M. J.
Organometallics 1998, 17, 1369. (g) Parks, D. J.; Piers, W. E.; Yap G.
P. A., Organometallics 1998, 17, 5492. (h) Sun, Y.; Piers, W. E.; Rettig,
J. R. Organometallics 1996, 15, 4110.
(5) (a) Ishihara, K.; Yamamoto, H. Eur. J. Org. Chem. 1999, 527.
(b) Ishihara, K.; Kurihara, H.; Yamamoto, H. J. Org. Chem. 1997, 62,
5664. (c) Ishihara, K.; Kurihara, H.; Yamamoto, H. Synlett 1997, 597.
(6) Recent patents dealing with synthesis or uses of 1 are: (a) Ikeno,
I.; Mitsui, H.; Iida, T.; Moriguchi, T. (Nippon Shokubai Co) Patent WO
0248156, 2002. (b) Ikeno, I.; Mitsui, H.; Iida, T.; Moriguchi, T. (Nippon
Shokubai Co) Patent WO 0244185, 2002. (c) Schottek, J.; Fritze, C.
(Targor) Patent DE 10009714, 2001. (d) Kratzer, R. (Basell Polyolefins)
Patent DE 10059717, 2001. (e) Kratzer, R.; Fritze, C.; Schottek, J.
(Targor) Patent DE 19962814, 2001. (f) Frances, J. M.; Deforth, T.
(Rhodia Chimie) Patent WO 0130903, 2001. (g) Schottek, J.; Fritze,
C.; Bohnen, H.; Becker, P. (Targor) WO 0020466, 2000. (h) Bohnen,
H.; Hahn, U. (Aventis R&T GMBH) Patent DE 19843055, 2000. (i)
Bohnen, H. (Hoechst A.-G.) Patent DE 19733017, 1999.
(10) The hypothesis that the observed association equilibrium
consisted in the formation of adducts with adventitious Lewis bases
(such as water) was immediately ruled out, due to the points i-iv and
to the fact that signals B are observed even in strictly anhydrous
conditions (see below).
(7) (a) Kratzer, R. (Basell Polyolefins) Patent WO 04041871, 2004.
(b) Kratzer, R. (Basell Polyolefins) Patent WO 04007570, 2004. (c)
Kratzer, R.; Fraaije, V. (Basell Polyolefins) Patent WO 04007569, 2004.
(11) (a) Fraenk, W.; Klapo¨tke, T. M.; Krumm B.; Mayer, P. Chem.
Commun. 2000, 667. (b) Parks, D. J.; Spence, R. E. von H.; Piers, W.
E. Angew. Chem., Int. Ed. Engl. 1995, 34, 809.

Role of Water in Oligomerization Equilibria
Organometallics, Vol. 23, No. 23, 2004 5495
Table 1. 1H and 19F NMR Data (δ and T1) for Compounds 1-3
compounda
m (A)
1t (B)
2 (C)
3 (D)
T (K)
δ 19F ortho [T1]b
δ 19F para [T1]b
δ 19F meta [T1]b
δ 1H [T1]b
1
283
283
283
184
-132.78 [1.95]
-138.27 [0.36]
-132.12 [0.55]
-132.5(1)d
-133.2(1)
-148.28 [1.89]
-151.24 [0.44]
-145.99 [0.55]
-151.5(1)
-152.0(1)
-153.0(1)
-160.70 [2.62]
-161.56 [0.57]
-161.14 [0.55]
-161.2(4)d
7.41 [1.6]
7.78 [0.5]
5.62(2)c
8.07(2)c
18.6(1)c
-162.2(1)
-138.3(1)
-139.9(1)
-141.9(1)
a In parentheses the label used in the text to indicate each set of resonances. b Longitudinal relaxation times T1 (s) measured at 253
K. c Attribution of the protonic signals, according to the labels of Chart 2: Ha 18.6, Hb 5.62, Hc 8.07. d The ortho and meta 19F resonances
here reported have been attributed on the basis of NOE and exchange cross-peaks in a 19F-19F EXSY/NOESY experiment and of the
integrated intensities. One ortho and one meta signal have not been identified with certainty.
following). In strictly anhydrous conditions, many hours
are required to attain the equilibrium even at a rela-
tively high temperature such as 283 K (see Figure S2
in the Supporting Information). From the equilibrium
values measured at different temperatures in the
experiment of Figure 2b, a good van’t Hoff plot (shown
in Figure S3, Supporting Information) was obtained.
Unfortunately, these experimental equilibrium values
are slightly biased toward 1t, due to the presence of
traces of water. Therefore, the thermodynamic values
estimated from the van’t Hoff plot (∆H° ) -69 kJ mol-1
,
∆S° ) -228 J K-1) do not correspond to the true
thermodynamic parameters of equilibrium 1, whose
determination was hampered by the sluggishness of the
equilibrium under anhydrous conditions.
Solvent Effect on the Oligomerization Equilib-
rium. As mentioned in the Introduction, in toluene
solution, versus CD2Cl2, the concentration of the trimer
1t is below detection limits, even at low temperatures.
The reasons for the different position of equilibrium 1
on varying the solvent are not obvious.
The different polarity should act in the opposite way,
because 1t has a lower polarity than 1m: B3LYP/6-
31G(d,p)//PM3 computations provided a value of 2.48
D for the monomer (Cs) and 0.75 D for the trimer (C2).
In agreement with this, we have found that in the
homologous series CH2Cl2, CHCl3, CCl4, the equilibrium
concentration of the trimer increases on decreasing
solvent polarity, according to the good linear free-energy
correlation shown in Figure 3 (K ) 3.9, 10.2, 44.5,
respectively, at 283 K, on passing from CH2Cl2 to CHCl3
to CCl4, with ET(30)12 ) 40.7, 39.1, 32.4, respectively).
Figure 2. Variation of the relative intensity of the signals
of 1t with respect to those of 1m (R, equilibrium values) on
varying (a) the overall concentration of 1 (at 278 K) and
(b) the temperature (in a 0.071 M solution).
We suppose that the stabilization of 1m in toluene,
despite the low polarity of this solvent, may be at-
tributed to preferential solvation of 1m with respect to
1t, due to B-OH‚‚‚arene interactions or to the higher
availability of its aryl rings for interacting with the
solvent aromatic rings.
Variable-Temperature Behavior: Spontaneous
Partial Dehydration. On lowering the temperature,
the 19F spectra revealed three main changes (Figure 4).
The first consists of splitting each of the signals of 1m
into two separate resonances of equal intensity, as
previously observed in toluene solution.4 This arises
from the freezing of the hindered rotation around the
B-OH bond, due to the partial double-bond character
Figure 3. Plot of the constant of equilibrium 1 versus the
polarity parameter ET(30) of the solvent.
of the B-O interaction,4 which removes the dynamic
equivalence of the two aryl rings of 1m. Band shape
analysis provided the kinetic parameters for the rotation
around the B-O bond, which are identical to those
estimated in toluene solution: ∆H# ) 41.8(5) kJ mol-1
(vs 42.3(3) in toluene), ∆S# ) 14(3) J K-1 mol-1 (vs 13(2)
in toluene), in agreement with the intramolecular
nature of the process.
(12) ET(30) is an empirical parameter of solvent polarity derived
from the wavelength of an intramolecular charge transfer band of a
suitable standard: Reichardt, C. Angew. Chem., Int. Ed. Engl. 1979,
18, 98.

5496 Organometallics, Vol. 23, No. 23, 2004
Beringhelli et al.
Figure 5. 1H DOSY spectrum of a CD2Cl2 solution of 1 (#
) solvent, 0.16 M, 170 K, 11.7 T).
of even small amounts of water to these solutions caused
the instantaneous disappearance of these signals, with
a corresponding increase of the signals of 1m.
2 Ar2BOH a Ar2BOBAr2 + H2O
(2)
In the absence of external water scavengers, the
driving force for dehydration of 1 is the formation at
low temperature of the species responsible for reso-
nances D. Such species has been formulated as the
adduct [(Ar2BOH)3‚H2O] (3) depicted in Chart 2, on the
basis of the experimental information summarized
below.
Figure 4. Variable-temperature 19F spectra of a solution
of 1 in CD2Cl2 (0.14 M).
The presence of three separate para signals indicates
that 3 contains three Ar2BOH units.16 The oligomeric
nature of this species has been confirmed by a 1H DOSY
experiment (performed at 170 K), which showed that
the diffusion coefficient of 3 was very similar to that of
Chart 2
1
1t (Figure 5). The H spectrum of 3 (three signals, in
the ratio 1:2:2, Table 1) suggested the presence of one
water molecule. In agreement with this, the addition
of B(C6F5)3 as water scavenger caused the instantaneous
and complete disappearance of the signals of 3, while a
(small) addition of water at low temperature (see below)
increased their relative intensity. The 1:2:2 pattern of
the protonic resonances implies (true or dynamically
generated) equivalence of two Ar2BOH units, i.e., C2
symmetry. The very low-field resonance of the singular
proton (δ 18.6) indicates that it is involved in a strong
hydrogen bond.
Structure and Dynamics of 3. The above evidence
would agree with both structures 3 and 4 of Chart 2.
However 2D 19F (Figure 6) and 1H (Figure 7) EXSY
experiments at 184 K showed exchange cross-peaks
between the resonances of 3 and those of 1m, while
negligible exchange cross-peaks were observed with 1t
at all the temperatures where species 3 was detectable.
This ruled out structure 4: indeed, if water was simply
H-bonded to the trimer, exchange with 1t and not with
1m should occur. Therefore in 3 the ring closure neces-
sary to account for the C2 symmetry occurs via a
hydrogen-bond interaction between a proton of coor-
dinated water and the oxygen of the Ar2BOH unit at
the opposite end of the trimeric chain. This hydrogen
bond, even if very strong (δ 18.6), is labile enough to
allow exchange of the terminal Ar2BOH unit with free
1m.
The other two changes observed in the 19F spectra
consist of the appearance, at temperatures lower than
210 K, of two new sets of resonances, labeled with C
and D in Figure 4 and Table 1. Even if their intensities
were always small (and somewhat variable in different
experiments), such signals are important, because they
give a hint of the complex equilibria in which the title
compound is involved in dichloromethane solution. The
appearance of resonances C and D was perfectly revers-
ible, and they always disappeared on increasing the
temperature.
Signals C (for which no protonic counterpart could
be identified) are attributable to the Ar2BOBAr2 anhy-
dride (2, Ar ) C6F5, Chart 1).13-15 Indeed these signals
appeared, even at room temperature, when we treated
dichloromethane or toluene solutions of 1 with powerful
water scavengers, such as B(C6F5)3 or P4O10, able to
drive to the right equilibrium 2. Moreover, the addition
(13) The synthesis of 2, by reaction of 1 with ClB(C6F5)2, and its
NMR data are reported in ref 14. 19F (toluene-d8): -132.6, -144.3,
-160.2 ppm. A recent patent6g gives the following values (C6D6
solution): -134, -146, -161 ppm.
(14) Tian, J.; Wang, S.; Feng, Y.; Li, J.; Collins, S. J. Mol. Catal. A:
Chem. 1999, 144, 137.
(15) The structure of 2 has been determined by X-ray crystal-
lography, and some metrical parameters are cited in: Piers, W. E.;
Irvine, G. J.; Williams, V. C. Eur. J. Inorg. Chem. 2000, 2131.
(16) The related ortho and meta signals are hardly detectable (Figure
S5 in Supporting Information), since they are broadened for the slowing
of the hindered rotation around the boron-carbon bonds, and some of
them are also buried under the stronger 1m and 1t resonances.

Role of Water in Oligomerization Equilibria
Organometallics, Vol. 23, No. 23, 2004 5497
Figure 6. 2D19F EXSY (region of the para fluorine atoms)
of a 0.076 M CD2Cl2 solution of 1 at 184 K.
Figure 8. 1H NMR spectra at 178 K of (a) a 0.10 M
solution of 1 in wet CD2Cl2 (#); (b) the same, after the
addition of 0.1 µL of water at 193 K.
that the lowest field para resonance of 3 has cross-peaks
with 1m weaker than those (of equal volumes) of the
other two resonances. This signal is then attributed to
ArC. Fast rotation of the water molecule around the
B-O bond accounts for the observed exchange between
proton Ha (δ 18.6) and proton Hb (δ 5.8) (Figure 7). The
exchange of both these protons with Hc was slower and
possibly mediated by the exchange with 1m (which in
turn exchanges faster with Ha and Hb than with the
“inner” protons Hc, Figure 7).
A {19F}1H experiment at 173 K showed that the large
bandwidth of the 1H signal of the two equivalent protons
Hc of 3 (δ 8.05) is due to coupling with 19F: this suggests
the presence of BO-Hc‚‚‚F-Caryl hydrogen bonds in-
volving ortho fluorine atoms of adjacent C6F5 rings, as
observed in the solid state structure of 1t.
Interaction of 1 with Water. If dehydration of 1m
was the only source of water present in 3, the anhydride
2 and the water adduct 3 should form in a 1:1 ratio. In
fact, the signals of 3 were always slightly more intense
than those of 2 (even if the ratio between 2 and 3 was
somewhat variable in different experiments). This can
be attributed to the unavoidable presence of adventi-
tious water. Indeed, the addition of water (at 193 K)
strongly increased the amount of 3, as depicted in
Figure 8 for 1H and in Figure S5 of the Supporting
Information for 19F. These figures show also that the
formation of 3 was accompanied by the decrease of the
concentration of 1m, while the concentration of 1t
remained unchanged: this clearly indicated that 3 was
formed at the expenses of 1m (eq 3).
Figure 7. 2D 1H EXSY spectrum of 1 (184 K, CD2Cl2 (#),
0.12 M). The asterisk marks an unidentified impurity.
There is evidence that this B(H)O‚‚‚Ha‚‚‚O(H)B hy-
drogen bond is not symmetrical (Chart 2); that is, the
C2 symmetry is not true, but dynamically generated.
Indeed at the lowest temperatures the 1H signal due to
Hb broadened and decoalesced into two very broad
separated resonances (as shown in Figure S4 of the
Supporting Information).17 Therefore, the observed C2
symmetry results from rapid proton migration between
the two oxygen atoms. This rapid oscillation makes in
turn each of the two involved Ar2BOH groups a chain
terminal (if the H-bond is ignored), ready to exchange
with free 1m. This mechanism implies that the aryl
groups ArA and ArB of Chart 2 should have a higher
exchange rate with 1m than those of the “inner” Ar2BOH
moiety (ArC). In agreement with this, Figure 6 shows
(17) Broadening and collapse was observed also in the para 19F
region, even if in this case the conditions for decoalescence were not
achieved.
3 Ar2BOH + H2O a [(Ar2BOH)3‚H2O]
(3)

5498 Organometallics, Vol. 23, No. 23, 2004
Beringhelli et al.
slightly but significantly increased at the initial titration
stages (see Figure 10a). The interaction of water with
1t most likely occurs through H-bonding, to give adduct
4 of Chart 2, rather than species 3. Indeed the formation
of the latter species is expected to shift the signals of
1
m rather than those of 1t, since at these temperatures
its exchange with 1m (detectable even at 173 K) should
be so fast to dynamically average their signals.20
(iii) The stabilization of the trimeric form resulting
from its interaction with water was not strong enough
to allow the complete transformation of the monomer
into the trimer (at least at 283 K). Then, on continuing
with the titration, the initial trend was reversed: the
relative intensity of the signals of the trimeric species
(1t plus 4) progressively decreased, and strong (upfield)
shifts were shown by the 19F and 11B signals of 1m
(Figures 9 and 10). Therefore at this point of the
titration added water was captured mainly by 1m, so
that its adduct/adducts with water became the domi-
nant species.21 Above 2 equiv the concentration of
trimeric species was negligible.
Figure 9. 19F NMR monitoring of a titration with water
of a CD2Cl2 solution of 1 (0.18 M) at 283 K (only the region
of the para fluorine atoms is shown, for the sake of clarity).
The figures on the right indicate the equivalents of added
water at each titration step.
The plots of Figures 10b-d suggest that 1m may be
able to bind up to two water molecules. As to the nature
of these adducts, the 11B spectra showed that, as soon
as water uptake by 1m became significant, the 11B signal
of the monomer began shifting from the region of tri-
to that of tetracoordination (Figure 10d). Therefore 1m
binds water through a covalent Ar2(OH)B-OH2 Lewis
acid-base bond, rather than a hydrogen-bond interac-
tion, to give the adduct 5 of Scheme 1.22 Most likely,
the hydrogen bond is responsible for the binding of the
second water molecule, as shown in Scheme 1.23
Moreover, the simultaneous formation of the anhy-
dride 2 and of 3 can be depicted with the overall
equilibrium 4, taking into account that the trimer-
monomer interconversion is frozen (even in the presence
of water) at temperatures where such reaction is
observed. The position of this equilibrium shifts to the
right only at temperatures low enough to make less
important the very unfavorable entropic term. This
explains the reversibility of the formation of 3 and 2 on
varying the temperature.
Role of Water in the Oligomerization Equilibri-
um. The titration experiments have shown that the
amount of water present in solution affects the posi-
5 Ar2BOH a Ar2BOBAr2 + [(Ar2BOH)3‚H2O] (4)
Further information on the interaction of 1 with water
was provided by a titration performed directly into a
NMR tube, at 283 K, monitored by 19F (Figure 9), 11B,
(20) The 1H NMR spectra provided another piece of evidence
supporting the idea that the traces of water are preferentially H-bonded
to 1t. Even if the 1H spectra (283-253 K) are useless at relatively high
water content (due to the presence of broad averaged resonances),18
separated resonances for trimeric and monomeric species can be
observed when water content is really low. In these cases the change
of their δ on varying the amount of water was significantly different
(see for instance Figure S6): the signal of the trimer shifted to higher
frequency, as expected for the increase of the fraction of 1t involved in
hydrogen-bonding interaction. By contrast, the signal of the monomer
became very broad and shifted to higher field. The latter behavior,
observed even in the absence of significant 1m‚H2O adduct formation
(as revealed in the case of Figure S6, step b, by the comparison with
the corresponding 19F spectrum), must be attributed to fast proton
exchange with the water molecules (mostly H-bonded to 1t, but in
minor amount covalently bonded to 1m or even free). Noteworthy, it
has been impossible to identify the resonances of water interacting
with either 1m or 1t, most likely due to such easy exchange. The
comparison of the 1H and 19F integration ratios between 1m and 1t
supports the idea that the protonic signal of 1m incorporated the water
signal. The addition of molecular sieves directly into the NMR tube
reversed the effects of water addition: the signal of the trimer shifted
back to higher field, and that of the monomer sharpened.
(21) According to a (probably oversimplified) model involving equi-
librium 1 (K1) and the equilibria of water uptake by 1t (K2) and 1m
(K3), the ratio between trimeric (1t+4) and monomeric (1m plus its
water adduct) species should be proportional to [1m]2(1 + K2[H2O])/(1
+ K3[H2O]), the proportionality constant being (K1K2/K3). In the course
of the titration, the fractional term containing [H2O] slightly increases
(if 1 was not negligible with respect to Ki[H2O] and K2 > K3), but [1m]
decreases, and eventually the latter effect becomes dominant, making
progressively smaller the concentration of the trimeric species.
(22) The concentration of 3 (and of the other oligomeric species
arising from the association equilibria involving 5 discussed below and
depicted in Scheme 2) is negligible at a temperature as high as 283 K.
Moreover, the importance of these oligomerization equilibria is obvi-
ously progressively reduced as the increase of the amount of water
reduces the concentration of free 1m.
1
and H NMR. The scenario appearing from these data
is as follows.18
(i) Water addition did not cause the appearance of
novel signals,19 but rather a strong shift of the reso-
nances of 1m and 1t. This implies that 1m and 1t are in
fast exchange with their respective water adducts, so
that molar fraction weighted averaged resonances are
observed.
(ii) At the very beginning of the titration (up to ca.
0.3 equiv of water) the 19F δ shift was stronger for 1t
than for 1m. This indicates that water, when present in
low concentration, interacts preferentially with 1t,
resulting in the shift of equilibrium 1 toward the
trimeric form. Indeed, the relative amount of the trimer
(18) The same behavior was observed in an analogous titration
experiment at lower temperature (253 K). At both temperatures, the1H
NMR monitoring provided poor information, since the addition of even
very small amounts of water produced one broad averaged resonance,
which progressively sharpened on further increasing the amount of
water (see Figure S6): the comparison of the 1H and 19F data shows
that water promotes proton exchange processes, whose rate increases
with water content. This hampered the identification of the resonances
of the water molecules interacting with borinic acid.
(19) The only novel species observed was pentafluorobenzene,
formed in very low concentration from the transformation of some 1
into (C6F5)B(OH)2 and then into B(OH)3. This reaction is known to
slowly occur in the presence of water (Bradley, D. C.; Harding, I. S.;
Keefe, A. D.; Montevalli, M.; Zheng, D. H. J. Chem. Soc., Dalton Trans.
1996, 3931). The 19F signals of the boronic acid are not observed due
to its poor solubility.

Role of Water in Oligomerization Equilibria
Organometallics, Vol. 23, No. 23, 2004 5499
Scheme 1. Stepwise Addition of Water on 1m
tion24 and the bandwidth25 of the signals (since they
become molar fraction weighted averaged resonances)
and also modifies the position of the 1m/1t equilibrium.
As mentioned above, the determination of the true
values of the constants of this equilibrium has been
hampered by the extreme slowness of the monomer-
trimer interconversion in the absence of water.
Therefore water plays a catalytic role in this multistep
equilibrium, which can be explained as follows. The
oligomerization process implies nucleophilic attack on
a boron atom by the oxygen atom of another monomeric
unit. The nucleophilicity of the oxygen atom of 1m is low,
due to the intramolecular pπ-pπ OfB electron dona-
tion.4 Water coordination on 1m destroys this interac-
tion, making the nonbonding electrons of the OH group
more available for intermolecular interaction with
another boron center. Indeed, a comparison of the
electrostatic potential maps for the monomer 1m and
the water adduct 5 (Figure 11) clearly evidences an
enhanced nucleophilicity of the oxygen atom in 5.
This opens alternative pathways for the transforma-
tion of 1m into 1t: the mechanism for this oligomeriza-
tion process, both in the absence and in the presence of
water, has been investigated by means of PM3 calcula-
tions. Scheme 2 reports the proposed steps and inter-
mediates, together with relative energies computed at
the B3LYP/6-31G(d,p)//PM3 level of theory. We must
remark that computed energy values have to be con-
sidered with caution, due to the known limitations of
PM3 in describing hydrogen-bond and transition struc-
ture geometries.26 Moreover, reported energies do not
take into account the entropic contribution, which is
expected to be important for an association process, but
difficult to estimate, particularly for solution phase
reactions. However, the overall qualitative picture
depicted by the above computations well agrees with the
main experimental findings.
The absence of enthalpic barriers for the formations
of the water adducts 5 and 4 is in line with the fast
(23) It is worthwhile to remark that the formation of the two water
adducts did not occur in the rigorous stepwise manner observed for
the adducts between B(C6F5)3 and water.1 Indeed, the shift of the 11
B
signal was not complete during the first titration step (Figure 10d)
and the change of the slope of the 19F titration plots in correspondence
with 1 equiv of water was less sharp than in the case of the analogous
titration of B(C6F5)3 with water.1 Furthermore, the δ shift of the 19F
and 11B signals did not stop exactly in correspondence with 2 equiv of
water, indicating that an excess of water is necessary to completely
drive to the right the association equilibria. At further variance with
respect to what is observed for B(C6F5)3, the water molecule in 5 was
labile enough to allow very fast water exchange between 1m molecules,
resulting in averaged resonances for 1m and its water adduct 5. By
contrast, B(C6F5)3 and (C6F5)3B-OH2 gave separated signals at tem-
peratures lower than 273 K.1,2a
(24) The 19F and 1H δ values of 1m and 1t reported in Table 1 have
been determined in the presence of B(C6F5)3, as water scavenger.
(25) The progressive sharpening of the averaged resonances of 1m
observed in the course of the titration monitored at 283 K (Figure 9)
is likely attributable both to the increased water mobility and to the
decreased ∆δ between the exchanging sites (revealed by the decrease
of the slopes of Figure 10), as previously observed for B(C6F5)3.1
(26) Jensen, F. Introduction to Computational Chemistry; Wiley:
Chichester, U.K., 1999; Chapters 3 and 11.
Figure 10. Titration of 1 with water, at 283 K: (a) change
of the relative intensity of the 19F signals of 1t with respect
to those of 1m (R) in the titration course; (b-d) simulta-
neous variation of the chemical shifts of the 19F para (b),19F
ortho (c), and 11B resonances (d) of the monomeric ([) and
of the trimeric species (2) (averaged signals for 1m and its
water adducts, and for 1t and its water adducts).

5500 Organometallics, Vol. 23, No. 23, 2004
Beringhelli et al.
Scheme 2. Oligomerization Equilibrium in the
Absence (left) and Presence (center and right) of
Watera
Figure 11. Space-filling models and electrostatic potential
mapped onto the electron density isosurfaces (0.0004 au)
for the monomer 1m and the water adduct 5 (B3LYP/6-
31G(d,p)//PM3). Dark red (blue) corresponds to -0.03
(+0.03) au.
exchanges (leading to averaged NMR signals) observed
between 1m or 1t and their respective adducts 5 and 4.
The water adduct 5 bears both a reactive OH group
(see Figure 11) and a coordinated water molecule prone
to be involved as a donor in a hydrogen bond: the
presence of these two reactive sites makes easier the
association of monomeric units and leads at first to the
formation of the dimeric intermediate I1 of Scheme 2
and eventually to the trimer 3.
The species 3 appears kinetically inert with respect
to its transformation into the closed cyclic trimer 1t or
4. Indeed its isomerization into I2 is accompanied by a
quite high energy barrier. This can be explained observ-
ing that the strong hydrogen bond present in 3 has been
replaced in the intermediate I2 by a weaker O-H‚‚‚OH2
interaction. The subsequent substitution reaction at the
boron atom that leads to 4 shows a quite high energy
barrier as well. Therefore 3 is not a good intermediate
in the monomer to trimer pathway.
A better pathway for the 1m a 1t interconversion is
provided by the formation of the oligomeric open species
I3 and I4 depicted in the center of Scheme 2. In
particular, in I4 the water molecule and the OH group
show the correct relative position to undergo a nucleo-
philic substitution: the energy barrier is much lower
(by 75 kJ mol-1) than computed in the absence of water
(left part of Scheme 2) and fully accounts for the
observed catalytic role of water. Attempts to find a direct
interconversion pathway between the open I4 and the
hydrogen-bonded species 3 failed, because the opening
requires rotation around the H2OB-O(H)B bond, ham-
pered by the presence of the two bulky pentafluoro-
phenyl substituents on the boron atom.
a
[B] refers to the (C6F5)2B moiety. Relative energies (kJ
mol-1) computed at the B3LYP/6-31G(d,p)//PM3 level are given
in parentheses.
very low temperatures only, for entropic reasons) from
1m, while its formation by cycle opening from 1t (or 4)
is strongly unfavored. Curiously, the formation of 3 is
generally accompanied by a decrease of the concentra-
tion of 4, but this results from water transfer and not
from direct water insertion into the cycle. Indeed, at
room temperature (or slightly lower) adventitious water
necessary for the formation of 3 is bound to the trimer
(giving 4). So, on lowering the temperature below 230
K, we observe the progressive shift of the 19F and H
1
signals of the trimer (averaged signals 1t plus 4) toward
their “anhydrous” values (as shown in Figure S7 in the
Supporting Information), due to the uptake of the
released water by 3. The formation of 3 from adventi-
tious water is then better described by equilibrium 5
rather than equilibrium 3. On increasing back the
temperature, the chemical shifts varied in the opposite
way, indicating the reversal of this equilibrium.
The relative kinetic barriers and enthalpic stabilities
of Scheme 2 also agree with the fact that 3 was the only
detected species, among all the dimeric and trimeric
water-containing intermediates of Scheme 2.
4 + 3 1m a 1t + 3
(5)
The experimental data clearly indicated that at low
temperatures the 1m a 1t equilibrium became exceed-
ingly slow. In other words, in these conditions the
communications between closed trimers (1t plus 4) and
monomers (1m plus 5) are blocked. The latter species,
however, maintain their communication with 3 (and the
easy exchange between 3 and 1m, revealed by the EXSY
at 173 K, indicates that at least the step from I1 to 3
must be fast).
Conclusions
The present investigation has shown that the trimeric
form of the title compound is not peculiar to the solid
state but can also be present in solution, the trimeriza-
tion being favored in nonaromatic solvents of low
polarity. A solvent of relatively high polarity like
dichloromethane permits trimer concentrations suf-
ficiently high to allow the study of the oligomerization
equilibrium.
Finally, both the experimental data and computa-
tional results agree with the fact that 3 is formed (at

Role of Water in Oligomerization Equilibria
Organometallics, Vol. 23, No. 23, 2004 5501
mmol) was dissolved in CD2Cl2 (0.6 mL) in a NMR tube, and
19F NMR spectra were acquired at 283 K at different times
until the 1m/1t ratio remained constant. The sample was
diluted by CD2Cl2 addition (0.32 mL), and the equilibration
was then monitored at the same temperature. Then the
temperature was lowered (at first at 273 K, and successively
at 263 and 253 K), and at each temperature the equilibration
was monitored, resulting in the data of Figure 3.
To study the effect of concentration, the equilibration
process was monitored by 19F NMR at 278 K on another sample
of 1 (initially 0.0771 M), treated with increasing amounts of
1, affording solutions of 0.110, 0.140, 0.186, and 0.202 M
(Figure 2).
The ratio between monomeric and trimeric forms
present in CD2Cl2 solution was found very variable,
depending not only on thermodynamic (temperature,
concentration) but also on kinetic factors (the elapsed
time at a given temperature, and even the rate by which
this temperature was reached). Indeed, the attainment
of equilibrium is very slow, and this greatly hampers
the determination of the equilibrium concentration at
temperatures lower than room temperature. This is
attributable to the intramolecular oxygen-to-boron π-
donation, which reduces both the Lewis acidity of boron
and the Lewis basicity of oxygen.
Dynamics of 1m. Variable-temperature 19F spectra were
recorded on a 0.141 M sample of 1. The kinetic constants for
the restricted rotation around the B-O bond of 1m were
obtained by band shape analysis, using the coupling constants
JFF, estimated from the simulation of a room-temperature
spectrum with Bruker WINDAISY software (J23 ) -23.0, J24
) -4.2, J25 ) 9.8, J26 ) -4.8, J34 ) -20.5, J35 ) 0.2 Hz): k,
s-1 (T, K): 24 (183), 90 (192), 280 (201), 1090 (211), 3400 (221).
In the band shape analysis the contribution of the exchange
between 1m and 3 was neglected, since the rate of the latter
process, as estimated from 2D EXSY volumes, was about 1
order of magnitude smaller than that of the intramolecular
process in 1m.
Moreover, the presence of water affects both the
kinetics and the thermodynamics of this equilibrium.
Water catalyzes the attainment of equilibrium due to
the increased nucleophilicity of the oxygen atom upon
water coordination. Semiempirical computations have
fully confirmed this experimental finding, allowing us
to depict a trimerization pathway with a lower kinetic
barrier than under anhydrous conditions. The same
computations have also confirmed the observed stabi-
lization of the trimeric form in the presence of traces of
water, due to the formation of a stable hydrogen-bonded
adduct containing exocyclic water.
The preferential H-bond interaction of water with the
trimer rather than with the monomer is attributable to
the higher effective electronegativity of an oxygen atom
bound to two acidic boron centers. On the other hand,
the preferential formation of an H-bond interaction
(with 1t) rather than a Lewis acid-base covalent bond
(with 1m) confirms the low Lewis acidity of the boron
atom in 1m.
The capability of modifying both the rate and the
position of the oligomerization equilibrium should be a
property that water shares with other Lewis bases. This
will be the subject of future investigations.
Peculiar of water, on the contrary, is the capability
of generating the trimeric adduct 3 containing a water
molecule within an octaatomic O-B-O-B-O-B-O-H
cycle, closed by a very strong hydrogen bond. The
stabilization of the (simultaneously H-bonded and B-
bonded) water molecule in 3 is strong enough to promote
water formation, by dehydrating borinic acid to its
anhydride. Low-temperature CH2Cl2 solutions of the
title compound can therefore be considered water scav-
engers more powerful than the anhydride 2 itself.
Addition of Water Scavengers to Two Solutions of 1.
A 0.10 M solution of 1 in CD2Cl2 was treated in a NMR tube
with a few molecular sieves. 1H and 19F spectra were recorded
at 278 K every 5 min for 3 h. The spectra (data not shown)
1
revealed (i) a progressive slight upfield shift of the H signal
of the trimer (due to the slow conversion of 4 into 1t); (ii) the
simultaneous decrease of the relative intensity of the signal
of the trimer with respect to the monomer (due to the loss of
the stabilization of the trimer upon hydrogen bonding with
water); and (iii) the sharpening of the 1H signal of 1m (the
broadening being due to intermolecular proton exchange
processes). No significant formation of anhydride was observed
in this time. Then, solid B(C6F5)3 (ca. 0.3 equiv) was added
directly in the NMR tube and 1H and 19F were recorded,
showing (i) the formation of some B(C6F5)3‚OH2 adduct; (ii) a
slight further shift upfield of the protonic signal of 1t; and (iii)
the formation of a significant amount (0.12 equiv) of anhydride
2. In another experiment, different aliquots of P4O10 were
added directly into a NMR tube containing a 0.069 M solution
of 1 in CD2Cl2: each addition caused a sudden increase of the
amount of the anhydride 2, up to the conversion of ca. 65% of
1
1 into 2, as revealed by the H and 19F spectra at 270 K.
Titration with Water. A 0.18 M sample of 1 was prepared
by dissolving 35.5 mg of 1 in 0.54 mL of CD2Cl2 in a NMR
tube. Water additions (0.1-0.5 µL, up to 3.5 equiv) were made
at room temperature, and 1H, 19F, and 11B NMR spectra were
recorded at 283 K after each addition (Figures 9 and 10). An
analogous experiment (without 11B monitoring) was performed
at 253 K.
Computational Study. Geometry optimizations have
been performed at the PM3 semiempirical level, employing
SPARTAN 02.28 No symmetry has been imposed; however 1m
and 1t possess Cs and C2 symmetry, respectively, and the
overall structures of I1, 3, and 4 deviate only marginally from
C2 symmetry (in I1 and 3 this is due to the asymmetry of the
hydrogen bond and in 4 to the position of the water molecule
with respect to the pseudo-C2 axis). Optimizations have been
done employing the Hessian matrix computed at the PM3
level; the convergence criteria in geometry optimization for the
maximum gradient component and the maximum change in
a bond length have been set to 10-5 au and 10-4 Å. The nature
(true minima or first-order transition structures) of all the
stationary points found have been determined by computing
Experimental Section
All manipulations were performed under N2 or Ar using
oven-dried Schlenk-type glassware. CD2Cl2 (C.I.L.) was an-
hydrified on activated molecular sieves. B(C6F5)2OH was a gift
from Basell Polyolefins.
All NMR spectra were acquired on a Bruker AVANCE DRX-
300 spectrometer, equipped with a 5 mm TBI probe (90°
pulse: 1H 8.5 µs, 19F 24 µs, 11B 9 µs) or with a 5 mm QNP
probe (90° pulse: 1H 7.0 µs, 19F 8.6 µs) and on a Bruker AC300
spectrometer equipped with a 5 mm BBO probe (90° pulse:
1H 7.2 µs, 11B 6.6 µs). 19F NMR spectra were referenced to
external CFCl3. The DOSY experiment and a low-temperature
1H spectrum shown in Figure S4 have been recorded at 11.7
T on a Bruker AMX-500 WB spectrometer. The temperature
was calibrated with a standard CH3OH/CD3OD solution.27
Equilibration Experiments at Different Concentra-
tions and Temperatures. A sample of 1 (23.8 mg, 0.066
(27) Van Geet, A. L. Anal. Chem. 1970, 42, 679.
(28) SPARTAN 02; Wavefunction Inc.: Irvine, CA, 2002.

5502 Organometallics, Vol. 23, No. 23, 2004
Beringhelli et al.
the Hessian matrix. Some of the transition states have been
located driving some internal coordinates (mainly bond dis-
tances) corresponding to the expected reaction coordinates.
Successively, the eigenvector associated with the negative
eigenvalue has been followed in both directions (by minimizing
two slightly distorted structures) to ascertain that the transi-
tion structure links the two expected minima. Single-point
computations of the energies at the B3LYP/6-31G(d,p) level
have been done with GAUSSIAN 98.29
edge Prof. Maurizio Sironi for the use of SPARTAN 02.
Thanks are also due to Italian CNR (ISTM) for provid-
ing facilities for inert atmosphere and low-temperature
experiments and to Mrs. M. Bonfa` for the experiments
at 11.7 T.
Supporting Information Available: Seven figures show-
ing details of the NMR experiments. This material is available
free of charge via the Internet at http://pubs.acs.org.
OM0494519
Acknowledgment. T.B. and G.D. thank Basell Poly-
olefins for partial funding of this work and for providing
samples of the title compound. P.M. and A.S. acknowl-
(29) GAUSSIAN 98 (revision A.11.3); Gaussian Inc.: Pittsburgh, PA,
2002.