Substituent effects on aryltrifluoroborate solvolysis in water: Implications for Suzuki-Miyaura coupling and the design of stable 18F-labeled aryltrifluoroborates for use in PET imaging

Article abstract of DOI:10.1021/jo800681d

(Chemical Equation Presented) Whereas electron withdrawing substituents retard the rate of aryltrifluoroborate solvolysis, electron-donating groups enhance it. Herein is presented a Hammett analysis of the solvolytic lability of aryltrifluoroborates where log(K_solv) values correlate to a values with a ρ value of approximately -1. This work provides a predictable rubric for tuning the reactivity of boron for several uses including ¹⁸F-labeled PET reagents and has mechanistic implications for ArBF₃-enhanced ligandless metal-mediated cross coupling reactions with aryltrifluoroborates.

Full text of DOI:10.1021/jo800681d

Substituent Effects on Aryltrifluoroborate Solvolysis in Water:
Implications for Suzuki-Miyaura Coupling and the Design of Stable
18F-Labeled Aryltrifluoroborates for Use in PET Imaging
Richard Ting, Curtis W. Harwig, Justin Lo, Ying Li, Michael J. Adam, Thomas J. Ruth, and
David M. Perrin*
Chemistry Department, UniVersity of British Columbia, 2036 Main Mall, and TRIUMF VancouVer,
B.C. V6T-1Z1, Canada
dperrin@chem.ubc.ca
ReceiVed March 26, 2008
Whereas electron withdrawing substituents retard the rate of aryltrifluoroborate solvolysis, electron-donating
groups enhance it. Herein is presented a Hammett analysis of the solvolytic lability of aryltrifluoroborates
where log(ksolv) values correlate to σ values with a F value of approximately -1. This work provides a
predictable rubric for tuning the reactivity of boron for several uses including 18F-labeled PET reagents
and has mechanistic implications for ArBF3-enhanced ligandless metal-mediated cross coupling reactions
with aryltrifluoroborates.
Introduction
arylboronic acid and free fluoride, for which the free energy of
solvation in water is calculated to be at least on par with that
of B-F bond formation.8
For nearly two decades, aryltrifluoroborates (ArBF3s) have
received considerable attention in Suzuki-Miyaura cross-
coupling reactions.1–7 As boronic acids/ester equivalents, ArBF3s
are increasingly preferred for their facile preparation from the
former in aqueous KHF2, their ready isolation via precipitation
from water- and ether-miscible solvents (e.g., alcohols, aceto-
nitrile, acetone), and their resistance to either oxidation or
boroxine formation. Many of these desirable characteristics have
been attributed to the alleged stability of the ArBF3, which, in
turn, is ascribed to the exceptional strength of the B-F bond
(∼140 kcal/mol). Nevertheless, under sufficiently dilute condi-
tions, an isolated ArBF3 must suffer thermodynamically favor-
able and kinetically irreversible solvolysis that restores the
In addition to the extensive reports on the use of ArBF3s in
synthetic chemistry, at least one report has suggested that
ArBF3s may act as protease inhibitors,9 whereas another
suggested that arylboronic acids could be useful captors of
aqueous [18F]-fluoride that would conveniently provide 18F-
labeled PET-imaging agents by virtue of a pendant 18F-
containing ArBF3.10 Because in ViVo use of ArBF3s as phar-
maceuticals and radiopharmaceuticals demands dilution to
micromolar levels or lower, resistance to solvolytic defluori-
dation would be of paramount importance. Conversely, for Pd-
catalyzed cross-couplings, which are run in aqueous cosolvents
such as methanol, dimethylformamide, tetrahydrofuran, or
acetonitrile at moderate-to-high concentrations of ArBF3 (10-100
mM), it has been postulated that defluoridation necessarily
precedes metalation.1,11
(1) Molander, G. A. Biolatto B. J. Org. Chem. 2003, 68, 4302–4314.
(2) Wright, S. W.; Hageman, D. L.; McClure, L. D. J. Org. Chem. 1994,
59, 6095–6097.
(3) Vedejs, E.; Chapman, R. W.; Fields, S. C.; Lin, S.; Schrimpf, M. R. J.
Org. Chem. 1995, 60, 3020–3027.
(4) Molander, G. A.; Bernardi, C. R. J. Org. Chem. 2002, 67, 8424–8429.
(5) Batey, R. A.; Quach, T. D. Tetrahedron Lett. 2001, 42, 9099–9103.
(6) Darses, S.; Genet, J.-P. Chem. ReV. 2008, 108, 288–325.
(7) Stefani, H. A.; Cella, R.; Vieira, A. S. Tetrahedron 2007, 63, 3623–
3658.
(8) Zhan, C.-G.; Dixon, D. A. J. Phys. Chem. A 2004, 108, 2020–2029.
(9) Smoum, R.; Rubinstein, A.; Srebnik, M. Org. Biomol. Chem. 2005, 3,
941–944.
(10) Ting, R.; et al, J. Am. Chem. Soc. 2005, 127, 13094–13095.
(11) Barder, T. E.; Buchwald, S. L. Org. Lett. 2004, 6, 2649–2652.
4662 J. Org. Chem. 2008, 73, 4662–4670
10.1021/jo800681d CCC: $40.75 2008 American Chemical Society
Published on Web 05/20/2008

Substituent Effects on Aryltrifluoroborate SolVolysis in Water
SCHEME 1. Minimal Kinetic Scheme for ArBF3 Solvolysis Where All Steps Are Taken To Be Reversiblea
a Not shown are all ionized states for 3-5, or a tautomer of 4, all of which may also react such that this scheme is complicated by branched pathways.
The pH-dependent hydration of the arylboronic acid to the arylborate is also not shown.
Thus, characterization of substituent effects that govern the
kinetics of solvolytic loss of fluoride ion would have broad
mechanistic significance for the use of ArBF3s in numerous
applications. Whereas Anbar and Guttmann found the notori-
ously inert tetrafluoroborate to solvolyze very slowly in water
(t1/2 ≈ 168 h, 100 °C) in a study that highlighted the strength
of the B-F bond,12 to date, the aqueous stabilities of a few
ArBF3s have been but briefly described with only brief mention
made of their allegedly high aqueous stability.13–15 There exists
neither a quantitative measure of the solvolytic lability of an
ArBF3 nor any knowledge as to how electronic structure would
influence it. Herein, we use both NMR spectroscopy and 18F/
19F-fluoride isotopic exchange autoradiography in this first study
reasons, we used phosphate buffer at a pH close to neutrality
to minimize the effects of HF evolution, which would also
enhance a back reaction and only diminish the observed rate of
solvolysis. Our interests in the biological applications of ArBF3s
led us to investigate solvolysis in aqueous 192 mM phosphate
buffer pH 6.9-7.0. In this initial study, we did not treat the
effects of less polar cosolvents such as methanol, THF, and
DMF, which are often used in metal coupling reactions. Under
these chosen conditions, the kinetics simplify to the reaction
shown in Scheme 2 (see the Supporting Information for a more
elaborate pathway).19
SCHEME 2. Irreversible Kinetic Scheme for ArBF3
Solvolysis under Buffered Conditions at Relatively High
Dilutiona
that quantitatively measures the rate constant for solvolysis, kobs
,
which can be fit to a simple first-order equation and which
depends quite reliably on the electronic nature of substituent
effects at the meta and para positions.
Results
In studying the solvolysis of ArBF3s, we initially considered
both the forward and reverse reactions shown schematically in
Scheme 1. A priori, at high concentrations of ArBF3, one must
entertain a difficult set of presteady-state differential equations
to account for product inhibition due to HF evolution and the
feedback effects of solvent pH depression in the absence of
buffer. To simplify this scheme, we chose to study solvolysis
under conditions that would (i) be relevant to those commonly
used by researchers working with ArBF3s (e.g., aqueous buffered
conditions) and (ii) simplify the kinetic analysis.
We chose not to treat the kinetics of ArBF3 formation because
ArBF3s are generally formed under one set of conditions (usually
molar concentrations of ArB(OR)2 and KHF2, low pH, R ) H,
alkyl) and then used for coupling or in ViVo targeting under
other more dilute conditions: for the former at moderate dilution
(1-100 mM) in mixed aqueous media (pH 8-10 adjusted with
K2CO3), or for the latter at very high intravenous dilution (blood,
pH 7.4). Thus for both uses, solvolytic loss of fluoride, starting
with the first fluorine atom, would arguably be considered the
most important step because this step is on the one hand thought
to be essential for transmetalation in Pd-catalyzed cross
coupling16–18 and on the other thought to be undesirable if an
ArBF3 is to be contemplated for in ViVo imaging.
a Not shown are the fleeting intermediates 2-5 from Scheme 1, nor
additional ionized states for 3-5 which may react differently according to
the ionization state.
In Scheme 2, fluoride release is governed by a single rate-
limiting step, without detection of any intermediates shown in
Scheme 1. If any of intermediates 2-5 were to accumulate due
to at least one more slow step then (i) the rate of fluoride release
could be stepwise, (i.e., biphasic or triphasic), sigmoidal, or even
more complicated, and (ii) 19F NMR spectroscopy would
identify the steady-state accumulation of another boron-bound
fluoridated intermediate. Nonetheless, no intermediate is ob-
served and both the time-dependent appearance of free fluoride
and the disappearance of the ArBF3 is pseudofirst order where
kobs approximates k1, which must be much smaller than all other
rate constants (i.e., k1 , k2 ≈ k3 ≈ k4 ≈ k5) that govern
intermediate solvolysis. The use of low concentrations of ArBF3
(qualitatively reflective of in ViVo conditions) not only simplified
the kinetics but ensured that the kobs we report represents the
maxmium observable value since it is not diminished by the
presence of intermediate products that might accumulate at
higher initial concentrations of ArBF3 (e.g., ∼100 mM used
during Pd-coupling).
Thus we chose conditions of reasonable dilution (∼5 mM
ArBF3) that would favor the forward hydrolysis reaction and
disfavor the back reaction of fluoride recapture. For similar
We measured solvolysis by 300 MHz 19F NMR spectroscopy
that is sensitive enough to detect 19F-chemical shifts that would
correspond to the starting ArBF3, any possible mono- and
difluorinated intermediates, and free fluoride. 11B NMR spec-
troscopy proved less reliable because considerably more sample
was required and we wished to work at low millimolar
concentrations. In this study, all ArBF3s were prepared from
(12) Anbar, M.; Guttmann, S. J. Phys. Chem. 1960, 64, 1896–1899.
(13) Yuen, A. K. L.; Hutton, C. A. Tetrahedron Lett. 2005, 46, 7899–7903.
(14) Moreno-Manas, M.; Perez, M.; Pleixats, R. J. Org. Chem. 1996, 61,
2346–2351.
(15) Smoum, R.; Rubinstein, A.; Srebnik, M. Org. Biomol. Chem. 2005, 3,
941–944.
(16) Molander, G. A.; Biolatto, B. J. Org. Chem. 2003, 68, 4302–4314.
(17) Molander, G.; Biolatto, B. Org. Lett. 2002, 4, 1867–1870.
(18) Barder, T. E.; Buchwald, S. L. Org. Lett. 2004, 6, 2649–2652.
(19) Although 2 could lose a fluoride directly via a borfluoronium ion, which
would hydrate to 4; likewise 4 could lose a fluoride directly via a boroxyl, which
would hydrate to 6 (see the Supporting Information).
J. Org. Chem. Vol. 73, No. 12, 2008 4663

Ting et al.
FIGURE 1. Left: 19F NMR spectra of p-sulfonamidophenyltrifluoroborate solvolysis at 1, 6, 28, 76, and 144 min. Right: Exponential fit of the
NMR spectral data.
1
arylboronic acids, purified, and characterized by H and 19F
NMR spectroscopy in CD3OD or DMSO-d6 (see the Supporting
Information). At 2-10 mM ArBF3, the buffer (192 mM
phosphate pH 6.9-7.0) neutralized the 2 equiv of evolved HF.
An example of such kinetics is shown in Figure 1 where the
integration values for p-NH2SO2-ꢀ-BF3 (ca. -65 ppm) and free
fluoride (ca. -40 ppm) were fit to the equation ([ArBF3]/
with p-sulfonamidophenyltrifluoroborate to investigate the effect
that excess fluoride, which represents product, would have on
the reaction. The value for kobs in the presence of fluoride was
approximately 60% of that in the absence of excess fluoride
(Figure 2).
This small rate depression may be explained either by a
specific reaction of fluoride with intermediate 2 forcing it back
to starting material or by a medium effect related to an increase
in ionic strength, or both. Although medium effects were not
investigated further, it is clear that the addition of excess fluoride
at 100 mM did not reveal any intermediates (2-5) and had little
overall effect on the kinetic or thermodynamic stability of the
ArBF3.
To independently address the rate of fluoride loss, we
prepared the p-sulfonamidophenyltrifluoroborate in the presence
of trace [18F]-fluoride. After an hour of synthesis, the crude
reaction was diluted 200-fold in 100 mM [19F]-KF and 192 mM
phosphate buffer pH 6.9-7.0 and allowed to solvolyze for
various periods of time. A volume of 1 µL from each exchange
reaction was then spotted on a TLC plate. The large excess of
[19F]-KF present prevented any recapture of [18F]-fluoride by
the hydrolyzed arylboronic acid as the TLC plate dried.
In the absence of any antecedent report on chromatographic
resolution of an ArBF3 on silica, we were delighted to find that
5:95 NH4OH:EtOH cleanly separated all [18F]-ArBF3 (Rf of
∼0.9) from free [18F]-fluoride on standard silica with the
exception of the p-trimethylammoniophenyltrifluoroborate.29
This clean separation permitted autoradiography and phospho-
rimaged densitometry analysis of the time-dependent decrease
in [18F]-ArBF3 and increase in the relative amount of free [18F]-
([ArBF3] + [F]))t ) ([ArBF3]/([ArBF3] + [F]))0e-kt
.
In no case did we observe any steady-state intermediate
corresponding to 2-5 by 19F NMR spectroscopy. That 2 went
undetected was expected since it is a difluoroborane, a species
that readily smolders in moist air.20–26 Thus, 2 must hydrate
instantaneously to give intermediate 3 at pH 7, which was also
not observed as it would eliminate HF to give an arylfluorobo-
ronic acid (i.e., 4), which should be equally reactive as 2 due
to the fact that 4 shares many electronic similarities with
benzoylfluoride, its carbon analogue. Because it is difficult to
characterize that which cannot be readily observed, these
intermediates (and possibly others not drawn) are fleeting, at
least on the 19F NMR time scale, and consistent with a kinetic
model where the total solvolysis of a dilute ArBF3 is governed
by dissociation of the first fluoride in a single rate-limiting step
that is best approximated by pseudofirst order kinetics.27
Since some free fluoride (e1 equiv) was occasionally present
in some of the ArBF3 preparations from the outset,28 we
deliberately added 100 mM fluoride to the solvolysis reaction
(20) McCusker, P. A.; Kilzer, S. M. L. J. Am. Chem. Soc. 1960, 82, 372–
375.
(21) McCusker, P. A.; Glunz, L. J. J. Am. Chem. Soc. 1955, 77, 4253–4255.
(22) McCusker, P. A.; Makowski, H. S. J. Am. Chem. Soc. 1957, 79, 5185–
5188.
(23) Kinder, D. H.; Katzenellenbogen, J. A. J. Med. Chem. 1985, 28, 1917–
1925.
(28) Most of the ArBF3s were prepared by simple precipitation from
methanol, acetone, or acetonitrile, and consequently some residual fluoride was
also present.
(24) Vedejs, E.; Chapman, R. W.; Fields, S. C.; Lin, S.; Schrimpf, M. R. J.
Org. Chem. 1995, 60, 3020–3027.
(29) A ∼5 mg synthesis of trace [18F]-labeled p-carboxyphenyltrifluoroborate
was prepared by running the crude fluoridation reaction through a 0.6 g, 230-
400 mesh, silica plug and collecting 100 µL fractions. Fractions 4-10 were
verified by autoradiography following TLC analysis in the same solvent and
were pooled, decayed and analyzed by 19F NMR spectroscopy to demonstrate
that these fractions were indeed the ArBF3 (data not shown).
(25) Frohn, H.-J.; et al, J. Organomet. Chem. 2000, 598, 127–135.
(26) Chase, P. A.; Henderson, L. D.; Piers, W. E.; Parvez, M.; Clegg, W.;
Elsegood, M. R. J. Organometallics 2006, 25, 349–357.
(27) An independent run was conducted in phosphate buffer at pH 7.5 with
no appreciable difference in kobs: kobs ) (2.88 ( 0.08) × 10-2 min-1
.
4664 J. Org. Chem. Vol. 73, No. 12, 2008

Substituent Effects on Aryltrifluoroborate SolVolysis in Water
FIGURE 2. Left:19F NMR spectra of p-sulfonamidophenyltrifluoroborate solvolysis in the presence of 100 mM KF at 1, 11, 29, 41, 67, and 105
min. Right: Exponential fit of the NMR spectral data.
FIGURE 3. Left: Autoradiogram of time-dependent loss of p-sulfonamidophenyl-[18F]trifluoroborate at 31, 46, 80, 134, and 199 min. Right: First-
order exponential fit of the autoradiographic densitometry data.
fluoride as featured in Figure 3. The autoradiographic density
corresponding to the ArBF3 spot was divided by the sum of the
total density, comprising ArBF3 and free fluoride, to calculate
the fraction of density corresponding to the ArBF3. This
treatment normalizes against variance in the total 18F-decay
intensities in each sample that is spotted on the TLC plate. The
ratio of [18F]-ArBF3 to the total [18F]-fluoride signal for each
sample is plotted against time in the following equation: [P]t )
assays identified similar rate constants for the overall solvolysis
reaction that proceeds without observable intermediates.
In contemplating the mechanism of hydrolysis, we hypoth-
esized that loss of fluoride with concomitant vacation of the
empty p-orbital on boron to give an ephemeral difluoroborane
(2) would be stabilized by electron-donating groups at the para
position. Hence, we hypothesized that such an effect bore
analogy to other systems that had been previously investigated.
For instance, the benzotrifluoride carbon analogues, namely
p-triflluoromethylaniline [CAS registry no. 455-14-1] and p-
trifluoromethylphenol [CAS registry no. 402-45-9], are com-
mercially available and relatively shelf stable.30 Although their
sensitivity to solvolytic defluorination, particularly under basic
[P]0(e-k t), where [P]t and [P]0 are the ratios of boron-bound
obs
[18F]-fluoride/total [18F]-fluoride at time t and 0 min, respec-
tively, and kobs is the observed first-order rate constant for loss
of [18F]-fluoride from the [18F]-ArBF3.
The value of kobs determined from the 18F-autoradiographic
analysis nicely corroborated that determined by 19F NMR
spectroscopy in the presence of 100 mM fluoride. Thus, both
(30) Stahly, G. P.; Bell, D. R. J. Org. Chem. 1989, 54, 2873–2877.
J. Org. Chem. Vol. 73, No. 12, 2008 4665

Ting et al.
conditions, was noted as early as 1947,31 scant mention of their
base sensitivity is found.32–36 Only very recently have molecular
dynamics calculations corroborated the difluoroquinonemethide
resonance form hypothesized early on by Roberts37 for these
systems and for which kinetic analysis showed that defluori-
nation of p-trifluoromethylphenol is first order.38 These results
mirror our findings. Nevertheless, to the best of our knowledge,
no study has systematically studied the effects of para and meta
substitution on the rate of benzotrifluoride hydrolysis, perhaps
in part because the trifluoromethyl group is particularly resistant
to solvolysis and only EDG-modified compositions can be
readily studied. On the other hand, more comprehensive
investigations had been undertaken for the solvolysis of ben-
zyltosylates39 and cumylchlorides,40,41 which is enhanced via
resonance effects of electron-donating groups (EDGs) at the
para position, and retarded via respective inductive and
resonance effects of electron-withdrawing groups (EWGs) at
meta and para positions (Scheme 3).
FIGURE 4. Hammett plot log(kobs) ) σ(m+p)F + log(k0) for data in
Table 1 (Supporting Information). Linear regression analysis of the
log(kobs) values from the 18/19F exchange TLC (b, black line) gave a F
value of -0.96 ( 0.15 (R ) 0.85); of the log(kobs) values from 19F
NMR spectroscopy kinetics (O, blue line) gave a F value of -1.21 (
0.13 (R ) 0.93); plotting σ+ values,43 for which there are fewer given
values, gave higher R values (0.91-0.94) and lower F values: F )
-0.89 (b, black line) and F ) -1.10 (O blue line) (see the Supporting
Information).
SCHEME 3. (Left) p-EDG Resonance Stabilization of the
Difluoroborane 2 and (Right) Analogous Para-Substituted
Resonance Stabilization of Either a Difluorobenzyl Cation or
a Cumyl Cation via a Quinonemethide-like Form
obtained from the TLC-autoradiography to those obtained by
19F NMR spectroscopy by adding the value of -0.21 (i.e.,
log(0.6)) to each and then replotting σ+ vs both sets of values
for log(kobs), which gave a value of F ) -1.0 ( 0.08, and a
very respectable value of R of 0.92 (data not shown). Finally,
because it seemed reasonable that EDGs at the para position
would distribute electron density into an empty p-orbital in the
transition state, we treated the data in terms of a Yukawa-Tsuno
plot (log(kobs) ) (Fσ+ - R′∆σ), where R′ ) Fr and ∆σ ) σ+
- σ). Subjecting the 19F NMR spectral data to a Yukawa-Tsuno
analysis afforded the following values: F ) -1.14 ( 0.13, R′
) 0.36 ( 0.53, and a correlation coefficient of 0.94. Likewise,
the 19F TLC data afforded the following values: F ) -0.81 (
0.16, R′ ) -0.43 ( 0.56, and a correlation coefficient of 0.91.
Subjecting both sets of data, where the TLC data were
normalized to those of the NMR spectral data by a factor of
0.6, to a Yukawa-Tsuno analysis gave the following values: F
) -1.01 ( 0.1, R′ ) 0.03 ( 0.38, and a correlation coefficient
of 0.92. The implications of these findings are briefly discussed
in the next section.
The effect of ortho-substitution with EDGs was briefly
investigated and remains complicated: resonance activates while
sterics and induction deactivate. Whereas the p-MeO-ꢀ-BF3
solvolyzed so rapidly that reliable values for ksolv could not be
obtained, the o-MeO-ꢀ-BF3 and the 2,4-diMeO-ꢀ-BF3 exhibited
t1/2 values of ∼20 and ∼4 min, respectively. Likewise, the 2,4-
difluoro-ꢀ-BF3 behaved similarly to the 2-fluoro-ꢀ-BF3 (see
Table 1, Supporting Information). Although these ortho effects
were not included in the σ-plot, qualitatively, ortho-substitution
with certain groups (e.g., -F, -CF3, -CN) should impart
significantly increased aqueous stability to the ArBF3.
We asked if the same effect would be seen for ArBF3s. To
test this hypothesis, we prepared over 15 mono/disubstituted
ArBF3s (e.g., -Cl, -F, -CF3, -NO2, -CN, etc., see the
Supporting Information) and followed their solvolysis to
completion. In all cases, no ArBF3 remained at equilibrium and
no intermediates were observed. For most samples, the rate
constants from both 19F NMR and 18F-positron TLC autorad-
iography were tabulated in Table 1 (p 47 of the Supporting
Information), and the log(kobs) was plotted vs σp, σm, or σp +
σm for disubstituted compositions as shown in Figure 4.42
Further consideration of these data, and in particular the
somewhat lower correlation coefficients (R values), suggested
several other treatments. The first retreatment of the data was
to recognize that the rate constant for solvolysis in 100 mM
fluoride was depressed by about a factor of 0.6 compared to
that observed in buffer alone, at least for the p-sulfonamidophe-
nyltrifluoroborate. This led us to normalize the rate constants
(31) Jones, R. G. J. Am. Chem. Soc. 1947, 69, 2346–2350.
(32) Roberts, J. D.; Webb, R. L.; McElhill, E. A. J. Am. Chem. Soc. 1950,
72, 408–411.
(33) Bornstein, J.; Leone, S. A.; Sullivan, W. F.; Bennett, O. F. J. Am. Chem.
Soc. 1957, 79, 1745–1748.
(34) Filler, R.; Novar, H. J. Org. Chem. 1961, 26, 2707–2710.
(35) Sakai, T. T.; Santi, D. V. J. Med. Chem. 1973, 16, 1079–1084.
(36) Tickner, D.; Kasthurikrishnan, N. Org. Process Res. DeV. 2001, 5, 270–
271.
(37) Roberts, J. D.; Webb, R. L.; McElhill, E. A. J. Am. Chem. Soc. 1950,
72, 408–411.
(38) Reinscheid, U. M.; Vervoort, J.; Zuilhof, H. Chemosphere 2006, 65,
318–323.
(39) Kochi, J. K.; Hammond, G. S. J. Am. Chem. Soc. 1953, 75, 3445–3451.
(40) Okamoto, Y.; Inukai, T.; Brown, H. C. J. Am. Chem. Soc. 1958, 80,
4969–4972.
Discussion of Results
(41) Brown, H. C.; Okamoto, Y. J. Am. Chem. Soc. 1958, 80, 4979–4987.
(42) Hansch, C.; et al, J. Med. Chem. 1973, 16, 1207–1216.
(43) Hansch, C. J. Med. Chem. 1970, 13, 964–966.
This work represents the first full investigation of the effects
of meta- and para- substitution on the solvolytic lability of an
4666 J. Org. Chem. Vol. 73, No. 12, 2008

Substituent Effects on Aryltrifluoroborate SolVolysis in Water
ArBF3, paralleling prior studies on the solvolysis of EDG-
substituted benzotrifluorides and EDG/EWG-substituted cumyl-
chlorides. Whereas the F value for cumylchloride solvolysis is
ca. -4,8 and that for benzyltosylate solvolysis in ethanol:water
is ca. -2.0, the F value for a ArBF3 in 100% buffered water is
ca. -1. Indeed, a lower F value might be expected in part
because of the greater stability of the sp2-hybridized borane (1),
which draws less electron density from the aromatic ring
compared to a highly deficient sp2-hybridized carbocation.
As the B-F bond is one of the strongest single bonds between
any two atoms on the periodic table, the lability of EDG-
substituted ArBF3s was unforeseen, even if not entirely surpris-
ing once fully investigated. In light of these data, and by analogy
to EDG-substituted benzotrifluorides, benzyltosylates, and cumyl-
chlorides, resonance overlap from the EDG into the aromatic
ring weakens the B-F bond strength significantly, and hence-
forth predictably. The effects of para- and meta-substitution
with EWGs on ksolv also trend in accordance with σ values.
Thus, Hammett plots of log(kobs) vs both σ and σ+ values
were undertaken for each set of data, and in addition, a plot of
σ+ vs a combined set of log(kobs) values comprising both the
rate constants generated by 19F NMR spectroscopy and those
generated by 18F TLC-autoradiography that were normalized
to the former. In all of these analyses, F values were ap-
proximately -1.0 and correlation coefficients were about 0.9.
We also performed a Yukawa-Tsuno analysis44–46 to address
the hypothesis that partial electron density delocalized through
the aromatic ring into the empty p-orbital on boron to enhance
the rate of fluoride dissociation. This linear regression analysis
improved correlation coefficients slightly but provided little
insight into a meaningful quantitative relation between σ/σ+
values and the extent of transition state electron density
delocalization into the boron p-orbital as the errors attributed
to R′ were as large as R′ itself.
In the treatments we attempted, the correlation coefficients
describing goodness of fit are roughly 0.9, which indicates a
respectable fit, although not as good as others. Nevertheless,
we examined 13-16 different samples which represent a
somewhat larger number of substituents than is often investi-
gated in other Hammett analyses. Values that deviate from the
correlated line can be attributed to several factors: (1) errors
associated with measuring kobs, (2) mechanistic differences (e.g.,
SN1 vs SN2) that might be attributed to a deviant analogue in
question, or (3) the possibility that the σ/σ+ values which had
been assigned for the ionization of a series of benzoic acids are
simply less applicable to this case.
In terms of error related to the calculation of kobs, greater
error is likely to be associated with the kobs values obtained
from compounds that solvolyze rapidly on the NMR/TLC time
scalesin fact the p-methoxyphenyltrifluoroborate and p-meth-
ylthiophenyltrifluoroborate solvolyzed so rapidly that their rate
constants could not be reliably measured (data not shown).
Nevertheless, plotting log(kobs) vs σ/σ+ values for only the slow-
to-solvolyze compounds did not greatly improve the correlation
coefficients. Interestingly, the p-trifluoromethylphenyltrifluo-
roborate and the m-di(trifluoromethyl)phenyltrifluoroborate,
which are slow to solvolyze, deviate quite distinctly from the
correlated line. No plausible explanation appears forthcoming
for this observation other than the Hammett σ/σ+ values may
not be fully appropriate in this analysis. Furthermore, as
previously noted, a Yukawa-Tsuno treatment did not shed light
on the extent to which electron donation resonates across the
aromatic ring to fill the vacant p-orbital on boron. Thus, we are
left to conclude that the correlation is somewhat less pronounced
than that for other reaction classes that are more neatly correlated
because of a greater connection to substituted benzoic acids.
Nevertheless, the correlations are sufficiently high to be of
predictive value regarding the aqueous stability of para- and
meta-substituted ArBF3s stability of ArBF3s, which have
heretofore been undefined.
Conclusions Regarding Metal Couplings and PET Agent
Design. The proclivity by which an ArBF3 loses a fluoride ion
has implications for several applications including metal cross
couplings where solvolysis is thought to precede metalation,47,48
and the design of humorally stable PET-imaging agents. With
regards to metal coupling, this work demonstrates that for cases
involving p-EDG-modified ArBF3s in buffered water, e.g., (p-
CH3O-phenyltrifluoroborate), the ArBF3 is likely to be entirely
defluoridated within minutes of solvation.49 On the other hand,
the stability of certain EWG-modified ArBF3s will be formi-
dable, and these stabilities are likely to be further enhanced by
the presence of less polar cosolvents that are commonly used
in mixed aqueous conditions.
Indeed, such is not altogether inconsistent with observations
that EWG-modified ArBF3s are generally poorer substrates for
Pd-catalyzed cross couplings. For instance, Kabalka et al. have
found that whereas a Baylis-Hillman-type coupling of a
phenyltrifluoroborate yields 86% in 3 h., the same coupling
using the 2,6-difluorophenyltrifluoroborate yields 68% in 48 h.50
Similar differences have been seen by Molander et al. for
another Suzuki-Miyaura coupling with EWG-substituted
ArBF3s.51 This observed rate depression is consistent with both
our findings and the suggestion that defluoridation precedes
transmetalation. Nevertheless, if in certain cases rate constants
for coupling exceed the value of kobs identified here for
solvolysis, one may have to reconsider this assumption.
The Hammett relationship between kobs for solvolysis and
substituent effects also bodes well for the rational design of an
assortment of PET-suitable ArBF3s, which builds on our earlier
work toward that goal. In 2005, we suggested the use of ArBF3s
for use in PET imaging based on earlier reports that had
emphasized the aqueous stability of ArBF3s in water, and from
inference of Anbar and Guttmann’s early work on the extreme
-
sluggishness of BF4 solvolysis. Our communication in the
Journal of the American Chemical Society employed an 18F-
labeled biotinylated-ArBF3 to suggest a radically different
approach to 18F-labeling that exploited the fluorophilicity of
arylboronic acids/esters as captors of aqueous [18F]-fluoride that
would circumvent the standard practice of generating arylfluo-
rides in multistep labeling procedures.52
Thus, our initial investigation used a convenient bioconjugate,
biotinamidophenylboronyl pinacolate,53 which we designed to
address the capture of radioactive [18F]-fluoride whereby the
(47) Molander, G.; Biolatto, B. Org. Lett. 2002, 4, 1867–1870.
(48) Barder, T. E.; Buchwald, S. L. Org. Lett. 2004, 6, 2649–2652.
(49) Sakurai, H.; Tsunoyama, H.; Tsukuda, T. J. Organomet. Chem. 2007,
692, 368–374.
(50) Kabalka, G. W.; Venkataiah, B.; Dong, G. Org. Lett. 2003, 5, 3803–
3805.
(51) Molander, G. A.; Fumagalli, T. J. Org. Chem. 2006, 71, 5743–5747.
(52) Lasne, M. C.; Perrio, C.; Rouden, J.; Barre, L.; Roeda, D.; Dolle, F.;
Crouzel, C. Top. Curr. Chem. 2002, 222, 201–258.
(53) Reference 10.
(44) log(kobs) )Fσ+- R′∆σ, where R′ )Fσr and ∆σ ) σ+-σ.
(45) Tsuno, Y.; Fujio, M. Chem. Soc. ReV. 1996, 25, 129–139.
(46) Yukawa, Y.; Tsuno, Y. Bull. Chem. Soc. Jpn. 1959, 32, 965–970.
J. Org. Chem. Vol. 73, No. 12, 2008 4667

Ting et al.
biotin appendage enabled affinity chromatography on avidin-
magnetic particles such that radioactive free [18F]-fluoride could
be removed by repeated washings. In that context we observed
specific labeling of the avidin- magnetic particles in a manner
that was resistant to washing and did not exchange its [18F]-
fluoride in the presence of added [19F]-KF.
Following up on our communication, we sought alternate
means besides avidin- affinity chromatography for removing
free fluoride from small-molecule compositions containing
an [18F]-ArBF3. Having discovered that the NH4OH:EtOH
TLC system separated the ArBF3 from free fluoride, we
recognized that the p-biotinamidophenyltrifluoroborate ex-
changed its fluoride very rapidly, consistent with the data
presented herein, and in stark contrast to the conclusions we
had initially reported.
Although our initial communication indicated that we had
prepared a substance derived from a boronic acid that labeled
in the presence of aqueous [18F]-fluoride and manifested suitable
stability, in light of the data we report herein, we speculate that
the high molar ratios of p-biotinamidophenyltrifluoroborate:
avidin in our earlier report may have contained trace impurities,
on the order of 1%, which may have adventitiously bound to
the avidin magnetic particles leading us to overestimate the
aqueous stability of the target p-biotinamidophenyltrifluorobo-
rate. Nevertheless, the conclusions of that preliminary work
proved conceptually and qualitatively correct: to apply this
approach for in ViVo imaging we needed to identify a class of
ArBF3s that would be considerably more stable over the time
that a physiological process is imaged. Thus, the results and
analysis herein, which reflect a reinvestigation of earlier claims,
disclose the scientific process and theory that now allow the
development of solvolytically robust ArBF3 compositions.
Indeed, based on the findings now assembled here, we have
recently reported the synthesis of a fluorescently (BODIPY)
linked 1,3,5-trifluoro-2-carboxamidophenyl-6-[18F]-trifluorobo-
regards to isotopic purity and yields, the reader must note
that a no-carrier-added [18F]-fluoride is neVer carrier-free
because of the adventitious presence of contaminating
environmental [19F]-fluoride, which, in practice, dramatically
reduces the specific activity to ∼100 Ci/µmol, if not
considerably lower.55 At ∼100 Ci/µmol, only 5 of 86 fluoride
ions are radioactive such that the resulting ArBF3 will
represent a statistical mixture of inseparable istopologs where
99% of all species are unlabeled and monolabeled, with less
than 1% containing more than 1 radioactive 18F-atom.56 Such
a distribution is not dissimilar from other labeling strategies
that rely on C-F bond formation. In addition, 1 Ci will
contain >10 nmol total fluoride, which means that the
reaction can be run in a tenable volume of 100 nL, using a
microfluidic reactor to ensure a solution of 100 mM fluo-
ride.57 Furthermore, the 3-to-1 stoichiometry of conversion
stipulates that the specific activity of the resulting ArBF3 will
be thrice that of the fluoride used, irrespective of either the
fluoride concentration or the yield of the resultant ArBF3.
Besides the question of yield is one of purity; this work also
demonstrates that no stable mono- or difluoridated species
are produced in this labeling scheme, which provides for a
chemically pure aryltrifluoroborate salt.
Finally, it is essential to note that the most important concern
for imaging an extracellular receptor with a radiolabeled ligand
is the final specific activity of the labeled ligand, which must
be >1 Ci/µmol, and not the specific activity of the fluoride used
initially. Thus, a synthesis in which 1 Ci of fluoride is
supplemented with e1µmol of [19F]-fluoride still satisfies the
specific activity requirements for imaging yet allows the
radiochemist to expect reasonable yields when working in
volumes as high as 10 µL.
Experimental Section
rate, which exhibits extraordinary in Vitro stability (kobs for
ArBF3 Synthesis. For nonradioactive samples, ArBF3s were
converted from their corresponding commercially available boronic
acids or boron pinacol esters in the presence of KHF2 and were
isolated through acetonitrile or methanol washes according to
procedures described by Vedejs58 and Molander.59 Certain ArBF3s
could not be completely separated from free fluoride without
concomitantly large losses of ArBF3. As these are considered
commonly in use, the spectra are reported in the Supporting
Information.
54
solvolysis ∼10-4 min.-1
)
and will report on the in ViVo
stability of a biotinylated derivative thereof in the near future.
For those who are skeptical of preparing viable trifluorobo-
rate-based PET-imaging agents in the presence of added carrier,
we reiterate our original claims that ArBF3s can be generated
under no-carrier-added conditions but not in carrier-free form.
To dispel this skepticism, we review the underlying thermody-
namics of ArBF3 synthesis given in equation 1.
-
The p-(CH3)N+PheBF3 (p-trimethylammoniophenyltrifluo-
roborate) was unknown in the literature and synthesized as with
the other ArBF3s from its precursor pinacolate: 1H NMR (400 MHz
DMSO-d6) δ 7.6 (d, J ) 8 Hz, 2H), 7.5 (d, J ) 8 Hz, 2H), 3.3
ppm (s, 9H); 19F NMR (400 MHz DMSO-d6) δ -62.5 ppm
(quadrapolar coupling to boron not observed). The precursor was
the p-trimethylammoniophenylboronylpinacolate, which was pre-
pared by reacting commercially available p-N,N-dimethylaminophe-
nylboronylpinacolate (700 mg) in neat CH3I (1 mL) for 12 h. The
[ArBF3]
[ArB(OH)2][F-]3
Keq(formation pH 4) )
(1)
On the basis of eq 1, radiosynthetic yields exhibit third-order
dependence on fluoride concentration. Yield notwithstanding,
le Chatelier’s principle holds that some ArBF3 must neces-
sarily form irrespective of the fluoride concentration used.
At 100 mM fluoride, ArBF3 formation is relatively fast, and
in our hands, chemical yields with respect strongly EWG-
containing arylboronic acids have varied from 10% to 50%
over the period of 1 h and higher yields are obtained over
longer times. Nevertheless, because relatively high concentra-
tions are needed, low volumes may be necessary to ensure
high yields when using no-carrier-added fluoride.
(55) Reference 52.
(56) The statistical breakdown is where approximately 83.55% will be
unlabeled, 15.47% will be mono-labeled, 0.951% will be bis-labeled, and
0.0195% will be tris-labeled.
(57) If one considers the hypothetical case of using 1 Ci of carrier free [18F]-
fluoride (581 pmol, specific activity 1720 Ci/µmol), one would have to employ
5 pL reaction volumes to achieve the same yieldssa veritable challenge that
may be met with microfluidic reactors. Whatever amount of ArBF3 that forms
with carrier-free [18F]- fluoride will necessarily be 100% 18F-labeled. In this
hypothetical case, the resulting ArBF3 will contain three atoms of [18F]-fluoride
and not only one.
In critically evaluating the potential of this approach for
labeling with no-carrier-added fluoride, in particular with
(58) Vedejs, E.; Chapman, R. W.; Fields, S. C.; Lin, S.; Schrimpf, M. R. J.
Org. Chem. 1995, 60, 3020–3027.
(54) Ting, R. et al. J. Fluorine Chem. 2008, 129, 349-358.
(59) Molander, G. A.; Bernardi, C. R. J. Org. Chem. 2002, 67, 8424–8429.
4668 J. Org. Chem. Vol. 73, No. 12, 2008

Substituent Effects on Aryltrifluoroborate SolVolysis in Water
precipitate was collected, washed with CH2Cl2, pumped dry of
residual CH3I, washed with Et2O, and then placed on high vacuum
overnight; H NMR (400 MHz CD2Cl2) δ 8.0 (d, J ) 8 Hz, 2H),
To assay the solvolysis rate of [18F]-fluoride from [18F]-ArBF3,
the [18F]-ArBF3s featured in Table 1 (Supporting Information) were
prepared as described for “Reaction A”. One hour (t ) 60 min)
following fluoridation, a volume of 1.5 µL was taken from
“Reaction A” and diluted 200 times into 300 µL of a 192 mM PO4
pH 6.9-7.0, 100 mM [19F]-KF solution. A series of dilution
experiments corresponding to different solvolysis times were made
from the same ArBF3 cocktail at times that were staggered such
that all reactions were terminated at the same end point and resolved
by TLC. At this end point, 0.5 µL of each reaction was spot on a
TLC plate, which was run ∼5 cm in a 5% NH4OH/EtOH
developing solution.
Kinetic Analysis with 18F/19F TLC Autoradiography. To
ensure that all samples were resolved at the same time by TLC,
kinetics of solvolysis in excess [19F]-fluoride were initiated for
different lengths of time, e.g., 199, 134, 80, 62, 46, and 31 min,
prior to TLC resolution (see Figure 3). Of note, while complete
conversion of the arylboronic acid/ester to the ArBF3 is not required
1
8.0 (d, J ) 8 Hz, 2H), 4.0 (s, 9H), 1.3 ppm (s, 12H); HRMS (M
+ K)+ calcd C9H13BNF3K+ 242.0730 m/z, found 242.0724.
19F NMR Spectroscopic Kinetic Analyses. All 19F NMR spectra
are referenced to TFA. A small quantity of the isolated ArBF3 salt
(∼0.3-1 mg or 1-5 µmol) was added to a Norell 507-HP NMR
tube. At t ) 0 min, the start of the solvolysis reaction, the ArBF3
was dissolved in 500 µL of a 192 mM pH 6.9-7.0 phosphate buffer
(resulting in a 2-10 mM solution) and decomposition was
monitored by 19F NMR spectroscopy. 19F NMR spectra were
acquired at different times. 19F NMR spectral integrations were
calculated on obtained spectra, using MestReC NMR spectral data
processing software. Integrals corresponding to the ArBF3 peak were
divided by the sum of the ArBF3 and free fluoride integrations to
calculate the fraction of 19F existing as the ArBF3. The ratio of 19
F
signal existing as the ArBF3 to the total 19F signal was plotted
against time in eq 1 to calculate, kobs, the observed first-order rate
constant for loss of [19F]-fluoride from ArBF3. A half-life for this
process was calculated by setting t1/2 ) (ln 2/kobs). The kinetic
constants reported in this text are the result of nonlinear regression
analyses of the cumulative plot of all data sets for identical
experiments using the Sigma Plot 2001 v7.101.
for this assay, it is essential that the ratio of ArBF
3:fluoride in
“Reaction A” be constant for each time point represented in the
18F/19F-exchange reaction and not be increasing over time. This
condition was thus verified even though previous reports suggested
that boronic acids or boron pinacol esters are quantitatively
converted to ArBF3s in aqueous solutions of 200 mM KFH2 and
this conversion occurs between 1 and 60 min. To verify that
fluoridation was rapid, kinetic profiles of ArBF3 formation were
first monitored by TLC-autoradiography in the following manner:
three different reactions were initiated and allowed to proceed for
various times, t ) 240, 60, and 30 min. All three reactions were
then stopped within a minute or two of each other by spotting 0.3
µL of each reaction on a TLC plate, which was run 5-6 cm with
5% NH4OH/EtOH as the running solvent. All arylboronic acids/
esters gave rise to ArBF3s that exhibited Rf values of 0.75-0.95 in
this developing solution, with the exception of the p-(CH3)3N-
PheBF3, which did not migrate appreciably from the baseline (data
not shown). Autoradiographic assays on very early time points, e.g.,
1-5 min, showed variable but usually minor amounts of ArBF3;
however, by 30 min ArBF3 synthesis was generally complete and
no more ArBF3 had formed at longer time points. Furthermore, in
no case was any intermediate observed by autoradiographic analysis.
On the basis of these tests, we concluded that ArBF3 formation for
all members of the assayed library was complete within 30 min
(data not shown).60 Furthermore, spotting and drying on the TLC
plate for 30 min had little effect compared to freshly spotted
reactions. Thus, we verified that the concentration of components
upon drying on the TLC plate did not result in substantially more
conversion of the ArBF3.
Autoradiography of [18F]-ArBF3s and [18F]-fluoride was ac-
complished by exposing TLC plates on Molecular Dynamics
phosphor screens overnight and scanning the screens on a Molecular
Dynamics Typhoon 9200 phosphoimager. Polygons were drawn
around distinct spots with ImageQuant v. 5.2. Time values noted
in the figures concerning the isotopic 18F/19F-fluoride exchange
experiments were the times that [18F]-ArBF3s remained diluted in
the phosphate solution.
Data were entered into Excel Spreadsheets and graphed in
either Excel or SigmaPlot. Using the LINEST function in Excel,
Hammett plots were generated from a simple one-variable linear
regression and Yukawa-Tsuno plots were generated with a
double variable linear regression according to log(kobs) ) Fσ+
- R′∆σ, where R′ ) FR and ∆σ is the difference between
reported values of σ+ and σ.
[18F]-Fluoride Preparation. The desired activity of [18F]-fluoride
was prepared via bombardment of 1 mL of H218O water with 13
MeV protons in a niobium target. To recover unconverted H218O,
the resulting [18F]-fluoride/H218O solution was passed through a
2-
short anion-exchange resin (∼10 mg, CO3 form). Resin-bound
fluoride was eluted with 200 to 300 µL of 1 mg/mL aqueous
NaClO4 and recovered in a glass vial. Depending on the desired
activity, this sample could be diluted with deionized water or used
as is, or the sample could be concentrated through heating under
vacuum. Usually 1-6 mCi were used to which carrier [19F]-fluoride
was added and then diluted accordingly. Hazard Note: Care must
be taken in acidifying and concentrating [18F]-fluoride containing
water. If the pH is adjusted lower than 3, a radioactive and volatile
HF vapor is produced. Use of a fumehood with filtered air flow is
recommended.
Although NaClO4 is a known oxidant, it was used instead of
carbonate as an elutant because (i) it is a more powerful elutant
than carbonate and (ii) minimized the presence of eluted carbonate
that often rendered the resulting [18F]-fluoride solution quite basic,
i.e., pH >9. Since 20 µCi was used (Vide infra), this represented a
>50-fold dilution of the perchlorate. As such, were all of the
perchlorate to pass through the column, the maximum amount of
perchlorate available in the fluoridation reaction would be 12
mMsfar less than the boronic acid used. At these concentrations
the extent to which perchlorate could interact (and thus diminish)
with the arylboronate species was considered negligible.
[18F]-ArBF3s were synthesized by incubating 5 µL of a 200 mM
boronic acid/ester solution in DMF with 5 µL of 400 mM KHF2
solution pH 3-4, containing a minimum of ∼20 µCi of [18F]-
fluoride (specific activity at the start of reaction). The mixing of
these solutions resulted in a 10 µL cocktail containing 100 mM
boron and 200 mM KHF2 containing [18F]-fluoride. This solution
will be referred to herein as “Reaction A” and the time at which
these two solutions were mixed will be referred to as the start of
the fluoridation reaction (t ) 0 min). All described experiments
requiring [18F]-fluoride did not exceed t ) 6 h or ∼ 3 half-lives of
decay. Typically the 400 mM KHF2 solution is made up in a
cocktail format, where 40 µL of 4 M KHF2 and 20 µL of 2 M HCl
2-
in order to neutralize any CO3 that eluted as a counterion from
the anion exchange resin are added to 340 µL of ∼3 mCi [18F]-
fluoride in H2O. With the large excess of [19F]-fluoride carrier added
in the conditions described, the [18F]-ArBF3 species is constituted
(>99.99%) by a uniquely labeled isotopolog ArB[18F][19F]2 con-
taining a single atom of [18F]-fluorine.
(60) Indeed, had this not been the case and the ArBF3 concentration were
increased between the start of the 18F/19F-exchange reaction representing a long
time point (e.g., 199 min) and the end of the exchange reaction representing a
short time point (e.g., 31 min), such would actually lead to an overestimation of
the rate constant for solvolysis.
J. Org. Chem. Vol. 73, No. 12, 2008 4669

Ting et al.
Acknowledgment. R.T. was the recipient of a Gladys Estella
Laird and a Michael Smith Foundation for Health Research
Ph.D. traineeship. D.M.P. received a Michael Smith Career
Scholar Award. D.M.P. thanks Prof. Charles L. Perrin and Dr.
Marilyn H. Perrin for insightful discussions. This work was
supported by the Canadian Institutes for Health Research.
Supporting Information Available: Detailed methods,
spectra, autoradiography, and Table 1 containing data for
Hammett plots. This material is available free of charge via the
Internet at http://pubs.acs.org.
JO800681D
4670 J. Org. Chem. Vol. 73, No. 12, 2008