An overlooked yet ubiquitous fluoride congenitor: Binding bifluoride in triazolophanes using computer-aided design

Article abstract of DOI:10.1021/ja500125r

Despite its ubiquity during the binding and sensing of fluoride, the role of bifluoride (HF₂^-) and its binding properties are almost always overlooked. Here, we give one of the first examinations of bifluoride recognition in which we use computer-aided design to modify the cavity shape of triazolophanes to better match with HF₂^-. Computational investigation indicates that HF₂^- and Cl^- should have similar binding affinities to the parent triazolophane in the gas phase. Evaluation of the binding geometries revealed a preference for binding of the linear HF₂^- along the north-south axis with a smaller Boltzmann weighted population aligned east-west and all states being accessed rapidly through in-plane precessional rotations of the anion. While the ¹H NMR spectroscopy studies are consistent with the calculated structural aspects, binding affinities in solution were determined to be significantly smaller for the bifluoride than the chloride. Computed geometries suggested that a 20° tilting of the bifluoride (stemming from the cavity size) could account for the 25-fold difference between the two binding affinities, HF₂^- < Cl^-. Structural variations to the triazolophane's geometry and electronic modifications to the network of hydrogen bond donors were subsequently screened in a stepwise manner using density functional theory calculations to yield a final design that eliminates the tilting. Correspondingly, the bifluoride's binding affinity (K ～ 10⁶ M^-1) increased and was also found to remain equal to chloride in the gas and solution phases. The new oblate cavity appeared to hold the HF₂^- in a single east-west arrangement. Our findings demonstrate the promising ability of computer-aided design to fine-tune the structural and electronic match in anion receptors as a means to control the arrangement and binding strength of a desired guest.

Full text of DOI:10.1021/ja500125r

Article
pubs.acs.org/JACS
An Overlooked yet Ubiquitous Fluoride Congenitor: Binding
Bifluoride in Triazolophanes Using Computer-Aided Design
Raghunath O. Ramabhadran,† Yun Liu,† Yuran Hua, Moira Ciardi, Amar H. Flood,*
and Krishnan Raghavachari*
Department of Chemistry, Indiana University, 800 East Kirkwood Avenue, Bloomington, Indiana 47405, United States
S
* Supporting Information
ABSTRACT: Despite its ubiquity during the binding and
−
sensing of fluoride, the role of bifluoride (HF2 ) and its
binding properties are almost always overlooked. Here, we give
one of the first examinations of bifluoride recognition in which
we use computer-aided design to modify the cavity shape of
−
triazolophanes to better match with HF2 . Computational
investigation indicates that HF2 and Cl− should have similar
−
binding affinities to the parent triazolophane in the gas phase.
Evaluation of the binding geometries revealed a preference for
binding of the linear HF2− along the north−south axis with a smaller Boltzmann weighted population aligned east−west and all
1
states being accessed rapidly through in-plane precessional rotations of the anion. While the H NMR spectroscopy studies are
consistent with the calculated structural aspects, binding affinities in solution were determined to be significantly smaller for the
bifluoride than the chloride. Computed geometries suggested that a 20° tilting of the bifluoride (stemming from the cavity size)
−
could account for the 25-fold difference between the two binding affinities, HF2 < Cl−. Structural variations to the
triazolophane’s geometry and electronic modifications to the network of hydrogen bond donors were subsequently screened in a
stepwise manner using density functional theory calculations to yield a final design that eliminates the tilting. Correspondingly,
the bifluoride’s binding affinity (K ∼ 106 M−1) increased and was also found to remain equal to chloride in the gas and solution
−
phases. The new oblate cavity appeared to hold the HF2 in a single east−west arrangement. Our findings demonstrate the
promising ability of computer-aided design to fine-tune the structural and electronic match in anion receptors as a means to
control the arrangement and binding strength of a desired guest.
of metal complexes9 for fluorine-related chemical trans-
INTRODUCTION
The importance of anions in chemistry and biology1 continues
to motivate studies of their recognition. For instance, during
■
formations10 and recently as a synthon in the emerging area
of self-assembled networks.11 Industrially, it is perhaps more
widely known where it is used as buffered oxide etch (BOE)12
−
the remediation of radionuclides (99TcO4 ) and counteranions
for the fabrication of silicon semiconductors, for digestion,13
(SO42−) present in the nuclear fuel cycle. Triazolophanes2,3 are
extraction,14 and as a cleaning agent in automatic car washers.15
among the newest receptors in this area that are characterized
A few toxicity studies have also emerged16 suggesting that it
by their modularity and use of activated CH hydrogen bonds4
acts like fluoride while showing fewer of the telltale signs of
with a size selectivity for chloride. We are now pushing the
exposure.17 Thus, during the course of our study, we were
limits of the cavity’s spherical shape to bind more complex
surprised to discover that bifluoride is not rare at all. Instead, it
anions, a situation stemming from the fact that polyatomic
is frequently observed in studies of fluoride recognition and
anions can be pseudospherical under the right circumstances.
sensing due to the fact that bifluoride forms readily18−22 by
abstraction of protons:
For instance, we showed that spherical Cl− and linear diatomic
CN− have equal binding affinities3b to tetraphenylene
triazolophanes, since the cyanide’s in-plane orientation is
rotationally averaged. Taking steps towards polyatomic anions
led us to consider the binding of bifluoride (HF2 ), the smallest
2F− + RH ⇌ HF + R− + F− ⇌ HF − + R−
(1)
2
−
In fact, we found that the only situation when bifluoride is not
present in appreciable amounts (>1 mol %), relative to F−, is in
aqueous solutions at concentrations less than 10 mM or in
basic solutions, i.e., pH >7. Despite bifluoride’s prevalence, it is
consistently unnoticed or overlooked in the field of anion
recognition, though it is almost always capable of serving as a
of the linear triatomic anions. This ion also attracted our
attention since its internal hydrogen bonds are the strongest5
known, making it a symbolic antonym of the so-called weak
CH hydrogen-bonding systems like triazolophanes.
While little appears to be known about bifluoride, we believe
that it may be more prevalent than first thought. Binding
studies are raretwo to date6,7 and two known crystal
structures.6a,8 Bifluoride has seen some use in the preparation
Received: January 6, 2014
© XXXX American Chemical Society
A
dx.doi.org/10.1021/ja500125r | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
a
Scheme 1. Computer-Aided Design on Receptor 1 Leads to the Development of Macrocycle 2
a
−
−
Simplified structures (3·HF2 and 4·HF2 ) were optimized using M06-2X/6-31+G(d,p) and shown to have different preferred bifluoride
orientations.
competing interferant in fluoride binding.23 As noted by
Cametti and Rissanen in their review on fluoride binding,23 we
too wonder how many of the hundreds of papers devoted to
fluoride recognition in the past 10 years carry a systematic error
associated with the bifluoride present. It is clear that careful
studies of bifluoride binding are long overdue.
to appropriately accommodate HF2− with an enhanced binding
energy. Such computer-aided-design approaches represent an
ideal situation, since virtual experiments can be conducted in
lieu of observational ones to quickly advance understanding and
to direct the rational design of molecular targets. In the field of
anion recognition, Hay and co-workers advocate the use of
HostDesigner27 for the de novo design of anion receptors.28
For instance, the creation of receptors for self-assembled
cages29a,h and sulfate29h comprised of metals with geometry-
and symmetry-matched ligands was made possible by recruiting
known supramolecular synthons, such as hydrogen-bond
donors29b,c,e,h and metal-binding sites,29a,d and utilizing the
rules of symmetry to screen a database of linkers that could be
used to connect them together.29 The outcomes of this
methodology are ligands that can be readily synthesized and
that self-assemble directly into polyhedra with an ability to bind
anions with matched shapes. An alternative method by Mascal
involved the computer-aided design of a fluoride-selective
receptor30 capable of supporting anion−π interactions that was
later realized experimentally.31 Finally, other prominent
examples of computer-aided strategies applied to supra-
molecular host−guest complexes include Houk’s extensive
studies on gating in hemicarcerands.32 Here, we take a
complementary computer-aided-design approach to prescreen
and then create anion receptors that can control the binding
geometry of the bifluoride guest within the cavity of the
triazolophane. This capability allows us to enhance the binding
Bowman-James’s bicyclic azacryptand,6 Eichen’s cyclic
benzoimidazole derivative,7 and Severin’s metallo-macrocycle8
are the only studies of HF2− binding that we are aware of. The
first two studies arose by serendipity during fluoride binding. In
particular, the elliptical cavity of the bicyclic azacryptand was
−
−
shown to bind linear anions (HF2 , N3 ) with greater strength
than spherical anions (Cl−, Br−, I−).6b Consequently, significant
−
and selective binding toward HF2 in dimethyl sulfoxide
(DMSO) was achieved. The cyclic benzoimidazole was shown
to bind HF2− strongly in DMSO (K ∼ 105), yet its F− binding
strength was not determined on account of the formation of
−
HF2 . Lastly, Severin’s metallo-macrocycle was found to bind
8
−
−
HF2 as a contact ion pair with Li+. Even though HF2 was
added intentionally to characterize the host−guest complex in
the solid state, solution-based titrations were not performed.
Thus, a detailed study focusing primarily on the character-
−
ization of the binding properties of HF2 will be crucial to aid
in an improved understanding of this anion’s recognition
properties and to help deliver better guidance to the role of
bifluoride in the active area surrounding the chelating and
sensing of fluoride.24
In this work, bifluoride was chosen to test the shape
selectivity of anion binding to the rigid triazolophane
macrocycle. On the basis of the detailed understanding
provided by the prior use of coupled experiment-theory studies
on triazolophanes,2f,3,25,26 we decided that a computer-aided-
design approach can be fruitfully used to reshape the receptor
−
affinity of HF2 to the macrocycle by altering the shape of the
cavity and its electropositive character.
In our study, we focused on binding bifluoride using CH
hydrogen bonds in a rigid two-dimensional macrocyclic
triazolophane. We started with a study of the triazolophane 1
in the gas-phase (theory) and solution (experiment) to bind
B
dx.doi.org/10.1021/ja500125r | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
1
−
Figure 1. H NMR spectra (2 mM, CD2Cl2, 298 K, 500 MHz) of 1 and 1·HF2 (35 equiv added). Triazole (T) protons (HT) are colored in red,
north−south (NS) phenylene protons (HN) are blue, and east−west (EW) phenylene protons (HC) are magenta.
binding energies. Implicit solvent single-point computations were
performed using the SMD model for the optimized geometries in the
gas phase at the M06-2X/6-311++G(3df,2p) level of theory. Finally,
Grimme’s dispersion corrections to the M06-2X functional were also
added to the computed energies, and the effects were found to be
minimal (Supporting Information).
NMR Titrations. Titrations of macrocycles 1 and 2 with anions in
the forms of tetra-n-butylammonium (TBA) salts were followed by 1H
nuclear magnetic resonance (NMR) spectroscopy using a 500 MHz
Varian Inova spectrometer at room temperature (298 K). In a typical
experiment, 400 μL of 2 mM (or 0.5 mM) macrocycle solutions were
prepared in screw-cap NMR tubes equipped with PTFE/silicone septa.
Aliquots of 80 mM (or 20 mM) TBA salt solutions in scew-cap vials
equipped with PTFE/silicone septa were added via 10 and 100 μL
gastight microsyringes.
UV−Vis Titrations. Titrations using ultraviolet−visible (UV−vis)
spectroscopy were conducted utilizing a Varian Cary 5000 UV−vis−
NIR spectrophotometer at room temperature (298 K). Host solutions
(5 μM, 4 mL) were prepared in 1 cm screw-cap quartz cell equipped
with PTFE/silicone septa. Aliquots of 1 mM TBA salt solutions in
scew-cap vials equipped with PTFE/silicone septa were added via 10
and 100 μL gastight microsyringes.
bifluoride using the macrocycle’s set of CH hydrogen-bond
donors constituted by four extrinsically activated 1,2,3-triazoles
and four phenylenes (Scheme 1). In order to correlate the
results of calculations with experimental observations, we used
the triazolophane’s binding energies to Cl− for internal
calibration. In the gas-phase calculations, we found the binding
energies for both anions to be similar. Given the anion-to-cavity
mismatch in shape and size (vide supra), HF2− was found to be
stable in multiple binding geometries. Among these, we noted a
−
20° tilting of the HF2 anion out of the triazolophane’s mean
plane in the gas-phase calculations. We hypothesized that the
tilting remained in solution and that it caused the reduction of
the binding affinities relative to that of Cl−. To test this idea,
computer-aided design was employed to sift through three
modifications that could be incorporated synthetically into the
original triazolophane. The goal was to utilize modifications to
the triazolophane that can minimize or remove the tilting angle
−
of HF2 as a means to improve the experimental binding
affinity without having to synthesize a whole series of receptors
in a trial and error approach. Consequently, this strategy
accelerated greatly the design of macrocycle 2. Experimental
studies then yielded an enhanced bifluoride binding affinity
which was found to be equal to that of Cl− in both the gas-
phase calculations and as observed in the solution-phase
experiments (∼106 M−1).
RESULTS AND DISCUSSIONS
■
−
Binding Modes of HF2 in Triazolophanes. Several
different bifluoride−triazolophane binding modes can be
envisioned. Binding can be along the north−south (NS),
east−west (EW), or diagonal directions. A perpendicular mode
of binding is also possible. The anion can either be nestled
down in the plane of the macrocycle or oriented out of the
plane.
COMPUTATIONAL AND EXPERIMENTAL DETAILS
■
DFT Calculations. All computations have been carried out using
the Gaussian suite of programs.33 All the geometries have been
optimized using the M06-2X functional.34 In our previous study on
cyanide binding,3b calculations performed on the parent tetrapheny-
lene triazolophanes revealed that the use of a hydrogen atom in place
of a methoxy group did not significantly affect the geometries and
anion binding energies (<2 kcal/mol). On this basis, triazolophane 1
was modeled using 3 (Scheme 1), which has hydrogens in the place of
the tert-butyl and triethylene glycol (OTg) groups. Macrocycle 2 was
modeled by 4 (Scheme 1) with an alkoxy-linked OTg group replaced
by an OH and the tert-butyl groups on the amide substituents
truncated to hydrogens. All other macrocyclic models are truncated
similarly with hydrogens. Vibrational frequencies, zero-point correc-
tions, and thermal corrections have been evaluated using the 6-
31+G(d,p) basis set. Unless stated, all the structures are confirmed to
be minimum-energy structures with no imaginary frequencies. The
basis set superposition errors (BSSE) on the computed interaction
energies were evaluated using the standard counterpoise method.35
Single-point energy calculations were carried out subsequently using
the significantly larger 6-311++G(3df,2p) basis sets to obtain the anion
1H NMR titrations of 1 with tetrabutylammonium bifluoride
(TBAHF2) in CD2Cl2 were conducted to first evaluate the
binding strength and the likely binding modes. The chemical
1
shifts of all the triazolophane protons observed during the H
NMR titration (Figure 1) with bifluoride indicate binding.2 In
particular, the inner protons (HT, HN, and HC) have noticeable
downfield movements consistent with significant hydrogen-
bond interactions between triazolophane 1 and bifluoride. Both
the chemical shifts and peak widths stop changing following
addition of 35 equiv of TBAHF2, indicating that triazolophane
−
1 is presumably saturated as the 1:1 complex, 1·HF2 . Upon
HF2 binding, the protons HN along the north−south axis
−
displayed the largest shift (1.1 ppm), which were more than
twice as large as that of the HC (0.5 ppm) located along the
east−west axis. By contrast, the complexation-induced chemical
shifts (Δδ, Figure 2) of the three inner protons (HN, HC, and
C
dx.doi.org/10.1021/ja500125r | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
fluorine atoms and the two southern triazoles point down
toward the other fluorine (Figure S11, Supporting Informa-
tion). Both the north−south phenylene protons and all four
triazole protons engage (Figure 3) the terminal fluorines of
bifluoride with six short CH···F hydrogen bonds (ca. 2.0 Å).
Interestingly, these distances are comparable to the NH···F
contacts (1.8−2.0 Å) seen in the crystal structure of the caged
HF2 complex reported by Bowman-James.6a
−
The fully planar D2h-symmetric complex with bifluoride
oriented along the EW direction is also a second-order saddle
point. Perturbing along its imaginary frequencies results in the
−
Cs-symmetric complex 3·HF2 (Figure S11, Supporting
Information) again with a ca. 21° tilting angle. The hydro-
gen-bond distances in the EW mode resemble those from the
NS mode (Table S2, Supporting Information), i.e., short 2 Å
contacts. We believe that this binding mode is also present in
solution and it accounts for the 0.5 ppm migration in the east−
west protons upon bifluoride complexation (Figure 1).
Figure 2. Representation of the complexation-induced chemical shifts
(Δδ, vertical axis) for triazolophane 1 versus spatial location of each
2b
2b
−
inner proton upon binding (a) Cl−, (b) Br−, and (c) HF2 . The
shifts for triazole protons (T) are colored in red, those of north−south
(NS) phenylene protons (HN) are blue, and those of east−west (EW)
phenylene protons (HC) are magenta. Structures of anion complexes
for 3 were optimized with M06-2X/6-31+G(d,p).
Since both the NS and EW binding modes were found to be
−
minimum-energy structures of 3·HF2 , we were curious about
−
HT) in the symmetrically bound halide complexes,2b such as 1·
Cl− and 1·Br−, are all similar (ca. 0.8 ppm). Since the shapes of
the rigid tetraphenylene triazolophanes change very little upon
anion binding,3a these results indicate that bifluoride prefers to
bind along the NS direction. This outcome matches with our
previous theoretical3a and experimental3b findings that the
nitrogen-linked phenylenes, located along the north−south axis
in 1, allow for greater stabilization of anions than the carbon-
linked ones located east−west.
the nature of the other binding modes. For instance, HF2
could potentially be pointed diagonally toward the triazoles
−
(T). Structural optimization produces a Ci-symmetric 3·HF2
structure with one imaginary frequency. Perturbation along one
direction of the imaginary normal mode leads to the NS
binding arrangement and to the EW geometry in the other
direction. This diagonal structure thus appears to be the
transition state for the in-plane precessional interconversion
between the two local minima (Figure 4). Finally, the
perpendicular bifluoride binding mode, with the bifluoride
bisected by the macrocycle’s mean plane, is found to be a
higher-order saddle point with several imaginary frequencies.
Systematically perturbing each of these imaginary frequencies
eventually results in the NS and EW modes as the only two
The binding modes were also examined by gas-phase
calculations utilizing a simplified triazolophane, 3 (Figure 3),
Figure 3. Hydrogen-bonded distances (red, Å) in 3·HF2− in (a) front
view and (b) side view. The structure of the complex is optimized with
M06-2X/6-31+G(d,p).
with hydrogen atoms in place of solubilizing substituents (OTg
or tert-butyl group). The fully coplanar, D2h-symmetric form of
the bifluoride complex 3·HF2− in the NS binding geometry was
investigated first. It turns out to be a second-order saddle point.
Its two imaginary frequencies correspond to the puckering of
the macrocycle and a tilting motion of the bifluoride out of
plane with retention of the north−south binding direction. The
optimized structure obtained after perturbing along these
imaginary frequencies has C2h-symmetry wherein the bifluoride
is tilted ca. 21° relative to the triazolophane’s mean plane and
the bifluoride’s hydrogen atom is situated perfectly at the
complex’s center of mass. The two northern triazole protons
(HT) are seen to point up toward one of the bifluoride’s
Figure 4. (a) Free energy profile of HF2− gyrating inside the cavity of
triazolophane 3 starting from one of the two degenerate NS modes
(0°). The geometry was optimized with M06-2X/6-31+G(d,p) and
energy calculations were optimized with M06-2X/6-311++G(3df,2p).
(b) Structures of the precessional states and the transition state. The
energy difference in solution phases in favor of the NS mode is
calculated from experiment (dichloromethane).
D
dx.doi.org/10.1021/ja500125r | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 5. 1H NMR titration (2 mM, CD2Cl2, 298 K, 500 MHz) of 1 with increasing equivalents of TBAHF2. The signal of the triazole (T) proton
(HT) is labeled in red, that of north−south (NS) phenylene proton (HN) in blue, that of east−west (EW) phenylene protons (HC) in magenta, and
2−
that of α-TBA+ protons in green. Proton peaks associated with 12·SiF6 are labeled with * (see the explanation in the text).
minima. The hydrogen-bonding distances (HT···F, 2.8 Å; HN···
F, 3.1 Å; HC···F, 3.2 Å) in the complex with bifluoride bound in
a perpendicular manner are larger than the sum of the van der
Waals radii (2.7 Å). Consequently, it is not surprising that the
binding free energy of this state is one-third less than that of the
others.
estimated to be 1.0
0.1 kcal mol−1 with a population
distribution of 74% in the two NS modes and 26% in the four
EW modes [Figures 3b and S13 and Tables S3, S4 (Supporting
Information)]. The difference is smaller than in the gas phase,
which we attribute to solvation. Computations using an implicit
solvation model (see Supporting Information) are consistent
with the ∼1 kcal mol−1 difference between the two binding
modes.
−
Overall, the different optimized structures of 3·HF2
obtained using computation are consistent with our exper-
imental findings in the solution phase that the NS geometry
represents the dominant mode of bifluoride binding.
−
Solution Binding Affinity of HF2 with Triazolophane
1. 1H NMR titrations of 1 with TBAHF2 in CD2Cl2 (Figure 5)
were utilized to establish the dominant equilibria present at
millimolar concentrations. In dichloromethane, ion pairing
between the cation (TBA+) and the anionic species are usually
significant.25 Since the chloride salt TBACl has an ion-pair
association constant (Kip) similar to that of TBAHF2 (see
−
Precessional Motion of HF2 in Triazolophanes. As
mentioned above, the NS binding geometry was found to be
the global minimum with the EW mode lying 2.5 kcal mol−1
higher in energy. The population of each geometric arrange-
ment is expected to be a Boltzmann average. The free energy
barrier between the NS and EW complexes via the proposed
transition state, T‡, was calculated to be 3.6 kcal mol−1 at room
temperature. This energy barrier is within the range for a C−C
rotation (3−5 kcal mol−1),36 and therefore, the precessional
Supporting Information), the ion-pair complex (1·HF2 ·TBA+)
−
between TBA+ and the 1:1 anionic complex was also included.
1
Unexpectedly, the H NMR peaks of the TBA+ cation, e.g., α-
methylene proton, were found to shift slightly upfield initially
and then turn downfield after the addition of just 0.5 equiv of
bifluoride (Figure S9, Supporting Information) rather than at
1.0 equiv, as seen with chloride titrations.25 On the basis of our
previous studies,25 we attribute this phenomenon to pairing of
−
movements of HF2 within the binding cavity are “free rotors”
at room temperature. Consequently, we expect the exchange
between minima to occur much faster than the NMR time
scale.
Since the phenylene hydrogen atoms have similarly short
hydrogen-bond distances (ca. 2 Å) in the NS and EW modes,
the greater shift in the north−south proton’s NMR resonance
indicates that the NS mode is more populated in solution. To
evaluate the energy difference, ΔΔGsol, between NS and EW
binding modes, we applied the Boltzmann distribution shown
in eq 2
the cation with the 2:1 complex to produce 12·HF2 ·TBA+. The
−
1
significant peak broadening seen at 0.5 equiv in the H NMR
spectra is consistent with the existence of higher-order species.
While the formation of 2:1 ion-pair complexes with
triazolophane has been previously hypothesized,25 the obser-
vation described above is the first direct solution-phase
evidence for this species. To rationalize its presence, we
−
compared the HF2 binding behavior to that previously seen
g N/g N = exp(−ΔEij/kBT)
with spherical halides. When a 2:1 species is formed with
halides, the structure of the sandwich complexes2c is expected
to shield the encapsulated anion from forming further
intermolecular contacts. By comparison, the tilted conforma-
tion of bifluoride within the 1:1 complex could presumably be
inherited by the 2:1 complex. Even though the structure of the
2:1 complex is prohibitively large to be accurately optimized
using DFT calculations, we propose a binding mode where
each macrocycle interacts with one fluorine terminus of the
tilted HF2− (2:1 complexes in Figure 6), and its linearity points
i
j
j
i
(2)
where kB is the Boltzmann constant, T is the temperature, g is
the degeneracy of levels having energy E, and N is the
population. Calculations indicate that the degeneracies are
different in the NS and EW modes (Figure S12, Supporting
Information). In the NS binding mode, two degenerate states
were found, while in EW there are four on account of the fact
that there is a 0.3 Å displacement of HF2− away from the center
of mass of the triazolophane. Consequently, the ΔΔGsol was
E
dx.doi.org/10.1021/ja500125r | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
2−
Binding of SiF6
:
1 + 1 + SiF 2− ⇌ 12·SiF 2−
(8)
6
6
Fortunately, this menagerie of complexes can be greatly
simplified by diluting out the weaker equilibria38 at lower
−
concentration. Consequently, the 1:1 binding of HF2 with 1
can be accurately determined by fitting the UV−vis titration
data (see Supporting Information) in CH2Cl2 (Table 1) using
an equilibrium-restricted factor analysis on the entire wave-
length range (250−340 nm) of data with Sivvu.39 At 0.5 μM,
equilibria involving higher-order species (eqs 4, 6, 7, and 8) can
be safely removed from the solution binding model (see
Supporting Information), leaving us to determine the
association constant for eq 3 after independently measuring
−
Figure 6. Cartoons showing solution-phase equilibria associated with
triazolophane 1 binding HF2 (2 mM, CD2Cl2, 298 K).
ion-pairing (eq 5). The 1:1 HF2 binding affinity (290 000
−
10 000 M−1) was found to be more than 1 order of magnitude
lower than that of chloride (7 400 000 700 000 M−1).2d
−
Rationale for Decreased HF2 Binding Affinity in
−
Solution. Examining the difference between HF2 and the
−
negative electron density into solution. Consequently, HF2 in
other anions that have been previously studied3 helped us to
rationalize the lowering of the bifluoride binding affinity in
solution relative to Cl−. Triazolophane 1 normally flattens out
its ruffled conformation when chloride,3a bromide,3a or
the sandwiched complex is still able to attract the TBA+ cation
as a 2:1 ion pair complex.
To further support our model, we monitored how the
prevalence of each species changes with concentration. Upon
dilution from 2 to 0.5 mM, the turning point of the α-
cyanide3b is bound. Upon HF2 binding, however, both the
−
NS and EW modes remain ruffled. Deformation energies40 can
be used to evaluate whether HF2− binding gives rise to a larger
penalty (Figure 7). The deformation energy for HF2− is actually
1
methylene proton’s H NMR signal moved toward 1.0 equiv,
−
which indicates that the 2:1 ion pair complex 12·HF2 ·TBA+ is
−
being diluted away and the 1:1 ion pair complex 1·HF2 ·TBA+
becomes the dominant ion-pair complex.
In addition, a minor species (asterisks in Figure 5) was
1
observed in the H NMR titration. Multiple characterization
methods confirmed the emergence and disappearance of a
sandwich complex involving the hexafluorosilicate dianion,
2−
12·SiF6 (see Supporting Information). The presence of this
silicate has been reported in the literature37 and was attributed
to the reaction between fluoride species, the glass walls, and a
−
proton source. In our case, HF2 is a capable etchant by itself
(see Supporting Information). Overall, the observations of the
1H NMR titrations led us to the binding model of eqs 3−8:
Figure 7. Deformation energies of triazolophane 3 (side view) for
binding to the different anions. The energies are calculated by M06-
2X/6-311++G(3df,2p).
Complexation:
1 + HF − ⇌ 1·HF −
(3)
2
2
1·HF − + 1 ⇌ 12·HF −
(4)
2
2
the lowest (2.5 kcal mol−1) among the NS complexes, even
after taking the 74:26 Boltzmann distribution into account
(∼2.8 kcal mol−1). Thus, the possibility that a larger
Ion-pairing:
+
TBA + HF − ⇌ TBAHF
(5)
(6)
(7)
−
2
2
deformation penalty is associated with HF2 binding can be
excluded.
+
+
TBA + 1·HF − ⇌ 1·HF −·TBA
The counterintuitive fact that three different anions, Cl−,
2
2
CN−, and HF2 , each have very similar binding energies with 3
−
+
+
TBA + 12·HF − ⇌ 12·HF −·TBA
2
2
in gas-phase calculations regardless of their orientations or
Table 1. Calculated and Experimental Binding Energies of Triazolophanes with Cl− and HF2 According to Eq 3
−
a
a
3
1
4
2
d
ΔG°gas (kcal mol−1
)
ΔGsoln (kcal mol−1
)
Ksoln (×105 M−1
74
2.9 0.1
)
ΔG°gas (kcal mol−1
−59
)
ΔGsoln (kcal mol−1
−7.4 0.1
)
Ksoln (×105 M−1
)
b
Cl−
−55
−53
−50
−49
−9.4 0.1
7
3.0 1.0
c
−
NS·HF2
EW·HF2
[D·HF2
−7.5 0.1
−
−57
−7.9 0.1
6.2 1.2
−
‡
]
a
b
Counterpoise corrected at M06-2X/6-311++G(3df,2p). Data reported in refs 2b and 2d were re-evaluated by including the formation of an ion-
c
d
pair complex and a 2:1 complex. Determined by UV−vis titration (CH2Cl2, 0.5 μM). Determined by UV−vis titration (CH2Cl2, 5 μM)
F
dx.doi.org/10.1021/ja500125r | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
hydrogen-bond configurations led us to an alternative
−
hypothesis: When the anion is tilted, e.g., HF2 (Figure 8),
Figure 8. Cartoon representations of tilted and coplanar HF2− inside a
−
macrocycle. HF2 is represented by a red, cylindrical rectangle with
two electronegative ends. The cross-section of a generic macrocycle is
represented by a box with two electropositive sites in blue, indicating
hydrogen-bond donors.
the negatively charged termini of the anion are more exposed to
solvent. Solvation can then more effectively stabilize the excess
charge on the anion and reduce its availability for hydrogen
bonding. Furthermore, the dielectric effect significantly reduces
the ionic contribution to the hydrogen bond between the
−
triazole CH and the fluorines from HF2 . These features might
explain why Cl− and HF2 are equally bound in gas-phase
−
calculations but they differ markedly in solution-phase
characterization. Computations were then performed to
complement our hypothesis. Interestingly, the gas-phase
−
calculations suggest that HF2 binding of 3 (ΔG°gas = −53
kcal mol−1) is just as strong as Cl− binding (−55 kcal mol−1).
However, after including implicit solvation, the calculations
qualitatively reproduced the experimental results. That is,
chloride binding (computed ΔGsoln = −8 kcal mol−1) was
found to be much stronger than that of bifluoride (computed
ΔGsoln = −3 kcal mol−1).
Computer-Aided Design of New Macrocyclic Recep-
tors. The tilting of HF2− within the cavity of 1 is believed to be
responsible for its lower binding affinity. Consequently, a
redesign of the receptor was considered as a means to generate
a nontilted binding mode (Figure 8). This geometric factor is
−
expected to enhance the HF2 affinity and to ensure that its
binding is equal to that of Cl− in both gas and solution phases.
Elongated designs,6b which have been shown to have good
−
−
binding with linear anions, e.g., N3 and HF2 , are not
appropriate candidates, as these receptors usually weakened Cl−
affinity due to the size mismatch. In order to best test our
hypothesis, new molecular designs are needed where two
Figure 9. Chemical structures of macrocycles 4−6 and their molecular
design features based on 3. Optimized structures using M06-2X/6-
31+G(d,p) of 3−6 are oriented in EW side views.
−
criteria are met: (1) HF2 is bound nontilted and (2) similar
incorporating a flexible 1,3-propylene linker, thereby allowing
triazolophane 5 to accommodate a slightly larger anion. Prior
work with this substitution showed a negligible lowering of the
triazolophane’s Cl− affinity.2f Consistent with our expectations,
structural optimization of its EW mode showed a ∼50%
reduction in the tilting angle to 10°. In addition, the cavity size
along the east−west axis became 0.4 Å larger than that of the
NS mode, suggesting that the EW mode in triazolophane 5 may
binding affinity exists toward Cl− and HF2 . The second
−
criterion allows the impact of the tilting behavior on solution
binding affinities to be verified; if HF2− is bound in a nontilted
mode, we would then observe similar binding affinity to Cl− in
solution. While it is not feasible to determine the tilting angle of
−
a bound HF2 by observation, we employed computer-aided
design on a series of triazolophanes and monitored the
resulting tilting angle within the structurally optimized complex
(Figure 9).
The first feature investigated was size- and shape-matching.
The rigidity of the triazolophane was relaxed slightly by
−
now be preferred by HF2 .
The second strategy involved tuning the electronics by
adding carboxylic acid as electron-withdrawing groups onto the
EW phenylenes. In addition, a hydroxyl group was introduced
G
dx.doi.org/10.1021/ja500125r | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
to the north phenylene to mimic any glycol ether chains that
would be needed for solubility. However, calculations on
triazolophane 6 yielded no significant improvement. Presum-
ably, the inductive effect of esters is too weak to affect the
electropositive character of triazole protons. An insignificant
change in cavity size for the EW and NS modes was observed,
which may also reflect the minimal change in electronics (Table
S4, Supporting Information).
coincides with the ∼30 kJ mol−1 increase in the maximum
electrostatic potential (B3LYP/6-31+G*) inside the cavity from
225 kJ mol−1 for 1 to 255 kJ mol−1 for 4. Finally, gas-phase
calculations showed that the binding energies between
−
macrocycle 4 and Cl− and HF2 are equal (Table 1). These
attributes marked 4 as a suitable target molecule for further
experimental binding studies.
Overall, by using computer-aided design we examined how
various molecular design features could impact the binding
geometry of bifluoride. Elongation alone, by introducing a
propylene linker, was insufficient to accommodate bifluoride
anion in 5. Neither will the enhanced electropositive character
improve the performance if the substituents are introduced on
the wrong positions. Installing electron-withdrawing groups in
the EW phenylenes in 6 fails to provide significant improve-
ment, although their CH groups are linearly aligned with the
bound bifluoride in the local minimum structure. Use of
electron-withdrawing amides groups on the phenylenes that
also interact directly with the triazoles has a much better
performance. This observation indicates that the binding
energy and geometry is largely dependent on the triazoles’
interactions with the anion. Even then, in triazolophane 4, the
size and shape are still mismatched: the length of bifluoride is
∼1 Å larger than the length of the cavity along the EW
direction; the aspect ratio of the cavity between the EW and NS
directions is ∼1.1, i.e., almost spherical. Therefore, the subtle
elongation in EW directions (∼0.4 Å) must cooperate
positively with the enhanced electropositive character (∼30
kJ mol−1) in 4 to allow the bifluoride anion to be bound in
plane and without any tilting.
To more significantly enhance the electropositive character
of the cavity, four flanking amides were introduced onto the
phenylenes along the east−west axis to yield amide
triazolophane 4. The intramolecular NH···N hydrogen bonds
formed between amide NH donors, and the four triazoles were
previously discovered to polarize CH donors.26 As computa-
tional studies have shown, the cavity size of 4 is further
expanded so that the length along the EW is 0.8 Å larger than
that of the NS. The expansion of the cavity along the east−west
direction may have resulted from the increased repulsions
between the more electropositive triazole protons. Interest-
ingly, the largest dimension of the pocket along the east−west
axis of 4 (r = 3.9 Å) is only 0.3 Å larger than that of
triazolophane 3 and is still smaller than the van der Waals
−
radius of HF2 along the F−H−F axis (r = 5.3 Å). With this
macrocycle, we verified computationally that the desired tilt-
−
angle of 0° was achieved. The nontilted conformation of HF2
binding with triazolophane 4 is attributed to the enhanced
attraction between 4 and HF2− (Figure 10). The enhancement
−
Solution Binding Studies of HF2 in Triazolophane 2.
The semirigid macrocycle 2 was synthesized2b as the
experimental analog of 4 to test the effectiveness of the
computer-aided design (Scheme S1, Supporting Information).
1H NMR titrations of 2 were performed with TBAHF2 (Figure
11) and TBACl (Figure 12) in CDCl3. In the Cl− titrations, the
2:1 complex persists (red boxes, Figure 12), and it requires ca. 6
equiv of Cl− to shift the preferences to favor the 1:1 complex.
Figure 10. Electrostatic potential maps of triazolophanes 3 and 4.
Figure 11. 1H NMR titration (2 mM, CDCl3, 298 K) of 2 with increasing equivalents of TBAHF2. The signal of the triazole protons (HT) is labled
−
in red, that of north phenylene proton (HN) in blue, and that of EW phenylene protons (HC) in magenta. 22·SiF62− is labeled with asterisks. 22·HF2
is shown in gray boxes.
H
dx.doi.org/10.1021/ja500125r | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 12. 1H NMR titration (2 mM, CDCl3, 298 K) of 2 with increasing equivalents of TBACl. The signal of the triazole protons (HT) is labeled in
red, that of north phenylene proton (HN) in blue, and that of EW phenylene protons (HC) in magenta. 22·Cl− is shown in red boxes.
1
Figure 13. Low-temperature H NMR spectra of macrocycle 2 (2 mM) in CDCl3 with TBAHF2 (20 equiv, 400 MHz, 223 and 298 K). Silicate
2−
complex 22·SiF6 was promoted by lowering the temperature and is marked with asterisks.
+
+
TBA + 2·X− ⇌ 2·X−·TBA
(12)
In comparison to the parent triazolophane, the overall
formation (β ∼ 109) of the 2:1 complex (gray boxes, Figure 11)
was determined to be 1 order of magnitude weaker. The
bifluoride anion inside the 2:1 species is instead replaced by
silicate with the formation of 22·SiF6 seen from 1.6 to 15
equiv as marked by the asterisks in Figure 11. The silicate
2−
Binding of SiF6
:
2 + 2 + SiF 2− ⇌ 22·SiF 2−
2−
(13)
6
6
Upon saturation with HF2 , the triazole (HT) and phenylene
protons (HC) remained broadened into the baseline, indicative
of the dynamics associated with the bound bifluoride. These
peaks can be resolved by lowering the temperature (Figure 13)
with the two inequivalent triazole peaks, Hd and He, now
evident. Interestingly, phenylene proton (HN) in the north
behaves distinctly differently from the similar north−south
protons in 1 (Figure 5) and the other hydrogen-bonded
protons in 2 (HT and HC): It remains sharp throughout the
titration. Therefore, HN is not participating in the dynamic
−
sandwich complex is ultimately replaced by the 1:1 bifluoride
−
complex beyond 15 equiv of HF2 .
The formation of ion pairs is still observed in chloroform and
included in the binding models. A significant turning point
observed in the α-methylene proton of TBA+ upon HF2
−
addition (Figure 11) and a more subtle turning point in the
case of TBACl (Figure 12) are both seen at ca. 1.0 equiv. For
the Cl− titration, the higher population of the 2:1 sandwich
complex at 1.0 equiv reduced the significance of the 1:1 ion-pair
complex (Figure S32, Supporting Information). These features
are summarized by the following equilibria (X = HF2− or Cl−):
−
behavior and it likely has a small role to play in stabilizing HF2 .
The very small complexation-induced chemical shift of HN
Complexation:
−
associated with 2·HF2 , which is 0.7 ppm smaller than that in
2 + X− ⇌ 2·X−
(9)
chloride complex, supports this last idea. These observations
suggest that the bound bifluoride has been locked into the EW
binding mode, presumably as a result of three factors: (1) The
NS cavity is now much smaller than the EW one; (2) the
2·X− + 2 ⇌ 22·X−
(10)
Ion-pairing:
−
nontilted EW binding mode results in HF2 occupying the
cavity so that the solvent effect is neutralized; and (3)
+
TBA + X− ⇌ TBAX
(11)
I
dx.doi.org/10.1021/ja500125r | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
polarization of the CH bonds along the EW direction shifts the
orientational preference of bound HF2− away from the original
NS mode. Having created a binding mode that we believe is
more competent for bifluoride, we are in a position to test our
hypothesis that the solution binding will be consistent with the
gas-phase calculation, i.e., the binding affinity of triazolophane 2
with HF2− is expected to match the Cl− affinity in both gas and
solution phases.
ASSOCIATED CONTENT
* Supporting Information
Computational methods, syntheses, NMR data, optimized
structures, UV−vis data, and titration analyses. This material
is available free of charge via the Internet at http://pubs.acs.org.
■
S
AUTHOR INFORMATION
Corresponding Author
kraghava@indiana.edu; aflood@indiana.edu
■
−
The 1:1 binding affinities of HF2 and Cl− with
triazolophane 2 were quantitatively determined by equili-
brium-restricted factor analysis of the UV−vis titrations
Author Contributions
†R.O.R. and Y.L. contributed equally.
(CH2Cl2, 5 μM, Table 1).41 The 1:1 HF2 binding energy
−
Notes
(−7.9 0.1 kcal mol−1) was found to match the Cl− binding
The authors declare no competing financial interest.
affinity with 2 (−7.4
0.1 kcal mol−1). This similarity
demonstrates that the tilted binding model can account for the
ACKNOWLEDGMENTS
■
−
reduced binding affinity of HF2 to the parent tetraphenylene
R.O.R., Y.L., Y.H., A.H.F., and K.R. gratefully acknowledge
support by the Chemical Sciences, Geosciences and Biosciences
Division, Office of Basic Energy Sciences, Office of Science,
U.S. Department of Energy (DE-FG02-09ER16068). M.C.
thanks the Ministerio de Educacion y Ciencia, FPU fellowship
AP2008-02355 for support. A.H.F. and M.C. thank Pablo
Ballester at the Institute of Chemical Research of Catalonia
(ICIQ), Tarragona, Spain for support.
triazolophane macrocycles, like 1, in solution phases. The
computer-aided molecular design can successfully reinforce the
−
binding strength of HF2 to triazolophane macrocycles.
The only three previous studies6−8 on bifluoride binding
showed how the anion can be bound in the oval-shaped cavities
of neutral receptors6 and the Li+ in charged receptors7 and the
fact that bifluoride can compete with the binding of fluoride.
Our study reveals for the first time how bifluoride can also be
bound in a mismatched host like 1. In contrast to previous
studies, where conclusions were established by varying the
guests, we took a complementary approach by modifying the
structure of the host. Herein, a new lesson in receptor design is
thus learned: the binding affinity of bifluoride does not depend
entirely on size and shape matching. In the gas phase, where only
the host−guest interactions are considered, bifluoride was
shown to be bound with the same affinity as chloride, albeit in a
tilted geometry as a result of shape and size mismatch. In
solution, however, a completely different story ensues.
Computer-aided design was then used to inspect the role of
solvent, where the tilting was examined in different macrocycles
and was eventually removed in the final design. Overall, we
learn that the tilting of bifluoride is a compromise between
minimizing the hard sphere repulsions between the bifluoride
anion and the macrocycle cavity and maximizing the anion’s
hydrogen-bond interactions with the triazoles.
REFERENCES
■
(1) Sessler, J. L.; Gale, P. A.; Cho, W.-S. Anion Receptor Chemistry;
RSC Publishing: Cambridge, U.K., 2006.
(2) Synthesis and anion binding studies of tetraphenylene
triazolophanes: (a) Li, Y.; Flood, A. H. Angew. Chem., Int. Ed. 2008,
47, 2649−2652. (b) Li, Y.; Flood, A. H. J. Am. Chem. Soc. 2008, 130,
12111−12122. (c) Li, Y.; Pink, M.; Karty, J. A.; Flood, A. H. J. Am.
Chem. Soc. 2008, 130, 17293−17295. (d) Li, Y.; Vander Griend, D. A.;
Flood, A. H. Supramol. Chem. 2009, 21, 111−117. (e) Hua, Y.; Flood,
A. H. Chem. Soc. Rev. 2010, 39, 1262−1271. (f) Hua, Y.;
Ramabhadran, R. O.; Karty, J. A.; Raghavachari, K.; Flood, A. H.
Chem. Commun. 2011, 47, 5979−5981. (g) McDonald, K. P.; Hua, Y.;
Lee, S.; Flood, A. H. Chem. Commun. 2012, 48, 5065−5075. (h) Lee,
S.; Flood, A. H. Top. Heterocycl. Chem. 2012, 28, 85−107.
(3) Computational studies of anion binding with tetraphenylene
triazolophanes: (a) Bandyopadhyay, I.; Raghavachari, K.; Flood, A. H.
ChemPhysChem 2009, 10, 2535−2540. (b) Ramabhadran, R. O.; Hua,
Y.; Flood, A. H.; Raghavachari, K. Chem.Eur. J. 2011, 17, 9123−
9129.
(4) Selected literature for works in C−H hydrogen-bond donors:
(a) Farnham, W. B.; Roe, D. C.; Dixon, D. A.; Calabrese, J. C.; Harlow,
R. L. J. Am. Chem. Soc. 1990, 112, 7707−7718. (b) Desiraju, G. R.;
Steiner, T. The Weak Hydrogen Bond in Structural Chemistry and
Biology; Oxford University Press: New York, 1999. (c) Lee, C.-H.; Na,
H.-K.; Yoon, D.-W.; Won, D.-H.; Cho, W.-S.; Lynch, V. M.; Shevchuk,
S. V.; Sessler, J. L. J. Am. Chem. Soc. 2003, 125, 7301−7306.
(d) Chmielewski, M. J.; Charon, M.; Jurczak, J. Org. Lett. 2004, 6,
3501−3504. (e) Kwon, J. Y.; Jang, Y. J.; Kim, S. K.; Lee, K. H.; Kim, J.
S.; Yoon, J. J. Org. Chem. 2004, 69, 5155−5157. (f) Ilioudis, C. A.;
Tocher, D. A.; Steed, J. W. J. Am. Chem. Soc. 2004, 126, 12395−12402.
(g) Maeda, H.; Kusunose, Y. Chem.Eur. J. 2005, 11, 5661−5666.
(h) Bryantsev, V. S.; Hay, B. P. Org. Lett. 2005, 7, 5031−5034.
(i) Bryantsev, V. S.; Hay, B. P. J. Am. Chem. Soc. 2005, 127, 8282−
8283. (j) Maeda, H.; Haketa, Y.; Nakanishi, T. J. Am. Chem. Soc. 2007,
129, 13661−13674. (k) Fujimoto, C.; Kusunose, Y.; Maeda, H. J. Org.
Chem. 2006, 71, 2389−2394. (l) Zhu, S. S.; Staats, H.; Brandhorst, K.;
Grunenberg, J.; Gruppi, F.; Dalcanale, E.; Lutzen, A.; Rissanen, K.;
Schalley, C. A. Angew. Chem., Int. Ed. 2008, 47, 788−792. (m) Yoon,
D. W.; Gross, D. E.; Lynch, V. M.; Sessler, J. L.; Hay, B. P.; Lee, C. H.
Angew. Chem., Int. Ed. 2008, 47, 5038−5042. (n) Berryman, O. B.;
Sather, A. C.; Hay, B. P.; Meisner, J. S.; Johnson, D. W. J. Am. Chem.
Soc. 2008, 130, 10895−10897. (o) Lee, S.; Chen, C.-H.; Flood, A. H.
CONCLUSIONS
■
Computer-aided molecular design was shown to accelerate the
creation of a new anion receptor capable of complexing the
often overlooked, yet surprisingly common, bifluoride anion in
a nontilted arrangement with high strength. This binding
geometry was believed to be critical for keeping the binding
energy high by preventing solvent-induced destabilization of
the complex. In the parent triazolophane, with its spherical
cavity, calculations show that the bifluoride is tilted while the
shape-matched chloride fits snugly inside. In this case, the gas-
phase affinity is the same for both anions, but it is not so in
solution, where HF2 < Cl−. When the triazolophane is
−
customized to allow the bifluoride complex to mimic the
chloride’s coplanar geometry, their stabilities are observed to
−
equalize, HF2 ≈ Cl−. Interestingly, the bifluoride was bound
with reasonable strength, indicating that it could compete for
receptor binding whenever fluoride is present. These findings
indicate that future fluoride-based studies of binding should
take these chemical species into consideration.
J
dx.doi.org/10.1021/ja500125r | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Nat. Chem. 2013, 5, 704−710. (p) Hua, Y.; Liu, Y.; Chen, C.-H.;
Flood, A. H. J. Am. Chem. Soc. 2013, 135, 14401−14412.
(5) (a) Frisch, M. J.; Del Bene, J. E.; Binkley, J. S.; Schaefer, H. F., III
J. Chem. Phys. 1986, 84, 2279−2289. (b) Emsley, J. Chem. Soc. Rev.
1980, 9, 91−124.
(6) (a) Kang, S. O.; Powell, D.; Day, V. W.; Bowman-James, K.
Angew. Chem., Int. Ed. 2006, 45, 1921−1925. (b) Kang, S. O.; Day, V.
W.; Bowman-James, K. Inorg. Chem. 2010, 49, 8629−8636.
(7) Abraham, Y.; Salman, H.; Suwinska, K.; Eichen, Y. Chem.
Commun. 2011, 47, 6087−6089.
(8) Lehaire, M.-L.; Scopelliti, R.; Severin, K. Inorg. Chem. 2002, 41,
5466−5474.
(17) Gormley, J. Professional Carwashing and Detailing 2001, 25, issue
1.
(18) Bifluoride is the most important congenitor of fluoride. Its
formation begins when basic fluoride anion abstracts a proton from an
appropriate source to form HF, which is a competent Lewis acid.
Subsequently, a strong hydrogen bond5 is formed between the Lewis
−
acidic HF and Lewis basic F− generating bifluoride, HF2 , as the
adduct. The proton sources capable of driving this equilibrium are
everywhere. First, solvents contain activated hydrogen atoms, e.g.,
chloroform and acetonitrile.19 In these solvents, bifluoride formation
will ideally be observed upon addition of fluoride and might even
become the dominating fluoride-containing species when equilibrium
is achieved.6a Second, trace amounts of water may be present. Third,
fluoride salts, like the tetrabutylammonium fluoride that is commonly
used in F− binding studies, normally contain finite waters of
crystallization, and anhydrous forms of fluoride20 are not straightfor-
ward to generate. Fourth, the hydrogen-bond donors present in anion
receptors are also sources.21 Consequently, the formation of bifluoride
is commonly observed when more than 1.0 equiv of fluoride is added
during a titration. In water, the stability of bifluoride is subject to
equilibration22 such that it is only present in acidic media and at higher
concentrations. At a concentration above 10 mM, bifluoride is present
from pH 1 to 6. Bifluoride will dissociate into fluoride in basic aqueous
media, where the hydroxide ions dominate over hydronium ions.
These alkali solutions and at low concentrations are probably the only
exceptions where bifluoride is not present in the same solutions as
fluoride. Other than these conditions, bifluoride is ubiquitous in the
media where fluoride binding studies are conducted.
(9) Selected papers in using bifluoride as ligands: (a) Murphy, V. J.;
Hascall, T.; Chen, J. Y.; Parkin, G. J. Am. Chem. Soc. 1996, 118, 7428−
7429. (b) Whittlesey, M. K.; Perutz, R. N.; Greener, B.; Moore, M. H.
Chem. Commun. 1997, 2, 187−188. (c) Roe, D. C.; Marshall, W. J.;
Davidson, F.; Soper, P. D.; Grushin, V. V. Organometallics 2000, 19,
4575−4582. (d) Jasim, N. A.; Perutz, R. N.; Foxon, S. P.; Walton, P.
́
H. J. Chem, Soc., Dalton Trans. 2001, 11, 1676−1685. (e) Jose, V.; Gil-
Rubio, J.; Bautista, D.; Sironi, A.; Masciocchi, N. Inorg. Chem. 2004,
43, 5665−5675. (f) Vergote, T.; Nahra, F.; Welle, A.; Luhmer, M.;
Wouters, J.; Mager, N.; Riant, O.; Leyssens, T. Chem.Eur. J. 2012,
18, 793−798.
(10) Selected papers in using bifluoride as reagents: (a) Bosch, P.;
Camps, F.; Chamorro, E.; Gasol, V.; Guerrero, A. Tetrahedron Lett.
1987, 28, 4733−4736. (b) Sanghera, J. S.; Hart, P.; Sachon, M. G.;
Ewing, K. J.; Aggarwal, I. J. Am. Ceram. Soc. 1990, 73, 1339−1346.
(c) Vedejs, E.; Chapman, R. W.; Fields, S. C.; Lin, S.; Schrimpf, M. R.
J. Org. Chem. 1995, 60, 3020−3027. (d) Molander, G. A.; Biolatto, B. J.
Org. Chem. 2003, 68, 4302−4314. (e) Hayashi, T.; Ueyama, K.;
Tokunaga, N.; Yoshida, K. J. Am. Chem. Soc. 2003, 125, 11508−11509.
(f) Ooi, T.; Takada, S.; Doda, K.; Maruoka, K. Angew. Chem., Int. Ed.
2006, 45, 7606−7608. (g) Vergote, T.; Nahra, F.; Peeters, D.; Riant,
O.; Leyssens, T. J. Oganomet. Chem. 2013, 730, 95−103.
(11) (a) Manson, J. L.; Conner, M. M.; Schlueter, J. A.; Lancaster, T.;
Blundell, S. J.; Brooks, M. L.; Pratt, F. L.; Papageorgiou, T.; Bianchi, A.
D; Wosnitzae, J.; Whangbo, M.-H. Chem. Commun. 2006, 42, 4894−
4896. (b) Manson, J. L.; Schlueter, J. A.; Funk, K. A.; Sutherland, H. I.;
Twamley, B.; Lancaster, T.; Blundell, S. J.; Baker, P. J.; Pratt, F. L.;
Singleton, J.; McDonald, R. D.; Goddard, P. A.; Sengupta, P.; Batista,
C. D.; Ding, L.; Lee, C.; Whangbo, M.-H.; Franke, I.; Cox, S.; Baines,
C.; Trial, D. J. Am. Chem. Soc. 2009, 131, 6733−6747. (c) Manson, J.
L.; Warter, M. L.; Schlueter, J. A.; Lancaster, T.; Steele, A. J.; Blundell,
S. J.; Pratt, F. L.; Singleton, J.; McDonald, R. D.; Lee, C.; Whangbo,
M.-H.; Plonczak, A. Angew. Chem., Int. Ed. 2011, 50, 1573−1576.
(d) Manson, J. L.; Lapidus, S. H.; Stephens, P. W.; Peterson, P. K.;
Carreiro, K. E.; Southerland, H. A; Lancaster, T.; Blundell, S. J.; Steele,
A. J.; Goddard, P. A.; Pratt, F. L.; Singleton, J.; Kohama, Y.;
McDonald, R. D.; Del Sesto, R. E.; Smith, N. A.; Bendix, J.; Zvyagin, S.
A.; Kang, J.; Lee, C.; Whangbo, M.-H.; Zapf, V. S.; Plonczak, A. Inorg.
Chem. 2011, 50, 5990−6009.
(12) (a) Carr, D. W.; Craighead, H. G. J. Vac. Sci. Technol. B 1997,
15, 2760−2763. (b) Kim, J.-H.; Lim, D. H. Y.; Gye, M. J. Vac. Sci.
Technol. B 1998, 16, 558−560. (c) Doughty, C.; Knick, D. C.; Bailey, J.
B.; Spencer, J. E. J. Vac. Sci. Technol. A 1999, 17, 2612−2618.
(13) (a) Zhang, W.; Hu, Z.; Liu, Y.; Chen, H.; Gao, S.; Gaschnig, R.
M. Anal. Chem. 2012, 84, 10686−10693. (b) Kabangu, M. J.; Crouse,
P. L. Hydrometallurgy 2012, 129, 151−155.
(14) (a) Goldcamp, M. J.; Goldcamp, D. M.; Ashley, K.; Fernback, J.
E.; Agrawal, A.; Millson, M.; Marlow, D.; Harrison, K. J. Occup.
Environ. Hyg. 2009, 6, 735−744. (b) Li, G.; Rao, M.; Li, Q.; Peng, Z.;
Jiang, T. Trans. Nonferrous Met. Soc. China 2010, 20, 1517−1520.
(15) Bhatti, Z. A.; Mahmood, Q.; Raja, I. A.; Malik, A. H.; Khan, M.
S.; Wu, D. Phys. Chem. Earth 2011, 36, 465−469.
(19) Christe, K.; Wilson, W. W.; Wilson, R. D.; Bau, R.; Feng, J.-A. J.
Am. Chem. Soc. 1990, 2, 7619−7625.
(20) (a) Sun, H.; DiMagno, S. G. J. Am. Chem. Soc. 2005, 127, 2050−
2051. (b) Aragay, G.; Frontera, A.; Lloveras, V.; Vidal-Gancedo, J.;
Ballester, P. J. Am. Chem. Soc. 2013, 135, 2620−2627.
(21) (a) Boiocchi, M.; Del Boca, L.; Gomez, D. E.; Fabbrizzi, L.;
Licchelli, M.; Monzani, E. J. Am. Chem. Soc. 2004, 126, 16507−16514.
(b) Gu, R.; Depraetere, S.; Kotek, J.; Budka, J.; Wagner-Wysiecka, E.
W.; Biernat, J. F.; Dehaen, A. Org. Biomol. Chem. 2005, 3, 2921−2923.
(c) Evans, L. S.; Gale, P. A.; Light, M. E.; Quesada, R. New J. Chem.
2006, 30, 1019−1025. (d) Cametti, M.; Rissanen, K. Chem. Commun.
2009, 20, 2809−2829. (e) Juwarker, H.; Lenhardt, J. M.; Castillo, J. C.;
Zhao, E.; Krishnamurthy, S.; Jamiolkowski, R. M.; Kim, K.-H.; Craig, S.
L. J. Org. Chem. 2009, 74, 8924−8934.
(22) Haque, R.; Reeves, L. W. J. Am. Chem. Soc. 1967, 89, 250−252.
(23) Cametti, M.; Rissanen, K. Chem. Soc. Rev. 2013, 42, 2016−2038.
(24) Selected reviews regarding fluoride binding and sensing:
(a) Snowden, T. S.; Anslyn, E. V. Curr. Opin. Chem. Biol. 1999, 3,
740−746. (b) Sessler, J. L.; Camiolo, S.; Gale, P. A. Coord. Chem. Rev.
2003, 240, 17−55. (c) Gunnlaugsson, T.; Glynn, M.; Tocci, G. M.;
Kruger, P. E.; Pfeffer, F. M. Coord. Chem. Rev. 2006, 250, 3094−3117.
(d) Cametti, M.; Rissanen, K. Chem. Commun. 2009, 20, 2809−2829.
(e) Wade, C. R.; Broomsgrove, A. E. J.; Aldridge, S.; Gabbai, F. P.
Chem. Rev. 2010, 110, 3958−3984.
(25) Hua, Y.; Ramabhadran, R. O.; Uduehi, E. O.; Karty, J. A.;
Raghavachari, K.; Flood, A. H. Chem.Eur. J. 2011, 17, 312−321.
(26) McDonald, K. P.; Ramabhadran, R. O.; Lee, S.; Raghavachari,
K.; Flood, A. H. Org. Lett. 2011, 13, 6260−6263.
(27) Hay, B. P.; Firman, T. K. Inorg. Chem. 2002, 41, 5502−5512.
(28) Hay, B. P. Chem. Soc. Rev. 2010, 39, 3700−3708.
(29) (a) Hay, B. P.; Firman, T. K.; Lumetta, G. J.; Rapko, B. M.;
Garza, P. A.; Sinkov, S. I.; Hutchison, J. E.; Parks, B. W.; Gilbertson, R.
D.; Weakley, T. G. R. J. Alloys Compd. 2004, 374, 416−419. (b) Hay,
B. P.; Gutowski, M.; Dixon, D. A.; Garza, J.; Vargas, R.; Moyer, B. A. J.
Am. Chem. Soc. 2004, 126, 7925−7934. (c) Hay, B. P.; Firman, T. K.;
Moyer, B. A. J. Am. Chem. Soc. 2005, 127, 1810−1819. (d) Hay, B. P.;
Oliferenko, A. A.; Uddin, J.; Zhang, C. G.; Firman, T. K. J. Am. Chem.
Soc. 2005, 127, 17043−17053. (e) Bryantsev, V. S.; Hay, B. P. J. Am.
Chem. Soc. 2006, 128, 2035−2042. (f) Bryantsev, V. S.; Hay, B. P. J.
Phys. Chem. A 2006, 110, 4678−4688. (g) Reyheller, C.; Hay, B. P.;
Kubik, S. New J. Chem. 2007, 31, 2095−2102. (h) Custelcean, R.;
(16) (a) Chuang, R.; Exelbert, E. J.; Heard, K. J. Clin. Toxicol. 2009,
47, 741−741. (b) Maddry, J.; Brozovich, D.; Heard, K. Clin. Toxicol.
2012, 50, 601−602. (c) Toledo, F.; Silvestre, J. F.; Cuesta, L.; Banuls,
J. Pediatr. Dermatol. 2013, 30, 145−147.
K
dx.doi.org/10.1021/ja500125r | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Bosano, J.; Bonnesen, P. V.; Kertesz, V.; Hay, B. P. Angew. Chem., Int.
Ed. 2009, 48, 4025−4029.
(30) Mascal, M.; Armstrong, A.; Bartberger, M. D. J. Am. Chem. Soc.
2002, 124, 6274−6276.
(31) (a) Mascal, M. Angew. Chem., Int. Ed. 2006, 45, 2890−2893.
(b) Mascal, M.; Yakovlev, I.; Nikitin, E. B.; Fettinger, J. C. Angew.
Chem., Int. Ed. 2007, 46, 8782−8784.
(32) (a) Houk, K. N.; Nakamura, K.; Sheu, C.; Keating, A. E. Science
1996, 273, 627−629. (b) Wang, X.; Houk, K. N. Org. Lett. 1999, 1,
591−594. (c) Helgeson, R.; Heyden, A. E.; Houk, K. N. J. Org. Chem.
2010, 75, 570−575. (d) Wang, H.; Liu, F.; Helgeson, R.; Houk, K. N.
Angew. Chem., Int. Ed. 2012, 52, 655−659. (e) Edwards, N.; Helgeson,
R.; Zhao, Y.; Houk, K. N. Mol. Cryst. Liq. Cryst. 2006, 456, 175−192.
(33) Frisch, M. J. et al. Gaussian Development Version, Revision H.08;
Gaussian Inc.: Wallingford, CT, 2010.
(34) Zhao, Y.; Truhlar, D. G. Theor. Chem. Acc. 2008, 120, 215−241.
(35) Mayer, I.; Surjan, P. R. Chem. Phys. Lett. 1992, 191, 497−499.
(36) Huheey, J. E.; Keiter, E. A.; Keiter, R. L. Inorganic Chemistry
Principles of Structure and Reactivity; Harper Collins: New York, 1993.
(37) (a) Morgan, G.; McKee, V.; Nelson, J. J. Chem. Soc., Chem.
Commun. 1995, 16, 1649−1652. (b) Mason, S.; Llinares, J. M.;
Morton, M.; Clifford, T.; Bowman-James, K. J. Am. Chem. Soc. 2000,
122, 1814−1815.
(38) The qualitative analysis performed in HySS to simulate the
titrations under UV−vis concentrations shows that the 2:1 complex,
−
12·HF2 , is present at low relative amounts (Figure S16, Supporting
Information). Although there are three absorbing species, triazolo-
−
phane 1, the 1:1 complex 1·HF2 , and a residual amount of the 2:1
−
complex 12·HF2 , present during UV−vis titration as based on an
unrestricted factor analysis of the data, the sandwich complex was
present at such low amounts that including it did not result in
−
reasonable fitting outcomes. Consequently, 12·HF2 was excluded
from the data fitting.
(39) Vander Griend, D. A.; Bediako, D. K.; DeVries, M. J.; DeJong,
N. A.; Heeringa, L. P. Inorg. Chem. 2008, 47, 656−662.
(40) The deformation energy measures the energy difference
between the geometry of the macrocycle in its lowest energy
conformation and after it has changed shape to accommodate the
guest.
(41) The solubility of 2 is limited in CH2Cl2, which only allows for
UV−vis titrations. On account of the fact that CH2Cl2 has similar
solvent property as CHCl3, the models established in NMR titrations
are transferrable to UV−vis studies.
L
dx.doi.org/10.1021/ja500125r | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX