Towards redox-switchable organocatalysts based on bidentate halogen bond donors

Article abstract of DOI:10.1039/d0cp06612e

Redox-active bidentate halogen bond donors based on halopyridinium groups as halogen-bond donating units were synthesized and their structures were elucidated by X-ray diffraction analyses and DFT calculations.Viareversible twofold reduction, these dicationic species can be transformed to neutral compounds which should be much weaker Lewis acids. The corresponding electrochemical data were obtained, and CV as well as UV-vis and NMR techniques were also used to determine binding constants of these halogen bond donors to halides. While all titrations agree on the relative order of binding strengths (with chloride being bound strongest), there are marked deviations in the overall affinity constants which are discussed. In contrast to earlier azo-bridge analogues, the ethylene-linked variants presented herein do not oxidize halides, and thus the novel halogen bond donors could also be used as Lewis acidic organocatalysts in a halide abstraction benchmark reaction, yielding a performance similar to bis(haloimidazolium)-derived catalysts.

Full text of DOI:10.1039/d0cp06612e

PCCP
View Article Online
View Journal | View Issue
PAPER
Towards redox-switchable organocatalysts based
on bidentate halogen bond donors†
Cite this: Phys. Chem. Chem. Phys.,
a
b
a
c
b
2
021, 23, 4344
E. Engelage, ‡ H. Hijazi, ‡ M. Gartmann, L.-M. Chamoreau, B. Schollhorn,*
¨
a
b
S. M. Huber * and C. Fave
*
Redox-active bidentate halogen bond donors based on halopyridinium groups as halogen-bond
donating units were synthesized and their structures were elucidated by X-ray diffraction analyses and
DFT calculations. Via reversible twofold reduction, these dicationic species can be transformed to
neutral compounds which should be much weaker Lewis acids. The corresponding electrochemical data
were obtained, and CV as well as UV-vis and NMR techniques were also used to determine binding
constants of these halogen bond donors to halides. While all titrations agree on the relative order of
binding strengths (with chloride being bound strongest), there are marked deviations in the overall
affinity constants which are discussed. In contrast to earlier azo-bridge analogues, the ethylene-linked
variants presented herein do not oxidize halides, and thus the novel halogen bond donors could also be
used as Lewis acidic organocatalysts in a halide abstraction benchmark reaction, yielding a performance
similar to bis(haloimidazolium)-derived catalysts.
Received 22nd December 2020,
Accepted 26th January 2021
DOI: 10.1039/d0cp06612e
rsc.li/pccp
halogen bond donor to form elemental bromine and the
hydrazine form of the donor. Careful comparison experiments
Introduction
Recently we have demonstrated the possibility to tune and could still demonstrate that the azo-bis-pyridinium derivatives
control the strength of halogen bonding (XB) by electrochemi- -2 (Scheme 1) were active through halogen bonding, but the
X
4
cally switching the redox state of the respective electroactive XB observed redox activity does of course strongly limit the
donors or acceptors. The observed shift of the standard poten- potential application of these halogen-based Lewis acids in
tials can be attributed to a stabilization of the oxidized or halide abstraction reactions. The situation should be different,
reduced redox probes via charge-assisted halogen bonding. The however, for the C analogous tetrachloro, tetrabromo and
first proof of concept was brought for p-quinones playing the tetraiodo stilbene derivatives X -1 (Scheme 1), which should
4
1
role of electrochemically active XB-acceptors. Since then it has be much less sensitive to reduction. In fact, purposeful
been applied to various types of redox active XB-donors such as reduction of these compounds by electrochemistry would allow
2
3
4
iodinated TTF, ferrocene or viologen derivatives confirming to basically switching off their halogen bonding capabilities.
the interesting potential of electrochemical techniques as a Changing the XB-donor strength of such bidentate ligands
5
powerful tool to detect and control XB in solution. In parallel via a reversible electrochemical reaction can lead to a redox-
we described the use of halogenated azo-bis-pyridinium deri- switchable receptor. In this work, we report the preparation of
vatives as activators in a proof-of-principle reaction (Scheme 1, the novel receptors X -1, the investigation of their XB-donor
4
6
X4-2 derivatives). Even though the activity of these halogen properties depending on the oxidation state and the evaluation
bond donors was very promising, the analysis of the reaction of their catalytic properties. Their interesting potential as
7
,8
was severely complicated by a side reaction, in which the redox-switchable catalysts
in organocatalysis is discussed.
N
bromide liberated from the S 1 substrate was oxidized by the
a
Fakult a¨ t f u¨ r Chemie und Biochemie, Ruhr-Universit a¨ t Bochum, Universit a¨ tsstrasse
50, 44801 Bochum, Germany. E-mail: stefan.m.huber@rub.de
1
b
c
Universit ´e de Paris – Laboratoire d’Electrochimie Mol ´e culaire, CNRS, F-75006
Paris, France. E-mail: claire.fave@u-paris.fr, bernd.schollhorn@u-paris.fr
Institut Parisien de Chimie Mol ´e culaire, IPCM, CNRS – Sorbonne Universit ´e ,
4
place Jussieu, 75252 Paris, France
†
Electronic supplementary information (ESI) available. See DOI: 10.1039/
d0cp06612e
‡
Contributed equally.
4 4
Scheme 1 Stilbene and azo derivatives X -1 and X -2.
4344 | Phys. Chem. Chem. Phys., 2021, 23, 4344ꢀ4352
This journal is the Owner Societies 2021

View Article Online
Paper
PCCP
In-depth studies on the determination of the thermodynamic
binding constants towards anions have been performed using
multiple analytical techniques affording valuable quantitative
data on XB strength in solution, a domain in which systematic
9
data are still longed for.
Results and discussion
1.
Synthesis and characterization
Different synthetic strategies have been undertaken for the
preparation of the target compounds X -1 (X = H, Cl, Br, I)
Scheme S1, ESI†). In order access the target compounds we
followed two separate pathways, starting either from 4-picolin
1) or 3,5-dihalopyridines (4a–b). For the non-halogenated
species 3 (Scheme 2), 4-picolin (1) was transformed into
-picolinyl-trimethylsilane (2) by lithiation with LDA at low
4
(
(
4
Scheme 3 Synthesis of the 3,5-dihalogeno-isonicotinaldehyde 5a, 5b
and 5c.
temperatures and subsequent quenching with TMSCl. In a
following Peterson olefination, 2 was lithiated with LDA and
1
0
treated with isonicotinaldehyde. After several successive pre-
cipitations diaza-stilbene 3 was obtained as the pure E-isomer.
2
+
Alkylation with octyltriflate afforded H -1 , following a pub-
4
6
lished procedure. For the synthesis of the halogenated
compounds two distinct procedures were used, depending
on the nature of the halogen atom (Scheme 3). Formylation
of the 3,5-dichloro and 3,5 dibromo pyridines 4a–4b afforded
1
1
the respective aldehydes 5a–5b. The analogue 3,5-diiodo-
1
2
isonicotinaldehyde (5c) could not be obtained with this
procedure, probably due to the buttressing effect of the flanking
13
2+
2+
4 4
iodines. After methylation of 4b with CH I and halide exchange Scheme 4 Synthesis of the targeted molecules Cl -1 , Br -1 and
3
2
+
4
I -1 .
on the resulting 3,5 dibromo-4-picolin (6), the diiodinated deriva-
14
tive 7 was obtained. Refluxing of 7 in DMF–DMA afforded
15
enamine 8 which could be oxidized with sodium periodate to
the key intermediate 5c.
The electrochemical behaviour of the three extended viologens
2+
2+
2+
H
4
(
(
-1 , Br
CV) in DMF with tetrabutylammonium hexafluorophosphate
nBu PF ) as supporting electrolyte. Two different behaviors were
observed. The CV of H
reduction waves (Fig. 1, black trace) with formal standard potentials
4 4
-1 and I -1 was investigated by cyclic voltammetry
4
The aldehydes 5a–c were reduced with NaBH affording the
corresponding carbinols 9a–c (Scheme 4). Conversion to the
picolinyl bromides 10a–c was conducted with thionyl bromide
in methylene chloride. An Arbuzov reaction of picolinyl bro-
mides with triethylphosphite in a sealed vessel at 110 1C yielded
quantitatively the corresponding diethyl phosphonates 11a–c.
After lithiation with LDA and subsequent addition of the
respective aldehydes 5a–c, the corresponding azastilbenes
4
6
2+
4
-1 exhibits two reversible single-electron
(
vs. SCE) of ꢀ0.46 and ꢀ0.66 V corresponding to the sequential
12a–c were isolated after filtration and trituration. Alkylation
with n-octyltriflate afforded the vinylogous bipyridinium deri-
2
+
2+
2+
vatives Cl -1 , Br -1 and I -1 .
4
4
4
2
+
2+
4 4 4
Fig. 1 CVs of 0.1 mM of I -1 (red trace), Br -1 (green trace) and H -
2+
1
(black trace) in 0.1 M nBu
(3 mm-diameter) at 298 K. n = 0.1 V s
4
PF
6
/DMF on a glassy carbon electrode
2+
ꢀ1
Scheme 2 Synthesis of the non-halogenated derivative H
4
-1
.
.
This journal is the Owner Societies 2021
Phys. Chem. Chem. Phys., 2021, 23, 4344ꢀ4352 | 4345

View Article Online
PCCP
Paper
ꢁ
-1 ) and the neutral fully Table 1 Halogen bond geometry in the crystal structures of compounds
4
2+ 2+
+
formation of the radical cation (H
reduced species (H -1), respectively (Table S1, ESI†). In contrast,
CVs recorded for Br -1 and I -1 , present only one single
ꢀ
4 4
I -1 and [I -1 , Cl ]: inter atomic distances (d) and angles (+)
4
2+
2+
2+
ꢀ
a
a
b
a
4
4
[I4-1 ꢂ(TfO ) ]
d
Iꢂ ꢂ ꢂO
% vdW
+X–Iꢂ ꢂ ꢂO
d
Cꢂ ꢂ ꢂI
2
reversible wave (Fig. 1, green and red traces) corresponding to a
C2–I1ꢂ ꢂ ꢂO1
2.94
3.45
3.43
84.0
98.6
98.0
173.5
147.0
164.5
2.09
2.09
2.09
two-electron transfer (ET) process respectively at ꢀ0.20 and ꢀ0.27 V
C4–I2ꢂ ꢂ ꢂO1
vs. SCE (Table S1, ESI†) leading directly to the neutral species. C4–I2ꢂ ꢂ ꢂO2
Owing to the electron withdrawing effect of the halo substituents,
2+
ꢀ
ꢀ
a
b
a
[
I
4
-1 ꢂ(Cl )(TfO )]
d
Iꢂ ꢂ ꢂCl
% vdW
+X–Iꢂ ꢂ ꢂCl
d
Cꢂ ꢂ ꢂI
the formal standard potentials of the halogenated derivatives
are significantly less negative compared to the parent compound C3–I1ꢂ ꢂ ꢂCl1
H -1 . The different mechanism (two single-electron vs. one two-
4
3.08
3.31
3.13
3.24
82.6
88.7
83.9
86.9
174.7
159.6
167.7
168.3
2.10
2.10
2.10
2.11
2+
C6–I2ꢂ ꢂ ꢂCl1
C17–I3ꢂ ꢂ ꢂCl1
C20–I4ꢂ ꢂ ꢂCl1
electron transfer process) can be attributed to steric hindrance.
The presence of four bromo or iodo substituents probably
causes a strong conformational torsion leading to a disruption
-1 +ꢂ(Clꢀ)
2
]
d
a
% vdW
+X–Iꢂ ꢂ ꢂCl
b
d
Cꢂ ꢂ ꢂI
a
[
I
4
2
Iꢂ ꢂ ꢂCl
of the p-conjugated system of the molecules. In the case of C2–I1ꢂ ꢂ ꢂCl1
3.15
3.31
85.6
89.9
172.8
160.9
2.10
2.11
2
+
C4–I2ꢂ ꢂ ꢂCl1
I
4
-1 , an additional wave was observed at ꢀ1.70 V, corres-
a
b
ponding to the signature of a dissociative C–I bond reduction
(
In Å. In 1.
1
6
Fig. S6 red trace, ESI†). In all cases, a linear dependence of
the peak currents on the square root of the scan rate confirms
diffusional electrochemical reactions without any adsorption to 17% shorter than the sum of the theoretical vdW radii
1
7
phenomena of the studied compounds (Fig. S7–S9, ESI†).
(3.74 Å) (Fig. 2 and Table 1). This fact and the large angles
2
+
2+
ꢀ
Crystals of I -1 triflate salt [I -1 ꢂ(TfO ) ] were obtained by (+C–Iꢂ ꢂ ꢂCl = 159.61–174.71) clearly suggest directional halogen
4
4
2
ꢀ
slow evaporation of a solution in an ACN/DMSO/benzene bonding. The tetra coordination of Cl certainly results in an
2
+
ꢀ
ꢀ
solvent mixture, the chloride complex [I
4
-1 ꢂ(Cl )(TfO )] was overall strong XB complex thus displacing the probably weakly
2+
crystallized from a DMF solution, and the dichloride [I
4
-1 ꢂ bonded triflates.
ꢀ
2+
(
Cl )
2
] by the slow solvolysis of benzhydryl chloride in acetoni-
The asymmetric unit contains two different moieties of I -1 , a
4
trile and evaporation of the solvent over months, since the chloride anion, and a disordered triflate with short contacts to the
direct addition of chloride salts only lead to precipitation (Fig. 2 aromatic pyridine protons. Infinite chains alternating between XB
2
+
ꢀ
and Fig. S10–S12, ESI†).
donors I -1 and acceptors Cl are formed. The 1 : 1 stoichiometry
4
2+
ꢀ
The three compounds crystallized in the triclinic space for I
4
-1 and Cl confirms the key role of halogen bonding in this
-1 and its dichloride contain only one bipyridine self-assembled structure. Triflate is here the second anion counter-
molecule in the unit cell, displaying an inversion center on the balancing the overall charge of the dications. The pyridinium rings
2
+
group P1%. I
4
2
+
ethylene bridge (Fig. 2 and Table 1). The triflate anion in [I
4
-1 ꢂ within each molecule are almost coplanar whereas the linking
ꢀ
(
TfO )
2
] is bound by I1 (2.94 Å, 84% vdW) while I2 interacts CQC double bond is strongly tilted of about 501 to 601. In the
2+
ꢀ
2
rather loosely with O1 and O2. The distances are about 3.44 Å, structure of [I
4
-1 ꢂ(Cl )
] the overall neutral dichloride complex
ꢀ
which is only 1% shorter than the sum of the vdW radii. The C– appears to be assembled via weak CHꢂꢂꢂCl hydrogen bonds.
I2ꢂ ꢂ ꢂO angles are 1471 and 164.51, so that the interaction cannot Details of the data collection and refinement are given in the ESI†
be considered a XB but a mere packing effect. In contrast, the (Tables S2–S10).
The strongly tilted conformation of the dications (tilt angles
four non covalent Iꢂ ꢂ ꢂCl contacts (dIꢂ ꢂ ꢂCl = 3.09–3.31 Å) of 11% of about 501 to 601) in the solid state is in agreement with the
2
+
ꢀ
ꢀ
crystal structure of the complex [I
4
-1 ꢂ(Cl )(TfO )] displays
2
+
ꢀ
2+
ꢀ
ꢀ
-12+ꢂ(Clꢀ)
Fig. 2 Crystal structures of [I
4
-1 ꢂ(TfO )
2
] (A), [I
4
-1 ꢂ(Cl )(TfO )] (B), and [I
4
2
] (C). Ellipsoids are depicted at their 50% probability level. C –
black, N – blue, O – red, F – light green, S – yellow, Cl – dark green, I – purple.
4346 | Phys. Chem. Chem. Phys., 2021, 23, 4344ꢀ4352
This journal is the Owner Societies 2021

View Article Online
Paper
PCCP
observation of a two electron transfer process during the
electrochemical reduction of the dication. Even in homoge-
neous solution the steric hindrance of the relatively big halide
substituents and the rigid H–CQC–H unit most probably lead
to a conformation in which the heteroaromatic systems are not
0
+
0
+
electronically coupled via the p-conjugated system and this Fig. 4 Spin densities of I -1 (A) and H -1 (B).
4
4
even in the XB complex.
Additionally, the formation of the infinite alternating donor
acceptor chains explains the solubility issues observed in ACN
and DMSO. The relatively less polar ACN is incapable of
separating the charges and upon chloride addition the poly-
meric complex precipitates. DMSO on the other hand is appar-
ently able to form competitive XB to the viologens, thus
occupying the second binding site, until getting replaced by
the first chloride anion, after exceeding 1 eq.
0
n+
0
+
in the series H
4
-1 (n = 2 to 0), the structures of I
4
-1 (tilt angle
0
4
31) and I
4
-1 (tilt angle 171) become more planar. In the latter
case, this approach to planarity is accompanied by slight
distortions of the carbon–iodine bonds, likely due to steric
repulsion between the iodines (see ESI†). The spin densities of
0
+
0
+
the radical intermediates I -1 and H -1 are depicted in Fig. 4.
4
4
A coarse comparison of the two images indicates that the spin
is similarly distributed in both compounds despite the differ-
ences in tilt angles (431 vs. 01) mentioned above. The latter do
not seem to have a strong influence on delocalization.
2.
DFT calculations
In order to obtain further information on the structures of
0
2
+
2
+
2+
In the optimized structure of I -1 (Fig. 3 left), the electro-
4
4 4
I -1 and H -1 (as well as their onefold or twofold reduced
philic carbon–iodine axes on different pyridine moieties are
almost perpendicular to each other, and thus a bidentate
coordination of Lewis acids like halides seems improbable at
first glance. In order to investigate the binding behavior
analogues), DFT calculations (M06-2X D3 def2-TZVP, for further
details and coordinates see the ESI†) were performed on model
compounds in which all octyl groups were replaced by methyl
2+
0 0
2+
(I -1 and H -1 ). Our main goal was to compare the geo-
4 4
0
2+
further, the chloride, bromide and iodide complex of I -1
metries of these compounds and to check whether bidentate
halogen bonding with halides was geometrically feasible for
4
were optimized by the same means as mentioned above. All
complexes feature a clear bidentate coordination of the halide,
as is illustrated for chloride in Fig. 5, in which the two pyridine
moieties of the halogen bond donor become almost co-planar.
There is, however, a slight distortion of the halogen bond
donors, as the iodines which are coordinated to the halide are
closer to each other than the ones on the opposite side of the
molecule (3.58 Å vs. 4.06 Å for the chloride complex, Fig. 5). In
all complexes, two slightly different halogen bonds are
observed, with the stronger one always also being the slightly
more linear one (see ESI†). Even though the determination of
adduct energies was not our primary goal, a comparison of the
iodine–halide distances seems to reveal a trend: compared to
the respective sum of the van der Waals radii, the distance is
shortest for the chloride complex (75% and 79% of van der
Waals radii) and the longest for iodide (79% and 82%), with
bromide in-between (77% and 80%). Thus, these gas-phase
calculations seem to predict that chloride is most strongly
bound by the halogen bond donors.
0
2+
I -1 , and thus the interpretation will focus entirely on those
4
aspects (as we are well aware that the exact calculation of redox
potentials and halogen bonding energies can be tricky). To
further verify the performance of the applied theoretical model,
0
2+
we compared the bond lengths in I -1 with the ones of the
4
2
+
ꢀ
2
cation in the crystal structure of [I
4
-1 ꢂ(TfO )
], and deviations
were (often far) below 0.01 Å. The optimized gas phase struc-
0
2+
0 0
+
tures obtained for all six compounds (H
4
-1 , H
-1 ) clearly indicated that all bond
lengths of the core structures were very similar (with deviations
4 4
-1 , H -1 as
0 0
2+
+
0
well as I -1 , I -1 , I
4 4 4
0
2+
o0.1 Å) between two comparable species (e.g. I -1
H -1 , Fig. 3). However, there is a clear difference concerning
and
4
0
2+
4
the planarity when comparing the iodinated and the non-
iodinated compounds. Due to the size of the four iodine
substituents, all iodinated compounds deviate more or less
strongly from planarity, while all non-iodinated compounds are
0
2+
more or less planar (with only H
4
-1
featuring a slight tilt
between the two planes of pyridines, tilt angle approx. 291 as
measured by the I–Cꢂ ꢂ ꢂC–I dihedral angle; see ESI†). In the
0
2+
iodinated compounds, I -1
shows the strongest deviation
4
from planarity (tilt angle 721), and similar to the progression
0
2+
0
2+
Fig. 3 Optimized gas phase structures of I
4
-1
(A) and H
4
-1
(B), with
0
2+
octyl groups replaced by methyl. Bond lengths in Å. Graphics by
CYLview.
Fig. 5 Optimized gas phase structure of the complex of I
chloride (front and side view). Distances for I–Cl and I–I in Å.
4
-1
with
18
This journal is the Owner Societies 2021
Phys. Chem. Chem. Phys., 2021, 23, 4344ꢀ4352 | 4347

View Article Online
PCCP
Paper
2
+
Fig. 6 CVs of 0.1 mM I
4
-1 in 0.1 M nBu
4
PF
6
/DMF in the absence and in the presence of increasing concentrations (from 0 to 30 mM) of nBu
4
ꢀ
X: (A) X =
ꢀ
ꢀ
ꢀ
ꢀ1
2+
ꢀ
Cl , (B) X = Br and (C) X = I . T = 298 K, n = 0.1 V s . (D) Variation of the formal reduction potential of I
4
-1 as function of LB added (Cl : red, Br :
ꢀ
ꢀ ꢀ ꢀ
orange, I : violet, NO
3 4
: blue, HSO : green and OTf : black). The error bars indicated for chloride, bromide and iodide are standard deviations and each
point is the average of at least three experiments.
0
The dependence of the observed formal potential shift DE1
3
.
XB detection
.1 Cyclic voltammetry. The electrochemical behaviour
2
+
of I -1 on the anion concentration is represented in Fig. 6D.
4
3
2
+
2+
2+
The titration curves have been fitted according to eqn (1) based
on a conventional square scheme mechanism (Scheme 5)
assuming a 1 : 1 stoichiometry of the XB complexes according
to the observed solid state crystal structure.
of the three derivatives I
by CV in the presence of various anions. The DMF solutions
contained 0.1 M nBu PF and in the titration experiments, the
total salt concentration was kept constant never exceeding
4 4 4
-1 , Br -1 and H -1 was studied
4
6
0
.11 M. In presence of 100 equivalents of chloride (nBu Cl)
4
2
+
ꢃ 0
ꢃ 0
RT ð1 þ Kox½LBꢄÞ
0
the formal standard potential of the reduction wave of I -1
4
DE ¼ E2 ꢀ E1 ¼ ꢀnF ln
(1)
0
ð1 þ Kred½LBꢄÞ
was shifted toward more negative potentials (DE1 = ꢀ21 mV)
(Fig. 6A).
ꢃ 0
Smaller but still significant potential shifts were recorded With E1 being the formal standard potential of the redox
ꢀ
ꢃ 0
2
+
when adding bromide (ꢀ15 mV for 100 eq. Br ) or iodide
couple I
4
-1 /I
4
-1 and E2 the nominal potential depending on
the Lewis Base concentration [LB] and Kox and Kred the thermo-
ꢀ
(
ꢀ10 mV for 100 eq. I ) (Fig. 6B and C and Table 2). These
2
+
observations are consistent with a stabilization of the dicationic
dynamic affinity constants for the interaction with I
4 4
-1 and I -
2
+
I -1 species via charge-assisted non-covalent XB interactions
4
1, respectively.
2
+
ꢀ
leading to the complex [I -1 , Cl ]. Such a reductive XB
4
deactivation has been already demonstrated for iodo-phenyl-
viologen derivatives in one of our previous articles. Reduction
to the neutral state (I -1) proved to significantly decrease the
4
XB-donor strength.
0
4
,5
Table 2 Table of standard potential shifts (DE1 ) for the redox couples
2+
2+
2+
dication/cation radical of I
nBu
constants Kox and Kred
4
-1 , Br
and NO
3
4
-1 and H
4
-1 upon addition of of
) and the corresponding affinity
4
X (X = Cl, Br, I, OTf, HSO
4
The shift intensity regarding the chosen halide can thus be
directly correlated with the charge density of the XB-acceptors
2
+
0
a
0
b
c
c
red
X
4
-1
Anion
DE1
DE1
K
ox
K
Kox/Kred
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
2+
ꢀ
(
Cl 4 Br 4 I ). The less basic oxo-anions OTf , HSO or
I
4
-1
Cl
ꢀ8
ꢀ4
ꢀ2
0
0
0
ꢀ21
ꢀ15
ꢀ10
0
255
135
55
60
35
7
—
—
—
4.2
3.8
7.8
—
—
—
4
ꢀ
ꢀ
2+
Br
NO3 have been also tested with I -1 but no significant
ꢀ
4
I
perturbations of the CVs could be detected (Fig. S20, ESI†)
OTf
—
—
—
suggesting a strong selectivity for halide anions. Replacing the
0
0
2
+
2+
NO
4 4
iodo-substituents with bromine (Br -1 ) or hydrogen (H -1 )
2
+
ꢀ
ꢀ
resulted in receptors with no detectable halide affinity (Fig. S21
Br
4
-1
Cl
0
0
0
ꢀ4
ꢀ3
0
14
8
—
6
3
—
2.3
2.7
—
and S22, ESI†). These results indicate that under the experi-
Br
ꢀ
ꢀ
I
mental conditions [C–Brꢂ ꢂ ꢂCl ] XB interactions are too weak
because of the lower polarizability of the XB-donor atom,
2+
ꢀ
H
4
-1
Cl
0
ꢀ5
—
—
—
decreasing for the homologous elements I 4 Br 4 Cl. Appar-
a
In mV, upon addition of 20 equiv. of nBu
4
X, standard deviation:
X standard
ꢀ
ently HB [C–Hꢂ ꢂ ꢂCl ] is also not competitive with the other non-
b
ꢅ1 mV. In mV, upon addition of 100 equiv. of nBu
4
c
ꢀ1
covalent interactions in the polar liquid electrolyte.
deviation: ꢅ2 mV. In M
.
4348 | Phys. Chem. Chem. Phys., 2021, 23, 4344ꢀ4352
This journal is the Owner Societies 2021

View Article Online
Paper
PCCP
Scheme 5 Square scheme mechanism.
Fitting this square scheme model to the experimental titra-
tion plots allowed for the determination of the respective
affinity constants depending on the oxidation state of the XB-
donor. The highest affinity constant has been calculated for an
2
+
Fig. 7 (A) UV-vis spectra of 0.1 mM I
presence of increasing equivalents of Cl (from 0, black trace up to 300
Tables 2 and 3). As expected, the value of the calculated Kred equivalents, red trace). (B) Variation of the absorbance at 334 nm as
4 4 6
-1 , in 0.1 M nBu PF /DMF in the
2
+
ꢀ
ꢀ1
association between I -1 and Cl with a Kox of 255 M
ꢀ
4
(
is quite low, indicating only weak interactions between the function of added equivalents of chloride. The dash curve was obtained
by numerical simulation representing the best fit to the experimental data
neutral I
-1 compared to I
corresponding s-holes, with a VS,max of 57 kJ mol for I
4
-1 with chloride. The markedly weaker Lewis acidity of
21,22
2
+
based on a 1 : 1 stoichiometry using BindFit.
I
4
4
-1 is also reflected by the size of their
ꢀ1
4
-1 and
84 kJ mol for I -1 . When DFT computations including the an isobestic point. The association constant between I
ꢀ
1
2+
2+
5
4
-1 and
SMD18 solvation model for DMF were employed to estimate Cl in the same solution as used for CV experiments was
the binding constants, a similar trend was observed, with K
determined using BindFit, an online tool for supramolecular
4
1
9
ꢀ
=
ox
ꢀ
1
ꢀ5
ꢀ1
21,22
6
2 M and Kred = 1.6 ꢆ 10
M
(even though the difference chemistry research and analysis.
a
A binding constant K of
ꢀ
1
between the two complexes was much more pronounced in this 187 M was calculated based on a 1 : 1 stoichiometry (Table 3).
case, for currently unknown reasons).
The values of Kox for bromide (135 M ) or iodide (55 M
are consistent with the relative Lewis basicity. Low affinity
It is one order of magnitude smaller than the value obtained in
ꢀ
1
ꢀ1
ꢀ1
)
the absence of nBu
4
PF
6
(1466 M , Fig. S23, ESI†).
This observation highlights the significant competition of
constants could be determined for Br -1 taking into account electrolyte ions at relatively high concentration (0.1 mM) with
2
+
4
the small potential shifts. The addition of chloride did not the species of interest in the following electrochemical experi-
2
+
show any effect neither on the first reduction wave of H -1 nor ments used to quantify halogen bonding.
4
2
+
ꢀ
on the second.
.2 UV-vis spectroscopy. The results obtained with CV were nBu
compared with titration experiments employing UV-vis absorp- be K
The association constant of I
4
-1 with Br in the absence of
3
4
PF was estimated from the UV-vis data (Fig. S24A, ESI†) to
6
ꢀ
1
a
= 365 M which is considerably smaller than the value
-1 was recorded in pure obtained with Cl under the same conditions. The affinity
2
+
ꢀ
tion spectroscopy. The spectrum of I
4
ꢀ
non-deuterated DMF at 298 K (Fig. 7A, black trace). Then the constant for I could not be determined (Fig. S24B, ESI†),
2
+
electrolyte salt nBu PF (0.1 M) was added in order to maintain probably due to its very weak interaction with I
-1 .
4
6
4
comparable ionic strength to the electrochemical experiments.
Titrations with the weak Lewis base, triflate, did not result in
The maximum absorption peak centred at 334 nm was attrib- any change of the absorption spectrum of I -1 (Fig. S24D,
4
2
+
2
0
uted to p–p* transitions. As illustrated in Fig. 7A (red trace), ESI†).
upon addition of chloride (nBu Cl), a decrease of the molar
Interestingly, the spectrum of the reference compound, the
absorption coefficient for the band at 334 nm, accompanied non-halogenated derivative H
with a slight bathochromic shift of 5 nm and the appearance of the presence of the Lewis basic chloride at high concentrations
Fig. S25, ESI†), confirming the XB predominant interaction in
4
2
+
4
-1 , was not at all perturbed by
(
2
+
the case of I -1 .
4
ꢀ
1
3.3 NMR spectroscopy. Taking advantage of the diamag-
-1 determined by various netic character of the dicationic receptors, H NMR spectro-
Table
(
techniques
3
Association constants (Kox/M
ꢅ
6)/binding free energies
ꢀ1
2+
1
DG/kJ mol ꢅ 1) of various anions with I
4
scopy could also be used as a technique in order to determine
1
affinity constants and to confront the results with those of the
other techniques described above.
Solvent
Anion CV
UV-vis
H NMR
ꢀ
ꢀ
0
.1 M TBAPF
6
in DMF Cl
Br
255 (ꢀ16.2) —1 87 (ꢀ15.3) 3— 4 (ꢀ10.3)
135 (ꢀ14.4)
2+
2+
2+
4 4 4
1 mM solutions of I -1 , Br -1 and H -1 in DMF and for
ꢀ
comparison two previously described strong XB-donors, iodo-
I
55 (ꢀ11.7)
—
—
—
—
ꢀ
23
24
TfO
nd
lium triflate 13 and bisimidazolium 14 (Scheme 6) were
1
ꢀ
titrated with increasing amounts of chloride. H NMR spectra
of the chloride titration of I -1 are shown in Fig. 8 (see also
4
66 (ꢀ12.3) Fig. S26, ESI†). The signal of the pyridine protons of I
DMF
Cl
—
—
—
—
1466 (ꢀ21.4) 313 (ꢀ4.6)
365 (ꢀ17.3) 59 (ꢀ12.0)
2
+
ꢀ
Br
ꢀ
2+
I
nd
nd
-1 is
4
ꢀ
TfO
nd
progressively shifting from 8.890 to 9.005 ppm upon the addi-
nd: not determined.
tion of about 83 equivalents of chloride. The corresponding
This journal is the Owner Societies 2021
Phys. Chem. Chem. Phys., 2021, 23, 4344ꢀ4352 | 4349

View Article Online
PCCP
Paper
bidentate XB donor 14 in this solvent. As expected the affinity
2
+
ꢀ1
constants for the interaction of I
4
-1 with bromide (59 M
)
ꢀ
1
and iodide (66 M ) are significantly smaller as with chloride
ꢀ
1
(
313 M ). The impact of the ionic strength of the solution was
also controlled by comparing the recorded spectra during the
titration with or without the supporting electrolyte nBu PF
23
6
.
Scheme 6 Structures of the XB-donors iodolium 13 and bisimidazo-
4
24
lium 14.
As expected and in accordance with electrochemical and
UV-vis results, addition of supporting electrolyte induced
ꢀ1
a
a decrease of K by one order of magnitude (313 M and
ꢀ
1
3
4 M , respectively).
These results indicate the significant role of Coulombic
interactions.
3.4 Discussion of the determined affinity constants. The
three different techniques, CV, NMR and UV-vis spectroscopy,
used for the determination of the binding free energies of
2
+
4
X -1 towards halides are complementary. CV experiments
require the presence of ionic species in form of a supporting
electrolyte involving multiple competing electrostatic interac-
tions in solution. UV-vis and NMR titrations can be realized
in presence or in absence of such electrolyte salts, thus helping
to discriminate the impact of XB among the different non-
covalent interactions. Furthermore, electrochemistry allows
determining elegantly the affinity constants of the reduced
2
+
Fig. 8 NMR spectra of the chloride titrations of I
4
-1 under standard
suppression conditions. The red trace indicates the shift of the signals of
interest. The signals at about 9 ppm and 11.1 ppm are residuals from the XB-donor species, named Kred in the discussed square scheme
DMF. Those appear probably due to interactions with the XB-donors and model (Scheme 5). These values are very difficult to obtain via
are absent in pure DMF. The latter one starts shifting at higher ion
UV-vis and NMR spectroscopy, requiring quantitative chemical
reduction of the XB-donors prior to the titration experiments.
concentrations.
Table 3 summarizes affinity constants and binding free ener-
2
+
ꢀ
gies for the interaction between I -1 and the four anions Cl ,
4
association constants (1/1 stoichiometry) were determined by
ꢀ
ꢀ
ꢀ
2
2
Br , I and OTf in DMF. It is noteworthy that the polar solvent
DMF was chosen for solubility reasons, thus giving rise to
relatively low formal affinity constants due to solvation of the
interacting species. Dependency of the affinity constant on the
nature of the solvent has been already reported but its impact
is difficult to explain and to predict.
using the BindFit program.
With an association constant of 313 M (i.e. DG = 14.6 kJ
ꢀ
1
ꢀ
1
2+
mol ), I
herein studied bidentate derivatives (Fig. 9A). Br
one order of magnitude smaller Ka value of about 28 M
These results follow the classic trend of the increasing XB
4
-1 exhibits the strongest binding to chloride of the
2
+
4
-1 displays a
2
6
ꢀ
1
.
2
5
All methods indicate the strongest interaction being
donor strength in the order Br o I. For comparison, mono-
dentate iodine(III)-based XB-donor 13 features a binding con-
stant to chloride of 123 M under the same conditions. Due to
signal overlap, binding affinities could not be determined for
2
+
between I -1 and chloride (Table 3), in agreement with the
4
ꢀ
1
DFT calculations in the gas phase. The corresponding binding
2
+
ꢀ1
energy of Br
4
-1 (ꢀ8.5 kJ mol ) is considerable weaker, which
can be attributed to the weaker polarization of the Br substi-
tuents and is not experimentally measurable for H
2
+
4
-1 . When
comparing the data from the spectroscopic experiments (UV-vis
2
+
ꢀ
and NMR) with [I -1 , Cl ] it can be stated that the competing
4
electrostatic interactions of the supporting electrolyte salt
significantly diminish the formal binding constants of up to
one order of magnitude. In both systems, with and without
nBu
4
PF
6
, the order of binding strength was observed as follows:
ꢀ
ꢀ
ꢀ
ꢀ
Cl 4 Br Z I c TfO . These experimental results together
with the obtained XRD structures for the complex [I -1 , Cl ]
4
2
+
ꢀ
constitute further evidence of halogen bonding being the main
driving force for the observed assemblies. Addition of triflate in
the solution did not perturb the signals whatever the method,
confirming the weak Lewis basicity of this anion.
When comparing the three employed analytical techniques
the corresponding results obtained by CV and UV-vis certainly
reflect the global electronic effects whereas NMR locally probes
ꢀ
1
ꢀ1
Fig. 9 Binding free energies (kJ mol ) and affinity constant (M ) deter-
2+
2+
mined by NMR-titration at 298 K in DMF-h7 for (1 mM) I
4
-1 , Br
-1 towards chloride, bromide,
iodide and towards chloride in the presence of 0.1 M nBu PF (right). All
compounds were fitted presuming a 1 : 1 stoichiometry. Association con-
4 4
-1 , H -
2+
2+
1
, 13 towards chloride (left), and for I
4
4
6
24–27
stants were calculated using the BindFit program.
4350 | Phys. Chem. Chem. Phys., 2021, 23, 4344ꢀ4352
This journal is the Owner Societies 2021

View Article Online
Paper
PCCP
1
The catalytic reaction was monitored via H NMR in MeCN-
2
+
d3. A stoichiometric amount of I
4
-1 reaches 97% conversion,
while half of this amount leads to 86% (Fig. 10). In parallel,
2
+
using Br -1 as XB-donor catalyst transforms 64% of the
4
2
+
benzhydryl bromide and H -1 only 16%. The measured back-
4
Scheme 7 XB activated solvolysis of benzhydryl bromide to
ground over the same timespan conduces to 6% of transforma-
3
benzhydrylacetamide-d .
2
+
2+
tion. The catalyst efficiency of I
4 4
-1 and Br -1 is significantly
2
9
higher than for the previously reported bis-imidazolium 4.
3
0
The iodolium derivative 13, still displays the highest catalytic
activity which is certainly due to the higher oxidation state of
the iodine(III) center.
distinct sites. However, all methods show the same tendency
and confirm the highest affinity constant for the interaction
2
+
ꢀ
4
between I -1 and Cl . To the best of our knowledge, such a
This result is in agreement with its high affinity constant
systematic quantification of XB strength in solution using
several analytical techniques has not been reported yet in
literature.
6
ꢀ1
to bromide in acetonitrile (K
= 2.21 ꢆ 10
M
a 301C
ꢀ
1
30,31
(
ꢀ8.8 kcal mol )).
4
. Substrate activation/solvolysis
Conclusions
The determined binding constants are lower than expected for
a dicationic chelating XB-donor, but are still high enough to
consider potential catalytic activity in organic reactions. The
Bidentate halogen bond donors with halopyridinium moieties
linked by an ethylene bridge were synthesized and structurally
characterized by X-ray diffraction analysis and accompanying
DFT calculations. Electrochemical measurements indicated
clean twofold reductions to the corresponding neutral species,
which should be much weaker XB-donors. Binding constants of
the dicationic XB-donors to halides were obtained by three
techniques: NMR, UV/vis and CV, with the latter also being able
to estimate the binding strength of the reduced species – a clear
advantage of this electrochemical approach. The data indicate
that the presence of electrolyte salts in CV measurements leads
to competition with the halides and a decrease of the observed
binding constant by about an order of magnitude. Despite the
strengths and weaknesses of the different methods, all three
techniques revealed the same tendency of relative binding
strength to halides, with chloride being the strongest. Com-
pared to CV and UV-vis, NMR spectroscopy seems to signifi-
cantly underestimate binding strength by almost one order of
magnitude even when performed in the same experimental
conditions (ionic strength, solvent, concentration. . .). We sug-
gest that the results obtained with CV and UV-vis represent the
effects on the global electronic structure of molecules, in
contrast to NMR spectroscopy, which mainly probes local
interactions. The Lewis acidity of the novel bidentate XB-
donors was applied in a proof-of-principle case for halide
abstraction, in which the chelating donors performed similarly
compared to previously established bis(haloimidazolium) com-
pounds. Furthermore, these new catalysts are chemically stable
2
+
2+
2+
catalytic properties of I -1 , Br -1 and H -1 were first tested
4
4
4
in the Diels–Alder reaction between methylvinylketone and
2
8
cyclopentadiene. However, we do not observe any catalytic
activity. As the activation of neutral substrates like carbonyl
derivatives has proven to be more difficult than the activation
N
of S 1-type substrates in halide abstraction reactions, we then
turned our attention towards the latter. In this second
approach, we investigated the bromide abstraction reaction of
benzhydryl bromide in the presence of acetonitrile (Scheme 7).
As mentioned in the introduction, we had previously tested
0
2+
2+
the 4,4 -azobis(halopyridinium) derivatives I
4
-2 , Br
4
-2 and
2
+
H
4
-2 in this benchmark reaction and we have demonstrated
6
their ability to act as XB-donor catalysts. However, with these
derivatives a side reaction occurred in which the halogen bond
donors acted as oxidants towards bromide. One of the main
goals of the present study was to prepare iodopyridinium-based
bidentate activators which would be stable against halides.
6
and did not decompose as their azo-bridged analogues. In
contrast to our previously employed activators and catalysts,
the new tetrahalo bipyridinium stilbenes are redox active, so
that in principle it should be possible to control their catalytic
activity by reversible switching the redox state. Future work on
this interesting concept is in progress.
Fig. 10 Solvolysis of benzhydryl bromide with MeCN-d
3
and H
2
O acti-
2+
2+
2+
vated by the synthesized X
4
-1 viologenes (I
4
-1 in blue, Br
4
-1 in
2+
2+
orange, H
4
-1 in grey and with 0.5 eq. of I
4
-1 in blue/white) over Conflicts of interest
1
00 h at 22.5 1C. For comparison 13 is in red, 14 in green and the
background in black.
There are no conflicts to declare.
This journal is the Owner Societies 2021
Phys. Chem. Chem. Phys., 2021, 23, 4344ꢀ4352 | 4351

View Article Online
PCCP
Paper
Acknowledgements
12 A. Krasovskiy, V. Krasovskaya and P. Knochel, Angew. Chem.,
Int. Ed., 2006, 45, 2958.
The work was supported by the Laboratoire d’Electrochimie
Mol ´e culaire (UMR 7591 CNRS – Universit ´e de Paris) and was
also funded by the Deutsche Forschungsgemeinschaft (DFG,
German Research Foundation) under Germany’s Excellence
Strategy – EXC 2033 – 390677874 – RESOLV. Special thanks
are going to Dr Sihem Groni for experimental assistance. CF
thanks the DAAD for a visiting scholarship (91708797) and a
Procop grant (44694PE).
1
1
1
1
1
3 J. Gorecka, C. Heiss, R. Scopelliti and M. Schlosser, Org.
Lett., 2004, 6, 4591.
4 J. J. Baldwin, K. Mensler and G. S. Ponticello, J. Org. Chem.,
1
978, 43, 4878.
5 I. Sasaki, J. C. Daran and G. G. A. Balavoine, Synthesis, 1999,
15.
8
6 S. Arnaboldi, A. Gennaro, A. A. Isse and P. R. Mussini,
Electrochim. Acta, 2015, 158, 427.
7 (a) A. Bondi, J. Phys. Chem., 1964, 68, 441; (b) S. C. Nyburg
and C. H. Faerman, Acta Crystallogr., Sect. B: Struct. Sci.,
Notes and references
1985, 41, 274.
1
S. Groni, T. Maby-Raud, C. Fave, M. Branca and 18 C. Y. Legault, CYLview, 1.0b, Universit ´e de Sherbrooke, 2009,
B. Sch o¨ llhorn, Chem. Commun., 2014, 50, 14616.
http://www.cylview.org.
(a) R. Oliveira, S. Groni, C. Fave, M. Branca, F. Mavre, D. Lorcy, 19 E. Engelage, N. Schulz, F. Heinen, S. M. Huber,
2
M. Fourmigu ´e and B. Sch o¨ llhorn, Phys. Chem. Chem. Phys.,
016, 18, 15867; (b) R. Oliveira, S. Groni, A. Vacher, F. Barri `e re,
D. G. Truhlar and C. J. Cramer, Chem. – Eur. J., 2018,
24, 15983.
2
D. Lorcy, M. Fourmigu ´e , E. Maisonhaute, B. Sch o¨ llhorn and 20 J. Deng, X. Fu, G. Wang, L. Wu and J. Hiang, Electrochim.
C. Fave, ChemistrySelect, 2018, 3, 8874. Acta, 2012, 85, 195.
3
(a) J. Y. C. Lim, M. J. Cunningham, J. J. Davis and P. D. Beer, 21 P. Thordarson, Chem. Soc. Rev., 2011, 40, 1305.
Chem. Commun., 2015, 51, 14640; (b) J. Y. C. Lim and 22 http://app.supramolecular.org/BindFit/.
P. D. Beer, Eur. J. Inorg. Chem., 2017, 220.
G. Creste, S. Groni, C. Fave, M. Branca and B. Sch o¨ llhorn,
Faraday Discuss., 2017, 203, 301.
23 F. Heinen, E. Engelage, A. Dreger, R. Weiss and S. M. Huber,
Angew. Chem., Int. Ed., 2018, 57, 3830.
24 F. Kniep, L. Rout, S. M. Walter, H. K. V. Bensch,
S. H. Jungbauer, E. Herdtweck and S. M. Huber, Chem.
Commun., 2012, 48, 9299.
4
5
6
7
C. Fave and B. Sch o¨ llhorn, Curr. Opin. Electrochem., 2019,
15, 89.
F. Kniep, S. M. Walter, E. Herdtweck and S. M. Huber, Chem. 25 (a) P. Metrangolo, F. Meyer, T. Pilati, G. Resnati and
–
Eur. J., 2012, 18, 1306.
A. M. Allgeier and C. A. Mirkin, Angew. Chem., Int. Ed., 1998,
7, 894.
J. Choudhury, Tetrahedron Lett., 2018, 59, 487.
G. Terraneo, Angew. Chem., Int. Ed., 2008, 47, 6114;
(b) A. C. Legon, Phys. Chem. Chem. Phys., 2010, 12, 7736;
(c) A. Priimagi, G. Cavallo, P. Metrangolo and G. Resnati,
Acc. Chem. Res., 2013, 46, 2686.
3
8
9
(a) T. M. Beale, M. G. Chudzinski, M. G. Sarwar and 26 S. M. Walter, F. Kniep, L. Rout, F. P. Schmidtchen,
M. S. Taylor, Chem. Soc. Rev., 2013, 42, 1667; (b) C. Laurence,
J. Graton, M. Berthelot and M. J. El Ghomari, Chem. – Eur. J.,
E. Herdtwevk and S. M. Huber, J. Am. Chem. Soc., 2012,
134, 8507.
2011, 17, 10431; (c) A. Caballero, F. Zapata, N. G. White, 27 D. Brynn Hibbert and P. Thordarson, Chem. Commun., 2016,
P. J. Costa, V. F ´e lix and P. D. Beer, Angew. Chem., Int. Ed.,
52, 12792.
2
012, 51, 1876.
28 S. H. Jungbauer, S. M. Walter, S. Schindler, L. Rout, F. Kniep
and S. M. Huber, Chem. Commun., 2014, 50, 6281.
1
1
0 (a) E. Gleich and Z. Warnket, Phosphorus, Sulfur Silicon
Relat. Elem., 1991, 55, 9; (b) K. A. Chudov, K. S. 29 S. M. Walter, F. Kniep, E. Herdtweck and S. M. Huber,
Levchenko, N. O. Poroshin, A. V. Shchegolkov, P. S. Shmelin Angew. Chem., Int. Ed., 2011, 50, 7187.
and E. P. Grebennikov, Russ. Chem. Bull., Int. Ed., 2019, 30 F. Heinen, E. Engelage, A. Dreger, R. Weiss and S. M. Huber,
8, 1565. Angew. Chem., Int. Ed., 2018, 57, 3830.
1 The chlorinated analogs were synthetized for the sake of 31 E. Engelage, N. Schulz, F. Heinen, S. M. Huber,
6
completeness, but due to the low halogen bonding capabil-
ities of the bromo derivatives never investigated any further.
D. G. Truhlar and C. J. Cramer, Chem. – Eur. J., 2018,
24, 15983.
4352 | Phys. Chem. Chem. Phys., 2021, 23, 4344ꢀ4352
This journal is the Owner Societies 2021