Recognition of Anions 13ThePoet05

http://www.nd-warez.
info/
http://www.bookstores.com/books/-/-/9783540790914.html?kbid=1010
129
Structure and Bonding
Series Editor: D. M. P. Mingos
Editorial Board:
P. Day · X. Duan · L. H. Gade · T. J. Meyer
G. Parkin · J.-P. Sauvage
Series Editor: D. M. P. Mingos
Recently Published and Forthcoming Volumes
Contemporary Metal Boron Chemistry I Non-Covalent Multi-Porphyrin Assemblies

Volume Editors: Marder, T. B., Lin, Z. Synthesis and Properties
Vol. 130, 2008 Volume Editor: Alessio, E.
Vol. 121, 2006
Recognition of Anions
Volume Editor: Vilar, R. Recent Developments in Mercury Science
Vol. 129, 2008 Volume Editor: Atwood, David A.
Vol. 120, 2006
Liquid Crystalline Functional Assemblies
and Their Supramolecular Structures Layered Double Hydroxides
Volume Editor: Kato, T. Volume Editors: Duan, X., Evans, D. G.
Vol. 128, 2008 Vol. 119, 2005
Organometallic and Coordination Chemistry Semiconductor Nanocrystals

of the Actinides and Silicate Nanoparticles
Volume Editor: Albrecht-Schmitt, T. E. Volume Editors: Peng, X., Mingos, D. M. P.
Vol. 127, 2008 Vol. 118, 2005
Halogen Bonding Magnetic Functions Beyond

Fundamentals and Applications the Spin-Hamiltonian
Volume Editors: Metrangolo, P., Resnati, G. Volume Editor: Mingos, D. M. P.
Vol. 126, 2008 Vol. 117, 2005
High Energy Density Materials Intermolecular Forces and Clusters II

Volume Editor: Klapötke, T. H. Volume Editor: Wales, D. J.
Vol. 125, 2007 Vol. 116, 2005
Ferro- and Antiferroelectricity Intermolecular Forces and Clusters I

Volume Editors: Dalal, N. S., Volume Editor: Wales, D. J.
Bussmann-Holder, A. Vol. 115, 2005
Vol. 124, 2007
Superconductivity in Complex Systems
Photofunctional Transition Metal Complexes Volume Editors: Müller, K. A.,
Volume Editor: V. W. W. Yam Bussmann-Holder, A.
Vol. 123, 2007 Vol. 114, 2005
Single-Molecule Magnets Principles and Applications

and Related Phenomena of Density Functional Theory
Volume Editor: Winpenny, R. in Inorganic Chemistry II
Vol. 122, 2006 Volume Editors:
Kaltsoyannis, N., McGrady, J. E.
Vol. 113, 2004
Recognition of Anions
Volume Editor: Ramón Vilar
With contributions by
P. Ballester · G. W. Bates · S. R. Bayly · P. D. Beer · S. L. Ewen
P. A. Gale · I. Hamachi · J. H. G. Steinke · S. Tamaru · R. Vilar
123
The series Structure and Bonding publishes critical reviews on topics of research concerned with
chemical structure and bonding. The scope of the series spans the entire Periodic Table. It focuses
attention on new and developing areas of modern structural and theoretical chemistry such as na-
nostructures, molecular electronics, designed molecular solids, surfaces, metal clusters and supra-
molecular structures. Physical and spectroscopic techniques used to determine, examine and model
structures fall within the purview of Structure and Bonding to the extent that the focus is on the
scientific results obtained and not on specialist information concerning the techniques themselves.
Issues associated with the development of bonding models and generalizations that illuminate the
reactivity pathways and rates of chemical processes are also relevant.
As a rule, contributions are specially commissioned. The editors and publishers will, however, always
be pleased to receive suggestions and supplementary information. Papers are accepted for Structure
and Bonding in English.
In references Structure and Bonding is abbeviated Struct Bond and is cited as a journal.
Springer WWW home page: springer.com

Visit the Struct Bond content at springerlink.com
ISBN 978-3-540-79091-4 e-ISBN 978-3-540-79092-1

DOI 10.1007/978-3-540-79092-1
Structure and Bonding ISSN 0081-5993
Library of Congress Control Number: 2008923960

c 2008 Springer-Verlag Berlin Heidelberg
This work is subject to copyright. All rights are reserved, whether the whole or part of the material
is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broad-
casting, reproduction on microfilm or in any other way, and storage in data banks. Duplication of
this publication or parts thereof is permitted only under the provisions of the German Copyright Law
of September 9, 1965, in its current version, and permission for use must always be obtained from
Springer. Violations are liable to prosecution under the German Copyright Law.
The use of general descriptive names, registered names, trademarks, etc. in this publication does not
imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
Cover design: WMXDesign GmbH, Heidelberg
Typesetting and Production: le-tex publishing services oHG, Leipzig
Printed on acid-free paper
9876543210
springer.com
Series Editor
Prof. D. Michael P. Mingos
Principal
St. Edmund Hall
Oxford OX1 4AR, UK
michael.mingos@st-edmund-hall.oxford.ac.uk
Volume Editor
Dr. Ramón Vilar
Imperial College London
Department of Chemistry
South Kensington
London, SW7 2AZ, UK
r.vilar@imperial.ac.uk
Editorial Board
Prof. Peter Day Prof. Thomas J. Meyer
Director and Fullerian Professor Department of Chemistry
of Chemistry Campus Box 3290
The Royal Institution of Great Britain Venable and Kenan Laboratories
21 Albermarle Street The University of North Carolina
London W1X 4BS, UK and Chapel Hill
pday@ri.ac.uk Chapel Hill, NC 27599-3290, USA
tjmeyer@unc.edu
Prof. Xue Duan
Prof. Gerard Parkin
Director
State Key Laboratory Department of Chemistry (Box 3115)
of Chemical Resource Engineering Columbia University
Beijing University of Chemical Technology 3000 Broadway
15 Bei San Huan Dong Lu New York, New York 10027, USA
Beijing 100029, P.R. China parkin@columbia.edu
duanx@mail.buct.edu.cn
Prof. Jean-Pierre Sauvage
Prof. Lutz H. Gade
Faculté de Chimie
Anorganisch-Chemisches Institut Laboratoires de Chimie
Universität Heidelberg Organo-Minérale
Im Neuenheimer Feld 270 Université Louis Pasteur
69120 Heidelberg, Germany 4, rue Blaise Pascal
lutz.gade@uni-hd.de 67070 Strasbourg Cedex, France
sauvage@chimie.u-strasbg.fr
Also Available Electronically
For all customers who have a standing order to Structure and Bonding, we
offer the electronic version via SpringerLink free of charge. Please contact
your librarian who can receive a password or free access to the full articles by
registering at:
springerlink.com
If you do not have a subscription, you can still view the tables of contents of the
volumes and the abstract of each article by going to the SpringerLink Home-
page, clicking on “Browse by Online Libraries”, then “Chemical Sciences”, and
finally choose Structure and Bonding.
You will find information about the

– Editorial Board
– Aims and Scope
– Instructions for Authors
– Sample Contribution
at springer.com using the search function.
Color figures are published in full color within the electronic version on
SpringerLink.
Preface
A large number of biologically relevant species are negatively charged, there-

fore it is not surprising that nature has developed sophisticated receptors to
recognise, detect and transform anions. For example, complex receptors such
as phosphate- and sulphate-binding proteins are employed by living cells to
selectively recognise these two geometrically analogous anions. In addition
to their roles in biological systems, some anions also have important envi-
ronmental impacts. For example, cyanide, pertechnetate and chromates pose
serious health problems if present in water supplies.
Because of their important biological roles and potential environmental
impact there is great current interest in developing molecular receptors to
selectively recognise anions and in doing so be able to sequester, transform or
sense them. The six chapters presented in this volume provide an overview of
anion recognition and the most recent advances in this fast-growing area of
supramolecular chemistry are highlighted.
The first chapter by Bates and Gale provides an overview of the coordination
of anions by synthetic organic hosts. The different organic functional groups
used to bind anions are presented and this provides an introduction to the
structural and electronic properties that hosts must have to recognise anionic
guests. On the other hand, Bayly and Beer give a detailed account of the use of
metal complexes as anion receptors. Besides the important structural features
that metals can confer to receptors, their optical and redox properties make
them attractive for the development of anion sensors.
Metal-based receptors have found particularly interesting applications in
the recognition of phosphorylated species of biological interest (e.g. phospho-
rylated amino acids and peptides). This area is reviewed in depth by Tamaru
and Hamachi with particular emphasis on a series of receptors based on
zinc(II) centres which have been shown to bind phosphates with very high
binding constants in aqueous media. The applications of this type of receptor
for the detection of samples of biological interest are also presented.
Ballester provides an interesting account of anion · · · π interactions and
their impact in host design. Over the past few years there has been mounting
evidence that this type of interaction plays an important role in anion recog-
nition. The chapter starts with a detailed overview of the theoretical aspects
of anion · · · π interactions which is followed by a discussion of the existing
X Preface
experimental evidence for this type of interaction. Both, solution and crystal-
lographic studies are analysed showing the potential impact that this type of
interaction could have in the design of new anion receptors.
The use of anions as templating agents is discussed by Vilar. The chapter
starts with a general overview of the area and a discussion of the applications
of anion templates in organic and coordination chemistry. The second part
of the chapter deals with examples where anions are employed as templates
in dynamic combinatorial libraries. This approach promises to provide an
efficient route for the synthesis of better and more selective anion receptors.
The last chapter by Ewen and Steinke also deals with the use of anions as
templates but in this case in the context of molecular imprinted polymers. The
first half of the chapter provides an introduction into molecularly imprinted
polymers and this is followed by a detailed discussion of examples where
anionic species have been used to imprint this class of polymeric materials.
The topics discussed in this volume provide an exciting and stimulating
overview of the most recent studies within anion recognition and templation.
Although the supramolecular chemistry of anions took a long time to develop,
it is now a mature area that provides solutions to challenging problems. There is
no doubt that its growth will continue yielding more sophisticated and efficient
receptors for the recognition of a wide range of negatively charged species.
London, February 2008 Ramón Vilar

Contents
An Introduction to Anion Receptors Based on Organic Frameworks

G. W. Bates · P. A. Gale . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Metal-Based Anion Receptor Systems

S. R. Bayly · P. D. Beer . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
Recent Progress of Phosphate Derivatives Recognition

Utilizing Artificial Small Molecular Receptors in Aqueous Media
S. Tamaru · I. Hamachi . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
Anions and π-Aromatic Systems. Do They Interact Attractively?

P. Ballester . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
Anion Templates in Synthesis and Dynamic Combinatorial Libraries

R. Vilar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
Molecularly Imprinted Polymers Using Anions as Templates

S. L. Ewen · J. H. G. Steinke . . . . . . . . . . . . . . . . . . . . . . . . . 207
Subject Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249

Struct Bond (2008) 129: 1–44
DOI 10.1007/430_2007_069
© Springer-Verlag Berlin Heidelberg
Published online: 10 November 2007
An Introduction to Anion Receptors

Based on Organic Frameworks
Gareth W. Bates · Philip A. Gale (u)
School of Chemistry, University of Southampton, Southampton SO17 1BJ, UK
philip.gale@soton.ac.uk
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
2 Neutral Receptors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2.1 Acyclic Amide and Sulfonamide-Based Receptors . . . . . . . . . . . . . . 2
2.2 Macrocyclic Amide Receptors . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.3 Urea and Thiourea-Based Receptors . . . . . . . . . . . . . . . . . . . . . . 12
3 Aromatic NH Donor Containing Neutral Receptors . . . . . . . . . . . . . 21

3.1 Pyrrole-Based Receptors . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.2 Carbazole and Indole-Based Receptors . . . . . . . . . . . . . . . . . . . . 29
4 Hydroxy (OH) Donors in Neutral Receptors . . . . . . . . . . . . . . . . . 32
5 Charged Receptors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
5.1 Imidazolium and Pyridinium-Based Receptors . . . . . . . . . . . . . . . . 34
5.2 Guanidinium-Based Receptors . . . . . . . . . . . . . . . . . . . . . . . . . 37
5.3 Ammonium-Containing Receptors . . . . . . . . . . . . . . . . . . . . . . . 39
6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
Abstract This review article provides a broad overview to the area of anion coordination
by synthetic organic receptors and includes examples of different functional groups used
to bind anions. The first section examines neutral anion receptors containing amide-,
sulfonamide-, urea- and thiourea-based receptors. Then aromatics such as pyrrole, car-
bazole and indole are discussed before concluding the discussion of neutral systems with
examples of hydroxy OH donors. A brief overview of charged systems is also provided.
Keywords Anion recognition · Complexation · Crystal structures · Hydrogen bonding ·

Supramolecular chemistry
1
Introduction
The development of new anion receptors based on organic frameworks con-

tinues to attract considerable research effort [1–4]. A wide variety of systems
have been published in the last 15 years with both macrocyclic and acyclic
2 G.W. Bates · P.A. Gale
systems functioning as effective and selective anion receptors. This review

is not comprehensive but provides examples of important classes of anion
receptor systems based on organic frameworks.
2
Neutral Receptors
2.1
Acyclic Amide and Sulfonamide-Based Receptors
Secondary amides are versatile and highly accessible hydrogen bond donors
that have been used in numerous synthetic receptors. In the biological arena,
there are many examples of proteins that employ amide NH· · ·anion interac-
tions to bind negatively charged guests [5–9]. The first example of a synthetic
amide containing receptor, published in 1986 by Pascal and co-workers, was
a crytpand-like tris-amide that was shown to interact with fluoride in DMSO-
d6 [10].
In 1993, Reinhoudt and co-workers described the synthesis and bind-
ing properties of a series of tris-amides and tris-sulfonamides based upon
the tren skeleton [11]. These receptors proved to be selective for phosphate
in acetonitrile solution and demonstrated, arguably for the first time, that
anion receptor systems need not be difficult to make but rather that sim-
ple organic compounds could function as very effective receptors. Stability
constants were calculated by conductivity experiments and showed that re-
ceptor 1f bound dihydrogenphosphate with the highest affinity (14 200 M–1 )
in acetonitrile presumably due to the preorganization of the receptor via π–π
interactions between the naphthyl groups.
Four years later, in 1997, Crabtree and co-workers reported that simple
isophthalamide receptors e.g. 2 can bind anions in organic solution [12].
These receptors, even simpler than Reinhoudt’s tren-based anion binders,
were found to bind smaller halides selectively in dichloromethane solution.
The X-ray crystal structure of the bromide complex of 2a shows the receptor
adopting the syn–syn conformation with the bromide anion coordinated to the
amide NH’s with N· · ·Br distances of 3.44 and 3.64 Å (Fig. 1). The crystal struc-
An Introduction to Anion Receptors Based on Organic Frameworks 3
ture also reveals that the bromide anion is positioned above the plane of the
central aryl ring. Solution studies revealed that receptors 2a and 2b have high
affinity for halide anions and form complexes with exclusively 1 : 1 host/guest
stoichiometry. Stability constants for 2b were determined by 1 H NMR titration
studies and found to be 6.1 × 104 M–1 for chloride, 7.1 × 103 M–1 for bromide
and 4.6 × 102 M–1 for iodide in dichloromethane-d2 .
Fig. 1 X-ray crystal structure of a bromide complex of 2a
Almost contemporaneously, B.D. Smith and co-workers reported the use of

functionalized isophthalamide receptors for the coordination of anions [13].
Smith appended boronate groups to the peripheral aryl groups in order to
form interactions between the Lewis acidic boron and the carbonyl oxygens
of the amides therefore “pre-organizing” the receptor into the syn–syn con-
formation (preferable for anion coordination) and presumably increasing the
acidity of the NH group. Proton COSY and NOE difference experiments indi-
cated that the receptor did indeed adopt the desired syn–syn conformation in
DMSO-d6 . NMR titration experiments in DMSO-d6 at 295 K showed that re-
ceptor 3 bound acetate with a stability constant of 2.1 × 103 M–1 as compared
to 1.1 × 102 M–1 found with the non-preorganized receptor 2a and acetate.
Recently J.T. Davis, Gale, Quesada and co-workers have shown that ap-
pended hydroxy groups on the central aryl ring of an isophthalamide can
pre-organize the receptor into the syn–syn conformation and again presum-
ably increase the acidity of the amide NH groups, which increases the re-
ceptor’s ability to bind chloride (Fig. 2) [14]. The preorganization occurs due
to intramolecular hydrogen bonds between the hydroxy groups and carbonyl
oxygen of the amides.
Proton NMR titration experiments in CD3 CN at 298 K revealed that re-
ceptor 4 bound chloride most strongly with a stability constant of 5230 M–1 ,
whereas a stability constant of 195 M–1 was obtained for the unfunctionalized
cleft 5. Model compound 6 contains methoxy groups in the 4- and 6-positions
and functions as a control. In this system the intramolecular hydrogen bonds
form between the amide NH groups and the methoxy oxygens and conse-
quently the compound does not interact with chloride (Fig. 2). Most interest-
ingly, it was shown that compound 4 functions as a highly efficient chloride
transport agent across EYPC lipid bilayers whilst the analogous isophtha-
lamide 5, and model compound 6 show no transport ability. Compound 4
seems to be the simplest synthetic lipid bilayer transport agent for chloride
studied so far. The origin of this ability is currently being investigated.
Thordarson et al. have studied the aggregation of pyromellitamide 7 and

its response to anions. It was found that compound 7 aggregates in non-
polar solvent by the formation of one-dimensional intermolecular hydrogen-
bonding networks. Upon the introduction of anions the aggregation of 7 is
disturbed [15]. Proton NMR titration experiments in d6 -acetone at 300 K re-
Fig. 2 X-ray crystal structures of 4 (left) and 6 (right)

vealed that compound 7 bound a range of anions with 2 : 1 anion/receptor

stoichiometry. Although compound 7 has two discrete binding sites the an-
ions were found to bind with negative cooperativity with the strength of
anion binding to 7 decreasing in the order Cl– < CH3 CO2 – < Br– < NO3 – ≈ I– .
Prohens and co-workers have synthesized compounds 8a and 8b, simple

squaramido-based receptors and investigated their ability to coordinate car-
boxylate anion in competitive solvents [16]. The amide NH groups of the
squaramide form a more open cleft (similar to ureas) than the isophthala-
mides. Receptors 8a and 8b therefore adopt a more suitable geometry for
the coordination of bidentate anions, such as carboxylate anions, through
two approximately linear hydrogen bonds. Proton NMR titration experiments
revealed association constants of 217 M–1 and 1980 M–1 for the binding of
acetate by 8a and 8b respectively in DMSO-d6 at 295 K.
A.P. Davis and co-workers have designed a number of acyclic receptors

using the steroid cholic acid as a scaffold upon which they appended sulfon-
amide and carbamate amide groups [17]. The inflexibility of the fused ring
system and the axial conformation of the functional groups in the 7α and 12α
positions results in the formation of a convergent hydrogen-bonding array,
ideal for anion binding.
Both the structural rigidity and the shape of the receptor result in the for-
mation of a relatively small, well-defined binding site that shows selectivity
for halide anions with particularly high affinities observed for fluoride with
receptor 9a (15 400 M–1 association constants were determined by 1 H NMR
titration experiments in CDCl3 at 298 K). In the case of compound 9b the
association constant for fluoride was too high to be determined however, as-
sociation constants for chloride and bromide were found to be much higher
than for compound 9a [17].
2.2
Macrocyclic Amide Receptors
Macrocyclic receptors often possess a higher degree of selectivity than acyclic

systems. Hamilton and Choi have described the synthesis and anion bind-
ing properties of a family of cyclic triamides 10a and 10b [18]. These C3
symmetric receptors were found to be selective for oxo-anions such as tosy-
late with association constants of 2.6 × 105 M–1 and 2.1 × 105 M–1 obtained
for compounds 10a and 10b respectively in CDCl3 /2% dimethylsulfoxide at
296 K. Hamilton also studied the binding ability of an acyclic analogue (11)
and found significantly lower stability constants for the anion complexes
formed. For example, in the case of nitrate, a stability constant of 620 M–1
was calculated for compound 11 whereas a stability constant of 4.6 × 105 M–1
(K2 = 2.1 × 103 M–1 ) was found with compound 10a.
Interestingly, the NMR data led the authors to suggest that receptor 10b
forms a 2 : 1 host/anion “sandwich” complex at low concentrations of iodide
(as evidenced by an initial up-field shift of the NH resonances until ca. 0.5
equivalents of iodide) then switching to a 1 : 1 complex at higher concentra-
tions of iodide (down-field shift of NH resonances after ca. 0.5 equivalents of
iodide) (Fig. 3). Titrations with 10a and 10b in CDCl3 /2% DMSO-d6 mixture
displayed complex binding behavior, thus titration experiments were con-
ducted in 100% DMSO-d6 at 296 K to simplify the equilibria occurring in
solution. All data for receptor 10b were fitted to a 1 : 1 binding model and
again the macrocycle was found to be oxo-anion selective with the highest
stability constants being found with dihydrogen phosphate and hydrogen sul-
fate (1.5 × 104 M–1 and 1.7 × 103 M–1 , respectively).
Fig. 3 Changes in the amide 1 H NMR proton resonance of 10b with increasing [I– ]
concentrations. Reprinted in part with permission from [18]
Chmielewski and Jurczak have reported a series of extended tetra-amide

macrocycles containing two pyridine-2,6-dicarboxamide “caps” linked via
short aliphatic chains. The macrocycles possess a well-defined cavity with all
the amide groups directed inwards [19, 20].
Proton NMR titrations in DMSO-d6 at 298 K were conducted in order to
determine the stability constants of receptors 12a, 12b and 12c with a range
of anions, added as their tetrabutylammonium salts. The strongest associa-
tion constants were obtained for the 20-membered macrocycle 12b with the
most significant increase in affinity between the macrocycles being observed
in the binding of chloride. Enlargement from the 18-membered macrocycle
12a to the 20-membered macrocycle 12b results in a 30-fold increase in the

association constant for chloride (65 M–1 for 12a against 1930 M–1 ) whereas
further enlargement to the 24-membered macrocycle 12c results in the reduc-
tion of the association constant by two orders of magnitude (1930 M–1 against
18 M–1 ). This suggests that the 20-membered macrocycle 12b has good size
complementarity with the chloride anion. Although the 24-membered macro-
cycle 12c was designed with a large enough cavity to accommodate two oxy-
gen atoms from oxo-anionic guests, the stability constants obtained were the
lowest of the receptors tested. This is evidence that the additional flexibility
introduced into the macrocycle via the longer aliphatic chain has a detrimen-
tal effect on the anion binding ability of the receptor.
More recently the same authors have studied the anion binding ability of
similar macrocyclic systems based on isophthalamides [21]. The isophthala-
mide moieties were introduced as previous studies have shown isophthala-
mide derivatives bind anions more strongly than the analogous pyridine-2,6-
dicarboxamides [22].
Stability constants were obtained for receptors 13a, 13b and 13c using anal-
ogous conditions to those employed for receptors 12a, 12b and 12c. As with
the pyridine-2,6-dicarboxamide macrocycles, the stability constants for the
isophthalamide macrocycles appear to be influenced by the size and flexi-
bility of the system with the higher constants observed in the 20-membered
receptor 13a with notable decreases in the association constants with the 22-
and 24-membered receptors 13b and 13c. The greatest decreases where ob-
served in the stability constants obtained with the carboxylate anions. In the
case of acetate the constants decreased from 3130 M–1 for 13a to 552 M–1
and 205 M–1 for 13b and 13c, respectively. In the case of benzoate a constant
of 601 M–1 was calculated for 13a decreasing to 302 M–1 and 82 M–1 for 13b
and 13c, respectively. Unexpectedly lower binding constants were obtained
for the isophthalamide macrocycles (13a–13c) compared to the pyridine-
2,6-dicarboxamide macrocycles (12a–12c), a result rationalized in terms of

the competition between the formation of intramolecular hydrogen bonds
(arising from the preferred syn–anti conformation of the isophthalamide in
solution) and complexation of the anion (Scheme 1).
Scheme 1 Intramolecular hydrogen-bonding vs. anion binding in compound 13a
Bowman-James and co-workers have designed polyamide cryptand-type

systems based on triamines, such as tren (e.g. 14) and trpn (e.g. 15), and
shown that they bind anions [23]. The crystal structure of the hydrochloric
acid and fluoride complexes of 14 reveal that the anions are encapsulated
within the cavity of the amidocryptand and bound to the six-amide NH
groups. In contrast the hydrochloric acid structure of the expanded trpn-
based amidocryptand 15 shows the encapsulation of two chloride anions
within the cryptand, bridged by a water molecule. Each chloride is bound to
the water molecule as well as a protonated bridgehead amine and two hydro-
gen bonds from the amides groups.
Stability constants for 14 and 15 with different anions (added as their tetra-
butylammonium salts) were obtained by 1 H NMR titrations in DMSO-d6 . In
both cases, a slow equilibrium was observed in the titrations with fluoride
with stability constants >105 M–1 . The expansion of the cavity from recep-
tor 14 to 15 results in a significant change in the binding and selectivity for
anions. In the smaller receptor, 14, chloride is bound much more strongly
(3000 M–1 ) as compared to 15 (180 M–1 ) whereas the receptor 15 has a much
higher affinity for hydrogen sulfate with a stability constant of 2700 M–1 as
compared to 68 M–1 for 14. These findings may be due to the size comple-
mentarity between the receptors and guests with 14 being an ideal size to
encapsulate chloride and 15 being ideal for hydrogen sulfate, as illustrated by
the crystal structures (Fig. 4).
Fig. 4 X-ray crystal structures of the chloride complex of 14 and the sulfate complex of 15
The authors have also synthesized 16, a tricyclic cryptand-like receptor,

and have studied its ability to bind anions. Proton NMR titration experiments
in DMSO-d6 at 23 ◦ C revealed that compound 16 was selective for bifluoride
(FHF– ) with an association constant of 5500 M–1 being calculated. Dihydro-
genphosphate, azide and acetate were also found to bind to 16 with stability
constants of 740 M–1 , 340 M–1 and 100 M–1 , respectively [24].
Kubik and co-workers have developed a series of highly effective anion

receptors based upon cyclic peptides. Cyclic hexapeptide receptors such as
17 consist of alternately linked l-proline and 6-aminopicolinic acid sub-
units [25]. A 1 : 1 binding stoichiometry for 17 and the sodium salt of ben-
zenesulfonate was confirmed by a Job plot but in the case of the halide and
sulfate sodium salts 2 : 1 host/guest complexes were found. This was con-
firmed by electrospray mass spectrometry and in the case of iodide a crystal
structure of the 2 : 1 complex was obtained where the iodide was “sand-
wiched” between two cyclic hexapeptide receptors.
The formation of the 2 : 1 complexes observed in 17 led the authors to

design and synthesize compound 18 where two cyclic hexapeptides are co-
valently linked. The new receptor binds anions in a 1 : 1 stoichiometry in
methanol/water mixtures efficiently, with high affinity and selectivity for sul-
fate being observed. Both 1 H NMR titrations and ITC experiments were
conducted in 50% methanol/water at 298 K and stability constants (log Ka ) of
5.54 and 4.55 ± 0.23 were found by 1 H NMR and ITC respectively for 18 with
sulfate (added as Na2 SO4 ) [26].
2.3
Urea and Thiourea-Based Receptors
Ureas and thioureas possess two parallel NH hydrogen bond donor groups
and have been shown in a wide variety of receptors to function as highly ef-
ficient binding sites for “Y-shaped” anions such as carboxylates. Thioureas
are more acidic than analogous ureas and on this basis might be expected to
form stronger complexes with anions. However, other effects can often mask
or reverse this expected trend.
There have been a number of reports of anions triggering the deproto-
nation of neutral NH groups in anion receptor systems. This is often due,
in the case of fluoride, to the formation of the stable HF2 – anion driving
the deprotonation process [27–30]. Fabbrizzi and co-workers have shown
that this process can occur in urea systems containing electron-withdrawing
groups. The interactions between a number of anions and the simple 1,2-
bis(4-nitrophenyl) urea 19 were investigated. Oxo-anions were found to bind
to the receptor with a 1 : 1 host/guest stoichiometry with the strength of the
interaction depending on the partial negative charge located on each oxygen
atom of the anion [31].
Stability constants were determined for compound 19 by UV-Vis spec-

trophotometric titrations in acetonitrile at 25 ◦ C and revealed that the
association constants increased with the increasing basicity of the anion
(CH3 COO– > C6 H5 COO– > H2 PO4 – > NO2 – > HSO4 – > NO3 – ). Addition of
fluoride appears to stabilize a strong 1 : 1 complex at low anion concentration,
however at higher anion concentrations deprotonation of the urea subunit oc-
curs resulting in the formation of HF2 – (confirmed by 1 H NMR). This process
was also characterized by the formation of a new band at 475 nm in the UV-
Vis spectrum upon the additions of the fluoride anion and was clearly present
after the addition of two equivalents of fluoride [32].
Gunnlaugsson and co-workers have studied several receptors containing
a thiourea group attached to an anthracene moiety [33]. These compounds
were designed to behave as fluorescent PET (photo-induced electron trans-
fer) sensors for the detection of anionic species. Proton NMR titration ex-
periments, conducted in DMSO-d6 , confirmed that the anions bind to the
receptors through the two NH protons of the thiourea group and form a 1 : 1
complex. The authors demonstrated that 20a–d act as ideal PET sensors (only
fluorescent quantum yield affected upon additions of anions) with quench-
ing of the fluorescence being observed with the addition of fluoride, acetate
and dihydrogen phosphate anions. Chloride and bromide did not induce any
changes in the fluorescence spectra.
Yoon and co-workers have reported a series of mono- and bis-func-

tionalized anthracenes and described their colorimetric and fluorescent
properties for the sensing of both fluoride and pyrophosphate anions. The
authors appended either phenylurea or p-nitrophenylurea groups through the
1-position (for the mono-functionalized derivatives 21a and 21b) and the
1- and 8-position (for the bis-functionalized derivative 22a and 22b) of the
anthracene [34].
Fluorescent titration experiments with receptors 21b and 22b were carried
out in DMSO with a variety of anions, added as the tetrabutylammonium
salts, in order to compare the stability constants of the mono- and bis-
functionalized receptors. The strongest anion binding was observed with the
bis-functionalized receptor (22b) with stability constants of 108 000, 9700
and 6000 M–1 calculated for fluoride, bromide and pyrophosphate, respec-
tively. For compound 21b a much lower binding constant of 4000 M–1 was
found with fluoride as compared to the strong anion complexation observed
with 22b, illustrating that there is a cooperative binding effect in operation
with the two urea groups in receptors 22a and 22b. Temperature-dependant
1 H NMR experiments in DMF-d also revealed that the anion complex stabil-
7
ity was enhanced by the formation of a hydrogen bond between the hydrogen
atom in the 9-position and both the fluoride and pyrophosphate guests in
receptors 22a and 22b.
Gale and co-workers have designed 23a, a bis-urea cleft based on o-
phenylenediamine, to selectively bind carboxylate anions [35]. The geometry
of the receptors provides a convergent cleft appropriate for the binding of
carboxylate anion through four hydrogen bonds. Stability constants of 3210,
1330 and 732 M–1 were calculated for acetate, benzoate and dihydrogen phos-
phate, respectively, after analysis of data from 1 H NMR titration experiments
conducted in DMSO-d6 /0.5% water at 298 K. The crystal structure of the ben-
zoate complex of compound 23a is shown in Fig. 5 revealing that the receptor
binds this carboxylate via four hydrogen bonds in the solid state. The sta-
bility constants for compound 23a were found to be greater than constants
Fig. 5 X-ray crystal structure of 23a with benzoate. Chem Commun, p 4696 (2005) repro-
duced by permission of The Royal Society of Chemistry
obtained with N,N -diphenylurea with a 2.5-fold increase observed for 23a
with acetate compared to the diphenylurea (3210 M–1 for 23a and 1261 M–1
for diphenylurea [36]).
The same authors appended electron-withdrawing groups onto both the
central aryl ring and peripheral aryl rings of the receptor in order to increase
the acidity of the urea NH groups. Titration studies were conducted under the
same conditions as for 23a and enhanced stability constants were observed
for both receptors 23b and 23c compared to 23a [37]. In the case of acetate the
stability constant increased from 3210 M–1 for 23a, to 4020 M–1 and 8080 M–1
for 23b and 23c, respectively. In the case of dihydrogenphosphate there is
a decrease in affinity from 732 M–1 for 23a, to 666 M–1 for 23b but a large
increase to 4720 M–1 for 23c. The authors proposed that the dihydrogenphos-
phate anion interacted most strongly with the central NH groups thus with
the increased acidity of these NH groups in 23c, due to the presence of the two
chloro-groups, stronger complexation is observed.
Naphthalene and binaphthalene appended with thiourea groups (24 and
25, respectively) have been synthesized by Kondo and co-workers in order
to investigate potential cooperative binding between two thiourea groups in
25 [38]. This group found that 1 : 1 complexes were formed between 25 and
fluoride, acetate and dihydrogen phosphate anion, which was confirmed by
Job plots in acetonitrile and ESI-MS.
UV-Vis spectroscopy titration experiments in acetonitrile solution were

carried out to ascertain the anion binding properties of the receptors with
acetate, dihydrogen phosphate, fluoride and chloride. The binding constants
revealed that the presence of the second thiourea group in 25 significantly im-
proves the receptor’s affinity for anions with respect to the mono-thiourea 24.
The most significant differences were obtained for titration with fluoride and
acetate where binding constants of 1.1 × 105 M–1 and 2.1 × 106 M–1 were elu-
cidated for 25 and 3.7 × 103 and 7.7 × 103 for compound 24 with acetate and
fluoride, respectively.
Pfeffer and co-workers have described the use of a highly rigid[3]poly-
norbornane as a scaffold on which to append electron-deficient thiourea
subunits [39].
The anion binding abilities of 26a and 26b were evaluated by 1 H NMR
titration techniques in DMSO-d6 with CH3 COO– , F– , H2 PO4 – and H2 P2 O7 2–
(added as their tetrabutylammonium salts). Additions of fluoride to the re-
ceptor resulted in a distinctive color change attributed to deprotonation. This
process was characterized by the loss of the thiourea NH proton resonances
and the appearance of the HF2 – resonance in the 1 H NMR during the titra-
tion. Analysis of the binding isotherms of receptors 26a and 26b with acetate
revealed that the anions were strongly bound by both 26a and 26b in a 1 : 2
receptor-to-anion complex with each of the thiourea units binding to a sin-
gle acetate anion. Binding constants (log β1 and log β2 values) of 3.5 (±0.1)
and 2.4 (±0.1) were calculated for 26a and 3.5 (±0.1) and 3.0 (±0.1) for 26b
with acetate. Titrations with H2 PO4 – were fitted to a 1 : 1 binding model and
constants of 3.9 (±0.1) and 3.6 (±0.1) (log β values) were calculated for re-
ceptors 26a and 26b, respectively. Pyrophosphate was then investigated to
evaluate the binding ability of 26a and 26b with a dianion. Analysis of the
titration curves for both 26a and 26b with pyrophosphate revealed the forma-
tion of a 2 : 1 receptor-to-anion stoichiometry in which each anion terminus
is accommodated by two urea groups of a single receptor.
Extending their work on “cholapods”, A.P. Davis and co-workers have ap-
pended urea and thiourea groups from the 7 and 12 positions of the steroid
scaffold and evaluated the ability of these receptors to bind chloride and bro-
mide (added as their tetraethylammonium salts) [40].
NMR data was found to be consistent with the formation of predomin-
antly 1 : 1 complexes of the receptors and anions. Stability constants were
determined by Cram’s extraction method in water-saturated chloroform at
30 ◦ C [40]. Affinities for both chloride and bromide anions increased through
the series 27a–d, reflecting the increase in acidity of the NH groups due
to the electron-withdrawing aryl substituents and the change from urea to
thiourea in 27d. In the case of chloride the association constant for the
“unsubstituted” derivative 27a was calculated to be 1.62 × 107 M–1 , with the
nitrophenyl substituted 27c the constant increased to 4.77 × 108 M–1 and the
constant increased further to 1.05 × 109 M–1 with the introduction of the
thiourea (27d). The addition of the nitrosulfonamide group in 28a–28c also
enhances the anion-binding affinities with the largest constants being ob-
served in the thiourea derivative 28c with constants of 1.03 × 1011 M–1 and
2.59 × 1010 M–1 calculated for chloride and bromide, respectively.
A further detailed study of these “cholapod” anion receptors was con-
ducted where the anion binding ability of several receptors with increas-
ing numbers of hydrogen bond donor groups was investigated [41]. It was
found that a combination of increasing numbers of hydrogen bonding groups
and increasing acidity of the NH groups via electron-withdrawing sub-
stituents had a significant effect on the anion stability constants. Receptor
29 was found to have the highest affinities for all the anions investigated
except acetate where the previously studied 28b and 28c had higher affini-
ties (2.6 × 1011 M–1 and 2.0 × 1011 M–1 , respectively) when compared to 29
(1.3 × 1011 M–1 ). These steroid-based receptors have also been studied as
transport agents for anions across vesicle and cell membranes. Electrochem-
ical, NMR and fluorescence techniques were employed and revealed that the
“cholapod” receptors act as mobile carriers and facilitate the transport of
chloride ions across vesicle membranes [42].
Reinhoudt and co-workers have synthesized both acyclic and cyclic recep-
tors containing multiple urea-binding sites (e.g. 30 and 31). Anion-binding
studies were conducted with these systems and a variety of putative anionic
guests (added as their tetrabutylammonium salts) using 1 H NMR titration ex-
periments in DMSO-d6 [43]. In the case of the cleft-like receptors dihydrogen
phosphate caused the largest shift in the NH group resonances of all the re-
ceptors however an association constant could not be obtained for 30a due
to the complexity of the binding processes in solution. Job plot analysis of
receptor 30b showed the formation of an exclusive 1 : 2 host/guest complex
with dihydrogen phosphate and an association constant of 5 × 107 M–2 was
calculated. The thiourea functionalized 30c cleft was also shown to bind di-
hydrogen phosphate with a 1 : 2 host/guest stoichiometry and chloride with
1 : 1 host/guest stoichiometry.
Macrocyclic receptors 31a and 31b were found to bind both dihydrogen
phosphate and chloride in exclusively 1 : 1 host/guest stoichiometries. Bind-
ing constants were calculated for 31a and 31b with dihydrogen phosphate and
chloride and revealed that dihydrogenphosphate was bound more strongly
(2.5 × 103 M–1 for 31a and 4.0 × 103 M–1 for 31b) than chloride (500 M–1 for
31a and <50 M–1 for 31b).
Gale and co-workers have combined urea and amide groups into a new
macrocyclic motif and studied the anion complexation properties of 32 [44].
Stability constants with a variety of anionic guests were elucidated by
1 H NMR titration techniques in both DMSO-d /0.5% water and DMSO-
6
d6 /5% water at 298 K. The macrocyclic receptor shows significant selectivity
for carboxylate anions over dihydrogen phosphate and chloride. Titrations
in DMSO-d6 /0.5% water resulted in high stability constants for acetate

(>104 M–1 ) and benzoate (6430 M–1 ) therefore the titrations were conducted
in 5% water, a much more competitive media, and stability constants of
5170 M–1 and 1830 M–1 were calculated for acetate and benzoate, respectively.
Interestingly, a crystal structure of a carbonate complex was obtained from
a crystallization with tetrabutylammonium fluoride (Fig. 6). It was presumed
that carbonate was gained via the fixation of atmospheric CO2 by the fluoride
salt-macrocycle solution.
Fig. 6 X-ray crystal structure of a carbonate complex of 32 reproduced by permission of

The Royal Society of Chemistry [43]
In 2000 Lee and Hong synthesized tris-thiourea macrocycles 33a and

33b and studied their anion recognition properties by 1 H NMR titration
experiments in DMSO-d6 at 25 ◦ C. It was found that macrocycle 33a was
selective for dihydrogen phosphate (800 M–1 ) over acetate (320 M–1 ) and
chloride (40 M–1 ). In contrast macrocycle 33b was found to be selective
for acetate (5300 M–1 ) over dihydrogen phosphate (1600 M–1 ) and chloride
(95 M–1 ) [45].
Recently, Tobe and co-workers have designed cryptand-like macrocycles
based on homobenzylic tripodal thiourea and compared their anion-binding
properties to a series of acyclic tripod-type receptors [46].
The proton resonances in the 1 H NMR spectra of cryptand-type recep-
tor 34b in various solvents were found to be very broad, possibly due to
conformational changes that are slow on the NMR timescale. Therefore, the
complexation of 34b with anionic species was evaluated by 1 H NMR titra-
tion experiments in CDCl2 CDCl2 at 373 K. Association constants of 116 M–1
and 112 M–1 were calculated for acetate and chloride, respectively, and were
found to be much lower than the tripodal receptor 35a under the same condi-
tion (3030 M–1 for acetate and 3700 M–1 for chloride). This low binding ability
of 34b was attributed to strong intramolecular hydrogen bonds between the
thiourea groups. Receptors 35a and 35c were then compared and the stability
constants (obtained from 1 H NMR titrations in DMSO-d6 at 303 K) revealed
that 35a has poor affinity towards all anionic species in DMSO solutions
whereas 35c has high affinity for dihydrogen phosphate and acetate.
3
Aromatic NH Donor Containing Neutral Receptors
3.1
Pyrrole-Based Receptors
Sessler and co-workers have pioneered the use of the pyrrole NH hydrogen
bond donor group in both charged and neutral anion receptor systems [47].
In 1992 they reported the anion-binding abilities and fluoride selectivity
of sapphyrin 36a, a pentapyrrolic macrocycle [48]. Fluorescence titration
experiments carried out in methanol revealed that 1 : 1 complexes formed
between the diprotonated sapphyrin 36a and halide anions and association
constants of 2.8 × 105 , ∼102 and <102 M–1 were calculated with fluoride, chlo-
ride and bromide, respectively. Four-years later Sessler reported the effective
binding of phosphate by receptor 36b. Phosphorus NMR titration experi-
ments were carried out in methanol-d4 at ambient temperature and revealed
that compound 36b bound both phosphoric acid and phenylphosphonic acid
with affinity (1.8 × 104 and 1.3 × 104 M–1 for H3 PO4 and C6 H7 PO3 , respec-
tively) [49].
Gale and co-workers have developed a number of receptors based on

the 2,5-dicarboxamidopyrrole skeleton where the combination of a pyrrole
and amide groups form convergent hydrogen-bonding arrays (e.g. 37a and
37b) [50].
Differences in the solubility of receptors 37a and 37b meant that their abil-
ity to bind anions was assessed in different solvents (DMSO-d6 /0.5% water
for 37a and CD3 CN for 37b at 298 K). Both receptors proved to be selec-
tive for oxo-anions, however 37a bound dihydrogen phosphate most strongly
(1450 M–1 ) whereas 37b bound benzoate most strongly (2500 M–1 ) (Fig. 7).
Continuing this work, Gale and co-workers have synthesized more acidic
diamidopyrrole derivatives 37c and 37d by including electron-withdrawing
Fig. 7 X-ray crystal structure of 37b benzoate complex
nitro groups on the peripheral phenyl rings and assessed their anion-binding
ability compared to the unfunctionalized 37a [30]. Proton NMR titrations
were carried out in DMSO-d6 /0.5% water solutions at 298 K and revealed that
the presence of the electron-withdrawing groups in receptors 37c and 37d
improved the systems affinity for anionic guests. In the case of benzoate the
association constants increased significantly from 560 M–1 (obtained for 37a
under identical conditions) to 4150 M–1 for 37c and 4200 M–1 for 37d. Upon
the addition of one equivalent fluoride to 37d the anion appears to coordinate
to the receptor. However, upon further additions of fluoride deprotonation
occurs (as indicated by the evolution of a blue color in solution due to the de-
protonated pyrrole). In the case of compound 37a fluoride was found to bind
to the receptor with a stability constant of 1245 M–1 .
Gale, Smith and co-workers have developed prodigiosin mimics based on

amidopyrroles in order to co-transport hydrochloride acid across a vesicle
membrane. The inclusion of the protonatable imidazole group allows the
receptor to carry the proton of the acid and in addition, once protonated, pro-
vides an extra hydrogen bond donor group to bind the chloride within the
cleft [51].
Sessler, Gale and co-workers have further developed receptors based on the
2,5-diamidopyrrole skeleton by appending 2-aminopyrrole groups (e.g. com-
pound 39) thus increasing the number of hydrogen-bonding groups, which
was hoped would increase the selectivity for oxo-anions compared to the pre-
viously studied 37a and 37b [52].
Proton NMR titration experiments were conducted in DMSO-d6 /0.5% wa-
ter at 298 K to investigate the solution phase anion complexation properties
of 39 compared to 37a. The stability constants revealed that 39 showed en-
hanced selectivity for both dihydrogen phosphate and benzoate compared to

37a however 39 was found to bind dihydrogen phosphate with higher affin-
ity (10 300 M–1 ) than benzoate (5500 M–1 ), the reverse selectivity observed for
37a.
Recently, Sessler and co-workers have appended the 2-aminopyrroles sub-
unit onto a pyridine-2,6-dicarboxamide spacer group to afford two acyclic
receptors (compounds 40a and 40b) [53].
Elucidation of the solution phase anion complexation properties of recep-

tors 40a and 40b (determined by UV-Vis spectrophotometric titrations in
dichloromethane at 298 K) revealed that 40b bound only acetate with a stabil-
ity constant of 13 900 M–1 whereas strong binding was observed for benzoate,
acetate, NO2 – and CN– with receptor 40a (43 000, 19 000, 13 000 and 5600 M–1 ,
respectively).
Gale and co-workers have also investigated 5,5 -dicarboxamido-dipyrrolyl-
methanes as anion receptors [54]. These systems demonstrated a remarkable
affinity and selectivity for dihydrogen phosphate in highly competitive sol-
vent media with stability constants of 234 M–1 and 20 M–1 being calculated
(by 1 H NMR titration experiments at 298 K) for 41a and 41b respectively in
DMSO-d6 /25% water. Receptors 41a and 41b were found to be unstable in
solution therefore analogous compounds containing two methyl groups at-
tached to the meso-carbon were synthesized (42a and 42b) in the hope they
would display increased stability in solution [55].
The meso-substituted derivatives showed lower affinities for anion com-

pared to 41a and 41b however 42a did display selectivity for dihydrogen phos-
phate over other anions with constants of 1092 M–1 for dihydrogen phosphate,
124 M–1 for fluoride and 41 M–1 for benzoate calculated in DMSOd6 /5% water
at 298 K.
Sessler, Ustynyuk and co-workers have incorporated dipyrromethane
subunits into 2,6-diamidopyridinedipyrromethane hybrid macrocyclic sys-
tems [56]. Initially 43 was synthesized and the anion-binding ability was
elucidated by UV-Vis spectroscopic titrations in CH3 CN at 23 ◦ C. Weak bind-
ing was observed for chloride, bromide, cyanide and nitrate which was
rationalized by the receptor forming a deep cavity that favors the formation
of well-oriented, directional NH-anion hydrogen bonds. Hydrogensulfate was
found to bind strongly to the receptor in a 1 : 1 host/anion fashion in ace-
tonitrile (Ka = 64 000 ± 2600 M–1 ) presumably due to the orientation of the
hydrogen-bonding groups within the macrocycle being ideal for tetrahedral
anions.
The same authors went onto “fine tune” the anion-binding properties of
the pyridine-2,6-dicarboxamide-dipyrromethane-hybrid macrocyclic system
and designed receptor 44 after DFT calculations suggested it to be more suit-
able for hydrogensulfate complexation [57]. The receptor’s affinity for a num-
ber of anions was determined by UV-Vis spectroscopic titrations in CH3 CN at
23 ◦ C and revealed that bromide, nitrate and chloride are not bound by com-
pound 44. Acetate and dihydrogen phosphate bound with a 1 : 1 stoichiometry
with constants of 12 600 ± 450 M–1 and 29 000 ± 1900 M–1 , respectively. An
enhanced affinity and selectivity was observed for 44 with hydrogen sulfate
(108 000 ± 17 000 M–1 ) compared to 43.
Recently Katayev, Sessler and co-workers have further developed the hy-
brid macrocycle systems by replacing the pyridine-2,6-dicarboxamide moi-
eties used in receptors 43 and 44 with bipyrrole and dipyrromethane subunits
(compounds 45 and 46) [58]. UV-Vis spectroscopic titrations in acetoni-

trile at 23 ◦ C showed that the smaller macrocycle 45 bound hydrogensulfate
with high affinity (2.7 × 106 M–1 ) similar to the previously studied receptors
43 and 44 whereas the larger bis-dipyrromethane macrocycle 46 exhibited
a different selectivity for anions with chloride being selectively bound with
high affinity (281 000 M–1 ). Interestingly, titrations with hydrogensulfate re-
sulted in the lowest association constant of all the anions investigated with 46
(2100 M–1 ).
In 1996, Sessler and co-workers reported the anion complexation proper-

ties of calix[4]pyrroles e.g. 47. Compound 47 (meso-octamethylcalix[4]pyrr-
ole) was originally synthesized by Baeyer in 1886 and is arguably the simplest
anion receptor to synthesize as it is formed in one step via the acid-catalyzed
condensation of pyrrole and acetone [59]. The macrocycle forms four hy-
drogen bonds to anionic guests and binds fluoride and chloride strongly.
Recently, it was discovered that this receptor actually functions as an ion-
pair receptor with large charge diffuse cations sitting in the cup-shaped cavity
formed by the pyrrole rings when complexed to an anion [60–62]. A wide var-
iety of functionalized calixpyrroles have been synthesized and are reviewed
extensively elsewhere [63, 64].
One approach to dramatically increase the affinity of calixpyrroles for

anions is to introduce a “strap” across the macrocycle containing addi-
tional hydrogen bond donor groups. For example, Lee, Sessler and co-
workers have recently reported the synthesis of isophthalamide strapped
calix[4]pyrroles [65]. The authors studied the effect of varying the length of
the strap on the anion complexation properties of the macrocycle and isother-
mal titration calorimetry was employed to investigate the anion-binding
abilities of receptors 48a–c (in CH3 CN at 30 ◦ C). Receptors 48a–c were found
to have high binding affinities toward halide anions however they failed to
show an appreciable size-dependence selectivity based on the increase of
strap length. This is illustrated in the case of chloride (the most strongly
bound anion) where association constants of 3.89 × 106 M–1 , 3.35 × 106 M–1
and 3.24 × 106 M–1 were calculated for 48a, 48b and 48c, respectively.
Recently, Lee and co-workers have shown that a binol-strapped calix[4]pyr-

role (49) can be used in the enantioselective recognition of carboxylate
anions [66]. Both the R- and S-enantiomers of the strapped calixpyrrole
were isolated and characterized. Detailed studies of the enantioselectivity
of the S enantiomer were carried out by isothermal titration calorime-
try experiments in acetonitrile with the chiral anions (R)-2-phenylbutyrate
and (S)-2-phenylbutyrate. Stability constants were determined and revealed
that receptor (S)-49 shows selectivity for (S)-2-phenylbutyrate over (R)-

2-phenylbutyrate with an order of magnitude difference in the constants
(Ka (R) = 9.8 × 103 M–1 and Ka (S) = 1.0 × 105 M–1 ).
In 2000 Kohnke and co-workers described the synthesis of meso-octame-
thylcalix[6]pyrrole via the conversion of calix[6]furan into dodecaketone,
which was then treated with ammonium acetate to obtain calix[6]pyrrole
50 [67].
Association constants were determined by Cram extraction methods,

which revealed that the macrocycle 50 formed a strong complex with chlo-
ride (12 800 ± 1300 M–1 ). Proton NMR titration experiments conducted in
CD2 Cl2 also revealed the size dependence selectivity of the calix[n]pyrroles
where bromide was found to bind approximately seven times stronger to
the larger calix[6]pyrrole compared to calix[4]pyrrole (710 ± 25 M–1 for 50
against 10 M–1 for 47) [68].
Recently Cafeo, Kohnke and co-workers have continued work on ex-
panded calixpyrroles and have reported the anion-binding properties of two
calix[2]benzo[4]pyrroles 51a and 51b [69].
Elucidation of the stability constants (determined by 1 H NMR titrations in
CD2 Cl2 at 20 ◦ C) showed that although chemically similar the two macrocy-
cles have significantly different anion-binding properties. Receptor 51b only
bound fluoride and acetate to an appreciable level (2246 ± 132 M–1 for flu-
oride and 597 ± 236 M–1 for acetate) whereas receptor 51a bound a number
of anions with higher affinities. 51a was found to have selectivity for fluo-
ride with a constant of approximately 20 000 M–1 estimated from competitive
binding studies in the presence of 51b. This selectivity is presumably due to
good size complementarity between 51b and the fluoride anion.
3.2
Carbazole and Indole-Based Receptors
Jurczak and co-workers have described 1,8-diamino-3,6-dichlorocarbazole as

a building block for anion receptor construction. Stability constants for re-
ceptors 52a and 52b with various anions, added as tetrabutylammonium salts,
were calculated by 1 H NMR titration experiments in DMSO-d6 /0.5% wa-
ter and it was found that compound 52b bound anions more strongly with
constants of 115 M–1 and 8340 M–1 for chloride and benzoate, respectively,
compared to 13 M–1 and 1230 M–1 with compound 52a with chloride and
acetate, respectively [70].
Sessler and co-workers have incorporated two carbazole subunits into

expanded calixpyrrole-type macrocycle 53 [71]. Fluorescence titration ex-
periments in dichloromethane at 0.5 µM concentration of host revealed that
compound 53 shows selectivity for acetate (Ka = 229 000 M–1 ) over a number
of other carboxylate-type anions (benzoate, oxalate and succinate).
Beer and co-workers have reported the anion-binding ability of a num-

ber of indolocarbazoles sensors (54a–c) [72]. The highest association con-
stants were obtained with receptor 54c presumably due to the presence of
the electron-withdrawing bromide groups increasing the acidity of the NH
groups. UV-Vis spectroscopic titrations in acetone at 25 ◦ C revealed that
compound 54c bound benzoate most strongly (log Ka = 5.9) followed by phos-
phate (log Ka = 5.3) then fluoride (log Ka = 5.0) and chloride (log Ka = 4.9),
a trend observed in both 54a and 54b.
Recently, indole-based receptors have attracted increasing attention.
Sessler and co-workers have described the use of indole subunits for an-
ion recognition within diindolylquinoxalines receptors (55a and 55b) [73].
Stability constants were determined by UV-Vis spectroscopic titrations in
dichloromethane at 22 ◦ C and showed that both receptors have apprecia-
ble selectivity for dihydrogenphosphate with constants of 6800 M–1 and
20 000 M–1 calculated for compounds 55a and 55b, respectively. Receptor 55b
was found to be highly colored and upon the addition of dihydrogenphos-
phate a visible change in color was observed.
A number of biindoyl-based systems have been developed by Jeong and

co-workers for the binding of anionic species. This group have prepared mo-
lecular clefts based on the 2,2 -biindoyl scaffold with amide groups attached
via alkyne linkers and compared the anion-binding ability of 56 to the more
rigid ethyno-bridged receptor 57 [74].
Stability constants were determined by UV-Vis spectroscopy titration ex-
periments in CH3 CN at 22 ◦ C and revealed that receptor 57 did indeed bind
the anionic species with higher affinities than 56, the less rigid receptor.
Chloride was bound 22-times more strongly by receptor 57 compared to
56 (1.1 × 105 M–1 for 57 against 5.1 × 103 M–1 for 56) whereas bromide was
bound 41-times more strongly by 57 compared to 56 (8.7 × 103 M–1 for 57
against 2.1 × 102 M–1 for 56) illustrating the importance of preorganization in
the binding of small anionic guests.
Jeong and co-workers have extended the research into biindolyl scaffolds
as a building block for anion receptors by synthesizing new macrocyclic sys-
tems 58 and 59 [75]. Both compounds 58 and 59 were found to strongly
bind various anions (stability constants determined by UV-Vis spectroscopy
titrations in CH3 CN at 295 K). In the case of the halide anions a size comple-
mentarity was observed for both macrocycles where the smaller fluoride and
chloride anions were found to bind more strongly to 58 and 59 than the larger
bromide and iodide anions.
Recently, Gale and co-workers have synthesized simple clefts with indole
subunits appended to isophthalamide and pyridine-2,6-dicarboxamide spac-
ers and described their anion-binding properties [76].
Proton NMR titrations in DMSO-d6 /0.5 water at 298 K were conducted to
elucidate stability constants for a number of anions, added as their terabuty-
lammonium salts. Fluoride affects the largest change on the proton reson-
ance however an association constant was only calculated for 60a (>104 M–1 )
therefore titrations were repeated in more competitive media (DMSO-d6 /5%
water). Both receptors showed selectivity for fluoride however 60a bound
with a 1 : 1 binding stoichiometry (1360 M–1 ) whereas the data for 60b could
only be fitted to a 1 : 2 receptor/anion model (K1 = 940 M–1 and K2 = 21 M–1 ).
The crystal structures of 60a with chloride and fluoride are shown in Figs. 8
and 9 respectively.
Fig. 8 X-ray crystal structure of a chloride complex of 60a reproduced by permission of

Fig. 9 X-ray crystal structure of a fluoride complex of 60a reproduced by permission of

4
Hydroxy (OH) Donors in Neutral Receptors
In 2003, D.K. Smith showed that simple aromatic hydroxides can complex
chloride anions. Smith compared stability constants (obtained from NMR
competition experiments in CD3 CN) of phenol, 61, resorcinol, 62 and cat-
echol, 63, with chloride (added as its tetrabutylammonium salt) and found
that 63 bound chloride with greater affinity than 61 and 62 (1015 M–1 for 63
against 125 M–1 for 61 and 145 for 62) [77].
Recently, Smith and Winstanley have further explored aromatic hydroxides

as anion receptors by studying the effect of ortho-substituents on the chloride
binding affinity of catechols 64a, 64b and 65 [78]. Binding constants were elu-
cidated by proton NMR titrations in CD3 CN : DMSO-d6 (9 : 1) solutions and
showed that 65 bound chloride with the highest affinity (235 M–1 ) presum-
ably due to the additional hydrogen bonding provided by the amide groups.
Although amide groups are present in 64a and 64b it appeared that they were
not involved in binding the anion and as a result compounds 64a and 64b
bound chloride with lower affinities (110 M–1 for 64a and 115 M–1 for 64b).
Row, Maitra and co-workers have linked two steroid subunits to synthe-
size macrocycle 66. The fluoride-binding properties of compound 66 were
then investigated by a 1 H NMR titration experiment in CDCl3 at 22 ◦ C, which
found that the receptor bound fluoride with a 1 : 2 receptor/anion stoichi-
ometry (a result confirmed by Job plot analysis) and stability constants of
K1 = 1.8 (±0.1) ×103 M–1 and K2 = 2.5 (±0.35) ×102 M–1 were found [79].
5
Charged Receptors
The incorporation of charged groups into receptors designed for anion recog-
nition allows for the receptors to bind the anion with both electrostatic in-
teraction and additional interactions dependent on the group and receptor
design. In the case of imidazolium groups, the cation can stabilize the anion
complex with additional CH· · ·A– type hydrogen bonds.
5.1
Imidazolium and Pyridinium-Based Receptors
Kang and Kim have synthesized the fluorescent anion receptor compound
67 where two methylene bridged bis-imidazolium subunits are attached to
a naphthalene backbone through the 1- and 8-position [80].
Molecular modelling showed that the receptor forms a convergent con-

cave cavity with all the imidazolium C(2)-H’s pointing inwards. Modelling
studies led the authors to suggest that the shape of the cavity was predis-
posed for the binding of halide anions. Fluorescence titration experiments in
90 : 10 CH3 CN : DMSO solutions were carried out with chloride, bromide and
iodide anions (added as their tetrabutylammonium salts) and stability con-
stants were calculated that showed that compound 54 had highest affinity for
I– (5000 ± 470 M–1 ) followed by Br– (243 ± 15 M–1 ) then Cl– (185 ± 13 M–1 ).
Yoon, Kim and co-workers have reported a highly effective fluorescent sen-
sor for dihydrogen phosphate based on a 1,8-disubstituted-anthracene-dimer
macrocycle bridged by two imidazolium subunits (68) [81].
Fluorescence titration experiments were conducted in 9 : 1 acetoni-

trile : DMSO solutions in order to elucidate association constants for 68 with
dihydrogenphosphate, fluoride, chloride and bromide. The results confirmed
that compound 68 selectively binds to dihydrogenphosphate over the other

anions tested with a stability constant >1 300 000 M–1 . Fluoride also bound
to compound 68 with high affinity (340 000 M–1 ) and competition studies
of dihydrogenphosphate and fluoride with respect to compound 68 clearly
showed that no interference to the dihydrogenphosphate binding occurred in
the presence of fluoride.
Beer and co-workers have shown how a number of tetrakis(imidazolium)
macrocyclic receptors, 69a–d, can be used for anion binding [82]. Proton
NMR titration investigations revealed that the macrocycles bind halide anions
strongly with fluoride being most strongly bound by 69b and 69c (>104 M–1
for both 69b and 69c). Good size complementarity is seen for iodide with 69d
as it gave the highest stability constant (900 M–1 ) compared to the other re-
ceptors (370 M, 560 and 470 M–1 for 69a, 69b and 69c, respectively). Benzoate
anions were found to bind to the receptor in a 1 : 2 host/anion stoichiometry,
a result rationalized by the relative size of the benzoate anion compared with
the spherical halides, thus the anion is only partially bound within the cavity
allowing a second anion to interact with the cavity.
Alcalde and co-workers have reported that imidazolium-based hetero-

phanes, such as 70, are capable of anion recognition [83].
Proton NMR spectroscopy was employed in order to examine the anion-
binding behavior of receptor 70. Upon the addition of a number of anionic
guests (added as their tetrabutylammonium salts) to compound 70, signifi-

cant changes in the C2 proton resonance of the imidazolium ring were ob-
served in both CD3 CN and DMSO-d6 solutions. Proton NMR titration experi-
ments in DMSO-d6 were then carried out and revealed that 70 binds acetate
most strongly and with a 1 : 1 binding stoichiometry (Ka = 359 ± 42 M–1 ).
Steed and co-workers have also utilized a tripodal backbone to construct
a number of tri-pyridinium “venus flytrap” receptors (71a–c) and investi-
gated their anion binding and sensing properties [84, 85].
Receptors 71a and 71b showed similar anion-binding behavior with both
receptors binding chloride most strongly (constants of >100 000 M–1 calcu-
lated for both receptors). In the case of compound 71b reduced affinities
were observed for both bromide and acetate (3953 M–1 and 2511 M–1 , re-
spectively) compared to 71a (13 800 M–1 and 10 500 M–1 , respectively) which
was attributed to the increased steric bulk provided by the benzyl groups.
For compound 71c chloride is bound stronger than bromide (similar to 71a
and 71b) however the affinities for halides are greatly reduced compared to
71a and 71b (5370 M–1 for chloride and 486 M–1 for bromide). Receptor 71c
was highly selective for acetate with the stability constant being almost an
order of magnitude higher than the chloride constant (49 000 M–1 against
5370 M–1 ). Variable-temperature 1 H NMR experiments were carried out and
showed that compound 71c selectively binds acetate over the spherical halide
anions due to mixtures of conformers being adopted in solution through
anthracene-anthracene mutual interactions. Further evidence of the confor-
mational behavior of compound 71c and its selectivity for acetate over other
anions was provided by UV spectroscopy and fluorescence studies.
Shinoda and co-workers have reported the one-step synthesis and anion
binding properties of macrocycle 72. Proton NMR titration experiments (in
D2 O) were used to determine the binding properties of 72 for tricarboxylate
anions and revealed that the tricarboxylate 72b was bound with the highest
affinity (log Ka = 5.1) [86].
5.2
Guanidinium-Based Receptors
Guanidinium groups may be regarded as charged analogues of ureas in

that they have two parallel NH groups and as a consequence often show
high affinities for oxyanionic species such as carboxylates binding these an-
ions by a combination of hydrogen bonding and electrostatic interactions.
Guanidinium-carboxylate and phosphate interactions occur in many biolog-
ical systems as a guanidinium group is present in the amino acid arginine [4]
Schmidtchen and co-workers have described the binding of benzoate to
the guanidinium-based receptors 73a and 73b [87]. ITC titrations were car-
ried with the iodide salts of 73a and 73b in acetonitrile at 30 ◦ C with benzoate
(added as its tetraethylammonium salt) and binding constants of 280 000 M–1
and 203 000 M–1 were calculated for 73a and 73b, respectively. Receptor 73a
was investigated further where ITC titration experiments (under identical
conditions to the previous titration) were carried with 73a and a variety of
counter anions. It was found that the change in counter anion had signifi-
cant effects upon the binding constant of benzoate observed, for example with
the tetrafluoroborate anion a binding constant of 414 000 M–1 was calculated
compared to a binding of 38 000 M–1 with chloride.
For several years Schmuck has investigated the binding affinities of guani-
dinium salts appended to hydrogen bonding pyrrole-containing motifs and
has shown how carboxylate anion binding is enhanced by the hybrid re-
ceptors. In 1999 Schmuck reported the binding ability of 74a with various
carboxylate anions in highly competitive media [88]. Proton NMR titration
in DMSO-d6 /40% H2 O at 25 ◦ C revealed that 74a formed stronger complexes
with acetate and Ac-L-Phe anions with binding constants of 2790 M–1 and
1700 M–1 , respectively.
Schmuck then investigated a series of guanidinium-appended pyrrole re-
ceptors and found that 74b bound Ac-l-Ala-O– more strongly than 74a
(1610 M–1 and 770 M–1 , respectively) [89]. The binding ability of 74b was
then assessed with a range of carboxylate anions by 1 H NMR titrations in
DMSO-d6 /40% H2 O at 25 ◦ C and showed that compound 74b formed stronger
complexes with the anions than 74a. High affinities were observed for 2-
pyrrole-COO– and acetate (5275 M–1 and 3380 M–1 ). A notable result was
that compound 74b displayed enantioselectivity in the case of Ac-Ala-O– an-
ions where a higher affinity was observed for the l-enantiomer over the
d-enantiomer (1610 M–1 vs. 930 M–1 ).
In 2005 Schmuck and Schwegmann reported the study of a tripodal “mo-

lecular flytrap” 75 where the pyrrole-guanidinium moieties were appended
to a triamide backbone. The receptor was designed to bind tricarboxylate an-
ions and UV and fluorescence titration experiments in water showed that 75
bound citrate and trimesoate with association constants >105 M–1 [90].
Recently, de Mendoza and co-workers have reported the complexation
of nitrate to an acyclic cleft and a series of macrocycles based on guani-
dinium [91].
Association constants were calculated by ITC titrations in acetonitrile at
303 K and it was found that the macrocyclic receptors bound the nitrate an-
ion more strongly than the acyclic system. The constants obtained for the
macrocycles 77a–c revealed a size dependence on the binding of nitrate with
the largest macrocycle 77c giving rise to the highest association constant of
73.7 × 103 M–1 , an order of magnitude greater than the smallest macrocycle
77a (7.26 × 103 M–1 ).
5.3
Ammonium-Containing Receptors
The work of Park and Simmons [92] has inspired many efforts into the re-
search of ammonium- and polyammonium-based anion receptors and are
the subject of numerous reviews [93, 94] Here we will only look at recep-
tors containing quaternary ammonium centers. These groups bind anions via
electrostatic interactions only.
An early pioneer in the area of ammonium-based anion receptors,
Schmidtchen synthesized the quaternary ammonium-based macrocyclic re-
ceptors 78a–c in 1977 [95].
NMR spectroscopy revealed that upon the addition of 1 equiv. of alkali

metal halide salts to receptors 78a–c, 1 : 1 complexes were formed. Stability
constants were measured using halide electrodes in water solutions, which re-
vealed that the receptors bound bromide anions (log Ka = 1.8, 2.45 and 2.45
for 78a, 78b and 78c, respectively) and iodide anions (log Ka = 2.2 and 2.4 for
78b and 78c, respectively).
Frontera, Anslyn and co-workers have described the binding of tricarboxy-

late salts with the tris-ammonium-squaramide-appended tripodal receptor
79 [96]. Isothermal titration calorimetry was used to study the association of
a number of tricarboxylate salts and 79 in 1 : 3 water:ethanol solutions at 294 K.
It was found that 79 bound the less rigid tricarboxylates citrate and tricaballate
more strongly than the rigid benzene-1,3,5-tricarboxylate (1.1 ± 0.1 × 105 M–1 ,
1.5 ± 0.2 × 105 M–1 and 4.5 ± 0.5 × 104 M–1 , respectively). Compound 79 was
also investigated as a receptor for the biscarboxylates gluterate and succinate.
The stability constants revealed that gluterate was bound with significantly
higher affinity than succinate (2.2 ± 0.2 × 104 vs. ∼ 2.8 × 102 , respectively).
Costa and co-workers have reported 80, a fluorescent squaramide-
containing macrocyclic receptor for monitoring sulfate in water [97]. Isother-
mal titration calorimetry was employed to characterize the host–guest
association of 80 with SO4 2– , PhOPO3 2– and C2 O4 2– dianions (titration
carried out in methanol at 294 K). The data was fitted to a 1 : 1 binding
model and it was found that 80 bound SO4 2– (4.6 ± 1.0 × 106 M–1 ) with the
strongest affinity followed by C2 O4 2– (3.2 ± 0.3 × 105 M–1 ) then PhOPO3 2–

(1.5 ± 0.2 × 104 M–1 ). A fluorescein-80 complex was then synthesized and
competitive fluorescent titration experiments in 9:1 methanol:water solutions
were conducted with the complex against sodium sulfate and an association
constant of 5.2 ± 1.2 × 106 M–1 was calculated.
Bowman-James and co-workers have synthesized a series of amide-based

macrocycles containing either tertiary amine spacer groups or quaternized
ammonium functionalities and have assessed their ability to bind a number
of anions in solution [98].
Stability constants were calculated by proton NMR titration experiments in
DMSO-d6 solutions and revealed that the quaternized macrocycles 82a and
82b showed higher affinities for anions compared to the neutral analogues
81a and 81b attributed to the additional electrostatic attraction of the qua-
ternary ammonium group. The pyridine analogues 81b and 82b were also
shown to have higher binding constants than the isophthaloyl derivatives 81a
and 82a attributed to the pyridine-assisted preorganization of the macrocy-
cle. All the anions binding data was fitted to 1 : 1 isotherms and receptor 82b
was found to have the highest affinity for anions with dihydrogenphosphate
being most strongly bound (log K = 5.32). Receptor 82b was found to bind
halide anions in the order Cl– > Br– > I– > F– whilst receptor 82a was also
found to bind chloride more strongly than other halide anions (log K = 3.23
and 2.14 for 82a with Cl– and Br– , respectively), illustrating the good size
complementarity of 82a and 82b to chloride.
6
Conclusions
The examples discussed in this review provide a broad overview of the variety
of synthetic organic receptors used for binding and sensing of anionic species.
As the understanding of the processes and factors that influence the effect-
ive binding of anions improves there is an increasing impetus to apply this
knowledge to solve real-world problems. Areas that are likely to benefit from
this knowledge are in the transport of anions, in separation processes and in
biological systems leading towards new treatments for disease and cancer.
Acknowledgements We would like to thank the EPSRC/Crystal Faraday for a project stu-
dentship (GWB).
References
1. Gale PA, Quesada R (2006) Coord Chem Rev 250:3219
2. Gale PA, García-Garrido SE, Garric J (2008) Chem Soc Rev. DOI 10.1039/b715825d
3. Gale PA (2006) Acc Chem Res 39:465
4. Sessler JL, Gale PA, Cho W-S (2006) Anion Receptor Chemistry. RSC Publishing,
Cambridge
5. Xu H, Strater N, Schroder W, Bottcher C, Ludwind K, Saenger W (2003) Acta Cryst
D59:815
6. Pflugrath JW, Quiocho FA (1985) Nature 314:257
7. Pflugrath JW, Quiocho FA (1988) J Mol Biol 200:163
8. He JJ, Quiocho FA (1991) Science 251:1497
9. Luecke H, Quiocho FA (1990) Nature 347:402
10. Pascal RA, Spergel J, van Engen D (1986) Tetrahedron Lett 27:4099
11. Valiyaveettil S, Engbersen JFJ, Verboom W, Reinhoudt D (1993) Angew Chem Int Ed
32:900
12. Kavallieratos K, de Gala SR, Austin DJ, Crabtree RH (1997) J Am Chem Soc 119:2325
13. Hughes MP, Smith BD (1997) J Org Chem 62:4492
14. Santacroce PV, Davis JT, Light ME, Gale PA, Iglesias-Sanchez JC, Prados P, Quesada R
(2007) J Am Chem Soc 129:1886
15. Webb JEA, Crossley MJ, Turner P, Thordarson P (2007) J Am Chem Soc 129:7155
16. Prohens R, Tomas S, Morey J, Deya PM, Ballester P, Costa A (1998) Tetrahedron Lett
39:1063
17. Davis AP, Perry JJ, Williams RP (1997) J Am Chem Soc 119:1793
18. Choi K, Hamilton AD (2001) J Am Chem Soc 123:2456
19. Szumna A, Jurczak J (2001) Eur J Org Chem, p 4031
20. Chmielewski MJ, Jurczak J (2005) Chem Eur J 11:6080
21. Chmielewski MJ, Jurczak J (2006) Chem Eur J 12:7652
22. Kavallieratos K, Bertao CM, Crabtree RH (1999) J Org Chem 64:1675
23. Kang SO, Powell D, Bowman-James K (2005) J Am Chem Soc 127:13478
24. Kang SO, Powell D, Day VW, Bowman-James K (2006) Angew Chem Int Ed 45:1921
25. Kubik S, Goddard R, Kirchner R, Nolting D, Seidel J (2001) Angew Chem Int Ed
40:2648
26. Kubik S, Kirchner R, Nolting D, Seidel J (2002) J Am Chem Soc 124:12752

27. Gunnlaugsson T, Kruger PE, Jensen P, Pfeffer FM, Hussey GM (2003) Tetrahedron Lett
44:8909
28. Camiolo S, Gale PA, Hursthouse MB, Light ME, Shi AJ (2002) Chem Commun, p 758
29. Gale PA, Navakhun K, Camiolo S, Light ME, Hursthouse MB (2002) J Am Chem Soc
124:11228
30. Camiolo S, Gale PA, Hursthouse MB, Light ME (2003) Org Biomol Chem 1:741
31. Boiocchi M, Boca LD, Gomez DE, Fabbrizzi L, Licchelli M, Monzani E (2004) J Am
Chem Soc 126:16507
32. Amendola V, Esteban-Gomez D, Fabbrizzi L, Licchelli M (2006) Acc Chem Res
39:343
33. Gunnlaugsson T, Davis AP, Hussey GM, Tierney J, Glynn M (2004) Org Biomol Chem
2:1856
34. Kwon JY, Jang YJ, Kim SK, Lee K-H, Kim JS, Yoon J (2004) J Org Chem 69:5155
35. Brooks SJ, Gale PA, Light ME (2005) Chem Commun, p 4696
36. Brooks SJ, Garcia-Garrido SE, Light ME, Cole PA, Gale PA (2007) Chem Eur J 13:3320
37. Brooks SJ, Edwards PR, Gale PA, Light ME (2006) New J Chem 30:65
38. Kondo S-I, Nagamine M, Yano Y (2003) Tetrahedron Lett 44:8801
39. Pfeffer FM, Gunnlaugsson T, Jensen P, Kruger PE (2005) Org Lett 7:5357
40. Ayling AJ, Perez-Payan N, Davis AP (2001) J Am Chem Soc 123:12716
41. Clare JP, Ayling AJ, Joos J-B, Sisson AL, Magro G, Perez-Payan MN, Lambert TN,
Shukla R, Smith BD, Davis AP (2005) J Am Chem Soc 127:10739
42. Koulov AV, Lambert TN, Shukla R, Jain M, Boon JM, Smith BD, Li H, Sheppard DN,
Joos J-B, Clare JP, Davis AP (2003) Angew Chem Int Ed 42:4931
43. Snellink-Ruel BHM, Antonisse MMG, Engbersen JFJ, Timmerman P, Reinhoudt DN
(2000) Eur J Org Chem, p 165
44. Brooks SJ, Gale PA, Light ME (2006) Chem Commun, p 4344
45. Lee KH, Hong J (2000) Tetrahedron Lett 41:6083
46. Hisaki I, Sasaki S-I, Hirose K, Tobe Y (2007) Eur J Org Chem, p 607
47. Sessler JL, Camiolo S, Gale PA (2003) Coord Chem Rev 240:17
48. Shionoya M, Furuta H, Lynch V, Harriman A, Sessler JL (1992) J Am Chem Soc
114:5714
49. Kral V, Furuta H, Shreder K, Lynch V, Sessler JL (1996) J Am Chem Soc 118:1595
50. Gale PA, Camiolo S, Tizzard GJ, Chapman CP, Light ME, Coles SJ, Hursthouse MB
(2001) J Org Chem 66:7849
51. Gale PA, Light ME, McNally B, Navakhun K, Sliwinski KE, Smith BD (2005) Chem
Commun, p 3773
52. Sessler JL, Pantos GD, Gale PA, Light ME (2006) Org Lett 8:1593
53. Sessler JL, Barkey NM, Pantos GD, Lynch VM (2007) New J Chem 31:646
54. Vega IED, Camiolo S, Gale PA, Hursthouse MB, Light ME (2003) Chem Commun,
p 1686
55. Vega IED, Gale PA, Hursthouse MB, Light ME (2004) Org Biomol Chem 2:2935
56. Sessler JL, Katayev E, Pantos GD, Ustynyuk YA (2004) Chem Commun, p 276
57. Sessler JL, Katayev E, Pantos GD, Scherbakov P, Reshetova MD, Khurstalev VN,
Lynch VM, Ustynyuk YA (2005) J Am Chem Soc 127:11442
58. Katayev E, Boev N, Khurstalev VN, Ustynyuk YA, Tananaev IG, Sessler JL (2007) J Org
Chem 72:2886
59. Gale PA, Sessler JL, Kral V, Lynch VM (1996) J Am Chem Soc 118:5140
60. Custelcean R, Delmau LH, Moyer BA, Sessler JL, Cho WS, Gross D, Bates GW,
Brooks SJ, Light ME, Gale PA (2005) Angew Chem Int Ed 44:2537
61. Sessler JL, Gross D, Cho WS, Lynch VM, Schmidtchen FP, Bates GW, Light ME,
Gale PA (2006) J Am Chem Soc 128:12281
62. Bates GW, Gale PA, Light ME (2006) Cryst Eng Comm 8:300
63. Gale PA, Sessler JL, Kral V (1998) Chem Commun, p 1
64. Gale PA, Anzenbacher P Jr, Sessler JL (2001) Coord Chem Rev 222:57
65. Lee C-H, Lee J-S, Na H-K, Yoon D-W, Miyaji H, Cho W-S, Sessler JL (2005) J Org
Chem 70:2067
66. Miyaji H, Hong S-J, Jeong S-D, Yoon D-W, Na H-K, Hong J, Ham S, Sessler JL, Lee C-H
(2007) Angew Chem Int Ed 46:2508
67. Cafeo G, Kohnke FH, La Torre GL, White AJP, Williams DJ (2000) Angew Chem Int
Ed 39:1496
68. Cafeo G, Kohnke FH, La Torre GL, Parisi MF, Nascone RP, White AJP, Williams DJ
(2002) Chem Eur J 8:3148
69. Cafeo G, Kohnke FH, White AJP, Garozzo D, Messina A (2007) Chem Eur J 13:649
70. Chmielewski MJ, Charon M, Jurczak J (2004) Org Lett 6:3501
71. Piatek P, Lynch VM, Sessler JL (2004) J Am Chem Soc 126:16073
72. Curiel D, Cowley A, Beer PD (2005) Chem Commun, p 236
73. Sessler JL, Cho D-G, Lynch V (2006) J Am Chem Soc 128:16518
74. Chang K-J, Chae MK, Lee C-H, Lee J-Y, Jeong K-S (2006) Tetrahedron Lett 47:6385
75. Chang K-J, Moon D, Lah MS, Jeong K-S (2005) Angew Chem Int Ed 44:7926
76. Bates GW, Gale PA, Light ME (2007) Chem Commun, p 2121
77. Smith DK (2003) Org Biomol Chem 1:3874
78. Winstanley KJ, Smith DK (2007) J Org Chem 72:2803
79. Ghosh S, Choudhury AR, Row TN, Maitra U (2005) Org Lett 7:1441
80. Kim H, Kang J (2005) Tetrahedron Lett 46:5443
81. Yoon J, Kim SK, Singh J, Lee JW, Yang YJ, Chellappan K, Kim KS (2004) J Org Chem
69:581
82. Wong WWH, Vickers MS, Cowley AR, Paul RL, Beer PD (2005) Org Biomol Chem
3:4201
83. Alcalde E, Mesquida N, Perez-Garcia L (2006) Eur J Org Chem, p 3988
84. Abouderbala LO, Belcher WJ, Boutelle MG, Cragg PJ, Dhaliwal J, Fabre M, Steed JW,
Turner DR, Wallace KJ (2002) Chem Commun, p 358
85. Wallace KJ, Belcher WJ, Turner DR, Syed KF, Steed JW (2003) J Am Chem Soc
125:9699
86. Shinoda S, Tadokoro M, Tsukube H, Arakawa R (1998) Chem Commun, p 181
87. Haj-Zaroubi M, Mitzel NW, Schmidtchen FP (2002) Angew Chem Int Ed 41:104
88. Schmuck C (1999) Chem Commun, p 843
89. Schmuck C (2000) Chem Eur J 6:709
90. Schmuck C, Schwegmann M (2005) J Am Chem Soc 127:3373
91. Blondeau P, Benet-Buchholz J, de Mendoza J (2007) New J Chem 31:736
92. Park CH, Simmons HE (1968) J Am Chem Soc 90:2431
93. Garcia-Espana E, Diaz P, Llinares JM, Bianchi A (2006) Coord Chem Rev 250:2952
94. Llinares JM, Powell D, Bowman-James K (2003) Coord Chem Rev 240:57
95. Schmidtchen FP (1977) Angew Chem Int Ed Engl 16:720
96. Frontera A, Morey J, Oliver A, Piña MN, Quiñonero D, Costa A, Ballester P, Deyà PM,
Anslyn EV (2006) J Org Chem 71:7185
97. Prohens R, Martorell G, Ballester P, Costa A (2001) Chem Commun, p 1456
98. Hossain MA, Kang SO, Powell D, Bowman-James K (2003) Inorg Chem 42:1397
Struct Bond (2008) 129: 45–94
DOI 10.1007/430_2007_073
Published online: 19 January 2008
Metal-Based Anion Receptor Systems

Simon R. Bayly (u) · Paul D. Beer (u)
Department of Chemistry, Inorganic Chemistry Laboratory, University of Oxford,
South Parks Road, Oxford OX1 3QR, UK
simon.bayly@chem.ox.ac.uk, paul.beer@chem.ox.ac.uk
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
1.1 Metal Complexes in Anion Sensing . . . . . . . . . . . . . . . . . . . . . . 46
1.2 Basis of Anion Sensing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
2 Redox-Active Transition Metal-Based Receptors

for Electrochemical Anion Sensing . . . . . . . . . . . . . . . . . . . . . . 47
2.1 Theory of Electrochemical Sensing . . . . . . . . . . . . . . . . . . . . . . 47
2.2 Metallocene Redox Anion Sensors . . . . . . . . . . . . . . . . . . . . . . . 48
2.3 Mixed Metal Metallocene-Lewis Acid Anion Receptors . . . . . . . . . . . . 53
2.4 Transition Metal Polypyridyl Anion Receptors . . . . . . . . . . . . . . . . 56
2.5 Transition Metal Dithiocarbamates as Receptors for Redox Anion Sensing . 59
2.6 Dendrimers as Redox Anion Sensors . . . . . . . . . . . . . . . . . . . . . 60
2.7 Surface Confined Redox Anion-Sensing Systems . . . . . . . . . . . . . . . 63
3 Optical Anion Sensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

3.1 Theory of Optical Anion Sensing . . . . . . . . . . . . . . . . . . . . . . . . 68
3.2 Metallocene Optical Anion Sensors . . . . . . . . . . . . . . . . . . . . . . 68
3.3 Ruthenium(II) Polypyridyl Complexes as Optical Anion Sensors . . . . . . 72
3.4 Optical Anion Sensing Using Reporter Groups
Based on Other Transition-Metal Complexes . . . . . . . . . . . . . . . . . 75
3.5 Transition Metals as Anion-Binding Groups and/or Structural Components 79
3.6 Optical Anions Sensing by Lanthanide(III) Complexes . . . . . . . . . . . . 86
3.7 Surface Confined Systems for Optical Anion Sensing . . . . . . . . . . . . . 88
4 Metal-Based Anion Receptors Without Reporter Groups . . . . . . . . . . 89
5 Conclusion/Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
Abstract Metal complexes play an important role in anion receptor chemistry. In the
majority of examples metal centres are used as optical and/or electrochemical re-
porter groups in anion-sensing applications. Metal centres can also act as Lewis acidic
anion-binding sites in their own right, and/or as structural components allowing the
self-assembly of anion-binding domains. This review describes the development of metal-
based receptors with regard to their anion-sensing properties, and is therefore divided
into sections on electrochemical and optical anion sensing. Within these sections cover-
age has been given to the diverse range of metals and anion-binding groups that have
been studied. Emphasis has been placed on recently described novel supramolecular,
nanoscale and surface confined anion receptor systems that give added functionality. The
last section describes metal-based anion receptors without reporter groups.
46 S.R. Bayly · P.D. Beer
Keywords Anion sensing · Anion recognition · Anion coordination · Anion receptors ·

Organometallic receptors
1
Introduction
1.1
Metal Complexes in Anion Sensing
Metal complexes have played an important role in anion receptor chemistry

since its earliest examples. The presence of a metal ion can introduce a range
of advantageous physicochemical properties to this class of receptor. In the
majority of examples the metal complex is incorporated as a reporter group,
whose photochemical or redox response is changed upon proximal binding of
an anion. Furthermore, the metal can contribute directly to anion binding, ei-
ther by using its positive charge to electrostatically attract the anion and/or
by acting as a Lewis acidic binding site. The metal complex motif can also
be utilised as a structural component in anion receptors, where its coordina-
tion geometric properties allow the self-assembly of receptor sites with a wide
range of topologies not possible with simple tetrahedral covalent bonds. By
exploiting these different properties, often in combination, metal complex
anion receptors achieve a range of functionality beyond the scope of purely
organic structures.
Previous reviews on this topic have included many aspects of anion recog-
nition by metal-based receptors [1–6]. This review does not seek to be com-
prehensive; instead it is designed to provide an introduction to the area
by highlighting notable examples and to bring the reader up to date with
significant recent results, especially in the application of metal-based anion
receptors in surface fabricated nanoscale sensor systems.
1.2
Basis of Anion Sensing
Molecular sensing refers to a remotely detectable change in the proper-

ties of a receptor molecule on binding of an analyte. The generic design
for an anion sensor utilises a spacer group to covalently link an anion re-
ceptor site to a signalling or reporter group. Provided the spacer allows
some degree of coupling between the components (through-space, through-
bond, or by conformational change), binding of an anion at the receptor
site perturbs the electronic properties of the signalling group in a way that
can be detected spectroscopically or electrochemically. Thus, when the sig-
nalling group is a suitable redox-active metal centre, binding can be probed
electrochemically (e.g. by voltammetry). If a suitable chromophore or flu-
Metal-Based Anion Receptor Systems 47
orophore is used, sensing can be accomplished via optical spectroscopy

(UV/Vis absorption or luminescence, respectively). The spectroscopic prop-
erties of many metal complexes make them particularly amenable for this.
NMR spectroscopy can also be used to detect anion binding to diamag-
netic metal-based receptors. However, this is not generally regarded as
a remote sensing technique since it obviously requires the sample to be
placed in an external magnetic field. This review is conveniently divided
into sections on metal-based receptors for electrochemical anion sensing,
receptors for optical anion sensing and anion receptors without reporter
groups.
2
Redox-Active Transition Metal-Based Receptors
for Electrochemical Anion Sensing
2.1
Theory of Electrochemical Sensing
When a redox-active transition metal is used as the signalling unit of a recep-

tor, anion binding is coupled to electron transfer, i.e. anion binding changes
the redox potential (couple) of the transition metal. This electrochemical
shift can be represented as ∆E0 , the difference in redox potentials between
the receptor : anion complex and the receptor alone. Concomitantly, elec-
tron transfer at the redox centre also changes the affinity of the receptor for
the guest species. These coupled processes are linked thermodynamically by
Eq. 1, where Kred and Kox are the stability constants of the reduced and oxi-
dised forms of the receptor:anion complex respectively [7].
nF(∆E0 ) = RT ln(Kox /Kred ) (1)
From a thermodynamic standpoint, the value of the shift in redox potential

is determined by the ratio of Kox /Kred , instead of the absolute value of either
Kox or Kred . As a consequence a receptor need not necessarily have a very
high binding strength for the anion to be sensed. If electron transfer leads
to a sufficiently large change in the stability of the receptor : anion complex,
a measurable change in redox potential can be observed.
Anion binding stabilises the oxidised form of the receptor, hence Kox /
Kred > 1 and the redox potential of the reporter group is shifted to a more
negative value (cathodic shift). The Kox /Kred ratio is also a measure of how
efficient the coupling is between the metal-based reporter group and the
anion-binding site.
2.2
Metallocene Redox Anion Sensors
Anion receptors incorporating the redox-active cobaltocenium group have

been studied extensively due to the combination of an accessible redox
couple and favourable electrostatic interactions of the cationic organometallic
metallocene motif with anions. The first anion receptors based on this
species were reported by Beer and co-workers in 1989 [8]. The macro-
cyclic bis-cobaltocenium receptor 1 was shown to bind bromide in ace-
tonitrile solution (due to electrostatic interaction). Electrochemical anion
sensing was also demonstrated, where bromide caused the potential of the
cobaltocenium/cobaltocene to undergo a cathodic shift. Augmentation of
cobaltocenium-based receptors with hydrogen-bond donor groups, such as
amides in receptors 2 and 3, generates both stronger and more selective an-
ion binding [9]. Proton NMR anion titration studies in CD3 CN reveal 2 and
3 to have selectivity for dihydrogenphosphate over chloride by approximately
an order of magnitude [10]. This is attributed not only to the greater basic-
ity of the dihydrogenphosphate anion, but also to complementary hydrogen
bonding between the receptor and the anion. In these receptors the elec-
trochemical sensing of anions is also enhanced, with chloride giving rise
to cathodic shifts in the cobaltocenium/cobaltocene redox couple of 30 and
85 mV for 2 and 3, respectively. The dihydrogenphosphate anion generates
cathodic shifts of 200 and 240 mV respectively, confirming that in this class
of amide hydrogen-bonding receptors the magnitude of the electrochemical
response directly mirrors the strength of the receptor-anion interaction.
Ferrocene has also been extensively exploited in redox responsive anion
receptor design. One advantage of ferrocene is that its synthetic chemistry
is highly developed, in particular in terms of its conjugation to organic

molecules. From the point of view of anion sensing the most relevant dif-
ference between the metallocenes is that ferrocene is neutral in charge and
therefore its derivatives have no inherent electrostatic interaction with anions
(until oxidised to ferrocenium) and therefore their complexes with anions
exhibit relatively lower stability constants.
Molecules 4–8 are a selection of ferrocene-based receptors which incorpo-

rate amide groups for the hydrogen-bonding of anions [11, 12]. In acetonitrile
solution dihydrogenphosphate induced cathodic shifts of up to 240 mV in
the ferrocene/ferrocenium couples of these receptors. Competition experi-
ments demonstrated the same shift even in the presence of a 10-fold excess
of chloride or hydrogensulfate. In these receptors it is largely the stability of
the electrostatically enhanced anion : ferrocenium complex which determines
the magnitude of the redox shift. Receptor 8 has the opposite selectivity, dis-
playing a hydrogensulfate induced shift of 220 mV which does not change
in the presence of excess dihydrogenphosphate. It is thought that binding of
HSO4 – leads to protonation of the amine group of the receptor. The result-
ing complex is cationic and this has a very high affinity for the residual SO4 2–
anion. Electrostatic interactions are particularly important in redox anion
sensing, and even very simple anion-binding motifs such as the ammonium
cation provide an increased redox response. This has been demonstrated
by Moutet and co-workers using 9 [13]. This molecule is able to sense di-
hydrogenphosphate and ATP2– in a range of solvents, displaying a shift of
470 mV in CH2 Cl2 with dihydrogenphosphate, solely due to a strong ion-
pairing interaction.
Sensing anions in aqueous conditions is a particular challenge which

must be met for molecular anion receptors to become a useful technology
in biological or environmental analysis. The high dielectric and compet-
itive hydrogen-bond donor capacity of water diminishes anion-receptor
interactions. In general strong electrostatic interactions are required to
overcome this. Beer et al. have developed a series of ferrocene-based re-
ceptors appended with various open chain and cyclic amine functional
groups, e.g. molecules 10 and 11, that bind ATP2– and dihydrogenphos-
phate in water [14–16]. The selectivity of this class of receptors is pH
driven. At pH 6.5 at least two of the amines are protonated and the 1 : 1
anion : receptor complexes formed show cathodic shifts of 60–80 mV in the
ferrocene/ferrocenium redox couple. Quantitative determination of phos-
phate and sulfate in the presence of competitor anions was demonstrated by
metallacyclic receptors 12 and 13. The electrochemistry of these receptors
was studied in 70 : 30 THF : H2 O over a range of pHs. A maximum selective
redox shift of 54 mV for phosphate over sulfate was observed at pH 4 for 12,
whereas 13 gave a maximum selective redox shift of 50 mV for phosphate over
sulfate at pH 7.
Recognition of fluoride in aqueous media is particularly difficult due to the

strongly hydrated nature of the anion. Shinkai and co-workers have demon-
strated that ferrocene-boronic acid 14 acts as a selective redox sensor for flu-
oride which operates in H2 O [17]. The favourable interaction between boron
and fluoride (a hard-acid and hard-base, respectively) generates a stability
constant of 700 M–1 for the fluoride-ferrocenium complex. Stability constants
for both the bromide and chloride complexes are < 2 M–1 .
In receptor molecules that contain multiple metal centres and anion bind-
ing groups the redox sensing properties are dictated by the precise spatial
arrangement of these groups. In receptors where the pendant reporter groups
are in close proximity to each other an increased redox response to anion
binding is often observed. For example, the cyclotriveratrylene amides 15 and
16 include closely spaced multiple amide anion-binding groups with pen-
dant ferrocene reporter units [18]. In CH2 Cl2 or acetonitrile solution cathodic
shifts of up to 260 mV were observed in the presence dihydrogenphosphate or
ATP2– . On oxidation 15 and 16 gain a triple positive charge. The magnitude of
the redox shift can be attributed to the increased electrostatic affinity of this
multiply charged oxidised species for a single anion compared to monoferro-
cenyl receptors which are monopositive on oxidation.
Receptor 17 has a similar topology—it comprises four cobaltocenium
amide groups attached to a porphyrin backbone as the cis-α,α,α,α-atrop-
isomer [19]. Proton NMR titrations in CD3 CN showed chloride and bro-
mide to be bound in 1 : 1 stoichiometry with stability constants of 860
and 820 M–1 , respectively, whereas nitrate exhibited weaker binding with
K = 190 M–1 . Electrochemical studies displayed cathodic shifts in the cobal-
tocene/cobaltocenium redox couple of 35–75 mV on addition of chloride or
hydrogensulfate, and 225 mV for dihydrogenphosphate in acetonitrile solu-
tion. Smaller shifts were seen in the porphyrin oxidation wave. The overall
selectivity trend was Cl– > Br– NO3 – .
The anion recognition properties of cobaltocenium calix[4]arene recep-
tors 18–20 were found to be dependent on the structure of the upper-rim of
the calix[4]arene [20]. In 1 H NMR studies in DMSO-d6 solution 18 shows
a greater affinity for acetate than for dihydrogenphosphate whereas its iso-
mer 19 displays the opposite trend. In 20 there is only a single cobaltocenium
group which bridges the upper rim of the calix[4]arene. This receptor displays
a significantly greater affinity for the above anions despite possessing only
a single positive charge. For example, the cobaltocene/cobaltocenium redox
couple of 20 was found to undergo a cathodic shift of 155 mV in the presence
of acetate. It is proposed that the surprising strength of the interaction is due
to the topology of the anion-binding cavity, in which the arrangement of the
two amide hydrogen bond donors is complementary to bidentate anions such
as carboxylates. The same selectivity is seen in the ferrocene-1,1 -bisamide
analogues 21–23 [21]. Results of 1 H NMR studies in CD3 CN show that these
receptors also preferentially bind carboxylate anions (acetate and benzoate)
over dihydrogenphosphate and chloride.
Other carboxylate selective redox sensors based on ferrocene include neu-
tral molecule 24, which utilises hydrogen bonding to bind mono and dicar-
boxylate anions with 2 : 1 and 1 : 1 guest : host stoichiometry [22]. However,

it cannot distinguish between these types of anion electrochemically, giving
a maximum cathodic shift of 150 mV for both acetate and phthalate. A more
recent example shows that selectivity for dicarboxylate anions over monocar-
boxylates and other simple anions can be achieved. Tetra-ammonium macro-
cycle 25 binds phthalate, isopthalate and dipicolinate with a 2 : 1 guest : host
stoichiometry, giving maximum cathodic shifts in the redox potential of
275, 193 and 168 mV, respectively [23]. In comparison the monoacid 4-
nitrobenzoate produced a maximum cathodic shift of only 49 mV.
The incorporation of crown ether units into a cobaltocenium receptor has
been shown to allow the switchable redox sensing of anions. Proton NMR
titrations of receptor 26 in CD3 CN solution gave log K values of 3.1 for chlo-
ride and 3.0 for bromide [24]. Electrochemical titrations showed cathodic
shifts of the cobaltocene/cobaltocenium redox couple of 60 and 30 mV for the
two anions, respectively. However, when either the NMR or electrochemical
titrations were carried out in the presence of K+ no significant anion induced
shifts were observed. It is proposed that the K+ ions form a 1 : 1 intramolec-
ular sandwich complex with the two benzocrown ether units of the receptor
causing a concomitant change in conformation of the amide groups which
reduces their availability for anion binding.
Molina and co-workers have investigated the urea-crown ether function-
alised ferrocene 27 [25]. This receptor-produced anion induced cathodic shifts
in the ferrocene/ferrocenium redox couple of 52 mV with fluoride and 190 mV
with dihydrogenphosphate. In acetonitrile solution on addition of 2 equiva-
lents of K+ ions a dramatic attenuation in the anion-induced cathodic shift was
observed, with dihydrogenphosphate giving rise to a shift of only 50 mV.
2.3
Mixed Metal Metallocene-Lewis Acid Anion Receptors
Lewis acidic metal centres can be utilized as anion-binding groups, which in

combination with ferrocenyl reporter groups provide enhanced redox anion
sensing. Ferrocene-amide receptor 28 utilises pendant phosphine groups to
allow the coordination of various transition metals to generate mixed-metal
complexes 29–32 [26]. 1 H NMR titrations carried out in CD2 Cl2 solution
revealed that the neutral molecules 29–31 bind chloride approximately an
order of magnitude more strongly than the parent phosphine. The affin-
ity of these complexes for bromide, iodide and hydrogenphosphate was also
found to be increased, but the effect was smaller. Cationic complex 32 was
found to bind the same anions an order of magnitude more strongly again,
due to the added influence of the electrostatic attraction. With chloride,
bromide and hydrogensulfate in acetonitrile/dichloromethane solution a sig-
nificant anion-induced cathodic redox shift was observed in both the fer-
rocene/ferrocenium couple and the irreversible oxidation of the secondary
metal centre.
Receptor 33 also incorporates a secondary Lewis acid anion binding

site. This molecule is the zinc metallated analogue of 17 with the cobal-
tocenium reporter groups replaced with ferrocenes [27]. The freebase pre-
cursor to 33 in dichloromethane solution shows no significant anion in-
duced shifts in the 1 H NMR signals of the amide protons, whereas the
metalloporphyrin binds bromide (K = 6200 M–1 ), nitrate (K = 2300 M–1 )
and hydrogenphosphate (K = 2100 M–1 ). Electrochemical studies in 3 : 2

dichloromethane/acetonitrile revealed anion-induced cathodic shifts in
both the porphyrin (∆E = 85–115 mV) and tetraferrocene oxidation (∆E =
20–60 mV) waves. The trend in magnitude of ∆E for the porphyrin oxidation
wave is hydrogensulfate > chloride > bromide > nitrate, reflecting the charge
density (charge to radius ratio) of the anionic guest species. Atropisomers of
33 (other than the α,α,α,α-atropisomer) were also studied and showed dif-
ferent anion-sensing properties. For instance the α,β,α,β-atropisomer was

found to be selective for nitrate (∆Eporphyrin = 175, ∆Eferrocene = 110).
Jurkschat and co-workers have investigated the redox anion-sensing
properties of metallomacrocycle 34 [28]. This comprises two ferrocene re-
porter units linked together covalently by two Lewis acidic organotin spac-
ers. Electrochemical measurements in dichloromethane solution showed
anion-induced cathodic shifts in the ferrocene/ferrocenium redox couple of
130 mV for chloride, 210 mV for fluoride and 480 mV for dihydrogenphos-
phate.
Another class of mixed-metal anion receptors has been investigated which
possess redox reporter groups based on two different metal complexes. This
enables the qualitative comparison of their comparative anion-sensing abili-
ties. Macrocycles 35 and 36 combine the {RuII (bpy)3 } moiety with a bridging
ferrocene or cobaltocenium unit [29]. Electrochemical experiments in ace-
tonitrile solution revealed that the RuII /RuIII redox potential was insensitive
to anion binding, whereas the ferrocene/ferrocenium (in 35) and cobal-
tocene/cobaltocenium (in 36) redox couples were shifted cathodically (by
60 mV and 110 mV respectively with chloride). However, the first reduction
of {RuII (bpy)3 }, a ligand-centred process based on the amide substituted
bipyridyl, was also found to undergo an anion induced cathodic shift (40 mV
and 90 mV with chloride for 35 and 36, respectively).
2.4
Transition Metal Polypyridyl Anion Receptors
Redox sensing of anions using {RuII (bpy)3 }-amides as a combined recep-

tor/reporter system has also been studied using complexes 37–40 [30–32].
The single crystal X-ray structure of the chloride complex of 37 clearly in-
dicates that the anion is bound tightly within the bipy amide ligand by six
hydrogen bonds. It forms hydrogen bonds not only to the two N–H groups
but also to four aromatic C–H groups with H–Cl distances ranging from 2.51
to 2.71 Å. 1 H NMR titrations in DMSO-d6 revealed strong binding of chloride
and dihydrogenphosphate. Four reversible redox couples (one metal-centred
oxidation and three ligand centred reductions are expected for {RuII (bpy)3 }
species) were observed in electrochemical studies. Of these only the least ca-
thodic ligand reduction was significantly shifted in the presence of anionic
guests. The assignment of this redox process to the relatively electron-poor
amide-substituted bipyridyl reaffirms the X-ray structure evidence that an-
ion recognition takes place at this site. It is interesting to note that the
calix[4]arene modified molecule 40 which can be compared with the cobal-
tocenium and ferrocene complexes 7 and 18–23, shows particular selectivity
for dihydrogenphosphate and is able to electrochemically sense this anion
(giving a cathodic shift of 175 mV) in the presence of a 10-fold excess of chlo-
ride or hydrogensulfate [30].
{RuII (bpy)3 } macrocycles 41–44 demonstrate the influence of the top-

ology of the anion-binding site on redox sensing [29]. The macrocycles with
the larger cavities 42–44 were shown by 1 H NMR studies in DMSO-d6 to
bind chloride preferentially to acetate and dihydrogenphosphate. This is the
opposite trend to 41, which has the smallest macrocyclic cavity and binds
acetate more strongly than either chloride or dihydrogenphosphate. Recep-
tor 43 displays outstanding selectivity; with a stability constant for chloride
of 40 000 M–1 and no measurable affinity for dihydrogenphosphate. The pres-
ence of a second positively charged metal centre in 43 and 44 leads to in-
creased overall anion affinity. The substitution of one Ru(II) centre for Os(II)
in 44 results in both an increased stability constant for the chloride com-
plex and a diminished affinity for acetate—in effect an increase in selectivity
between the two anions. Results of electrochemical measurements on the
complexes in acetonitrile solution confirmed the pattern of anion selectivity.
In 41 the cathodic shift in the first bpy redox wave was 30 mV for chloride
and 55 mV for acetate. In 43 and 44 cathodic shifts of 110 and 125 mV, re-
spectively, were recorded in the presence of chloride, whereas for acetate the
change was 15 and 20 mV, respectively. This class of anion receptor also ex-
hibits optical anion-sensing properties (see Sect. 3.2).
{RuII (bpy)3 } has also been used as the basis of a redox sensor for fluoride.
Receptors 45 and 46 were studied in acetonitrile solution using differential
pulse voltammetry. Addition of two equivalents of fluoride was found to pro-

duce a peak at 0.73 V vs. Ag/AgCl, ascribed to a ligand-centred redox process.
Other anions, including halides, nitrate and hydrogensulfate caused no sig-
nificant change in the electrochemistry [33].
A receptor based on {CoIII (bpy)3 } has been used by Sessler and co-
workers for redox fluoride sensing [34]. In the cyclic voltammetry of 47 in
DMSO solution addition of fluoride led to a complete disappearance of the
Co(II)/(III) reduction wave. Addition of water to this solution restored this re-
dox process, suggesting that the presence of a strongly bound fluoride anion
renders the complex redox inactive. Chloride and dihydrogenphosphate pro-
duced cathodic shifts of 160 mV and 70 mV, respectively. It is proposed that
the pyrrole NH protons of the quinoxaline phenanthroline ligand are made

more acidic by the electron-withdrawing effect of the coordinated metal cen-
tre, thereby giving the complex an increased affinity for fluoride compared to
the free ligand.
2.5
Transition Metal Dithiocarbamates as Receptors for Redox Anion Sensing
Another interesting development has been the self-assembly of metallo-

dithiocarbamate macrocyclic receptors for electrochemical anion sensing.
The naphthyl-based Cu(II) macrocycle 48 displays a cathodic shift of 85 mV
in the Cu(II)/(III) redox couple in the presence of dihydrogenphosphate or
perrhenate, but gives no response to halides in acetonitrile solution [35]. It
is proposed that this selectivity is controlled by cavity size. A related recep-
tor incorporating thiourea and hydrogen-bond donor groups, 49, revealed
cathodic shifts in the Cu(II)/(III) couple of up to 160 mV with hydrogenphos-

phate [36]. Cobalt (III) dithiocarbamate cryptands 50 and 51 also function as
redox active anion sensors [37]. In dichloromethane solution the irreversible
Co(IV)/Co(III) redox couple of the complexes was found to undergo signifi-
cant anion-induced cathodic perturbation; up to a maximum of 125 mV for
51 with dihydrogenphosphate.
2.6
Dendrimers as Redox Anion Sensors
Dendrimers have been investigated as a platform for enhanced anion sensing.

Astruc and co-workers have synthesised dendrimers 52–54, incorporating
up to 18 amido-ferrocene units. These multi-metallic multi-binding site re-
ceptors are able to electrochemically sense anions in dichloromethane so-
lution [38]. In 54 the selectivity trend is dihydrogenphosphate > hydrogen-
sulfate > chloride > nitrate. Evidence of a “dendritic effect” was observed
in the redox response of the consecutive dendrimer generations in the
presence of dihydrogenphosphate or hydrogensulfate. As the number of
amido-ferrocene units is increased, the magnitude of the cathodic shift in
the ferrocene/ferrocenium couple also increases. Stability constants for the

hydrogensulfate complexes of 53 and 54 were reported to be 9390 and
216 900 M–1 , respectively.
Kaifer and co-workers have studied an analogous series of dendrimers
based on a commercial DSM polyamine core with 4, 8, 16 and 32 periph-
eral ferrocenyl urea groups as the anion-sensing component [39]. No stability
constants are reported, but cathodic shifts in the redox response of the fer-
rocene/ferrocenium couple with various anions in DMSO show a similar
selectivity trend to 52–54. In this case the data suggests two ferrocene urea
arms are involved in binding a single dihydrogenphosphate anion. The den-
dritic effect was observed in the change from the first (4 ferrocene units) to
second (8 ferrocene units) generation dendrimers. No further increase in re-
sponse was seen in the third generation and the fourth generation dendrimer
showed a decreased response, presumably due to steric crowding.
Astruc and co-workers have also investigated five generations of penta-

methyl-amidoferrocene dendrimers using the DSM polyamine core [40]. The
pentamethyl-substituted ferrocene was chosen to overcome the irreversible
electrochemistry and electrode adsorption observed with 52–54. In this se-
ries the dendritic effect seen in the electrochemistry DMF solution varied
according to the anion studied. In changing from lower to higher den-
drimer generations modest increases in the anion-induced cathodic shift of
the ferrocene/ferrocenium couple were observed with dihydrogenphosphate,
whereas with ATP2– anion binding progressed from weak to relatively strong.
This is perhaps due to ATP2– adopting a 1 : 2 anion/ferrocene unit binding
stoichiometry, whereas dihydrogenphosphate binds 1 : 1.
2.7
Surface Confined Redox Anion-Sensing Systems
In a step towards the fabrication of prototype sensory devices organisation

of redox-active anion receptors on to electrode surfaces is being exploited.
Importantly, self-assembled monolayers or thin polymer films of metal-based
receptors can generate an amplified response to anion binding akin to the
dendritic effect and could potentially become the basis of robust anion-
sensing devices.
Beer and co-workers have investigated this concept using self-assembled

monolayers of the 1,1 -bis(alkyl-N-amido)ferrocene 55 on gold electrodes [41].
The pendant disulfide groups serve to covalently anchor the receptor
to the gold surface. In electrochemical experiments on 55 in acetoni-
trile/dichloromethane solution anion-induced cathodic shifts of the fer-
rocene/ferrocenium redox couple were observed for chloride (40 mV),
bromide (20 mV) and dihydrogenphosphate (210 mV). When confined to

a monolayer the anion-induced shifts measured were 100 mV for chloride,
30 mV for bromide and 300 mV for dihydrogenphosphate in the same solvent
system—consistently greater than for the solution-phase receptor. This rep-
resents a significant “surface sensing amplification”. The modified electrodes
were also able to selectively detect dihydrogenphosphate in the presence of
a 100-fold excess of halide. In aqueous solution the selectivity of the system
was altered, enabling the detection of the poorly hydrated anion perrhenate
in the presence of dihydrogenphosphate.
A number of groups have been exploring the anion-sensing properties
of thin polymer films which incorporate metal-based receptors. Monomer
56 consists of a cobaltocenium amide redox signalling group with a poly-
merisable pyrrole unit. Thin films of the receptor were prepared by elec-
tropolymerisation on a platinum or carbon electrode [42]. In electrochem-
ical experiments on 56 in acetonitrile solution significant anion-induced ca-
thodic shifts of the cobaltocene/cobaltocenium redox couple were observed
for dihydrogenphosphate (45 mV) and hydrogensulfate (20 mV) only. When
confined to a polymer film the cathodic shifts were amplified: 210 mV for
dihydrogenphosphate and 250 mV for hydrogensulfate. Chloride and bro-
mide could also be detected, both giving shifts of 20 mV. Polymerisation
of 56 as well as giving rise to surface sensing amplification also resulted
in a change in selectivity from dihydrogenphosphate to hydrogensulfate. It
was also found that film thickness influences the sensitivity of the sensor.
Thin films (Γ = 1.8 × 10–9 mol cm–2 ) exhibited higher sensitivity to dihydro-
genphosphate at low concentrations (< 50 µM), whereas thick films (Γ =
2.7 × 10–8 mol cm–2 ) extend the measurable concentration range to higher
levels (up to 2 mM).
Films of the analogous ferrocene monomer 57 have been studied by
Moutet and co-workers [43]. Anion-induced shifts of the ferrocene/ferro-
cenium couple were measured in acetonitrile for hydrogensulfate (30 mV),
ATP2– (180 mV) and dihydrogenphosphate (220 mV). Again this represents
a surface sensing amplification. The same group has also explored the
anion-sensing properties of viologen 58 in thin polymer films [44]. In aque-
ous solution poly-58 registered small anion-induced cathodic shifts of the
ferrocene/ferrocenium redox couple with hydrogensulfate (20 mV), S2 O4 2–
(10 mV), and ATP2– (35 mV).
Metal-based receptors that are able to form self-assembled monolayers on
planar electrodes can also be used to functionalise the surface of nanoparti-
cles, leading to a surface sensing amplification effect. The very large surface
area of nanoparticles may also allow greater overall sensitivity to anions.
Astruc and co-workers have prepared the amidoferrocenylalkylthiol
(AFAT)-gold nanoparticle system 59 [45]. The proportion of AFAT to do-
decanethiol obtained by ligand substitution on different batches of dode-
canethiol stabilised nanoparticles ranged from 7–38%, corresponding to an
average of 8–39 AFAT units per nanoparticle. Electrochemical measurements

in dichloromethane solution show a single reversible redox wave for the fer-
rocene/ferrocenium couple at identical potential in each case. Addition of
dihydrogenphosphate led to the appearance of a new redox wave (220 mV
cathodically shifted) with the attenuation of the initial wave, which was com-
pletely replaced at 1 equivalent of anion per AFAT branch, indicating of 1 : 1
anion/branch binding. The cathodic shift is the same irrespective of the AFAT
loading and is considerably larger than observed for the comparable amido-
ferrocene monomer FcCONHCH2 CH2 OPh (45 mV) or even a representative
ferrocene tripod PhC(CH2 CH2 CH2 NHCOFc)3 (110 mV).
The same group has also investigated the anion-sensing properties of gold
nanoparticles 60 and 61 substituted with dendrons comprising three amid-
oferrocene or silyl ferrocene branches [40]. The surface loadings of 60 and
61 were 3% and 4.8% respectively, corresponding to an average 3 and 5
dendrons per nanoparticle. Nanoparticles of type 60 show very similar prop-
erties to the AFAT-modified nanoparticles 59, with a dihydrogenphosphate-
induced cathodic shift of 210 mV in dichloromethane solution. Despite lack-
ing any hydrogen-bonding groups the nanoparticles dendronised with silyl
ferrocenes, 61, gave a dihydrogenphosphate-induced cathodic shift of 110 mV

in the same solvent.
A highly ambitious multicomponent surface-anchored anion-sensing ro-
taxane assembly has recently been described [46]. This comprises two re-
ceptor molecules, 62—an isophthalamide macrocycle with an exocyclic fer-
rocene reporter group, and 63 —a cationic pyridinium amide thread bearing
a disulfide tether for SAM formation at one terminus, and a pentaphenylfer-
rocene at the other as a combined redox reporter and bulky stopper group.
In low polarity solvents such as dichloromethane chloride is bound simul-
taneously by both receptors, causing the threading of 63 into the annulus
of 62 to form a pseudorotaxane (Fig. 1). Adsorption of the pseudorotaxane
onto a clean gold surface to form a SAM causes the components to be locked
together as a rotaxane. This rotaxane SAM shows a remarkable selective elec-

trochemical response to anions compared to 62 or 63 alone. The addition of
molar excesses of chloride, dihydrogenphosphate or hydrogensulfate to 63 in
acetonitrile solution or as simple SAMs resulted in only a small cathodic shift
(∆E < 10 mV) of the pentaphenylferrocene redox couple. The macrocycle 62
was more responsive to anions, undergoing a cathodic shift in the ferrocene
redox couple of 45 mV with dihydrogenphosphate, 15 mV with hydrogen-
sulfate and < 10 mV with chloride. In the rotaxane SAM the pentaphenyl-
ferrocene centre of 63 exhibits redox responses to chloride and oxoanions
broadly similar to SAMs of this receptor on its own. In contrast, within
the surface assembled rotaxane the ferrocene of the macrocyclic compon-
ent exhibits a markedly greater electrochemical cathodic response to chloride
Fig. 1 Self-assembly of anion templated rotaxane SAM (RC = redox centre). From [46],
reproduced by permission of The Royal Society of Chemistry
(40±5 mV), but gives no significant response to the oxoanions tested. This
suggests that chloride binding inside the interlocked cavity of the surface
confined rotaxane results in a conformation where the macrocycle’s pendant
ferrocene group is in proximity to the complexed halide anion, whereas the
oxoanions are too large to penetrate the rotaxane binding pocket. Preliminary
electrochemical competition experiments in acetonitrile solutions revealed
that these rotaxane SAMs are capable of selectively detecting chloride in the
presence of 100-fold excess amounts of dihydrogenphosphate and exhibit an
appreciably greater detection sensitivity than that shown by the free macro-
cycle. The superior electrochemical response of rotaxane SAMs to chloride
over dihydrogenphosphate mirrors the high degree of chloride anion selectiv-
ity of previous rotaxanes prepared via chloride anion templation.
3
Optical Anion Sensors
3.1
Theory of Optical Anion Sensing
Optical reporter groups signal anion binding through a change in their elec-
tronic absorption or emission spectra. The precise nature of the response in
the UV/vis absorption spectrum will largely depend on the energy differences
between the molecular orbitals of the receptor before and after anion bind-
ing. For changes of any magnitude to be observed the anion-binding site must
be strongly coupled to the metal centre. In the case of metal-based transitions
such as d-d transitions this typically requires the anion to bind directly to the
metal centre. In the case of MLCT or LMCT transitions it is advantageous if
the anion-binding site is π-conjugated to the ligand involved. Luminescence
spectroscopy can be a more sensitive technique for probing anion binding
to metal-based receptors. In addition to altering the energy of the emission
maxima, anion binding often causes significant changes in their intensity.
The intrinsic luminescence of a particular reporter group can be “switched
off” if the anion : receptor complex provides a more efficient pathway for
non-radiative energy loss. Similarly, in systems where the luminophore is
quenched by a nearby functional group, anion binding can “switch on” the
luminescent emission by blocking this non-radiative decay process.
3.2
Metallocene Optical Anion Sensors
The common reporter groups cobaltocenium and ferrocene have not fre-
quently been used in optical anion sensing, since these chromophores are
generally insensitive to anion binding. However, metallocene-based receptors
that incorporate a suitable organic chromophore or luminophore have been

shown to operate as combined optical and electrochemical anion sensors.
For instance the tetra-cobaltocenium porphyrin 17, exhibits the same selec-
tivity trend (Cl– > Br– NO3 – ) in UV-vis anion-binding experiments that
was observed by electrochemistry [19]. In acetonitrile solution the Soret band
(λmax = 425 nm, due to the porphyrin) of 17 was significantly bathochromi-
cally shifted on addition of dihydrogenphosphate (∆λmax = 15 nm), hyp-

sochromically shifted with C1– (∆λmax = 10 nm) and split into two maxima
(λ = 430,440 nm) with HSO4 – .
The novel series of ferrocene receptors 64–68, which incorporate thiobar-
biturate anion binding/chromophore groups, have been shown to operate as
UV-vis anion sensors in acetonitrile solution. Addition of basic anions such
as cyanide, acetate and benzoate causes the attenuation of the absorption
maximum at around 440 nm (due to a charge transfer transition between the
amine and the thiobarbiturate group), with the simultaneous occurrence of
a new band at 370 nm. In cases where the titration data could be fitted, 1 : 2
receptor : anion binding stoichiometries were found [47].
Another interesting example of a ferrocene-based optical sensor is 69,

which acts as a chromogenic molecular switch [48]. Appended to one cy-
clopentadienyl ring of the ferrocene of molecule 69 is a p-nitrophenyl urea
unit which acts as a combined anion-binding site and chromophore. A crown
ether is attached to the other cyclopentadienyl ring for cation binding. Tucker
and co-workers reported that on addition of fluoride to a solution of 69 in
acetonitrile a significant perturbation of the UV-vis spectrum was observed
including the appearance of a new absorption at 472 nm. The Ka for the
1 : 1 anion-receptor complex was determined as 9340 M–1 . The addition of
10 equivalents of KPF6 to the solution of 69 containing 10 equivalents of flu-
oride caused the complete disappearance of the 475 nm absorption. Ka of
the receptor with K+ in the presence of fluoride was calculated as 1460 M–1 .
Surprisingly model receptor 70, which lacks the crown ether moiety, exhib-
ited similar switching properties. Ka of this receptor for fluoride is equivalent
(9660 M–1 ) but the Ka for K+ in the presence of fluoride is far lower (230 M–1 ).
Inhibition by K+ of the response of these receptors to fluoride is therefore
thought to be due to the ion-pairing interaction between fluoride and K+ .
Aldridge et al. have demonstrated that boryl-ferrocene 71 can be used as
a selective colourimetric sensor for fluoride [49]. When fluoride was added
to a CH2 Cl2 solution of 71 under aerobic conditions a colour change from
orange to pale green was observed. This did not occur with any other anion
tested. Spectroscopic and electrochemical measurements suggest that com-
plexation of fluoride causes the spontaneous formation of a ferrocenium

species, i.e. the 150 mV anodic shift in the ferrocene/ferrocenium redox po-
tential caused by fluoride complexation reduces the redox potential enough
for the 71 : 2F– complex to be oxidised by atmospheric O2 . The pale green
colour is due to the characteristic absorption of ferrocenium. In this case the
optical response is the direct result of an anion-induced redox process and
does not require an additional chromogenic group.
Luminescence sensing of anions has also been achieved using ferrocene re-
ceptors. Example 72 uses amide groups for anion binding in conjunction with
naphthalene groups to provide the fluorescence signal [50]. Addition of fluo-
ride to a DMSO solution of 72 led to a 3-fold enhancement (at 5 equivalents)
of the intramolecular naphthalene-naphthalene excimer emission at 492 nm.
Dihydrogenphosphate also generated a significant response, causing a 2-fold
enhancement at 5 equivalents. In electrochemical studies in DMF electrolyte
solution fluoride generated a 120 mV cathodic shift in the redox potential.
The receptor 73, based on an azaferrocenophane structure bearing two
urea groups as linkers between the redox active (ferrocene) and fluorescent
(naphthalene) signalling subunits, also shows both fluorescent and electro-
chemical sensing of fluoride [51]. On addition of excess fluoride it displays
an enhancement factor of 13 in the naphthalene emission bands at 362 and
380 nm in DMF solution and a cathodic shift of the ferrocene/ferrocenium
couple of 190 mV in DMSO electrolyte solution.
The unusual new [3,3]ferrocenophane 74·H+ also acts as a selective fluor-

escent anion sensor—in this case for nitrate [52]. The protonated receptor is
weakly fluorescent (Φ = 0.043) in CH2 Cl2 solution and on addition of nitrate
the naphthalene-based emission at 354 nm is quenched (to Φ = 0.020). Add-
ition of acetate, hydrogensulfate and dihydrogenphosphate merely induced
deprotonation of the receptor. This receptor is also able to act as a redox
sensor for other anions and as a fluorescent sensor for group II cations.
3.3
Ruthenium(II) Polypyridyl Complexes as Optical Anion Sensors
The spectroscopic and redox properties of {RuII (bpy)3 } have allowed this
metal complex to be used for combined optical and electrochemical sensing
of anions without the need for additional chromophores or luminophores.
For example Sessler’s complex 75 gives a UV-vis response to fluoride in

DMSO solution, with a stability constant for the receptor : fluoride complex
of 12 000 M–1 [34]. The new {RuII (bpy)3 }-pyrrole 76 has also been found to
selectively sense fluoride in DMSO solution by UV/vis (Ka = 7000 M–1 ) [53].
The luminescent emission of {RuII (bpy)3 } is also very useful for signalling
anion binding. In the emission spectra of 37–40 both a blue shift (of up to
16 nm for 40) and an increase in intensity of the λmax of the MLCT emis-
sion band was observed on addition of dihydrogenphosphate. It has been
proposed that the conformational flexibility of the receptors is decreased by
complexation of the anion guest thus reducing the rate of non-radiative decay
through vibrational and rotational relaxation. Similarly, macrocyclic com-
plexes 41–44 and 77–79 were also found to sense chloride by luminescence
enhancement.
In other examples anion binding can cause quenching of the {RuII (bpy)3 }
MLCT emission. In aqueous solution polyaza receptors 80-82 bind phosphate
and ATP anions, producing up to a 15% reduction in the emission intensity of
λmax at 605 nm [16]. Similarly, 76 shows up to a 40% reduction in the intensity
of the luminescent emission at 630 nm in the presence of dihydrogenphos-
phate in DMSO solution.
The RuII bipyridylcalix[4]diquinone receptor 83 selectively binds and
senses acetate anions (from 1 H NMR titrations in DMSO-d6 solution
K = 9990 M–1 ) [54]. This receptor is only weakly luminescent because the
{RuII (bpy)3 } MLCT emission is partially quenched by oxidative electron
transfer to the electron-poor calix[4]diquinone. Addition of acetate to ace-
tonitrile solutions of 83 resulted in a five-fold increase in luminescence
intensity (60% for chloride) concomitant with a slight blue shift of the emis-
sion maximum. Anion binding causes this increase in emission intensity
by interrupting the electron transfer pathway from the {RuII (bpy)3 } to the
calix[4]diquinone, thus reducing its quenching effect.
A similar effect is seen in macrocycle 36 which incorporates the
{RuII (bpy)3 } moiety with a bridging cobaltocenium unit [29]. In acetoni-
trile solution the quantum yield of the {RuII (bpy)3 } emission of this complex
is relatively low. However, in the presence of chloride a 100% increase in
emission intensity is observed.
Complexes 45 and 46 are capable of selectively sensing fluoride in ace-
tonitrile solution both by UV-vis and luminescence spectroscopy [33]. This
anion causes a dramatic reduction in the intensity of the MLCT absorption in
the 350–450 nm range with a new absorption appearing in the 500–650 nm
range ascribed to a ligand-based CT process. The emission spectra of the
complexes (exciting at 465 nm) show no significant peaks, indicating that the
characteristic Ru-centred luminescence is quenched by the dinitrophenylhy-
drazone group. Upon addition of fluoride a strong peak at 625 nm develops,
with quantum yields for the 45 : F– and 46 : F– adducts of 8.0 × 10–5 and
4.0 × 10–4 , respectively. Again it is apparent that anion binding interrupts the
non-radiative decay pathway.
Deetz and Smith have prepared a heteroditopic {RuII (bpy)3 } receptor 84

incorporating both amide and boronic acid groups which selectively senses
certain phosphorylated sugars in aqueous solution [55]. Boronic acids are
known to form covalent complexes with the diol groups of saccharides,
whereas the adjacent amides are positioned to complex the anionic phosphate
component. Sensing was accomplished by measuring luminescence enhance-
ment, with fructose-6-phosphate generating the highest stability constant
(log Ka = 3.1). Non-phosphorylated saccharides gave much smaller changes
in emission intensity (log Ka < 1.2), showing that the anionic component of
the guest is essential for strong binding. A covalent attachment between the
anion and the saccharide is not required. In the presence of sodium phos-
phate buffer non-phosphorylated saccharides are bound with similar strength
to their phosphorylated counterparts. It is reported that this apparent coop-
erativity is a result of favourable hydrogen bonding between the phosphate
anion and the saccharide. Watanabe and co-workers have also shown that
anionic and neutral phosphodiesters can be sensed in acetone by the imi-
dazole functionalised receptor 85 by both UV-vis and luminescence spectro-
scopies [56].
3.4
Optical Anion Sensing Using Reporter Groups
Based on Other Transition-Metal Complexes
Zn(II) porphyrins are another class of complex which can operate both as
UV/vis and luminescent sensors for anions. Example 86 is a picket-fence por-
phyrin with four imidazolium anion-binding groups [57]. In DMSO solution
this receptor undergoes slight (5 nm) bathochromic shifts in the Q-bands of
the absorption spectrum on addition of anions—indicative of the anion bind-
ing directly to the Zn centre. Using UV/vis titrations in DMSO solution 86 was
found to be selective for hydrogensulfate over chloride and dihydrogenphos-
phate. In water : DMSO (5 : 95) solution the receptor was found to be selective
for sulfate and hydrogensulfate over ATP2– and dihydrogenphosphate. Under
similar conditions the Q bands in the luminescence emission spectrum of 86
(λex = 424 nm) are also bathochromically shifted with a concomitant decrease
in intensity.
It should be noted that 86 also functions as an electrochemical anion
sensor, where the Zn(II) porphyrin-based oxidation potential is sensitive to
anion binding. In acetonitrile solution a greater cathodic shift was observed

with chloride (175 mV) than hydrogensulfate (140 mV) or nitrate (95 mV).
This demonstrates that the anion selectivity of sensor systems is dependent
upon the mode of detection used and underlines the fact that optical and
electrochemical anion sensing operate by different mechanisms.
Bis-terpyridine Iridium(III) has been used as an optical reporter group in
anion sensing. In aqueous solution isomers 87 and 88 both exhibit halide-
induced luminescence quenching with selectivity for chloride [58]. Receptor
87, although less sensitive than 88, is reported to possess good characteristics
for sensing chloride at physiologically relevant concentrations.
A related series of novel cyclometallated iridium(III) polypyridine thiourea
complexes also display anion-induced luminescence quenching. The repre-
sentative complex 89 was tested in acetonitrile solution with fluoride, acetate
and dihydrogenphosphate, and gave log K values for the 1 : 1 complex of 3.38,
4.03 and 3.14, respectively. This selectivity trend is ascribed to the combined
effect of the basicity and geometry of the guest anions [59].
Complexes of rhenium(I)tricarbonylchloride with pyridyl ligands are lu-
minophores and hence in anion sensing have principally been used as re-
porter groups. For instance calix[4]diquinone receptor 90 selectively binds
and senses acetate in DMSO solution [54] (from 1 H NMR titrations K =
1790 M–1 in DMSO-d6 solution). The receptor exhibits relatively weak lumi-
nescence because calix[4]diquinone is an electron acceptor, quenching the
Re(I) bipyridyl emission by oxidative electron transfer. Addition of anions to
DMSO solutions of 90 resulted in a significant increase in luminescence inten-
sity. It is clear that the presence of the anion in the binding pocket between
the {ReI (bpy)} moiety and the quencher interrupts the oxidative electron-
transfer process.
The mixed Re(I)/Pd(II) molecular square 91 has been found to sense
perchlorate in acetone (giving K = 900 M–1 ) by enhancement of the Re(I) lu-
minescent emission [60]. In this case luminescence quenching by oxidative
transfer to the Pd(II) ion is inhibited by the bound anion. The Pd(II) ion
also plays a role as a structural element and charge carrier. Squares 92–96
are very similar, but incorporate a bis-phosphinylferrocene supporting lig-
and [61]. Again the normally strong luminescence of the Re(I) component is
partially quenched by the bimetallic corners. Binding studies of the squares
with different inorganic anions were carried out by luminescence titrations
in acetone solution. Of the anions investigated, only hexafluorophosphate and
tetrafluoroborate induced significant changes in luminescence. As these an-
ions were added an initial decrease in emission intensity was followed by
an increase to a plateau. This is taken to indicate the presence of two com-
peting quenching pathways which are inhibited to different extents by anion
binding.
Lees and co-workers have investigated Re(I) bipyridyl anion hosts based
on aryl bisamide skeletons 97–99 [62]. Measurement of anion-induced lumi-
nescence quenching in CH2 Cl2 showed 97 to have strong binding affinities for
halides, acetate and cyanide, weaker affinity for dihydrogenphosphate, and
even less affinity for nitrate and perchlorate. The iso- and terephthalamide re-
ceptors 98 and 99 possess smaller stability constants for all the anions tested.
It is proposed that the anion-sensing efficiency of 98 is due to intramolecular
hydrogen-bonding of the amido NH proton to the pyridyl nitrogen holding
the receptor in a “cleft” conformation.
A metal-templated approach has been used by Thomas and co-workers to
produce the Re(I) metallomacrocycle 100 [63]. In acetonitrile solution this
receptor displayed luminescence enhancement on addition of anions. Stabil-
ity constants for the 1 : 1 adducts with BF4 – (1575 M–1 ), SO4 2– (7135 M–1 )
and BPh4 – (2895 M–1 ) were determined. Since SO4 2– is of similar size to
BF4 – the comparatively high affinity of this anion for 100 is thought to be
due to its additional charge. The slightly higher stability of the 100 : BPh4 –
complex compared to 100 : BF4 – is attributed to π–π interactions with the

receptor.
Interlocked supramolecular assemblies such as rotaxanes and catenanes
have the potential to provide tailor-made binding cavities for guest species.
An example of this is 101, a rotaxane formed using anion templated synthe-
sis which incorporates the Re(I) bipyridyl fragment [64]. Addition of chlo-
ride, hydrogensulfate or nitrate to the receptor in acetone solution caused
an enhancement of the fluorescence emission. Curiously, although the ro-
taxane was formed using chloride as the template, it was found to be se-
lective for hydrogensulfate with which a stability constant of > 106 M–1 was
determined.
3.5
Transition Metals as Anion-Binding Groups and/or Structural Components
Complexes of the late transition metals are well studied in anion sensing,
much of the work has been pioneered by Fabbrizzi and co-workers. As well
as providing an optical signalling function the Lewis acid metal ion can act
as a binding site for the anion. In addition the metal ion often forms an or-
ganisational unit designed to create a receptor of a specific shape. In order to
harness the metal-anion interaction for anion sensing the binding properties
of the metal must be modulated by an ancillary ligand. In this way one or two
vacant coordination sites can be made available for anion binding, and other
elements appended to allow signalling or modified selectivity.
In complexes with simple tripodal amines such as 102 and 103 zinc(II)
forms five-coordinate metal complexes of trigonal bipyramidal geometry,
leaving one of the axial coordination sites of the metal available for an-
ion binding. The Zn(II) complex of 102 was found to undergo quench-
ing (by photoinduced electron transfer) of the anthracene fluorescent
emission in the presence of aromatic carboxylate anions such as 4-N,N-
dimethylaminebenzoate in ethanol solution [65]. Complete quenching was
observed at the 1 : 1 anion to receptor ratio (log K = 5.45). Likewise the Zn(II)
complex of 103 was found to form 1 : 1 adducts with carboxylate anions in
methanol solution, with log K values ranging from 4 to 5 [66]. Only aromatic
carboxylates induced quenching of the ligand luminescence.
The bis(boradiazaindacene) substituted bipyridine ligand 104 is highly

fluorescent in organic solvents whereas its Zn(II) complex is not [67]. It
was found that the complex progressively regained its fluorescent emission
when it was titrated with various anions in acetonitrile. Stability constants
for the anion:receptor complexes were calculated for fluoride (4160 M–1 ),
chloride (3230 M–1 ), bromide (2500 M–1 ), acetate (4760 M–1 ), and phosphate
(4000 M–1 ). Decomplexation of the chelated Zn(II) ion from 104 by the weakly
coordinating anions was ruled out as the sensing mechanism. It is proposed
that the quenching of the ligand luminescence by electron transfer to the
Zn(II) centre is inhibited by anion coordination.
Fabbrizzi and co-workers have demonstrated the use of bis-copper(II)
cryptates to sense ambidentate anions [68]. On titrating molecule 105 with
NaN3 in aqueous solution the colour changed from pale blue to bright green
and an anion-metal LMCT absorption appeared at 400 nm. X-ray diffraction
studies have shown that the azide anion is held colinear with the two Cu(II)
centres, coordinated through the two terminal sp2 hybridised nitrogen atoms.
Stability constants for 105 with a variety of anions in aqueous solution were
calculated and the selectivity of this anion sensor for the azide anion was
found to be determined by the bite distance between the two copper atoms.
Cryptate 106, in which the aryl spacer of 107 is replaced with a furanyl
unit, acts as a colourimetric sensor for anions. UV/vis titrations in aqueous
solution gave log K values for the 1 : 1 halide/receptor adducts of 3.98 for chlo-
ride, 3.01 for bromide and 2.39 for iodide. X-ray diffraction studies confirm
that bromide is held between the two copper atoms. Under the same con-
ditions 106 also interacts strongly with azide (log K = 4.7) and thiocyanate
(log K = 4.28) anions. This receptor is interesting because of its lack of se-
lectivity compared to 105. The complex appears to be able to expand and
contract its bite-length in order to accommodate anions of various dimen-
sions.
The use of this class of receptors in practical applications is limited by
the small changes in UV-vis absorption which indicate anion binding. To
overcome this problem of sensitivity a chemosensing ensemble approach has
been applied. The fluorescent indicator coumarine 343 carries a carboxylate
group which allows it to be bound by 105 in a 1 : 1 complex (log K = 4.8) with
complete quenching of the luminescent emission [69]. Titration of a solution
containing 0.2 mM 105 and 0.1 µM coumarine 343 with hydrogencarbonate,
azide or cyanate anions resulted in complete recovery of the indicator lu-
minescence. Anions with a lower affinity for the receptor were unable to
displace the coumarine 343 and produced only a slight luminescence en-
hancement. The usefulness of this chemosensing ensemble was demonstrated
by the quantitative determination of carbonate in mineral water. Using the
same principle Han and Kim have recently reported a chemosensing ensem-
ble made up of the dizinc complex 107 and pyrocatechol violet which is
selective for phosphate [70].
Anslyn and co-workers have developed a series of tripodal Cu(II) com-
plexes 108 and 109 in which the metal ion and three cationic organic groups
form a tetrahedral cavity designed to host phosphate [71]. Receptor 108 is
built from the tris(2-ethylamino)amine skeleton with appended benzylamine
groups. UV/vis anion titrations were carried out in aqueous solution at pH 7.4
where it can be assumed the terminal amines are all protonated. Hydrogen-
phosphate and its congener hydrogenarsenate were found to bind strongly
in a 1 : 1 anion/host ratio, both with a log K value of 4.40. Perrhenate was
bound an order of magnitude less strongly and the affinity for chloride was
too small to measure. Model complex 110, which has no ammonium groups,
gave a log K value of 2.95 with hydrogenphosphate, indicating that the Cu(II)-
anion interaction contributes significantly to anion binding. Receptor 109 fol-
lows the same design principle, this time incorporating guanidinium binding
units with a tris-[(2-pyridyl)methyl]amine skeleton. log K values for hydro-
genphosphate and hydrogenarsenate were found to be 4.18 and 4.23, respec-
tively. Other anions, including perrhenate had no significant affinity for 109.
It is apparent that the guanidinium groups are responsible for the improved
selectivity of this receptor for phosphate. In a separate study the driving force
for hydrogenphosphate binding was found to be entropic for receptor 108, but
enthalpic for receptor 109 [72]. Partnered with the colourimetric indicator
5-(and 6)-carboxyfluorescein receptor 109 provides an effective chemosens-
ing ensemble for the determination of inorganic phosphate in serum and
saliva [73].
Zinc has been used as a binding site for the detection of pyrophosphate
in aqueous solution by fluorescence. Complex 111 couples two tridentate Zn
centres to a fluorescent naphthalenediimide core. In HEPES buffer the ap-
pearance of a new emission band at 490 nm was observed on addition of
pyrophosphate, attributed to naphthalenediimide excimer formation. This
did not occur with other anions including halides, acetate and ADP. It is pro-
posed that 111 binds pyrophosphate in a 2 + 2 complex, which brings two
naphthalenediimides in close enough proximity to give the excimer emis-
sion [74].
The recent tripodal Cu(II) complex 112 has intriguing optical anion-
sensing properties [75]. This receptor has a cavity with two distinct anion-
binding sites—the vacant site on the Lewis acidic Cu(II) centre, and the
three favourably arranged nitrophenylurea fragments. On titration with up
to one equivalent of azide or dihydrogen phosphate in DMSO solution the

Cu(II) d–d bands in the region 600–900 nm increased markedly in intensity.
This indicates that the anion is binding at the metal centre. Upon addition
of a second equivalent of the anion the absorption associated with the nitro-
phenylurea groups (below 500 nm) increased in intensity, showing that these
hydrogen-bonding units are involved in binding the second anion. On their

own halide anions were found to give only 1 : 1 complexes (with the halide
bound at the Cu(II) centre). However, on titration of dihydrogenphosphate
into a solution of the receptor pre-saturated with chloride (i.e. dissolved in
a solution containing a 150-fold excess of chloride) formation of a 1 : 1 adduct
of the [112 : Cl] complex with dihydrogenphosphate was observed. The stabil-
ity constant of this species was found to be approximately 700 times higher
than the stability constant of the analogous [112 : H2 PO4 ] complex with a sec-
ond equivalent of dihydrogenphosphate.
Stepwise anion coordination equilibria are also observed in the Cu(II)
complexes of ligands 113 and 114 [76]. UV/vis titrations in acetonitrile so-
lution show that each Cu(II) complex binds two anions (chloride, bromide,
iodide, nitrate or thiocyanate), the first at the Cu(II) centre and the second
in the bis-imidazolium compartment. The Cu(I) complexes of these ligands
are able to host only one nitrate anion (in the bis-imidazolium cavity), while
other anions induce demetallation. Cyclic voltammetry and spectroelectro-
chemical experiments showed that in the presence of one equivalent of nitrate
the Cu(II)/Cu(I) redox change causes the anion to translocate quickly and
reversibly from the metal-based binding site in the Cu(II) complex to the im-
idazolium binding site in the Cu(I) system.
Another noteworthy example in which Cu(I) forms the basis of an opti-
cal anion sensor is 115, in which the metal complex acts both as a UV/vis
signalling group and as a structural component dictating the topology of the
urea anion-binding site [77]. The MLCT band within the CuI (phenanthroline)
complex at 282 nm is sensitive to halide ions, acetate and dihydrogenphos-
phate in 4 : 1 v/v THF/MeCN (a relatively low polarity solvent). However, in
DMSO solution, only acetate and dihydrogenphosphate produced a UV/vis re-
sponse. Crucially, dihydrogenphosphate was found to bind the receptor with

1 : 1 stoichiometry, whereas acetate bound 2 : 1 anion : receptor. It is pro-
posed that dihydrogenphosphate is able to bind simultaneously to the two
urea groups to form an overall helical structure. This is not the case with
the free ligand—the CuI centre is required to template the formation of the
binding cavity, and augments dihydrogenphosphate binding. Acetate can only
hydrogen-bond to one urea subunit at a time, and the affinity of 115 for this
anion is only slightly higher than that of the free ligand (presumably due to
electrostatic effects).
Cadmium(II), the heavier congener of zinc(II) can also act as a coor-
dination site for anion binding. Mizukami et al. employ a novel approach
Fig. 2 X-ray crystal structure of bromide encapsulated in the Fe : 117 complex. Repro-
duced with permission from [79]. © Wiley-VCH Verlag GmbH & Co. KGaA
with receptor 116, by covalently attaching the coumarin indicator group of

a chemosensing ensemble to a macrocyclic Cd(II) anion receptor [78]. It
is designed such that anions will displace the nitrogen of the aromatic flu-
orophore from the coordination sphere of the Cd(II) centre. In aqueous
solution as the receptor was titrated with pyrophosphate the luminescent
emission of the molecule was observed to shift gradually from 342 nm to
383 nm. Receptor 116 shows a high degree of selectivity for pyrophosphate,
citrate, ATP and ADP with log K’s in the range 4–5. It is worthy of note that
the Zn(II) analogue of 116 was found to be ineffective for anion sensing.
Iron(II) has been used as a supramolecular template for the formation
of a tris-imidazolium receptor from ligand 117 [79]. 1 H NMR studies and
X-ray crystal structure determination were used to demonstrate the encap-
sulation of bromide in the cavity of the receptor, with the anion coordi-
nated by six C–H fragments (Fig. 2). Spectrophotometric titrations in ace-
tonitrile solution revealed that this receptor binds halides with selectivity
for chloride > bromide > iodide, but has no affinity for dihydrogenphosphate
or hydrogensulfate. Presumably the restricted size of the receptor cavity ex-
cludes the binding of these larger tetrahedral anions. The linear anions azide,
cyanate and thiocyanate also produced a response in the UV/vis spectrum,
and azide was found to bind preferentially to 117 in comparison to the non-
symmetrical linear anions.
3.6
Optical Anions Sensing by Lanthanide(III) Complexes
Given their favourable luminescence properties and propensity to coordinate

oxy-anions it is perhaps surprising that complexes of lanthanide(III) ions
have received comparatively little attention as anion sensors.
Parker and co-workers have reported a novel method for the selective de-
tection of carboxy anions by time-delayed luminescence using the Eu(III)
and Tb(III) complexes of 118 [80]. Luminescence measurements on the co-
ordinatively unsaturated complexes in aqueous solution showed significant
increases in lifetime and emission intensity in the presence of anions. This be-
haviour is consistent with the anions displacing water from the non-ligated
coordination sites at the metal centre. Studies allowed the number of water
molecules (q) remaining coordinated in the presence of added anions to be
estimated. For the triflate salt of the Eu(III) receptor q = 2.14 whereas with
hydrogencarbonate q = 0.34 and with hydrogenphosphate q = 0.74. The selec-
tivity for hydrogencarbonate reflects the ability of this anion to chelate the
Eu(III) centre whereas hydrogenphosphate prefers to bind in a monodentate
fashion. This behaviour is also pH-dependent (pKa HCO3 – /H2 CO3 – = 6.38) so
that for a given starting bicarbonate concentration, a pH-dependent lifetime
and emission intensity was observed [81].
Other groups have subsequently reported anion receptors that work on the
same principle. For instance an Eu(III) complex of the bis-bipyridinephenyl-
phosphine oxide ligand 119 made by Ziessel and co-workers is able to sense
anions by luminescence enhancement in acetonitrile with stability constants
which follow the trend fluoride > acetate > chloride > nitrate [82]. Tsukube
and co-workers have investigated the properties of the Eu(III) and Tb(III)
complexes of the chiral ligand 120 [83]. Anion binding was assessed by profil-
ing luminescence enhancement in acetonitrile and it was found that the dif-
ferent metal centres provided different selectivities. The emission at 548 nm
of the Tb(III) complex was increased by 5.5 times in the presence of 3 equiva-
lents of chloride compared to 2.2 for nitrate and 1.1 for acetate. Conversely the
emission at 618 nm of the Eu(III) complex was increased 8.3 times by 3 equiv-
alents of nitrate, 2.5 times for chloride and 1.0 times for acetate. Stability
constants were not reported.
The same group most recently reported the use of neutral lanthanide(III)
tris-diketonates of type 121 for the determination of chloride [84]. The
response in luminescence of the Eu(III) complex for chloride in acetoni-
trile solution was large enough to be seen by the naked eye. Incorpora-
tion of the complexes in PVC membrane electrodes allowed measurement
of potentiometric selectivity coefficients. These showed the Eu(III) complex
to be the more selective for chloride than the Pr(III), Dy(III) or Yb(III)
analogues.
Gunnlaugsson and co-workers have developed an anion sensor based on
a ternary europium complex [85]. In 122 the naphthalene β-diketonato ligand
acts as a sensitising group for Eu(III) emission. Displacement of this antenna
group from 122 by competitor anionic species would therefore be expected
to decrease the intensity of this emission. In aqueous solution at pH 7.4 it
was found that iodide and dihydrogenphosphate reduced the intensity of the
emission at 616 nm by 20–40%. More pronounced changes, which could be
seen with the naked eye, were observed with highly competitive anions such
as tartarate and fluoride.
Gunnlaugsson and co-workers have also studied di-europium(III) com-
plex 123, which incorporates two different macrocyclic ligands for Eu(III) in
addition to a covalently bound antenna group [86]. Upon titrating 123 with
acetate, aspartate and succinate at pH 6.5, each of the Eu(III) emission bands
was quenched by up to 50% for acetate and aspartate. Malonate, however, pro-
duced a nearly two-fold enhancement in the emission intensity. It is thought
that this particular anion is able to displace coordinated water molecules from
both the Eu(III) centres thereby reducing the rate of non-radiative energy
transfer.
3.7
Surface Confined Systems for Optical Anion Sensing
Beer and co-workers have developed a surface-enhanced optical anion sensor

based on gold nanoparticles [87]. Dodecanethiol stabilised gold nanoparti-
cles were modified by ligand substitution with a disulfide-substituted zinc
porphyrin 124 to provide 30 and 80 receptors per nanoparticle. Titration of
both the free receptor and the modified nanoparticles with various anions
in dichloromethane or DMSO solution revealed significant changes in the in-
tense porphyrin absorption bands. Calculated stability constants are given in
Table 1 and reveal highly enhanced anion-binding affinities (up to two orders
of magnitude with chloride and dihydrogenphosphate in DMSO solution) for
Table 1 Association constant (log K) data of 124 and 124 modified nanoparticles in DMSO
solution determined at 293 K, errors ±0.1
Anion 124 124-nanoparticles a
Cl– <2 4.3

H2 PO4 – 2.5 4.1
a Association constant values for the 1 : 1 porphyrin–anion complex on the nanoparticle
surface
the surface-bound porphyrin receptor with respect to the free metallopor-

phyrin.
4
Metal-Based Anion Receptors Without Reporter Groups
Described below are a number of interesting examples from the literature of

metal-based anion receptors for which anion binding has only been studied
by NMR.
If one of the cyclopentadienyl rings of ferrocene is replaced with an arene
moiety a cationic complex is generated, thereby gaining the potential for elec-
trostatic interaction with anions. Beer and co-workers have exploited this
principle in the simple receptor 125 [88]. Similarly, Atwood and co-workers
have derivatised cyclotriveratrylene to generate receptor 126 [89]. Qualitative
1 H NMR studies demonstrate that 125 binds chloride and bromide in CH CN,
3
and 126 binds halides in acetone. It is proposed that the strength of the in-
teraction in 126, which does not possess amides or other hydrogen-bonding
groups, is due to the arrangement of positive charges around the upper-rim

of the hydrophobic cavity.
The anion-binding properties of a series of highly cationic metal-
lated calix[4]arenes have also been investigated. Host 127, in which the
calix[4]arene is coordinated to four Ru(η6 -p-cymene) units was found by
1 H NMR titration to bind chloride, bromide, iodide and nitrate in aqueous
solution [90]. Stability constants were determined with Ka = 551, 133, 51, and
49 M–1 , respectively.
Cr(CO)3 is an effective electron-withdrawing group [91]. The coordination
of this organometallic unit to the arene substituents of the isophthalamide in
128 increases the acidity of the NH protons and hence their affinity for an-
ions. 1 H NMR titrations in CD3 CN solution provided stability constants of the
same order of magnitude as those for the non-metallated receptor in the far
less competitive solvent CD2 Cl2 .
Loeb, Gale and co-workers have used Pt(II) as a structural template for the
self-assembly of a series of anion receptors. Receptors 129–132 were found
to bind a variety of anions in DMSO-d6 solution [92, 93]. Even the simple
tetrapyridine receptor 129 has an affinity for anions due to its electrostatic
charge. The addition of pyrrole hydrogen-bond donors increases the stabil-
ity of the receptor : anion complexes by more than five-fold in 130, but in 131
they are poorly orientated for anion coordination and the stability constants
are only marginally higher than in 129. Receptor 132 which uses urea groups
as hydrogen-bond donors binds anions with 1 : 2 receptor : anion stoichiom-
etry, with stability constants an order of magnitude higher than 131.
5
Conclusion/Outlook
The geometric, Lewis acidic, optical and electrochemical properties of metals

make them ideal for use as multifunctional components in the construction
of anion receptors. They are able to combine roles such as reporter groups,
anion-binding sites and structural components to form receptor systems with
sophisticated anion-sensing properties. In particular metal-containing recep-
tors form the basis of systems in which anion-binding can be switched on and
off, or in which the anion guest can even be moved from one binding site to
another. Another significant advance has been the development of nanoscale
structures such as dendrimers, nanoparticles, thin polymer films and self-
assembled monolayers which incorporate redox- or photoactive metal centres
as an anion-sensing component. The wide variety of added functionality and
sheer adaptability that metals bring to anion receptor chemistry means that
metal-based systems are set to continue at the forefront of research in this
area.
References
1. Beer PD, Cadman J (2000) Coord Chem Rev 205:131
2. Beer PD, Gale PA (2001) Angew Chem Int Ed 40:486
3. Beer PD, Hayes EJ (2003) Coord Chem Rev 240:167
4. Sun S-S, Lees AJ (2002) Coord Chem Rev 230:171
5. Amendola V, Fabbrizzi L, Mangano C, Pallavicini P, Poggi A, Taglietti A (2001) Coord
Chem Rev 219:821
6. Bayly SR, Beer PD (2005) Anion Sensing. Topics Curr Chem 255:125
7. Beer PD, Gale PA, Chen GZ (1999) J Chem Soc Dalton Trans, p 1897
8. Beer PD, Keefe AD, Drew MGB (1989) J Organomet Chem 378:437
9. Beer PD (1996) Chem Commun, p 689
10. Beer PD, Hesek D, Kingston JE, Smith DK, Stokes SE, Drew MGB (1995) Organo-
metallics 14:3288
11. Beer PD, Graydon AR, Johnson AOM, Smith DK (1997) Inorg Chem 36:2112
12. Beer PD, Chen Z, Goulden AJ, Graydon A, Stokes SE, Wear T (1993) J Chem Soc Chem
Commun, p 1834
13. Reynes O, Moutet J-C, Pecaut J, Royal G, Saint-Aman E (2002) New J Chem 26:9
14. Beer PD, Chen Z, Drew MGB, Kingston J, Ogden M, Spencer P (1993) J Chem Soc
Chem Commun, p 1046
15. Beer PD, Chen Z, Drew MGB, Johnson AOM, Smith DK, Spencer P (1996) Inorg Chim
Acta 246:143
16. Beer PD, Cadman J, Lloris JM, Martinez-Manez R, Padilla ME, Pardo T, Smith DK,
Soto J (1999) J Chem Soc Dalton Trans, p 127
17. Dusemund C, Sandanayake KRAS, Shinkai S (1995) J Chem Soc Chem Commun,
p 333
18. Reynes O, Maillard F, Moutet J-C, Royal G, Saint-Aman E, Stanciu G, Dutasta J-P,
Gosse I, Mulatier J-C (2001) J Organomet Chem 637-639:356
19. Beer PD, Drew MGB, Hesek D, Jagessar R (1995) J Chem Soc Chem Commun, p 1187
20. Beer PD, Hesek D, Nam KC, Drew MGB (1999) Organometallics 18:3933
21. Tomapatanaget B, Tuntulani T, Chailapakul O (2003) Org Lett 5:1539
22. Kim DS, Miyaji H, Chang BY, Park SM, Ahn KH (2006) Chem Commun, p 3314
23. Cui XL, Delgado R, Carapuca HM, Drew MGB, Felix V (2005) Dalton Trans, p 3297
24. Beer PD, Stokes SE (1995) Polyhedron 14:873
25. Oton F, Tarraga A, Espinosa A, Velasco MD, Molina P (2006) Dalton Trans, p 3685
26. Kingston JE, Ashford L, Beer PD, Drew MGB (1999) J Chem Soc Dalton Trans, p 251
27. Beer PD, Drew MGB, Jagessar R (1997) J Chem Soc Dalton Trans, p 881
28. Altmann R, Gausset O, Horn D, Jurkschat K, Schuermann M, Fontani M, Zanello P
(2000) Organometallics 19:430
29. Beer PD, Szemes F, Balzani V, Sala CM, Drew MGB, Dent SW, Maestri M (1997) J Am
Chem Soc 119:11864
30. Szemes F, Hesek D, Chen Z, Dent SW, Drew MGB, Goulden AJ, Graydon AR, Grieve A,
Mortimer RJ, Wear T, Weightman JS, Beer PD (1996) Inorg Chem 35:5868
31. Beer PD (1998) Acc Chem Res 31:71
32. Beer PD, Dent SW, Wear TJ (1996) J Chem Soc Dalton Trans, p 2341
33. Lin ZH, Zhao YG, Duan CY, Zhang BG, Bai ZP (2006) Dalton Trans, p 3678
34. Mizuno T, Wei WH, Eller LR, Sessler JL (2002) J Am Chem Soc 124:1134
35. Beer PD, Berry N, Drew MGB, Fox OD, Padilla-Tosta ME, Patell S (2001) Chem Com-
mun, p 199
36. Berry NG, Pratt MD, Fox OD, Beer PD (2001) Supramol Chem 13:677
37. Beer PD, Berry NG, Cowley AR, Hayes EJ, Oates EC, Wong WWH (2003) Chem Com-
mun, p 2408
38. Valerio C, Fillaut J-L, Ruiz J, Guittard J, Blais J-C, Astruc D (1997) J Am Chem Soc
119:2588
39. Alonso B, Casado CM, Cuadrado I, Moran M, Kaifer AE (2002) Chem Commun,
p 1778
40. Daniel M-C, Ruiz J, Blais J-C, Daro N, Astruc D (2003) Chem Eur J 9:4371
41. Beer PD, Davis Jason J, Drillsma-Milgrom DA, Szemes F (2002) Chem Com-
mun, p 1716
42. del Peso I, Alonso B, Lobete F, Casado CM, Cuadrado I, Losada del Barrio J (2002)
Inorg Chem Commun 5:288
43. Reynes O, Gulon T, Moutet J-C, Royal G, Saint-Aman E (2002) J Organomet Chem
656:116
44. Reynes O, Bucher C, Moutet J-C, Royal G, Saint-Aman E (2004) Chem Commun, p 428
45. Labande A, Astruc D (2000) Chem Commun, p 1007
46. Bayly SR, Gray TM, Chmielewski MJ, Davis JJ, Beer PD (2007) Chem Commun, p 2234
47. Basurto S, Riant O, Moreno D, Rojo J, Torroba T (2007) J Org Chem 72:4673
48. Miyaji H, Collinson SR, Prokes I, Tucker JHR (2003) Chem Commun, p 64
49. Aldridge S, Bresner C, Fallis IA, Coles SJ, Hursthouse MB (2002) Chem Commun,
p 740
50. Zhang BG, Xu J, Zhao YG, Duan CY, Cao X, Meng QJ (2006) Dalton Trans, p 1271
51. Oton F, Tarraga A, Espinosa A, Velasco MD, Molina P (2006) J Org Chem 71:4590
52. Oton F, Tarraga A, Molina P (2006) Org Lett 8:2107
53. Plitt P, Gross DE, Lynch VM, Sessler JL (2007) Chem Eur J 13:1374
54. Beer PD, Timoshenko V, Maestri M, Passaniti P, Balzani V (1999) Chem Commun,
p 1755
55. Deetz MJ, Smith BD (1998) Tetrahedron Lett 39:6841
56. Watanabe S, Onogawa O, Komatsu Y, Yoshida K (1998) J Am Chem Soc 120:229
57. Cormode DP, Murray SS, Cowley AR, Beer PD (2006) Dalton Trans, p 5135
58. Goodall W, Williams JAG (2000) J Chem Soc Dalton Trans, p 2893
59. Lo KKW, Lau JSY, Lo DKK, Lo LTL (2006) Eur J Inorg Chem, p 4054
60. Slone RV, Yoon DI, Calhoun RM, Hupp JT (1995) J Am Chem Soc 117:11813
61. Sun S-S, Anspach JA, Lees AJ, Zavalij PY (2002) Organometallics 21:685
62. Sun S-S, Lees AJ, Zavalij PY (2003) Inorg Chem 42:3445
63. de Wolf P, Waywell P, Hanson M, Heath SL, Meijer A, Teat SJ, Thomas JA (2006) Chem
Eur J 12:2188
64. Curiel D, Beer PD (2005) Chem Commun, p 1909
65. DeSantis G, Fabbrizzi L, Licchelli M, Poggi A, Taglietti A (1996) Angew Chem Int Ed
Engl 35:202
66. Fabbrizzi L, Licchelli M, Parodi L, Poggi A, Taglietti A (1999) Eur J Inorg Chem, p 35
67. Coskun A, Baytekin BT, Akkaya EU (2003) Tetrahedron Lett 44:5649
68. Fabbrizzi L, Licchelli M, Rabaioli G, Taglietti A (2000) Coord Chem Rev 205:85
69. Fabbrizzi L, Leone A, Taglietti A (2001) Angew Chem Int Ed 40:3066
70. Han MS, Kim DH (2002) Angew Chem Int Ed 41:3809
71. Tobey SL, Jones BD, Anslyn EV (2003) J Am Chem Soc 125:4026
72. Tobey SL, Anslyn EV (2003) J Am Chem Soc 125:14807
73. Tobey SL, Anslyn EV (2003) Org Lett 5:2029
74. Lee HN (2007) J Am Chem Soc 129:3828
75. Allevi M, Bonizzoni M, Fabbrizzi L (2007) Chem Eur J 13:3787
76. Amendola V, Colasson B, Fabbrizzi L, Douton MJR (2007) Chem Eur J 13:4988
77. Amendola V, Boiocchi M, Colasson B, Fabbrizzi L (2006) Inorg Chem 45:6138
78. Mizukami S, Nagano T, Urano Y, Odani A, Kikuchi K (2002) J Am Chem Soc 124:3920
79. Amendola V, Boiocchi M, Colasson B, Fabbrizzi L, Douton MJR, Ugozzoli F (2006)
Angew Chem Int Ed 45:6920
80. Dickins RS, Gunnlaugsson T, Parker D, Peacock RD (1998) Chem Commun, p 1643
81. Bruce JI, Dickins RS, Govenlock LJ, Gunnlaugsson T, Lopinski S, Lowe MP, Parker D,
Peacock RD, Perry JJB, Aime S, Botta M (2000) J Am Chem Soc 122:9674
82. Montalti M, Prodi L, Zaccheroni N, Charbonniere L, Douce L, Ziessel R (2001) J Am

Chem Soc 123:12694
83. Yamada T, Shinoda S, Tsukube H (2002) Chem Commun, p 1218
84. Mahajan RK, Kaur I, Kaur R, Uchida S, Onimaru A, Shinoda S, Tsukube H (2003)
Chem Commun, p 2238
85. Leonard JP, dos Santos CMG, Plush SE, McCabe T, Gunnlaugsson T (2007) Chem
Commun, p 129
86. Plush SE, Gunnlaugsson T (2007) Org Lett 9:1919
87. Beer PD, Cormode DP, Davis JJ (2004) Chem Commun, p 414
88. Beer PD, Dickson CAP, Fletcher N, Goulden AJ, Grieve A, Hodacova J, Wear T (1993)
J Chem Soc Chem Commun, p 828
89. Holman KT, Orr GW, Atwood JL, Steed JW (1998) Chem Commun, p 2109
90. Staffilani M, Hancock KSB, Steed JW, Holman KT, Atwood JL, Juneja RK, Burkhal-
ter RS (1997) J Am Chem Soc 119:6324
91. Camiolo S, Coles SJ, Gale PA, Hursthouse MB, Mayer TA, Paver MA (2000) Chem
Commun, p 275
92. Bondy CR, Gale PA, Loeb SJ (2004) J Am Chem Soc 126:5030
93. Vega IE, Gale PA, Light ME, Loeb SJ (2005) Chem Commun, p 4913
Struct Bond (2008) 129: 95–125
DOI 10.1007/430_2007_072
Recent Progress of Phosphate Derivatives Recognition

Utilizing Artificial Small Molecular Receptors
in Aqueous Media
Shun-ichi Tamaru1 (u) · Itaru Hamachi2
1 Department of Nano-science, Sojo University,
4-22-1 Ikeda, 860-0082 Kumamoto, Japan
stamaru@nano.sojo-u.ac.jp
2 Department of Synthetic Chemistry and Biological Chemistry, Kyoto University,
Kyotodaigaku Katsura, Nishikyo-ku, 615-8510 Kyoto, Japan
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
2 Phosphate Recognition by Electrostatic Interactions and Hydrogen Bonds 98

2.1 Cyclic/Acyclic Polyamine Type Artificial Chemosensors . . . . . . . . . . . 98
2.2 Artificial Chemosensors Containing Other Cationic Groups . . . . . . . . . 102
3 Phosphate Anion Recognition Utilizing Coordination Chemistry . . . . . 104

3.1 Zn-cyclen Type Artificial Chemosensors . . . . . . . . . . . . . . . . . . . . 105
3.2 Zn-Dpa Type Artificial Chemosensors . . . . . . . . . . . . . . . . . . . . . 106
3.3 Ratiometric Detection of Phosphate Derivatives . . . . . . . . . . . . . . . 109
4 Recognition of Phosphorylated Protein Surfaces . . . . . . . . . . . . . . . 112
5 High Throughput Sensing Systems for Phosphate Derivatives

Using Semi-wet Sensor Array . . . . . . . . . . . . . . . . . . . . . . . . . 119
5.1 Molecular Recognition of Chemosensors in a Supramolecular Hydrogel . . 119
5.2 Use of Hydrophobic Micro-domains of Supramolecular Hydrogel Fibers
for Discrimination of Phosphate Derivatives . . . . . . . . . . . . . . . . . 120
6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
Abstract Phosphate derivatives are ubiquitous in living organisms playing several import-
ant roles. Therefore, the development of sophisticated artificial phosphate receptors and
chemosensors that can work in aqueous conditions is currently an area of great inter-
est. There is a need to develop new methodologies for detection, separation or transport
of biologically relevant phosphate derivatives. In the past two decades, many artificial
chemosensors have been reported; these can be broadly classified into (1) chemosensors
utilizing electrostatic interactions and hydrogen bonds and (2) chemosensors utilizing co-
ordination chemistry. The development of these receptors is often inspired by the design
of substrate-binding centers in natural enzymes such as protein kinases, phosphatases
and phospholipases. More recently, the targets of such chemosensors have been broad-
ened to include the recognition of phosphorylated protein surfaces. Here we review the
recent progress in the development of molecular receptors and chemosensors that can
selectively detect phosphate derivatives in aqueous media.
96 S.-i. Tamaru · I. Hamachi
Keywords Phosphate · Molecular recognition · Phosphoprotein · Host-guest ·

Fluorescent sensing
Abbreviations
ADP adenosine 5 -diphosphate
AMP adenosine 5 -monophosphate
ATP adenosine 5 -triphosphate
AXP adenosine nucleotide
cAMP adenosine 3 ,5 -cyclic monophosphate
CD circular dichroism
CTP cytosine 5 -triphosphate
CoA coenzyme A
5-CF 5-carboxylfluorescein
diMeP dimethyl phosphate
GTP guanosine 5 -triphosphate
IP3 d-myoinositol 1,4,5-triphosphate
MAPK mitogen-activated protein kinase
MeP methyl phosphate
MLCT metal-ligand charge transfer
NPP 4-nitrophenyl phosphate
PDGFR-β platelet-derived grows factor receptor-β
PET photo-induced electron transfer
PhP phenyl phosphate
PPi pyrophosphate
PV pyrocatechol violet
p-Tyr O-phospho-L-tyrosine
UDP-Gal uridine 5 -diphospho-α-d-galactose
Zn-Dpa zinc(II)-dipicolylamine
1
Introduction
In nature, a variety of phosphate anion derivatives are ubiquitously dis-

tributed [1]. In addition to inorganic phosphate, there are many phosphates
found in biological molecules. For instance, the backbones of DNA and RNA
consist of phosphate diesters connecting 3 - and 5 -oxygens of ribose. Nu-
cleoside phosphates (nucleotides) are not only the structural units of RNA
and DNA, but also the essential parts of several cofactors such as coen-
zyme A (CoA), flavin adenine dinucleotide (FAD), and nicotinamide ade-
nine dinucleotide (NAD). They play important roles in metabolic cascades
of biological substances and in cellular signaling. Adenosine triphosphate
(ATP) is a primary energy currency molecule in all living organisms, which
is produced from ADP and inorganic phosphate by ATPase in cellular res-
piration or photosynthesis. The covalent bond connecting the phosphate
groups stores a high energy that is easily broken down by hydrolysis to re-
Phosphate Recognition Utilizing Artificial Small Molecular Receptors 97
lease an amount of energy sufficient to drive many biochemical reactions.

For example, the nucleoside pyrophosphate derivatives of saccharides such
as uridine 5 -diphospho-α-D-galactose (UDP-Gal) are universal substrates of
glycosyltransferase necessary to form a glycosyl bond for the synthesis of
glycoprotein and oligo-saccharides. Unlike the molecular elements in bioen-
ergetics, adenosine 3 ,5 -cyclic monophosphate (cAMP) and D-myoinositol
1,4,5-triphosphate (IP3) represent important second messengers playing cen-
tral roles in intra-cellular signal transduction [2, 3].
Phosphorylated proteins are other important phosphate derivatives in bio-
logical phenomena, which are produced by protein kinases, a family of en-
zymes catalyzing phosphorylation. Protein phosphorylation is one of the
most common post-translational protein modifications in eucaryotic cells.
The vast majority of phosphorylation occurs as a mechanism to regulate the
biological activity of a protein [4]. From the genomic sequence, the presence
of over 500 kinds of protein kinase in a human body is forecasted [5]. This
fact suggests that such post-translational CENOTE phosphorylation is widely
used to control diverse protein functions. In particular, phosphorylation on
serine, threonine or tyrosine residues by protein kinases is fundamental to
cell signaling networks. In these networks protein functions are controlled
via the reversible phosphorylation/dephosphorylation processes catalyzed by
corresponding kinase/phosphatase coupling with spatio-temporal fashions in
response to changes of the cellular environment [6]. In practice, the phos-
phorylation acts as a molecular switch to turn on the protein activity. A typ-
ical example of such allosteric control of protein functions is observed in
mitogen-activated protein kinase (MAPK) ERK2. In the deactivated state, two
domains of ERK2 (extracellular signal-regulated kinase 2) are separated from
each other by the loop strand. The structural consequences of phosphoryla-
tion at threonine (Thr 183) and tyrosine (Tyr 185) residues on the loop strand
include active site closure, alignment of key catalytic residues that interact
with ATP, and remodeling of the activation loop, resulting in recovery of the
kinase activity [7]. Protein phosphorylation also plays a crucial role in con-
trolling the reversible assembly of a cellular signaling complex [8, 9]. Phos-
phorylation of the appropriate peptide sequence generates new interaction
sites exposed on a protein surface which are recognized by the downstream
proteins containing phosphoprotein-binding domains, forming multibinding
domains such as SH2, PTB, FHA, and WW domains involved in many signal-
ing proteins [10].
Due to such significant roles of phosphate derivatives in biological events,
it is generally anticipated that development of sophisticated artificial recep-
tors toward phosphates would greatly contribute to the understanding of
biological phenomenon involving phosphate derivatives. Such advancement
in understanding should afford new methodologies for detection, separa-
tion, or transport of biologically important phosphates. In the following we
review the recent progress of molecular recognition of phosphate deriva-
tives utilizing artificial small molecular receptors that can work in aqueous
media.
2
Phosphate Recognition by Electrostatic Interactions and Hydrogen Bonds
2.1
Cyclic/Acyclic Polyamine Type Artificial Chemosensors
Polyamines that exist as polycationic species in water under appropriate pH

conditions are suitable scaffolds to capture phosphate derivatives by the for-
mation of multiple-hydrogen bonding and electrostatic interaction. It has
previously been shown that naturally occurring polyamines such as spermine
and spermidine bind phosphates or DNA [11]. Nakai et al. reported that the
apparent affinity constants, logKapp for the 1 : 1 complexation of spermine
with AMP, ADP, and ATP were found to be 2.6, 3.1 and 4.0, respectively, in
50 mM tris buffer (pH 7.5) [12]. Based on this moderate affinity, several ar-
tificial phosphate receptors containing polyamines were developed. Because
of the difficulty of capturing phosphate anions by non-covalent interaction
in aqueous media, however, successful examples of artificial small molecular
receptors or sensors for phosphates are still relatively scarce.
In a pioneering work, Czarnik and co-workers synthesized the branch-

ing polyamine-appended anthracene derivative as a fluorescent chemosensor
for phosphate anions. Compound 1, having a branching polyamine chain on
the 9-position of the anthracene moiety, formed a 1 : 1 complex with inor-
ganic phosphate or ATP in aqueous solution at pH 6 (logKeq = 0.82 for PO4 2– ,
4.2 for ATP) [13]. The complexation caused a fluorescence enhancement that
can be ascribed to the cancellation of quenching due to photo-induced elec-
tron transfer (PET), as shown in Scheme 1. In the absence of phosphate ions,
1 showed weak fluorescence due to the PET quenching by the free amine
group. The complexation of the anionic phosphate oxygens with ammonium
ions on 1 localized the remaining phosphate OH group in close proximity
to the free amine, and then favorable intracomplex proton transfer occurred.
As a result, the intramolecular PET quenching was eliminated and stronger
emission was observed.
Scheme 1 Proposed binding mechanism of Chemosensor 1 with inorganic phosphate.

The PET quenching is canceled upon the phosphate binding with 1 causing fluorescence
enhancement
The Czarnik group then prepared compound 2, which has two polyamine
chains at the 1- and 8-positions of anthracene, as a fluorescence chemosen-
sor toward pyrophosphate (PPi) [14]. To effectively bind both phosphates
of PPi simultaneously, the polyamine chains were geometrically preorga-
nized. The dissociation constant (Kd ) for the 2-PPi complex was found to be
2.9 µM, whereas that for 2-phosphate was 6.3 mM. In particular, chemosen-
sor 2 displayed a 2200-fold pyrophosphate/phosphate discrimination at pH 7
(Scheme 2). Such high ion selectivity allows a real-time assay of pyrophos-
phate hydrolysis catalyzed by inorganic phosphatase.
Since macrocyclic polyamines have a much greater charge density within
the molecular skeletons in the multi-protonated state as compared to the lin-
ear polyamines, they should have an entropic advantage for the complexation
of phosphates. On the basis of this advantage of macrocyclic polyamines over
the corresponding acyclic polyamines, Kimura et al. reported that a macro-
cyclic hexaamine 3 showed stronger affinity to AMP, ADP and ATP at pH 8.0
Scheme 2 Proposed binding mechanism of Chemosensor 2 with pyrophosphate. The PET

quenching is canceled upon the phosphate binding with 1 causing fluorescence enhance-
ment
(logKs = 3.2, 5.6 and 6.4, respectively) than those for corresponding spermine
complexes (mentioned above) [15].
Lehn and his co-workers reported that the oxygen-containing macrocyclic
polyamine 4 is a good receptor of nucleotides such as ATP and ADP [16, 17].
Due to the formation of multiple ionic interactions, 4 strongly bound ATP, es-
pecially under very acidic conditions (logKs = 4.8 at pH 7.6, 11.0 at pH 4.0).
Interestingly, they found that compounds 4 and 5 were capable of catalyzing
the hydrolysis of ATP to ADP (Scheme 3). In the complex of 4-ATP, a nitrogen
atom attacked the terminal phosphate to yield ADP and thus phosphorylated
4. The phosphorylated 4 was immediately hydrolyzed to regenerate the ori-
ginal 4 in the catalytic process. Compound 5, in which a fluorogenic acridine
ring was introduced to the polyamine ring, showed greater selectivity be-
tween ATP and ADP than the parent compound 4. Compound 5 bound such
nucleotides using two distinct forces, electrostatic interactions between the
macrocyclic polycationic moiety and the polyphosphate and π-π stacking
interactions between the acridine ring and nucleic base. The latter inter-
actions induced the fluorescence change of the acridine moiety of 5 upon
complexation.
Scheme 3 Schematic representation of the catalytic cycle for ATP hydrolysis by the chemo-
sensor 5 following the nucleophilic pathway
Other types of polycationic macrocyclic compound like 6 ∼ 8 were also

reported as water-soluble receptors for nucleotides [18–20]. The apparent
complexation constants, logKapp , toward ATP for 6 and 8 at neutral pH were
found to be 5.2 and 8.4, respectively, while the logKs for 7 was 4.8.
Coupling a suitable signal transducer to the binding scaffold is generally
crucial for designing efficient chemosensors. For example, redox-active metal
complexes such as metallocenes have been incorporated into various guest-
binding scaffolds to yield chemosensors that detect the corresponding guests
electrochemically. Beer et al. synthesized many artificial receptors contain-
ing redox active metal centers and evaluated their usefulness[21–23] In these
studies, they prepared water-soluble phosphate chemosensors. Chemosen-
sors 9 ∼ 11, tethering cyclic polyamines on a ferrocene moiety, bound ATP
or phosphate to form a 1 : 1 complex at pH 6.5. By cyclic voltammetric ex-
periments, it was revealed that the receptors were able to detect these phos-
phate anions using cathodic shifts of the redox potential of ferrocene by
60–80 mV [24, 25].
Ruthenium-(II)-tris-(2,2 -bipyridiyl), Ru(bpy)3 , has proven to be a useful

signal transducer by virtue of its ability to change its luminescent emis-
sion upon guest binding. Water soluble Ru(bpy)3 -appended polyamine re-
ceptors 12 ∼ 14 were prepared by Beer’s group. These receptors detected
ATP and phosphate in aqueous media by changing the metal-ligand charge
transfer (MLCT) luminescent emission [26]. The intensity of MLCT emission
(600–630 nm) was sensitive to pH and phosphate anions. The emission titra-
tion studies with phosphate and ATP showed that the MLCT emission of these
chemosensors was quenched by 15% upon complexation of the guest.
2.2
Artificial Chemosensors Containing Other Cationic Groups
In addition to polyamines, guanidinium groups have also shown to be use-

ful binding groups for phosphate anions. It is well-known that the active
site of some phosphatases such as Staphylococcal nuclease [27] and alkaline
phosphatase [28] contain guanidinium groups (from Arg residues) that are
actively involved in binding phosphate derivatives. Inspired by this approach,
several groups have designed guanidinium-based phosphate receptors.
Anslyn and co-workers have published several interesting papers on mo-
lecular recognition of phosphate by cleft-like receptor molecules containing
poly-guanidinium or -ammonium rims [29–31]. Phosphate receptors 15, 16
were designed for detection of D-myoinositol 1,4,5-triphosphate (IP3) [30]. To
evaluate the binding affinity, they devised a fluorescent competitive assay. In
this system, 5-carboxylfluorescein (5-CF) was initially bound to the receptors
(15, 16) to form a 1 : 1 complex. The fluorescence of 5-CF increased as a con-
sequence of the entrapment in the cleft of the receptor from aqueous media.
The addition of IP3 to an aqueous solution containing 15-(5-CF) complex
led the guest exchange to form the 15-IP3 complex and caused the concur-
rent release of 5-CF from the cleft to aqueous media, which in turn caused a
dramatic decrease of 5-CF emission. From the change in fluorescence it was
possible to calculate a logK value of 5.7 for the formation of 15-IP3.
On the basis of this cleft-type scaffold, Anslyn’s group prepared a chemo-

sensor library to discover a sophisticated nucleotide receptor possessing high
selectivity toward ATP [32]. The attachment of the scaffold on Wang resin
for solid phase synthesis, followed by tethering two tripeptide chains branch-
ing off the guanidium groups by the split-and-pool method yielded a small
library of potential candidates for chemosensors. Several “hit” molecules
were selected and picked up by the fluorescence assay with ATP derivative
(N-methylanthraniloyl-ATP: MANT-ATP). Among them, chemosensor 17 con-
taining Ser-Tyr-Ser sequences, was finally discovered as an ATP selective
chemosensor. The binding constant of 17 for ATP was found to be 3.4 × 103 M–1 ,
approximately 10 times greater than that of the reference compound 18.
Alternatively, Rebek et al. rationally designed and synthesized a water-
soluble nucleotide receptor 19 that is comprised of a carbazole backbone,
a cyclic guanidinium cation, and two Kemp’s triacidic imides [33]. Compound
19 bound cyclic AMP (cAMP) to form a 1 : 1 complex (logKs = 2.8) in aqueous
media by cooperatively accumulated interactions, specifically electrostatic in-
teraction between its guanidinium cation and the anionic phosphate diester,
π-π stacking interaction between its carbazole backbone and adenine ring,
and hydrogen bonding interactions (Fig. 1).
Fig. 1 Structure of 19 and the predicted lowest energy conformation of the complex
between 19 and cAMP
Amidinium cations are used as an alternative to guanidinium cation and

as they are more easily accessible in organic synthesis. Gramlich and co-
workers reported the complexation behavior of cationic “cavitands” having

four amidinium cations on the peripheral of resorcin[4]arene, toward nu-
cleotides [34]. Cavitand 20 possessed four ethylene glycol chains on the lower
rim of resorcin arene that enhance the water-solubility. The binding constants
of 20 to nucleotides were estimated to be 6.6 × 105 M–1 for ATP, 4.9 × 104 M–1
for ADP, 1.0 × 104 M–1 for AMP, and 1.4 × 103 M–1 for cAMP. 1 H NMR studies
suggested that the adenine ring was encapsulated in the hydrophobic cavity of
resorcin[4]arene (Fig. 2). The affinity of 20 to AXP was only slightly greater
than that to the corresponding nucleotides having other nucleobases, how-
ever, suggesting that electrostatic interactions between amidinium groups
and phosphates are predominant in the complexation between the cavitand
and nucleotides.
Fig. 2 Structure of 20 and the energy-minimized structure of the complex formed be-
tween chemosensor 20 and AMP
Similar to guanidinium, imidazole, a basic group of histidine side chains,

might be used as a binding unit for phosphate in artificial receptors. As far as
we know, however, there are no reported examples of such receptors that can
successfully bind phosphates in aqueous media. Known examples are limited
to receptors used in organic media. This again indicates the difficulty in the
development of sophisticated phosphate receptors usable in aqueous solutions.
3
Phosphate Anion Recognition Utilizing Coordination Chemistry
As mentioned above, many type of phosphate receptors containing polyamines

as binding sites have been developed. Because polyamines also have high
affinity toward various metal ions, however, the metal complexation with such
binding sites competes with the desired phosphate capturing under physio-
logical conditions. Therefore, the utilization of the polyamine type phosphate
receptors in biological applications may be seriously restricted. In addition,
the moderate binding affinity of polyamines toward phosphates under bi-
ological conditions is another drawback. On the other hand, coordination
chemistry provided by metal complexes is often employed to bind a phos-
phate unit of substrates in the enzyme active site [35, 36]. This gives one
an important clue for the design of phosphate-binding motifs that show the
stronger affinity in aqueous media.
3.1
Zn-cyclen Type Artificial Chemosensors
Kimura et al. reported a series of zinc(II) complexes of cyclic polyamine

ligands as biomimetic models of zinc-containing enzymes such as carbonic
anhydrase and alkaline phosphatase A [37]. In these biomimetic model sys-
tems, the phosphate group from the substrates coordinate to the zinc(II)
centers [38]. Initially, it became clear that the mono-nuclear Zn-cyclen com-
plex 21 formed a 1 : 1 complex with monodentate phosphate dianions such as
phosphate, phenyl phosphate (PhP, logKs = 3.5), and 4-nitrophenyl phosphate
(NPP, logKs = 3.1), whereas the metal-free cyclen showed considerably weaker
affinity with these phosphate dianions under the same conditions. This in-
dicated that the coordination bond between phosphate and metal center was
strong enough in aqueous media, thus such the non-covalent interaction
should be powerful candidate to capture phosphate anions in water. Ditopic
compound 22 bound PhP or NPP as a sandwich-shaped 1 : 1 complex with
moderate affinity (logKs = 4.6 for PhP, 4.0 for NPP) [40]. Trinuclear zinc com-
plex 23, a mimic of the active site of zinc enzymes such as phospholipase C,
showed the strongest affinity toward these phosphates (logKs = 5.8 for PP, 6.6
for NPP) among the Zn-cyclen type receptors, probably due to capturing the
phosphate moiety in its pseudo-cleft like binding site [41]. Three molecules
of a bipyridine derivative equipped with two Zn-cyclen complexes 24 were
Scheme 4 Cartoon of the structure of chemosensor (24)3 -Ru
assembled with ruthenium(II) to form another type of phosphate receptor

(24)3 -Ru (Scheme 4) [42]. Having two sets of the trinuclear binding site at the
top and bottom of the assembly, the receptor selectively bound IP3 or CTP3
in a 1 : 2 stoichiometry.
Interestingly, Kimura and coworkers found that receptor 21 selectively
formed a 1 : 1 complex (logKs =3.4) with deoxythymidine (dT) or uracil
analogues at physiological pH. In these systems, the deprotonated imide an-
ion of dT coordinated to the zinc (II) ion center and two sets of hydrogen
bonds were formed between carbonyl oxygens of dT and the amine protons of
21 [43]. This is the first dT (or U)-selective artificial receptor to work in aque-
ous solution. They subsequently accomplished the selective recognition of
thymidine nucleotide by using 22 with relatively high affinity (logKs = 5 ∼ 6).
In these complexes, the thymine moiety was bound to Zn-cyclen and the
phosphate anion moiety was concurrently coordinated to the other Zn center.
3.2
Zn-Dpa Type Artificial Chemosensors
Hamachi and co-workers developed a fluorescent ATP chemosensor con-

sisting of dinuclear zinc(II)-dipicolylamine (Zn-Dpa) appended to 2-acetyl
anthracene 25 [44]. The apparent complexation constants of 25, Kapp , toward

ATP, ADP, and AMP were found to be 2.2 × 106 , 2.2 × 105 , and 6.6 × 104 M–1 ,
respectively, at neutral aqueous pH, while the considerably weaker binding
of mononuclear analogue 26 to AXP was observed. This suggests that the
two Zn-Dpa moieties cooperatively captured the phosphate units of the nu-
cleotides. The pattern of the fluorescence change of 25 upon complexation
was dependent on the kind of nucleic bases of nucleotides. The binding of
ATP to 25 caused a dramatic fluorescence enhancement (I/I0 ≈3), whereas
a small increase of fluorescence was observed in the case of CTP. In con-
trast, the emission was diminished with the addition of GTP, probably due to
the electron transfer quenching by the guanine group. Similar nucleic base-
dependent fluorescence change was also reported in the nucleotides sensing
study utilizing a cyclic polyamine-type of receptor 5.
Hong et al. produced binuclear Zn complex type receptors 27, 28 as fluoro-

genic and chromogenic phosphate sensors, having high selectivity toward py-
rophosphate [45, 46]. Kim and his co-workers reported that a complex of 29
with pyrocatechol violet (PV) can colorimetrically detect phosphate deriva-
tives such as inorganic phosphate and AMP through the ligand exchange-
induced color change [47, 48]. In this system, the bound PV was effectively
Scheme 5 Schematic representation of the mechanism of chromatic detection by chemo-

sensor 29-PV toward inorganic phosphate. The effective displacement of PV with phos-
phate anion at the binding site results in color change of the solution from blue to yellow
displaced with phosphate anion at the binding site so as to cause the remark-
able color change of the solution from blue to yellow (Scheme 5).
Excimers, excited dimers of two planar fluorophores, often display sig-
nificantly red-shifted fluorescence relative to the corresponding monomer.
Hence, suitable molecules that can form excimers resulting from complexa-
tion with the target phosphate would become unique chemosensors that show
drastic fluorescence change during phosphate recognition. A chemosensor 30
attaching two Zn-Dpa units to both ends of naphtalenediimide formed a 2 : 2
complex with pyrophosphate resulting in an excimer formation (logKapp =
4.1 × 105 M–1 ), whereas the addition of other anions including AXP did not
show any meaningful fluorescence change [49]. This sensor can selectively
detect pyrophosphate in the presence of ATP or inorganic phosphate by the
characteristic red-shifted emission. In addition to the four Zn binding sites
for pyrophosphate, the favorable π-π interaction of the two flat aromatic
Scheme 6 Proposed binding mechanism of chemosensor 30 with PPi. The formation of

a 2 : 2 type binding between 30 and pyrophosphate leads to excimer formation
moieties allowed the formation of a 2 : 2 type binding between 30 and py-

rophosphate (Scheme 6).
3.3
Ratiometric Detection of Phosphate Derivatives
The detection of a specific anion by using the emission or excitation change

at two different wavelengths provides a significant advantage over conven-
tional measurements using a single wavelength change. Such a dual exci-
tation/emission system enables a ratiometric detection of an analyte, thus
allowing precise and quantitative analysis and imaging even in complex en-
vironments such as that inside living cells. Hamachi et al. successfully de-
veloped a bis(Zn-Dpa) appended xanthone type receptor 31 that is capable
of ratiometric detection of phosphate derivatives under physiological condi-
tions using excitation fluorescence [50]. As shown in Fig. 3, the addition of
ATP to an aqueous solution of 31 resulted in a see-saw type excitation spectral
change. The mechanism of such a excitation spectral change was proposed as
follows. In the absence of phosphate, the carbonyl oxygen atom of the xan-
thone fluorophore was coordinated to both Zn centers, as shown in Scheme 7.
When a phosphate derivative was bound, the fickle coordination bond be-
tween one of the Zn centers and the carbonyl oxygen atom was cleaved by
the strong interaction of the metal ion with the phosphate, which caused the
excitation change of the xanthone fluorophore.
Acridine type chemosensors 32, 33 were also designed by Hamachi and
coworkers as dual-emission chemosensors of nucleoside pyrophosphate
Fig. 3 Change in the excitation spectrum of 31 induced by ATP addition

Scheme 7 Phosphate anion-induced coordination rearrangement of 31 upon binding to

a phosphate derivative
(Fig. 4) [51]. In the resting state, these sensors are in equilibrium between
mono- and binuclear Zn complexes. In this state the sensors predominantly
exist as mononuclear Zn complexes, with binuclear Zn complexes as the
minor species, as illustrated in Scheme 8. The binding of anionic nucleo-
Fig. 4 Change in the fluorescence emission of 32 (a) and 33 (b) induced by ATP addition
Scheme 8 Schematic representation of the dual-emission sensing mechanism of the acri-

dine chemosensor for nucleoside PPs
side pyrophosphate with the sensor molecules induced complexation of the

second Zn, concurrently cleaving the originally formed coordination bond
between Zn and the acridine nitrogen atom. Such a guest-induced rearrange-
ment of the coordination chemistry resulted in the dual-emission change.
The chemosensors showed high selectivity for nucleotide di- or tri-phosphate
over nucleoside monophosphate, inorganic phosphate, and nucleotide deriva-
tives of monosaccharides such as uridine 5 -diphospho-α-D-galactose (UDP-
Gal). On the basis of such selectivity, these chemosensors were successfully
utilized for real-time monitoring of glycosyl transfer catalyzed by β-1,4-
galactosyltransferase. Since UDP-Gal is converted into UDP in the process of
the glycosyltransfer reaction, the detection of UDP by chemosensor 33 im-
plied a monitoring of the reaction progress. It should be noted that UDP-Gal
has a lower affinity toward 33 than UDP. This can be done in a dual-emission
change manner without any modification of the reactants such as UDP-Gal
and the glycosyl acceptors (Scheme 9).
Kikuchi et al. reported a cyclen-Cd complex containing 7-amino-4-
trifluorocumarin 34 as a fluorogenic chemosensor [52]. During the complex-
ation of 34 with phosphates, the coordination bond of the aromatic amino
group of the coumarin to Cd was cleaved, causing a dual-change of the ex-
citation spectrum (Scheme 10). The dissociation constant of 34-nucleotide
Scheme 9 Fluorescence sensing of the glycosyl-transfer reaction based on ratiometric

detection by 33
Scheme 10 Design concept of sensing phosphates utilizing cyclen-Cd complex 34
complexes was approximately 10–5 M. This chemosensor was utilized for the
real-time monitoring of the activity of a phosphodiesterase which cleaves
an undetectable cyclic nucleotide (e.g. cAMP) to produce the corresponding
detectable nucleotide (e.g. AMP).
4
Recognition of Phosphorylated Protein Surfaces
In addition to small molecular phosphate derivatives, phosphate anions

having large molecular weight recently emerged as important new sub-
jects in anion recognition research. Hamachi and his co-workers developed
the first artificial chemosensors that can recognize phosphorylated pro-

teins/peptides [53, 54]. These receptors 35, 36 in which the anthracene deriva-
tives were linked with two Zn-Dpa complexes, bound single phosphate
groups of peptides or proteins by the cooperative action of two Zn-Dpa
sites, so that the fluorescence intensity increased due to the anthracene fluo-
rophore. From the fluorescence change, the apparent affinity constants for the
1 : 1 complexation with inorganic phosphate or phosphate monoesters such as
o-phosphotyrosine (p-Tyr) and relevant species were found to be in the range
of 104 ∼105 M–1 in aqueous solution (Table 1). These chemosensors tightly
complexed with ATP or ADP with stronger affinity (Kapp > 107 M–1 ), whereas
the phosphate diesters such as dimethyl phosphate (diMeP) and cyclic AMP
did not cause any fluorescence change up to the millimolar concentration
range. No evidence for binding of the chemosensors was obtained when other
anions such as sulfate, nitrate, acetate, fluoride, or carboxylate were added, in-
dicating that the chemosensors possessed high selectivity toward phosphate
Table 1 Summary of the apparent binding constants (Kapp , M–1 ) of 35 and 36 to phos-
phate species by fluorescence change
Phosphate Chemosensor Phosphate Chemosensor

derivative a 35 36 derivative a 35 36
NaH2 PO4 b 4.2 × 105 2.9 × 105 ATP c > 107 4.0 × 105
PhP b 2.1 × 105 5.1 × 104 ADP c > 107 1.6 × 105
p-Tyr b 3.1 × 105 6.1 × 105 AMP c 2.3 × 105 9.1 × 103
MeP b 1.1 × 105 7.9 × 103 cAMP c –d –d
DiMeP b –d –d
a PhP = phenyl phosphate, p-Tyr = O-phospho-L-tyrosine, MeP = methyl phosphate,
diMeP = dimethyl phosphate, ATP = adenosine 5 -triphosphate, ADP = adenosine 5 -di-
phosphate, AMP = adenosine 5 -monophosphate, cAMP = adenosine 3 ,5 -cyclic mono-
phosphate.
b Measurement conditions: 10 mM HEPES, pH 7.2, 20 ◦ C.
c 50 mM HEPES, 50 mM NaCl, pH 7.2, 20 ◦ C.
d Since the fluorescence change was scarcely observed, the association constant cannot be
obtained.
species among various anions, and can distinguish phosphate and phosphate
monoesters from phosphate diesters.
Determination of the binding ability of the chemosensors 35, 36 to phos-
phorylated peptides was conducted as a model study of phosphorylated pro-
tein surface recognition. Similar to phosphate sensing, the fluorescence inten-
sity of 35 increased upon addition of a phosphorylated peptide in aqueous
solution. In the titration study with peptide-a (EEEI-pY-EEFD), a consensus
sequence phosphorylated by a protein kinase v-Src, the fluorescence of 35
was enhanced, with the intensity finally reaching a 2.5 fold increase relative
to that in the absence of the peptide. In contrast, the corresponding non-
phosphorylated peptide-g (EEEI-Y-EEFD) did not cause any emission change,
showing that the chemosensor can distinguish a phosphorylated peptide
from a non-phosphorylated one. Interestingly, the affinity of the chemosen-
sors depended on the number of negative charges located on the phospho-
rylated peptide. Among the tested peptides, both chemosensors showed the
strongest binding affinity (Kapp of 106 –107 M–1 ) for peptide-a, which has a
larger negative charge (– 8) (Table 2).
The significant emission enhancement of these chemosensors and the high
selectivity towards phosphorylated peptides enabled the detection of phos-
phorylated peptides by naked inspection of the emission change. This is
illustrated in the photograph shown in Fig. 5. Such fluorescence intensifica-
tion of the chemosensors is clearly ascribed to the phosphate-assisted binding
of the second Zn cation. A schematic illustration of the sensing mechanism
toward the phosphorylated peptide is depicted in Scheme 11. In the absence
of a phosphorylated peptide, the second Dpa site of the chemosensor is
Table 2 Amino acid sequences of peptides containing optimal consensus sequences phos-
phorylated by different protein kinases and apparent binding constants (Kapp , M–1 ) of 35
and 36 to the peptides as determined by fluorescence change
Consensus substrate Kinase Net 35 36

sequence charge
peptide-a Glu-Glu-Glu-Ile-pTyr-Glu-Glu-Phe-Asp v-Src –8 8.9 × 106 9.5 × 105

peptide-b Asp-Glu-Glu-Ile-pTyr-Gly-Glu-Phe-Phe c-Src –6 1.5 × 106 3.6 × 105
peptide-c Ala-Glu-Glu-Ile-pTyr-Gly-Val-Leu-Phe Lck1 –4 8.2 × 105 1.5 × 105
peptide-d a Lys-Ser-Gly-pTyr-Leu-Ser-Ser-Glu EGFR –2 5.8 × 104 1.2 × 104
peptide-e Ala-Arg-Arg-Gly-pSer-Ile-Ala-Ala-Phe PKA 0 –b –b
peptide-f Arg-Arg-Phe-Gly-pSer-Ile-Arg-Arg-Phe Bck1 +2 –b –b
peptide-g Glu-Glu-Glu-Ile-Tyr-Glu-Glu-Phe-Asp v-Src –b –b

aAmino acid sequence of ezrin (142–149) phosphorylated by EGFR.
bSince the fluorescence change was scarcely observed, the association constant cannot be
obtained.
Fig. 5 Photograph of the increased emission of 36 in the presence of phosphorylated

peptide-a (middle) compared to 36 only (left) and 36 with non-phosphorylated peptide-g
(right)
Scheme 11 Schematic representation of the sensing mechanism of the chemosensors 35,

36 toward the phosphorylated peptide
partially free from the Zn-complexation, so that the PET quenching by the
benzylic amine of Dpa lessened the fluorescence intensity of anthracene. In
the presence of phosphorylated peptide, on the other hand, the binding of
the second Zn cation to the free Dpa site was facilitated, and as a result the
PET quenching was canceled so as to recover the fluorescence intensity. The
careful thermodynamic study of this molecular recognition using isothermal
titration calorimetry (ITC) undoubtedly demonstrated that the binding was
an endothermic and entropy-driven event in the aqueous buffer solution.
These chemosensors were applied to two biological assays. Firstly, the real-
time fluorescence monitoring of phosphatase-catalyzed dephosphorylation
reactions was demonstrated. A phosphopeptide DADE-pY-LIPNNG (a frag-
ment (988–998) of EGFR) was dephosphorylated by phosphatase PTP1B
to yield the non-phosphorylated peptide (Fig. 6) [54]. The pre-binding of
the substrate peptide with 35 enhanced the fluorescence intensity of the
chemosensor. After the addition of PTP1B, the fluorescence declined in a
time-dependent manner during the progress of enzymatic dephosphoryla-
tion of the peptide. This method is much simpler than the conventional
one using a radio-active phosphorylated peptide as the enzyme substrate.
Secondly, the selective staining of phosphoprotein in SDS-PAGE was car-
ried out [55]. The chemosensor 36 was shown to work well as a fluorescent
staining reagent in the conventional gel electrophoresis of the protein mix-
tures. As shown in Fig. 7, only two distinct bands were fluorescently observed
under a UV trans-illuminator, those corresponding to phospho-ovalbumin
(MW = 45.0 kDa) and phospho-α-casein (MW = 23.6 kDa). In contrast, very
slight or no emission was observed in other bands corresponding to the
Fig. 6 Time trace of PTP1B-catalyzed dephosphorylation monitored by the emission of 35

with (circle) or without (square) PTP1B using a fluorescence plate reader
Fig. 7 Selective phosphoprotein detection in SDS-PAGE using 36. Each lane includes
two phosphoproteins (ovalbumin (45.0 kDa) and α-casein (24.0 kDa)), and four non-
phosphorylated proteins (β-galactosidase) (MW = 116.0 kDa), bovine serum albumin
(MW = 66.2 kDa), avidin (MW = 18.0 kDa), and lysozyme (MW = 14.4 kDa). Lane 1: CBB
staining of the six proteins. Lane 2, 3: Detection of the phosphoproteins with UV transil-
luminator after staining with 36. The amount of each protein in lane 2 and lane 3 is 5.0
and 2.5 µg, respectively
four non-phosphorylated proteins β-galactosidase (MW = 116.0 kDa), bovine

serum albumin (MW = 66.2 kDa), avidin (MW = 18.0 kDa), and lysozyme
(MW = 14.4 kDa). Interestingly, brighter fluorescence was observed at the
band of phospho-α-casein which has eight to nine phosphorylated amino
acid residues, compared to that of less phosphorylated phospho-ovalbumin
containing one to two phosphate units. This suggested that the chemosensors
can distinguish the degree of protein phosphorylation.
Recent advances in understanding of post-translational modification re-
vealed that multisite phosphorylation, so-called hyper-phosphorylation, is
a common mechanism for regulating protein functions in cell signaling path-
ways. For example, platelet-derived growth factor receptor-β (PDGFR-β),
a membrane-bound cytokine receptor, exposes multiple tyrosine residues in
the cytoplasmic domain. The binding of PDGF to the extracellular domain
induces auto-phosphorylation at these tyrosine residues, followed by recruit-
ing of specific signaling proteins containing SH-2 domains, consequently
triggering multiple signaling pathways. Thus, the development of artificial
receptors that can sense hyperphosphorylated proteins is currently another
important topic in biology and biochemistry. Toward this end Hamachi et al.
subsequently designed chemosensors 37, 38, in which two Zn-Dpa units
functioning as phosphate binding sites were connected with bipyridine spac-
ers [56, 57]. These chemosensors were able to bind a multiply-phosphorylated
protein surface. The Zn-Dpa units were juxtaposed at an appropriately dis-
tal position to enable the cross-linking interaction with two distinct phos-
phate groups on a protein surface (Scheme 12). The binding abilities of the
chemosensors with a series of bis-phosphorylated model peptides were evalu-
Scheme 12 Strategy for phosphorylated protein/peptide recognition by chemosensors 37,

38 applying a cross-linking interaction with two phosphate groups on a protein surface
ated by circular dichroism (CD) spectral studies (Fig. 8), in which the α-helix
content of the peptide was measured upon addition of the chemosensors.
Sensors 37, 38 induced the α-helix formation of peptides having two phos-
phoserine residues (pS-5,16, – 9,16, and – 12.16), whereas they did not affect
the secondary structure of the mono-phosphorylated peptide (pS-16). The
complexation process was also monitored by changes in the emission inten-
sity. The fluorescence titration of 38 with pS-9,16 gave the affinity constant
(Kapp = 2.0 × 106 M–1 ), the value of which is over 40-fold higher than that
of 38 with the mono-phosphorylated pS-16 peptide (Kapp = 4.8 × 104 M–1 ).
This indicated that 38 can discriminate the number of phosphorylation sites
Fig. 8 CD (θ) value change of phosphopeptides upon addition of bipyridiyl-type chemo-

sensor 37 ∼ 41
of peptides. More recently, it was found that similar binuclear type artificial
receptors could strongly bind phosphorylated CTD peptide by tight two-
point interactions between Zn-Dpa sites and phosphates on the peptide. Such
a complexation disrupted phosphoprotein/protein interactions in a phospho-
rylated CTD peptide and the Pin1 WW domain, a phosphoprotein-binding
domain [57]. The strategy based on a small molecular disruptor that directly
interacts with phosphoprotein is unique and should be promising in develop-
ing a designer inhibitor for phosphoprotein-protein interaction.
5
High Throughput Sensing System for Phosphate Derivatives
Using Semi-wet Sensor Arrays
5.1
Molecular Recognition of Chemosensors in a Supramolecular Hydrogel
In order to achieve a complete understanding of the functions and roles of

phosphate derivatives in biological systems, it is necessary to collect data
simultaneously for a large range of different phosphorylation events by ex-
haustive analysis. Rapid and high throughput methods based on chemosen-
sors are expected to significantly contribute to determining the phosphoryla-
tion level of biological samples. Similar to DNA arrays [58], a chemosensor
array immobilizing a number of chemosensors on a surface of solid sub-
strates is a promising candidate to carry out convenient and high-throughput
sensing. To establish the chemosensor array, the effective immobilization
of sensor molecules on a solid support is regarded to be of primary im-
portance. Hamachi et al. recently proposed a unique chemosensor array

utilizing a supramolecular hydrogel as a matrix in which the chemosen-
sors were non-covalently immobilized in the water-rich environment of
the gel, while practically retaining the original molecular recognition func-
tions [59, 60].
An artificial glycolipid mimic GalNAc-suc-glu-(O-methyl-cyc-hexyl)2 spon-
taneously formed into supramolecular fibers based on a bimolecular layer
structure in water (Fig. 9) [61–64]. The entanglement of the fibers gave
a transparent hydrogel consisting largely of aqueous media and well-
developed hydrophobic domains. Chemosensor 36 was non-covalently em-
bedded in the hydrogel and showed binding affinity/selectivity toward phos-
phate derivatives and consequent fluorescence responses comparable to those
observed in aqueous solution [59].
Fig. 9 Schematic representation of the hierarchal molecular assembly used to form

a supramolecular hydrogel. The artificial glycolipid mimic GalNAc-suc-glu-(O-metyl-cyc-
hexyl)2 forms incipient nano-fibers based on a bimolecular layer structure. Such fibers
contain extensive hydrophobic domains in their cores with oriented saccharide arrays ex-
posed at the interfaces. The incipient nano-fibers are bundled to give thicker fibrils whose
entangling results in the formation of a hydrogel
5.2
Use of Hydrophobic Micro-domains of Supramolecular Hydrogel Fibers
for Discrimination of Phosphate Derivatives
In addition to the uses above, the elaborate utilization of hydrophobic nano-

fiber domains in the supramolecular hydrogel allowed more selective dis-
crimination of the Zn-Dpa-based chemosensors among various phosphate
derivatives [60]. The chemosensor 42, newly synthesized by Hamachi and

coworkers, had two Zn-Dpa binding sites and an environmentally sensitive
fluorophore (dansyl). After the chemosensor was embedded in the supra-
molecular hydrogel, solutions including various anions were placed on it. As
shown in Fig. 10, roughly three patterns of the fluorescence spectral change
were observed: (1) the emission intensity of 42 (at 512 nm) increased with
a blue shift in the case of the relatively hydrophobic PhP, (2) the inten-
sity decreased with a red shift of the emission for the relatively hydrophilic
ATP, phosphate and phospho-Tyr, (3) addition of phospho-diesters and non-
phosphate anions caused no fluorescence changes. Fluorescence titration with
ATP (Fig. 10c) in the gel spot showed that the emission gradually decreased
with the concurrent red-shift of the emission maximum. In contrast, the
reverse type of spectral change was observed when PhP was added to the
hydrogel spot containing 42 (Fig. 10d). From these spectral changes, the bind-
ing constants were calculated to be 1.8 × 105 M–1 for ATP and 7.2 × 103 M–1
for PhP. Interestingly, such fluorescence change never occurred in homo-
Fig. 10 a Schematic illustration of chemosensor redistribution upon binding to a hy-

drophobic or hydrophilic phosphate derivative between the hydrophobic hydrogel
nanofiber and the hydrophilic cavity. The three patterns of change observed in the
fluorescence spectra of the hydrogel containing the dansyl-appended receptor 42.
b Fluorescence spectral change of 42 (60 µM, red line) embedded in the hydrogel upon
addition of PhP, ATP, phosphate, and phospho-Tyr. The emission intensity of 42 in-
creases with the blue shift for PhP, whereas the intensity decreases with the red shift for
ATP, phosphate, or phospho-Tyr. (c–d) Fluorescence spectral change and the fluorescence
titration plots (inset) of 42 embedded in the hydrogel upon addition of ATP or PhP, re-
spectively: [42] = 2 µM, [ATP] = 0–60 µM for (e), [42] = 6 µM, [PhP] = 0–600 µM for
(f) at λex = 322 nm
geneous aqueous solutions, indicating that the hydrophobic domain of the

present supramolecular hydrogel is crucial for the phosphate-induced change.
Most significantly, the pattern of fluorescence response depended on the hy-
drophilicity of the phosphate derivatives. Specifically, the strongly hydrophilic
ATP induced a red-shift in the emission of 42 with reduced intensity, whereas
the rather hydrophobic PhP caused an emission increase with a blue shift.
These spectral changes imply that the microenvironment of the 42-ATP com-
plex is more hydrophilic than that of 42 itself, whereas the 42-PhP complex
locates in the more hydrophobic microenvironment relative to the original 42.
Using the unique semi-wet hydrogel array, a variety of phosphate anion species
may be rapidly and conveniently discriminated from each other by both the
emission intensity change and the wavelength shift as shown in Fig. 11.
Fig. 11 Photo images of the sensing patterns of semi-wet molecular recognition (MR)
chips of the hydrogel of GalNAc-suc-glu(O-metyl-cyc-hexyl)2 containing (a) 36 (40 µM),
(b) 42 (60 µM)
6
Summary
Phosphates and their derivatives are abundant in biological systems including

inorganic phosphates, small molecular organic phosphates such as ATP, phos-
pholipids and phosphorylated organic intermediates, and phosphate deriva-
tives having large molecular weight such as DNA and phospho-peptides and
proteins. It is now clear that these derivatives play distinct roles in biologi-
Fig. 12 a Construction of the hybrid biosensor and fluorescence sensing for the doubly
phosphorylated peptide. b Structure of the chemosensor unit on the hybrid biosensor.
c Amino acid sequences of the WW domain and its mutant
cal phenomena under aqueous conditions, depending on their corresponding

forms. Therefore, development of sophisticated artificial chemosensors that
possess high binding affinity and selectivity toward the target phosphate is
keenly desired in order to significantly contribute the understanding of the
roles of phosphorylation in complex biological systems. Although many ar-
tificial chemosensors have been developed in the past two decades, several
problems still remain to be overcome. In particular, clear discrimination
among the phosphate anion families, such as phosphates, phosphate esters,
and ATP derivatives, etc., is thus far quite difficult and remains challenging.
In addition, molecular recognition in aqueous systems is generally difficult
to manage. Fundamental research in many areas is required for the rational
design of efficient artificial chemosensors. Most recently, Hamachi et al. pro-
posed a hybrid system of artificial sensors with a biological receptor scaffold
to achieve improved selectivity among various phosphates. They successfully
constructed a unique biosensor in which phosphoprotein-binding peptide
WW domains were attached to Zn-Dpa units on the side chains (Fig. 12) [65].
Such pattern recognition and detection may also be promising for the pre-
cise discrimination between various phosphates as proposed by Anslyn [66]
and Hamachi [67]. More creative ideas and concepts are required for progress
in this growing area because the design and synthesis of novel chemosen-
sors is envisioned to have a great impact on both basic research and practical,
diverse applications in the biological, diagnostic and medicinal fields.
References
1. Fraústo da Silva JJR, Williams RJP (1991) The Biological Chemistry of the Elements.
Clarendon Press, Oxford
2. Reitz AB (1990) Inositol Phosphates and Derivatives. American Chemical Society,
Washington DC
3. Woodgett J (ed) (2000) Protein Kinase Function. Oxford University Press, New York
4. Sefton BM, Hunter T (1998) Protein Phosphorylation. Academic Press, New York
5. Lander SE et al. (2001) Nature 409:860
6. Johnson LN, Lewis RJ (2001) Chem Rev 101:2209
7. Canagarajah BL, Khokhlatchev A, Cobb MH, Goldsmith EJ (1997) Cell 90:859
8. Yaffe MB (2002) Nat Rev 3:177
9. Yaffe MB, Elia AEH (2001) Curr Opin Cell Biol 13:131
10. Holmberg CJ, Tran SEF, Eriksson JE, Sistonen L (2002) Trends Biochem Sci 27:619
11. Labadi I, Jenei E, Lahti R, Lönnberg H (1991) Acta Chem Scand 45:1055
12. Nakai C, Glinsmann W (1979) Biochemistry 16:5636
13. Huston ME, Akkaya EU, Czarnik AW (1989) J Am Chem Soc 111:8735
14. Vance DH, Czarnik AW (1994) J Am Chem Soc 116:9397
15. Kimura E (1985) Top Curr Chem 128:113
16. Hosseini MW, Lehn JM (1987) Helv Chim Acta 70:1312
17. Hosseini MW, Blacker AJ, Lehn JM (1990) J Am Chem Soc 112:3896
18. Aguilar JA, García-España E, Guerrero JA, Luis SV, Llinares JM, Miravet JF, Rami-
rez JA, Soriano C (1995) J Chem Soc Chem Commun 21:2237
19. Bazzicalupi C, Bencini A, Bianchi A, Fusi V, Giorgi C, Granchi A, Paoletti P, Valtan-
coli B (1997) J Chem Soc Perkin Trans 2 4:775
20. Baudoin O, Gbonnet F, Teulade-Fichou MP Vigneron JP, Tabet JC, Lehn JM (1999)
Chem Eur J 5:2762
21. Beer PD, Gale PA, Chen Z (1999) Coord Chem Rev 185–186:3
22. Beer PD, Cadman J (2000) Coord Chem Rev 205:131
23. Beer PD, Gale PA (2001) Angew Chem Int Ed 40:486
24. Beer PD, Chen Z, Drew MGB, Kingston J, Ogden M, Spencer P (1993) J Chem Soc
Chem Commun 24:1046
25. Beer PD, Chen Z, Drew MGB, Johnson AOM, Smith DK, Spencer P (1996) Inorg Chim
Acta 246:143
26. Beer PD, Cadman J (1999) New J Chem 23:347
27. Cotton FA, Hazen EE Jr, Legg MJ (1979) Proc Natl Acad Sci USA 76:2551
28. Kim EE, Wyckoff HW (1991) Clin Chim Acta 186:175
29. Metzger A, Anslyn EV (1998) Angew Chem Int Ed 37:649
30. Niikura K, Metzger A, Anslyn EV (1998) J Am Chem Soc 120:8533
31. Suzanne LT, Anslyn EV (2003) J Am Chem Soc 125:14807
32. Schneider SE, O’Neil SN, Anslyn EV (2000) J Am Chem Soc 122:542
33. Kato Y, Conn MM, Rebek J Jr (1994) J Am Chem Soc 116:3279
34. Sabo L, Diederich F, Gramlich V (2000) Helv Chim Acta 83:93
35. Wilcox DE (1996) Chem Rev 96:2435
36. Lipscomb WN, Sträter N (1996) Chem Rev 96:2375
37. Kimura E (2000) Curr Opin Chem Biol 4:207
38. Aoki S, Kimura E (2002) Rev Mol Biotechnol 90:129
39. Koike T, Kimura E (1991) J Am Chem Soc 113:8935
40. Fujioka H, Koike T, Yamada N, Kimura E (1996) Heterocycles 42:775
41. Kimura E, Aoki S, Koike T, Shiro M (1997) J Am Chem Soc 119:3068
42. Aoki S, Zulkefeli M, Shiro M, Kohsako M, Takeda K, Kimura E (2005) J Am Chem Soc
127:9129
43. Aoki S, Kimura E (2000) J Am Chem Soc 122:4542
44. Ojida A, Park S, Mito-oka Y, Hamachi I (2002) Tetrahedron Lett 43:6193
45. Lee DH, Kim SY, Hong JI (2004) Angew Chem Int Ed 43:4777
46. Lee DH, Im JH, Son SU, Chung YK, Hong JI (2003) J Am Chem Soc 125:7752
47. Han MS, Kim DH (2003) Bioorg Med Chem Lett 13:1079
48. Han MS, Kim DH (2002) Angew Chem Int Ed 41:3809
49. Lee HN, Xu Z, Kim SK, Swamy KMK, Kim Y, Kim SJ, Yoon J (2007) J Am Chem Soc
129:3828
50. Ojida A, Nonaka H, Miyahata Y, Tamaru S, Sada K, Hamachi I (2006) Angew Chem
Int Ed 45:5518
51. Ojida A, Miyahara Y, Wongkongkatep J, Tamaru S, Sada K, Hamachi I (2006) Chem
Asian J 1:555
52. Mizukami S, Nagano T, Urano Y, Odani A, Kikuchi K (2002) J Am Chem Soc 124:3920
53. Ojida A, Mito-oka Y, Inoue M, Hamachi I (2002) J Am Chem Soc 124:6256
54. Ojida A, Mito-oka Y, Sada K, Hamachi I (2004) J Am Chem Soc 126:2454
55. Ojida A, Kohira T, Hamachi I (2004) Chem Lett 33:1024
56. Ojida A, Inoue M, Mito-oka Y, Hamachi I (2003) J Am Chem Soc 125:10184
57. Ojida A, Inoue M, Mito-oka Y, Tsutsumi H, Sada K, Hamachi I (2006) J Am Chem Soc
128:2052
58. Shena M, Shalon D, Davis RW, Brown PO (1995) Science 270:467
59. Yoshimura I, Miyahara Y, Kasagi N, Yamane H, Ojida A, Hamachi I (2004) J Am Chem
Soc 126:12204
60. Yamaguchi S, Yoshimura I, Kohira T, Tamaru S, Hamachi I (2005) J Am Chem Soc
127:11835
61. Kiyonaka S, Snikai S, Hamahci I (2003) Chem Eur J 9:976
62. Kiyonaka S, Sugiyasu K, Shinkai S, Hamachi I (2002) J Am Chem Soc 124:10954
63. Tamaru S, Kiyonaka S, Hamachi I (2005) Chem Eur J 11:7294
64. Kiyonaka S, Sada K, Yoshimura I, Shinkai S, Kato N, Hamachi I (2004) Nat Mater 3:58
65. Anai T, Nakata E, Koshi Y, Ojida A, Hamachi I (2007) J Am Chem Soc 129:6232
66. Lavigne JL, Anslyn EV (2001) Angew Chem Int Ed 40:3118
67. Koshi Y, Nakata E, Yamane H, Hamachi I (2006) J Am Chem Soc 128:10413
Struct Bond (2008) 129: 127–174
DOI 10.1007/430_2007_070
Anions and π-Aromatic Systems.

Do They Interact Attractively?
Pablo Ballester
ICREA, Pg. Lluís Companys 23, 08010 Barcelona, Spain
pballester@iciq.es
Present address:
Institute of Chemical Research of Catalonia (ICIQ), Avda. Països Catalans 16,
43007 Tarragona, Spain
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
2 Theoretical Investigations of the Interaction of Anions (Halides)

with π Aromatic Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
3 Experimental Evidence of the Anion-π Interaction . . . . . . . . . . . . . 152

3.1 Solution and Related Crystallographic Studies . . . . . . . . . . . . . . . . 152
3.2 Further Crystallographic Evidence of Anion-π Interactions . . . . . . . . . 163
4 Summary and Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
Abstract Hydrogen bonds and charge–charge interactions have been widely used (either
alone or in combination) in the design of efficient and selective synthetic receptors for
anions. Intuitively, the interaction between anions and π-aromatic systems is associated
with a repulsive force. Consequently, for many years, anion-π interactions have been
completely neglected as favorable non-covalent interactions for the construction of ef-
ficient anion receptors. Recently, however, theoretical studies indicate the existence of
an attractive interaction between anions and certain type of π-acidic aromatic systems.
These theoretical studies together with the observation of supramolecular complexes
in the solid state in which anions are included in deep aromatic cavities have encour-
aged an in depth study of the anion-π interaction. Nowadays, the anion-π interaction
is considered by several researchers a potential non-covalent interaction for the design
of anion receptors. This chapter will provide an overview of recent theoretical investiga-
tions, performed since 2002, on anion binding involving six-membered aromatic rings.
A series of experimental studies, carried out since 2004, also evidencing the existence of
a possible attractive interactions betweens anions and six-membered aromatic moieties
of host-guest systems in the solid state and in solution is also discussed.
Keywords Anion-π interactions · Anion coordination · Anion receptors · Anions ·

Host–guest systems · Supramolecular chemistry
Abbreviations
HB Hydrogen bond
RHF Restricted Hartree-Fock
128 P. Ballester
BSSE Basis set superposition error

MEP Molecular electrostatic potential
MP2 Møller Plesset
DFT Density functional theory
ZPE Zero point energy
SI International system of units
PCM Polarized continuum solvent model
AIM Atoms in molecules
GMIPp General molecular interaction potential with polarization
MIPp Molecular interaction potential with polarization
EPS Electrostatic potential surfaces
SAPT Symmetry-adapted perturbation theory
NICS Nucleus-independent chemical shift
RI-MP2 Resolution identity MP2
DFT Density functional theory
TCB 1,2,4,5-Tetracyanobenzene
CSD Cambridge Structural Database
bptz 3,6-bis(2 -pyridyl)-1,2,4,5-teatrazine
bppn 3,6-bis(2 -pyridyl)-1,2-pyridazine
1
Introduction
The driving forces controlling the formation of non-covalent complexes be-

tween molecular or ionic species are quite varied and include the so-called
hydrogen-bonding, stacking, hydrophobic, charge-transfer, van der Waals
and ionic interactions. There is good experimental evidence of the domin-
ant role of electrostatic in many of those intermolecular interactions [1]. An
electrostatic or Coulombic interaction by definition requires no adjustment of
the electronic properties around the ion or the molecule. Consequently, ion–
ion, ion–dipole, dipole–dipole, ion–quadrupole and quadrupole–quadrupole
interactions can be considered as mainly electrostatic non-covalent in-
teractions. A “conventional” hydrogen bond (HB) may be represented as
D – H· · ·A whereby D (donor) and A (acceptor) are both electronegative
atoms (usually N and O). Hydrogen bonds are a prominent and typical rep-
resentation of a dipole–dipole interaction. However, HBs are not necessarily
restricted to N and O, but may involve other electronegative donor and ac-
ceptor functionalities. For example, a hydrogen bond of the type N – H· · ·X– ,
where X– is an anion, is termed an ionic hydrogen bond. In fact, molecules
containing polarized N – H functionalities which behave as H-bond donors
towards anions are widely used as receptors for recognition and sensing pur-
poses in aprotic solvents [2]. The coordination of the anion by the N – H
functionality is an ion–dipole interaction clearly dominated by electrostatics.
The covalent bonding contribution to the interaction can be considered as in-
significant. Equation 1 represents the electrostatic potential created by a point
Anions and π-Aromatic Systems. Do They Interact Attractively? 129
charge q, at a distance r from the charge. Equation 2 corresponds to the elec-

trostatic potential at a distance r from the center of a dipole of q charges
having a 2d separation in space. θ is the value of the angle defined by the
lines that connect the center of the dipole with the point at r distance and the
two point charges. As deduced from Eqs. 1 and 2, the ion–dipole interaction
has the same dependence on the dielectric environment as the interaction be-
tween fully charged partners but is appreciably weaker on an absolute scale
falling off with distance more steeply.
1 q
V(r) = (1)
4πξ0 r
1 2dq
V(r, θ) = cos θ (2)
4πξ0 r2
The main virtue of the ion–dipole compared to the ion–ion interaction, from
the viewpoint of host–guest chemistry, is its directionality (i.e., the energy of
the interaction of an ion and the dipole depends on their mutual orientation
defined by the θ angle). This translates into a fundamental property to achieve
selective recognition of anions using hydrogen-bonding receptors, that is, the
topology of the targeted anion should be considered when deciding the place-
ment of the hydrogen bond donors in the receptor’s structure. Other virtues
of hydrogen bonding in the design of abiotic receptors for anions derive from
its electroneutrality, from its capacity to form simultaneously several bonding
interactions with the anion and from the rich chemistry available to incorpo-
rate this functionality into the molecular scaffold.
Scheme 1 Preparation of the diazabicycloalkanes and ion pairing in acidic aqueous media
130 P. Ballester
A clear analogy exists in the supramolecular chemistry of anions and the

coordination chemistry of main group cations and transition metals [3, 4]1 .
Fundamental design principles of cationic receptors (i.e., macrocyclic effect
and preorganization) have been directly applied to the supramolecular chem-
istry of anions. Many hosts capable of forming ion-pairs with anions have
been prepared by simply inverting the electronic nature of macrocyclic and
macrobicyclic receptors by protonation of a suitable Lewis-basic center. In
principle at least, the non-protonated receptor could be suitable for cation
binding. This was the case of the serendipitous discovery of the first recep-
tor of a guest chloride emerged during the preparation of a macrobicyclic
compound having two converging tertiary amines (Lewis-basic centers).
The protonation of the Lewis-basic centers afforded two protonated am-
monium groups (Lewis-acid centers) capable of maintaining the spherical
chloride anion within the receptor’s cavity through the formation of two
ion-pair-reinforced hydrogen-bonding interactions (salt-bridge) [5–7]. In the
case at hand, the protonation of the amine groups has a twofold effect: 1) adds
positive charge to the receptor and 2) converts a tertiary amine into an excel-
lent hydrogen bond donor (ammonium). Nevertheless, the addition of posi-
tive charge to an anionic receptor, even in the absence of additional hydrogen-
bonding influences, enhances anion binding. The great majority of positively
charged organic receptors for anions are based on nitrogen compounds. The
incorporation of cationic metal centers into the molecular scaffold of the re-
ceptor represents an interesting alternative to the introduction of positive
charge by protonation or quaternization of the amine functionality. Further-
more, when the coordination features of the metal are partially fulfilled they
can be exploited to assist the electrostatic binding with metal-anion covalent
bonding. The major drawback of using positively charged anion receptors is
the internal competition established between the anionic target and the coun-
teranion of the cationic host. This is an unavoidable disadvantage of positively
charged receptors when compared with electroneutral hosts for anions.
The “anticrown chemistry” term has been coined to describe the way in
which many electroneutral hosts were built to recognize anions. That is the in-
corporation of multiple and convergent Lewis-acid centers into preorganized
cyclic molecular scaffolds. The Lewis-acid centers, usually transition metals or
elements like B, Si, Sn or Hg, should expose their electron-deficient sites for
interaction with the lone electron pairs of the anions. The hydrogen-bonding
interaction of an anion with an amide N – H mentioned above has also been
widely used in the preparation of neutral receptors for anionic guests [8].
While the above-mentioned non-covalent interactions (hydrogen bonds,
charge–charge) have been widely used in the design of efficient and selective
synthetic receptors for anions either alone or combined, the use of attrac-
1The researchers working in the area of anion recognition having an inorganic chemistry back-
ground usually refer to the field as anion coordination chemistry
tive interactions between negative charges and aryl rings towards the same
end has remained dormant in the literature since they were first noticed by
Scheneider in 1993 [9].
Recently, the studies into the anion-π interaction have intensified and
gained interest among the scientific community. This is due, on the one
hand, to the publication in 2002 of several papers dedicated to exploring the
physical nature of the interaction of anions with aromatic compounds using
electronic structure methods and, on the other hand, to the observation since
2004 of supramolecular complexes in the solid state in which anions are in-
cluded in deep aromatic cavities. Furthermore, the recently noticed fact that
the orientation of negatively charged groups [10] and lone pairs [11] above
the plane of an aromatic is a frequently occurring structural motif in biopoly-
mers has added additional interest to the subject.
At first sight, a stabilizing effect for the interaction between a pair of
electrons or a negative charge and the face of a π system seems to be coun-
terintuitive. Not surprisingly, simple modelling studies indicated a repulsive
interaction between anions and a benzene ring. It is now clear that non-
covalent interaction between aromatics and positively charged cations, the
cation-π interaction, are prominent in a wide variety of systems and should
be considered as an important and general binding force [12]. The cation-
π interactions are expected simply from electrostatic arguments. In fact,
a simple electrostatic model based on the visual inspection of electrostatic
potential surface of the aromatic ring has been used to rationalize the major
binding trends of the cation-π interaction. In line with the principles outlined
above for the preparation of anion receptors through the inversion of the
electronic character of suitable Lewis-basic centers (amine to ammonium),
a possible explanation for the observed geometry that points lone pairs or
places negative charges into the face of the π system has to do with an inver-
sion of the electronic character of the aromatic ring [13].
Theoretical investigations have clearly established the existence of bind-
ing interactions between a negatively charged species and an electron-poor
π system. The theoretical results have encouraged the experimental obser-
Fig. 1 Inversion of the electronic character of Lewis-basic centers: a the lone pair of
an amine is a good binding site for a cation while the protonated ammonium group is
complementary to anions; b the π-system of an electronically rich aromatic compound
establishes attractive interactions with cations (cation-π interaction) the introduction
of electro-withdrawing substituents produces the depletion of electronic density of the
π-system giving rise to attractive interaction with anions
132 P. Ballester
vation and evaluation of this new non-covalent interaction. Several authors

have already claimed the potential use of anion-π interactions in the prep-
aration of selective receptors for anions. Solid-state structural examples of
anions included in deep aromatic cavities of supramolecular complexes have
been used as evidences for the existence of a substantial anion-π attractive
interaction. Few studies, however, have attempted to quantify the strength of
the interaction in solution. Since the anion-π interaction seems to be a direc-
tional non-covalent interaction it has potential and unexplored applications
in the design of selective neutral receptors for anion recognition. This chap-
ter will provide a short overview of recent theoretical investigations on anion
binding involving six-membered aromatic rings followed by a summary of
experimental studies evidencing or not the existence of attractive interactions
between anions and six-membered aromatic moieties of host–guest systems
in the solid state and in solution2 .
2
Theoretical Investigations of the Interaction of Anions (Halides)
with π Aromatic Systems
Since 2002 several groups have studied in detail the interaction of anions
with electron deficient aromatic rings using theoretical methods. These stud-
ies can be considered to derive from earlier fundamental work carried out
by Dougherty [14], Besnard [15] and Alkorta [16] on the interaction between
hexafluorobenzene and the heteroatom in molecules such as H2 O, HCN and
HF wherein the negative end of the dipole is directed toward the π-system
and aligned with the C6 axis of the ring.
In particular, Alkorta et al. [16] studied the interaction of C6 F6 with
HF using both MP2 and hybrid DFT methods (B3LYP) with 6-31-G∗∗ and
6-311++G∗∗ basis sets. The authors considered that these methods are more
accurate for weak complexes than methods without electron correlation.
A shortening effect in the calculated distance value between the interacting
atom and the centroid of the aromatic ring is clearly observed by the inclusion
of electron correlation (MP2 and B3LYP methods). Thus, for the minimum
structure of the C6 F6 · · ·FH complex the computed distance is 3.076 Å when
using the RHF method and the 6-31G∗∗ basis set and 2.858 Å and 2.814 Å,
respectively, when the B3LYP/6-31G∗∗ and MP2/6-31G∗∗ methodologies are
applied. The use of a more extended basis set B3LYP/6-311++G∗∗ produces
a larger elongation of the same distance to 3.127 Å while the calculated bind-
ing energy at this level of theory and basis set for the C6 F6 · · ·FH complex is
2 Complexes in which the anion is one component of an aromatic ring sandwich complex are out-
side of the scope of this chapter. Likewise, examples for the interaction of anions with five and seven
membered aromatic rings will not be presented
Fig. 2 Schematic representation of the geometry of the complexes of water, HCN, and HF
with C6 H6
Fig. 3 Calculated MP2/6-31+G∗ MEP surfaces for a 1,3,5-triazine, b trifluoro-1,3,5-tri-

azine, and c hexafluorobenzene. Electrostatic potential surfaces energies range from –19
(red) to +19 (blue) kcal/mol for 1,3,5-triazine, –39 (red) to +39 (blue) kcal/mol for
trifluoro-1,3,5-triazine, and –24 (red) to +24 (blue) kcal/mol for hexafluorobenzene
∆EBSSE = 1.23 kcal mol–13 . The interaction energy of the C6 F6 · · ·FH complex
is similar to the one observed in the formation of weak hydrogen bonds.
Several years later, in 2002, Mascal et al. [17], Alkorta et al. [18] and
Frontera, Deyà et al. [19] reported almost simultaneously the theoretical
demonstration of the existence of an attractive interaction between a formally
charged negative species and the π-system of an aromatic ring.
Mascal et al. [17] performed ab initio orbital calculations at the MP2
level of theory with the 6-31+G∗ basis set, including counterpoise correc-
tions for the basis set superposition error (BSSE), for the interaction of
both 1,3,5-triazine and trifluoro-1,3,5-triazine with chloride and fluoride. The
molecular electrostatic potential (MEP)4 maps of 1,3,5-triazine and trifluoro-
1,3,5-triazine clearly indicate an area of positive density concentrated on the
C3 rotational axis passing through the center of the hexagonal ring and being
perpendicular to the plane of the aromatic ring, similar to that observed for
C6 H6 on the C6 rotational axis.
Optimization from geometries that place the chloride near the area of
positive density found on the C3 rotational axis of 1,3,5-triazine demon-
3 BSSE stands for Basis Set Superposition Error and the interaction energies are corrected for this
inherent error
4 The MEP maps have been used for long time as a tool to identify both nucleophilic and elec-
trophilic regions in a molecule
134 P. Ballester
strated the existence of an attractive interaction between the chloride and the
π-system. The optimized structure obtained for the chloride-1,3,5-triazine
complex positioned the chloride anion exactly on the C3 rotational axis of the
1,3,5-triazine (Φ = 90) and at a distance of 3.2 Å of the aryl centroid. The fact
that the structure of the complex was a minimum was confirmed by calculat-
ing the corresponding frequencies (no negative vibrational frequencies were
found). The MP2/6-31+G∗ energy calculated for the geometry of the chloride-
aryl-centroid complex optimized using the same level of theory and basis set
was ∆EBSSE –1
0 K =– 4.8 kcal mol .
Mascal et al. [17] also located another geometry for the complexation of
the chloride and bromide anion with 1,3,5-triazine. This geometry is also
a minimum and involves a C – H· · ·Cl– hydrogen bond. The formed hydro-
gen bond shows good geometrical characteristics, having a C· · ·Cl– distance
of 3.4 Å and a C – H· · ·Cl– angle of 180◦ . The calculated MP2/6-31+G∗ en-
ergy of this interaction geometry optimized at the same level of theory and
basis set was ∆EBSSE –1 –1
0 K =– 7.4 kcal mol , that is 2.6 kcal mol more stable than
chloride-aryl centroid complex discussed above. The authors clearly state that
although the participation of ions in gas phase chemistry yields interaction
energies which seem to be exaggerated when compared to solution values,
the reported energies of these interactions may still be relevant in the inte-
rior of a receptor. The authors used the trifluoro-1,3,5-triazine as a model
to evaluate the extent to which further electron withdrawal in the triazine
Fig. 4 Side and top views of the minimized structures at the MP2/6-31+G∗ level of a the
triazine chloride aryl centroid complex and b the triazine chloride hydrogen-bonding
complex. The triazine and the chloride are shown in a scaled ball-and-stick representation
Fig. 5 Side view and top view of the “attack” structure for the fluoride 1,3,5-triazine
complex
Fig. 6 Molecular structures of the perfluoroaromatic compounds studied by Alkorta

et al. [18]
system would enhance complexation. Higher interaction energy and even

stronger chloride-aryl centroid complexes were located using the trifluoro-
◦
0 K =– 14.8 kcal mol , r = 3.0 Å and θ = 90 ). In fact, this
1,3,5-triazine (∆EBSSE –1
high energy and close interaction distance computed for the Cl– -trifluoro-
1,3,5-triazine complex is comparable to those of the potassium cation-π com-
plexes of benzene calculated at the same level of theory [20, 21]5 .
The corresponding fluoride-aryl centroid complexes with 1,3,5-triazine
and trifluoro-1,3,5-triazine are characterized by the presence of one negative
vibrational frequency and represent a shallow inflection point on a surface
connecting to a second geometry, the so-called “attack” structure which is
a real minimum structure. This structure was suggestive of a reactant com-
plex with the nucleophilic fluoride anion “attacking” one of the carbon atoms
with close to the Bürgi-Dunitz trajectory [22, 23]. The considerable stabiliza-
tion energy of the “attack” structure (∆EBSSE –1
0 K =– 18 kcal mol , r = 1.5 Å and
◦
θ = 106.6 ) together with the distance values for the C· · ·F interaction and the
lengthening of the adjacent C – N bonds points to a strong σ -complex.
5 Na+ · · ·C6 H6 interaction E0 = –15.0 kcal mol–1 , r=2.84 Å. K+ · · ·C6 H6 interaction E0 =–18.3 kcal mol–1
136 P. Ballester
The calculated energies for these interactions fall off sharply with increas-
ing polarity of the medium. The simple comparison of uncorrected MP2
single point energies of chloride-aryl centroid complex with triazine, using
Tomasi’s Polarized Continuum solvent Model (PCM), gas phase (–8.5), hep-
tane (–0.4), chloroform (2.1), ethanol (2.8), and finally water (3.2 kcal mol–1 )
suggest that the practical manifestation of these forces will be most likely in
the context of anion containment.
Alkorta et al. [18] studied the complexes formed by a variety of anions
with perfluoro derivatives of benzene, naphthalene, tiophene and furan using
DFT (B3LYP/6-31++G∗∗ ) and MP2 (MP2/6-31++G∗ and MP2/6-311++G∗∗ )
ab initio methods.
The minimum structures of hexafluorobenzene with chloride and bromide
show the anion interacting with the π-system with an anion-aryl centroid
geometry. In both cases, a C6v symmetry was assumed during the optimiza-
tion procedure. However, the C6 H6 · · ·F– complex structure that locates the
anion on the C6 symmetry axis and on top of the aromatic ring (r = 2.5 Å)
shows two degenerate imaginary frequencies. In this case, the minimum
structure corresponds to an “attack” geometry similarly to the interaction of
fluoride with 1,3,5-triazine discussed above.
The calculated interaction energy values for the anion-aryl centroid geom-
etry range between –18.7 and –12.19 kcal mol–1 , which is comparable, as
mentioned above, to the interaction energy of some benzene-cation com-
plexes [20, 21].
The centroid-to-anion distance varies from 2.554 to 3.230 Å (see Table 1).
The effects of the complexation of the anion to the C6 F6 molecule are short-
ening of the C – C bonds (0.004 Å) and a lengthening of the C – F bonds
(0.006 Å). On complexation, the fluorine atoms move slightly towards the
anion and the plane formed by the carbon atoms is about 0.02 Å further
away from the anion than that formed by the fluorine atoms. The hexafluoro-
benzene molecule adopts a “cup” conformation.
Non-covalent interactions have been characterized using Bader’s theory of
“Atoms In Molecules” (AIM) [24] which has been used successfully to under-
Table 1 Corrected interaction energies (kcal mol–1 ) and C6 F6 centroid-anion distance

(r, Å) calculated at the MP2/6-31++G∗ level together with the electron densities and their
Laplacian (au) for the calculated bond critical point
Complex ∆EBSSE r ρ ∇2ρ

MP2/6-31++G∗
C6 F6 · · · F– –18.63 2.554 0.0118 0.0470

C6 F6 · · · Cl– –12.76 3.159 0.0080 0.0238
C6 F6 · · · Br– –12.19 3.230 0.0087 0.0240
Fig. 7 Schematic representation of the location of a bond (red), a ring (blue) and the cage
critical point (black) originated from the interaction of hexafluorobenzene with fluoride
stand conventional [25] hydrogen bonds and cation-π interactions [26]. The
AIM analysis of the electron density of these complexes (C6 F6 · · ·X– , X = F, Cl,
Br) carried out by Alkorta et al. [18] indicated the formation of six degenerate
bond critical points (bcp) between the anion and each of the carbon atoms of
the C6 F6 molecule in a similar way to what was observed for the C6 F6 · · ·X – Y
neutral complexes [16] and the C6 F6 · · ·Na+ complex [20]. In addition, six-
ring critical points (rcp) and one cage critical point (ccp) are generated. The
rcp connect the anion with the middle of the C – C bond. The ccp is located
over the hexafluorobenzene molecules along the C6 axis, connecting the an-
ion with the center of the ring. The quantitative values for ρ and ∇ 2 ρ at the
cps give a hint on the character and strength of the interaction. The values of
electron density of the new bcp and its corresponding rcp are almost the same
values. The positive and small value of the Laplacian of the bcps indicates
a depletion of the electron density, as is common in closed shell interactions
like those found in hydrogen bonds, ionic and cation-π interactions.
Table 2 Corrected interaction energy at the MP2/6-31++G∗ level (kcal mol–1 ) and electro-
static, polarization and van der Waals contribution (kcal mol–1 ) to the interaction energy
calculated using GMIPp method
Complex ∆EBSSE Eele Epol EvdW Et
C6 F6 · · · Cl– –12.7 –11.8 –6.6 5.2 –13.2

138 P. Ballester
It is worth mentioning that the interaction energies calculated at the

MP2/6-31++G∗∗ level of the perfluoronaphtalene complexes with anions are
the largest of all the study (∆EBSSE =– 17.31 (F– ), –16.70 (Cl– ) and –16.51
(Br– ) kcal mol–1 ) approximately 4.5 kcal mol–1 more than the energies of the
corresponding C6 F6 · · ·X– complexes (see Table 1). However, the anions are lo-
cated over the C(4a)–C(8a) bond, that is, the anion interacts with one bond,
not with one aromatic ring. In the optimized structure, the naphthalene ring
bent away from the anion for the complexes with chloride and bromide.
In this case, instead of the “cup” observed for hexafluorobenzene, the per-
fluoronaphtalene adopts a “book” conformation pointing toward the anion.
Alkorta et al. also analyzed the interaction energies using the general molecu-
lar interaction potential with polarization partition (GMIPp) [27, 28] of all
the complexes of the aromatic compounds with chloride in order to calculate
the contribution of the electrostatic, polarization and van der Waals terms to
the overall interaction energy. The obtained results indicate that the polariza-
tion energy is 50 to 100% of the electrostatic energy. The sum of the three
terms calculated with the GMIPp method provides an energetic value very
close to the corrected interaction energy obtained at the MP2 level. Again, the
obtained results for the contribution of the polarization and the electrostatic
term to the overall interaction are analogous to those found for the cation-π
interaction.
Frontera, Deyà et al. [19] also studied the interactions of several anions
with C6 F6 using HF/6-31++G∗∗ and MP2/6-31++G∗∗ ab initio methods. In
all the complexes the anion is positioned over the aromatic ring along the C6
rotational axis passing through the center of the hexagonal ring and being
perpendicular to the plane of the aromatic ring, that is, having an anion-
aryl centroid geometry. Initially the geometries of all complexes were fully
optimized at the HF/6-31++G∗∗ level. The authors reported the correspond-
ing binding energies with and without basis set superposition error (BSSE)
and zero-point vibrational energy (ZPE) corrections. Frequency calculations
at the same level confirmed that the structures are at their energy minimum.
When they extended the calculations to the MP2/6-31++G∗∗ level, assum-
ing C6v symmetry for the complexes C6 F6 · · ·X– , X = F, Cl, Br, the frequency
calculations gave either one or more imaginary frequencies. This problem
was solved by performing the geometry optimizations without imposing
any symmetry constrains. In complete agreement with Alkorta’s report, the
minimum energy complex found for the interaction C6 F6 · · ·F– corresponds to
the nucleophilic attack of the anion at one carbon atom. In general, they con-
clude that the MP2-computed binding energies are more negative than the HF
ones and the equilibrium distances are shorter.
Frontera, Deyà et al. [19] also performed a topological analysis of the
charge-density distribution and properties of critical points in the complexes
using the AIM method providing an unambiguous definition of chemical
bonding between the anion and the π-system. These calculations were per-
Table 3 Interaction energies with (∆EBSSE , ∆EBSSE 0K kcal mol–1 ) and without (∆E,
kcal mol–1 ) the basis set superposition error (BSSE) and zero-point vibrational energy
correction (ZPE, 0 K) and equilibrium distances (r/Å) at HF/6-31G++G∗∗ and MP2/6-
31++G∗∗ (italics) levels of theory and selected electron-density topological properties for
the complexes of hexafluorobenzene with anions from reference [19]
Complex ∆E ∆EBSSE ∆EBSSE

0K r CPa n ρ ∇2ρ
C6 F6 · · · F– –18.8 –18.1 –17.8 2.669 bond 6 0.01001 0.04200

ring 6 0.00996 0.04226
cage 1 0.00702 0.04106
C6 F6 · · · Cl– –11.0 –10.8 –10.6 3.404 bond 6 0.00572 0.01637
–18.0 –13.2 –12.9 3.155 ring 6 0.00571 0.01637
cage 1 0.00450 0.01820
C6 F6 · · · Br– –13.2 –9.3 –9.5 3.479 bond 6 0.00618 0.01612
–20.7 –12.4 –11.9 3.214 ring 6 0.00616 0.01613
cage 1 0.00480 0.01887
a The electron density (ρ) and its Laplacian (∇ 2 ρ) in atomic units at the critical points
(CP) originated upon complexation are given, as well as, the number (n) of each type of
critical points
formed by means of the program AIMPAC using HF/6-31++G∗∗ wavefunc-

tion and the obtained results are completely coincident with the ones we have
previously discussed based on Alkorta’s report. The physical nature of the
anion-π interaction and the importance of the polarization were analyzed by
computing its contribution to the total interaction energy using the molecular
interaction potential with polarization (MIPp) [29]. The MIPp is an improved
generalization of the molecular electrostatic potential (MEP) and was also
used by Alkorta in the energetic analysis of the C6 F6 · · ·Cl– . In the MIPp
calculation three terms contribute to the total interaction energy: 1) an elec-
trostatic term identical to MEP, 2) a classical dispersion-repulsion term, and
3) a polarization term derived from perturbation theory. In the calculation
F– ion was considered as a classical nonpolarizable particle. The electro-
Table 4 Contribution to the total interaction energy (kcal mol) calculated with MIPp for
hexafluorobenzene interacting with F– at several distances (Å) from the center of the ring
Distance Eele Epol Evdw Et
1.5 –36.59 –43.84 1047.82 967.38

2.0 –22.40 –24.89 119.05 71.72
2.5 –16.39 –13.93 13.44 –16.89
3.0 –12.66 –7.96 –0.50 –20.12
3.5 –9.90 –4.74 –0.83 –15.48
140 P. Ballester
Fig. 8 Schematic representation (top and side view) of the positive molecular quadrupole
moment of C6 F6 as cylinders with positive charges on the ends and negative charges in
the center
static (Eele ), polarization (Epol ), van der Waals (Evdw ), and total interacting
energies were calculated when a fluorine ion approaches the hexafluoroben-
zene molecule perpendicular to the center of the aromatic ring. The obtained
results point out the importance of the polarization component, which is
similar to the electrostatic term in the 2.0 to 3.0 Å range, the equilibrium dis-
tance for the fluoride aryl centroid complex is 2.6 Å. The authors mention the
importance of the quadrupole moment for understanding intermolecular in-
teractions of aromatic system but they do not elaborate on this issue in this
work.
To date, most of the work reported on anion-π interactions using theor-
etical methods is between anions and electron deficient aromatics rings, i.e.,
hexaflurorobenzene, 1,3,5-trinitrobenzene, and 1,3,5-triazine. The electron-
deficient aromatic rings are also called π-acidic aromatic rings and are char-
acterized by having a permanent positive quadrupole moment value Qzz . The
value of the quadrupole moment is a measure of the distribution of charge
within a molecule, relative to a particular axis. In six-membered aromatic
rings, the Qzz quadrupole measures the distribution of charge relative to the
C6 rotational axis passing through the center of the hexagonal ring and be-
ing perpendicular to the plane of the aromatic ring. When the ring presents
a high degree of symmetry, one may relate the distribution of charge with re-
spect to the axis perpendicular to the aromatic plain to that along the main
rotational axis and gain a quick characterization of the charge distribution in
the molecule [30]. The SI value of the electric quadrupole moment of hex-
afluorobenzene is 31.7 × 10–40 C m2 (Qzz =+ 9.5 B); 1 B (Buckingham)6 . The
schematic representation of such a molecule as two like positive charges,
through which the main rotational axis (C6 ), passes, separated by the oppo-
site, balancing negative charges lying perpendicular to the main rotational
axis gives a clear picture of its molecular positive quadrupole value.
Topologically, the quadrupoles can be considered equivalent to d orbitals, as
dipoles are to p orbital [12]. In particular, the Qzz quadrupole of six-membered
6Debye suggested in the 1960s that the quantity 1 unit of charge distributed over 1 Å2 be termed
the Buckingham
aromatic rings can be considered equivalent to a dz2 orbital, having a non-

spherical charge distribution with regions of relative negative and positive
charges. Molecules like C6 F6 having no net dipole have a defined charge dis-
tribution, that is, the molecular quadrupole described above. Consequently,
the Qzz molecular quadrupole of aromatics has the adequate spatial orienta-
tion to be involved in the formation of non-covalent anion-π complexes in
which the halide is located above the arene centroid (anion-aryl centroid geom-
etry). The great majority of the theoretical studies on the anion-π interaction
deal with complexes displaying this type of geometry in the gas phase. It
has been shown conclusively and elegantly by Alkorta [18], and Frontera
and Deyà [19] using the molecular interaction potential with polarization
(MIPp) energetic partition scheme as discussed above that the two funda-
mental components contributing to the stabilization energy of non-covalent
anion-π complexes in the gas phase are electrostatic and anion-induced po-
larization. The electrostatic component correlates well with the magnitude
of the Qzz of the aromatic ring [31]. In molecules having a very positive
Qzz , the electrostatic contribution dominates the anion-π interaction but the
polarization energy is not negligible. For example, 27% of the total energy
(Et =– 16.5 kcal mol–1 ) calculated using MIPp at the MP2/6-31++G∗ for the
interaction of trifluoro-1,3,5-triazine (Qzz =+ 8.23 B) with Cl– when the an-
ion approaches perpendicular to the center of the aromatic ring is due to
polarization energy (Epol =– 4.5 kcal mol–1 ). The electrostatic term for this in-
teraction is the major component Eele =– 12.9 kcal mol–1 while the van der
Waals term is almost negligible Evdw = 0.9 kcal mol–1 . The computed molecu-
lar polarizability of the trifluoro-1,3,5-triazine aromatic ring is α|| = 30.26 a.u.
As the Qzz value diminishes, the term of the electrostatic energy (Eele )
becomes less important. Thus, the contributions to the total interaction en-
ergy (Et =– 6.6 kcal mol–1 ) of 1,3,5-triazine (Qzz =+ 0.9 B) with chloride are:
Eele =– 2.2 kcal mol–1 , Epol =– 4.1 kcal mol–1 and Evdw =– 0.3 kcal mol–1 . It is
worth to note that the computed molecular polarizability of the 1,3,5-triazine
aromatic ring is α|| = 30.34 a.u. very similar to the fluoride derivative. This re-
lationship corroborates the central role of the quadrupole moment value in
anion-π interactions when the anion approaches the aromatic compound per-
pendicular to the center of the aromatic ring. Since the molecular quadrupole
moment describes the electron density of the aromatics, and because a direct
correlation between the interaction energy of the anion binding of positive Qzz
value and the aromatic Qzz value has been observed, it is reasonable to suggest
that the anion-π interaction is governed by electrostatics. Frontera and Deyà
have also demonstrated that the contribution due to polarization (Epol ) to the
total energy calculated with MIPp increases linearly with the computed mo-
lecular polarizabilities of the compounds [31]. This becomes the predominant
term of the total interaction energy for compounds having Qzz values lower
than 1 B. Overall, however, the contribution due to polarization (Epol ) can be
considered as almost constant with a value of 4–6 kcal/mol.
142 P. Ballester
Based on the results of the theoretical calculations, the binding energy of

the anion-π interaction is not 100% electrostatic. In fact, as shown in Fig. 9
the fraction of the total binding energy that is electrostatic varies consider-
ably depending on the aromatic. However, the variation of the anion-π bind-
ing energies is faithfully mirrored by the electrostatic term (plots a and d).
To predict the trend in an anion-π interaction across a series of similar aro-
matics, all with the same anion, i.e., chloride, it is enough to consider the
electrostatic term. It is for this reason that the visual inspection of electro-
static potential surfaces (EPS), which are a good way to visualize the charge
distribution of aromatics and consequently the quadrupole moment, provides
a simple and reliable guide to the relative strength of anion-π interaction
across a series of aromatics interacting with the same anion. The EPSs are
also useful to locate the computed minimum geometry of anion-aryl centroid
complex stabilized by anion-π interaction, although this geometry is not al-
ways observed in the experimental studies (vide infra). The simple consider-
ation of the electrostatic term explains many trends derived from the theoret-
Fig. 9 Plot of the regressions between the quadrupole moments and molecular polariz-
abilities to: a the electrostatic contribution, b polarization contribution and c van der
Waals contribution of the total interaction energy calculated with MIPp for the com-
pounds interacting with chloride at the minimum. d Plot of the regression between the
quadrupole moments to the interaction energy at the MP2/6-31++G∗∗ level of theory for
the compounds interacting with chloride at the minimum. From reference [31]
ical studies although sometimes the major component of the binding energy
could be “non-electrostatic”. It is worth to note that across a series of aromat-
ics compound having quadrupole moment values from Qzz = 8.23 to 0.57 B
the “non-electrostatic” component of the interaction energy (Epol + EvdW )
makes a contribution of –3 to –7 kcal mol–1 . The binding energy of the anion-
π interaction decreases slightly with the increase of the ionic radius of the
anions which is also consistent with an electrostatic model.
Kim et al. [32] also studied the nature of the anion-π interactions using
the symmetry-adapted perturbation theory (SAPT) [33] to obtain a physi-
cal interpretation of the interaction energy. In this method, the interaction
energy is expressed as a sum of perturbative corrections in which each
correction results from a different physical effect. The different intermole-
cular terms obtained from this method can be summarized in electrostatic,
exchange-repulsion, induction, and dispersion contribution. The authors
show that for different halogen complexes with several π systems the elec-
trostatic term follows the same trend as the total interaction energy. This
tendency can be explained by taking in consideration the fact that both dis-
persion and induction energies can be ascribed, to a large extent, to the
interaction of the molecular orbitals of the anion and the π system. The at-
tractive dispersion and induction energies increase as the diffuse electron
cloud of the anion interacts with the substrate. The repulsive exchange inter-
action also depends on the molecular orbitals overlap. This has the effect of
establishing a balance between dispersion-induction and exchange-repulsion
energies resulting in a good correlation between the electrostatic energy and
the total interaction energy. One of the main conclusions of Kim’s work is
that the total interaction energies calculated for the anion-π complexes are
comparable to those obtained for the cation-π complexes. This is a funda-
mental statement also mentioned in other theoretical studies, if one wants
to hypothesise on the strength and importance of the anion-π interaction
in solution. It has been possible to estimate an energy value in the range of
–2 to –0.5 kcal mol–1 for a single cation-π interaction using supramolecular
model systems [34–36] and protein engineering studies [37]. As complement
to these experimental studies, other experimental gas-phase measurements
and high-level theoretical studies have been used to assign a binding energy
of approximately –10 kcal mol–1 for the interaction of tetramethylammonium
with benzene in the absence of solvent [12].
Frontera and Deyà [38] warn that although it is true that the interac-
tion energies of benzene with cations and hexafluorobenzene with anions
are similar, it is not possible to generalize that the interaction energies cal-
culated for the anion-π complexes are comparable to those obtained for the
cation-π complexes. These authors also indicate that the same is applicable to
Kim’s conclusion stating that the largest contribution in anion-π complexes
are electrostatic and induction, because as we have seen before, these con-
tributions sharply depend of the Qzz and α|| values of the aromatic system.
144 P. Ballester
Molecules with negligible Qzz values are expected to interact favorably with
either anions or cations, and the strength of the interaction would be com-
parable, especially if the ionic van der Waals radii of the charged species are
similar. Other conclusions from the work by Frontera and Deyà [38] com-
paring different aspects of the cation-π and anion-π interactions are: a) the
contribution of dispersion and correlation terms to the total interaction en-
ergy are small, but they are more important in anion-π complexes, b) the
density at the cage critical point generated upon complexation of the ion
is a useful parameter for measuring the strength of the interaction, even
when comparing anion-π to cation-π complexes, and c) a gain in aromatic-
ity of the ring is observed, based on the nucleus-independent chemical shift
(NICS) [39] criterion at the center of the ring, upon complexation of the an-
ion, and the contrary is observed for the cation. Many authors agree that the
contribution from the dispersion energy is more important in the anion-π
complexes than in the cation-π interaction. The induction energy emerges
from the interaction of the occupied p orbital of the halide anion and the
LUMO of the π-system. The inductive type of the MO interaction can also
be correlated to the extent of charge transfer from the anion to the π sys-
tem. In fact, several experimental studies have established a charge-transfer
character to anion-π complexes (see below). The degree of charge transfer
in several anion-π complexes has been evaluated using different quantum
chemical approaches. The computed charge transfer reported in the work of
Frontera and Deyà from the anion (F– ) to the π-system is in the range of 0.1 to
0.2|e| and 0.005 to 0.14|e| in Kim’s study. The results based on AIM method-
ology indicate that the charge transfer is almost negligible for all complexes
studied.
Table 5 Binding energies MP2(full)/6-31++G∗∗ //RI-MP2(full)/6-31++G∗∗ (kcal mol–1 )

with (∆EBSSE ) and without (∆E) basis set superposition error correction and anion-aryl
centroid distance (r, Å)
Complex ∆E ∆EBSSE r
TFZ· · ·Cl– –20.3 –15.0 3.008

(TFZ)2 · · ·Cl– –38.5 –28.5 3.006
(TFZ)3 · · ·Cl– –65.6 –41.0 3.019 a
TFZ· · ·Br– –21.8 –14.2 3.176
(TFZ)2 · · ·Br– –41.7 –26.8 3.170
(TFZ)3 · · ·Br– –75.3 –38.6 3.172 a
TAZ· · ·Cl– –9.0 –5.2 3.220
(TAZ)2 · · ·Cl– –17.4 –10.4 3.213
(TAZ)3 · · ·Cl– –39.6 –22.2 3.015 a
a Mean distance
Fig. 10 Schematic geometries of the anion-π complexes studied by Frontera and Deyà [38]
Fig. 11 RI-MP2(full)/6-31++G∗∗ fully optimized structure of the trimeric complex of

TAZ with chloride. The distances between the N atom and the C – H bonds of the 1,3,5-
triazine, as well as, between the anion and the centroid of the aromatic ring are shown
The additivity of the anion-π interactions has also been explored by Fron-
tera and Deyà using high-level ab initio calculations [38]. They optimized
chloride and bromide complexes with one, two, and three aromatic units,
such as trifluoro-1,3,5-triazine (TFZ) and 1,3,5-triazine (TAZ) – and analyzed
the interaction using the AIM theory and studied the charge transfer using
several methods for deriving atomic charges. The results revealed additivities
in the binding energies and complex geometries that are almost insensitive
Table 6 Computed binding energies (∆E, kcal mol–1 ) for TFZ complexes with chloride
simulating two solvents and in the gas phase at the MP2(full)/6-31++G∗∗ /RI-MP2(full)/
6-31++G∗∗ level of theory
Complex ∆E (CHCl3 ) ∆E (H2 O) ∆EBSSE (gas phase)
TFZ· · ·Cl– –5.9 –3.5 –15.0

TFZ2 · · ·Cl– –11.1 –7.0 –28.5
TFZ3 · · ·Cl– –31.7 –24.2 –41.0
receptor –15.8 –11.9 –31.0
146 P. Ballester
Fig. 12 CAChe optimized structure of the chloride complex based on the receptor
proposed by Frontera and Deyà [38]. The receptor binds the anion via four anion-π
interactions
to its stoichiometry (geometry additivity). To speed up the calculations pro-

cess, the authors used the resolution identity MP2 method (RI-MP2) [40, 41].
They state that the interaction energies and equilibrium distances obtained
with the RI-MP2 method in the study of anion-π and cation-π interac-
tions are almost identical to those obtained with the time consuming MP2
calculations.
Calculations simulating two solvents systems (CHCl3 and H2 O) within the
self-consisting reaction field PCM model using the RI-MP2(full)/6-31++G∗∗
geometries were also performed. The additivity for a set of complexes (TFZ
with chloride) is maintained in both solvents, i.e., the binding energy of the
1 : 2 complex is twice the value of the 1 : 1 complex. There is, however, a sig-
nificant reduction of energy in comparison with the gas phase although this
is less important in the 1 : 3 complex.
As we discussed in the introduction, the anion-π interaction has poten-
tial application in the field of molecular recognition of anions. In this sense,
several theoretical studies have proposed structures for novel receptors based
Fig. 13 Molecular structures of the anion receptors proposed by Mascal [13]. The anion is
bound by a combination of three ion-pair reinforced hydrogen-bonding and two anion-π
interactions
on multiple anion-π interactions. Thus, Frontera and Deyà propose a neutral

receptor for the binding of chloride that is composed of a neutral platform
of 1,3,5-triazine substituted by three 2,4-difluoro-6-methylene triazine groups
resulting in a tripodal architecture. The receptor adopts a cup-like cavity that
includes and binds the chloride via four anion-π interactions. The theoret-
ical study simulating water and chloroform has been extended to the receptor
indicating that this type of receptor could be adequate for the binding of chlo-
ride in organic solvents.
Mascal [13] has also proposed the synthesis of three novel receptors for the
binding of anions that take advantage of the high conformational stability of
cylindrophane, which can effectively discriminate guests based on size. The
receptors are designed to bind the anion through a combination of ion-pair
reinforced hydrogen bonds and two anion-π interactions.
Mascal performed a detailed theoretical study of the association of the tri-
azine cage and its cyanuric acid and boroxine analogues with the fluoride
and chloride anions. Since the principal motivation of the work is the se-
lective complexation of anions within the cages, the author compared the
bisector distance (Ar – Ar distance/2 = 2.30 Å average) for the empty re-
ceptors with the anion-aryl centroid distance of halide sandwich complexes
(raverage Ar···Cl– = 3.21 Å and raverage Ar···F– = 2.45 Å), which are analogous to
the 1 : 2 complex studied by Frontera and Deyà. The energies calculated for
the sandwich complexes are approximately additive, which is in agreement
with the work of Frontera and Deyà. The comparison of the bisector distances
with the anion-aryl centroid distance of the sandwich complexes suggest that
both chloride and fluoride are too large for all cage-like receptors. In the three
receptors, however, the optimal fluoride distance is better accommodated for
the empty cavity.
Table 7 Interaction energies (kcal mol–1 ) for the complexes of the triazine cages (TAZ)
and its analogues cyanuric acid (CNA) and boroxine (BOX) with fluoride and chloride
ion
a a
Complex ∆EBSSE ∆EF–Cl b ∆EBSSE
H2 O ∆EF–Cl(H2 O) b
TAZcage· · ·Cl– –237.5 –1.1

CNAcage· · ·Cl– –245.0 –13.4
BOXcage· · ·Cl– –245.8 –21.8
TAZcage· · ·F– –292.6 –55.1 –30.0 –28.9
CNAcage· · ·F– –296.3 –51.4 –36.9 –23.5
BOXcage· · ·F– –295.6 –40.7 –35.7 –13.9
a BSSE corrected B3LYP/6-31+G(d, p) interaction energies in the gas phase (∆EBSSE ) and
in an aqueous solvent model (∆EBSSE
H2 O )
b Differences between the energy values of the fluoride and chloride complexes
148 P. Ballester
The binding properties of the halide complexes (F– , Cl– ) were evaluated
in the gas phase and solution. Energies were also determined using the
conductor-like polarisable continuum model for water, with the molecular
cavity specified by the United Atom Topological Model applied on radii op-
timized for the PBE0/6-31g(d) level of theory. The complexation of fluoride
with the three receptors is calculated to be more energetically favorable than
chloride by >40 kcal mol–1 in the gas phase and >13 kcal mol–1 in water. It is
worth noting that the complexation energies calculated for F– ion are about
three times the experimental enthalpy of F– hydration. The author indicates
that the raw comparison of these values is of limited predictive significance
in regard to the potential of these receptors to extract F– from water, given
the absence of activation and entropy values, as well as, the assumptions in-
herent to nonexplicit, polarized continuum models of water. The comparison
of relative energetics of complexation, however, clearly indicates a preference
for fluoride binding over chloride, from the gas phase to water, in all three
receptors (see Table 7).
The principal interaction force in these complexes is the ion-pair rein-
forced hydrogen bond of the ammonium groups with the anions. It may be
supposed that the intrinsic stronger ammonium· · ·F– interaction could actu-
ally form the basis for the difference in binding energy. The isolation of the
relative binding contributions (ion-pair reinforced hydrogen bond and anion-
π interactions) was achieved by modelling the association of three NH4 + ions
in a trigonal plane around a chloride and fluoride ions. The obtained results
are shown in Table 8.
The high differences in binding energies calculated for the cage complexes
and the modelled NH4 + trigonal systems are clearly due to the fact that the
three ammonium groups are enforced in close proximity in the free recep-
tor. The point to note is that although the stabilization of three NH· · ·F–
bonds in the trigonal (NH4 + )3 · · ·F– complex is about 11 kcal mol–1 greater
than the corresponding (NH4 + )3 · · ·Cl– complex, in the triazine and cyanuric
acid cage complexes there is still >10 kcal mol–1 “additional” stability in the
fluoride complexes in aqueous solution. This observation suggests that the
discrimination has to do with a better fit of the anion at least in these two
Table 8 Interaction energies (kcal mol–1 ) for NH4 + · · ·X– complexes
a a
Aggregate ∆EBSSE ∆EF–Cl b ∆EBSSE
H2 O ∆EF–Cl(H2 O)
(NH4 + )3 · · · Cl– –130.0 –10.8

(NH4 + )3 · · · F– –161.9 –31.9 –21.6 –10.8
a BSSE corrected B3LYP/6-31+G(d, p) interaction energies in the gas phase (∆EBSSE ) and
in an aqueous solvent model (∆EBSSE
H2 O )
b Difference between the energy values of the fluoride and chloride complexes
Table 9 Interaction energies (kcal mol–1 ) and noncovalent bonded distances for selected
complexes of 1,3,5-triazine (TAZ) and boroxine (BOX) with chloride and fluoride anions
a
Complex ∆E a ∆EBSSE ∆EF–Cl b rc
TAZ· · ·Cl– –4.2 –4.0 3.48

TAZ· · ·F– –10.4 –9.0 –5.0 2.66
BOX· · ·Cl– –12.9 –12.5 2.94
BOX· · ·F– –31.9 –29.6 –17.1 1.93
TAZ· · ·Cl– · · ·TAZ –8.1 –7.6 3.49
TAZ· · ·F– · · ·TAZ –18.9 –16.5 –8.9 2.79
BOX· · ·Cl– · · ·BOX –23.1 –22.1 3.02
BOX· · ·F– · · ·BOX –47.2 –43.9 –21.8 2.23
TAZcage –55.1
BOXcage –40.7
a B3LYP/6-31+G(d, p) interaction energies without (∆E) and with (∆EBSSE ) BSSE correc-
tion
b Difference between the ∆EBSSE values of the fluoride and chloride complexes
c Noncovalent bond distance (r) for simple (Ar· · ·X– ), sandwich (Ar· · ·X– · · ·Ar). Corres-
ponding data for the cage complex can be found in Table 7
Fig. 14 Structure of the tweezers studied in [42]
Table 10 Calculated interaction energies (kcal mol–1 ) of the investigated tweezer-ion com-
plexes
Complex ∆Ecomplex a ∆Einter-MP2 b
6FT 5.20 –4.06

10FT –0.28 –9.72
14FT –5.95 –15.82
a Obtained by using B3LYP functional ∆Ecomplex = ∆Einter + ∆Edeform
b Obtained by performing single point MP2 calculations on the optimized geometries
150 P. Ballester
cages. Surprisingly, the order of complementarity or selectivity for F– vs Cl– is

TZNcage > CNAcage > BOXcage, which is completely reverse to the discrim-
ination of the halide ions observed in simple binding complexes.
Hermida-Ramón and Estévez [42] have recently conducted a theoretical
study showing that fluorinated tweezers, that derive from those synthesized
by Kläner [43] were able to bind anions. Kläner tweezers are generally used
for the molecular recognition of eletrodeficient aromatic and aliphatic sub-
strates as well as organic cations. Several complexes formed between the
fluorinated tweezers derivatives and an iodide anion were characterized. The
nature of the interaction was analyzed using the SAPT method while the en-
ergetics for the complexation process were computed using calculations at the
MP2 level on complex geometries that were optimized using density func-
tional theory (DFT) and the B3LYP functional. The differences between MP2
and B3LYP interacting energies can be considered as a rough approximation
of the importance of the dispersive interactions in the stability of the tweezer-
anion complex.
The values of the molecular electrostatic potential in the center of the cav-
ity are –10.25 (6 FT), –5.79 (10 FT) and –0.64 (14 FT) kcal mol–1 . Clearly, the
best receptor to accommodate iodine in its cavity is 14 FT, in which the MEP
is almost neutral.
An increase in the fluoride substitution produces a stabilizing trend in the
energetics of the complexes. A comparison between the MEP value and the
binding energies indicates that the stability comes from the depletion of the
electrostatic repulsion between the anion and the π-clouds of the aromatic
rings, which have a lower electron density owing to the influence of fluorine.
The decrease in electrostatic repulsion is accompanied by a decrease in the
distance between anion and the tweezer together with an increase in attrac-
tive energies (induction and dispersion), which gives more stability to the
complexes. Calculations that include a polarizable continuum model of the
solvent (H2 O) indicate that the complex is not stable. It must be noted that
for the I– @14 FT complex, a large attractive binding energy is calculated even
though the electrostatic potential inside the cavity is slightly negative.
Lewis and Clements [44] have performed quantum mechanical computa-
tions that show that negative Qzz aromatics bind anions in the gas phase,
dispelling the idea that the π electron density of negative Qzz aromatics is
only appropriate for cation binding. No correlation was observed, however,
between anion binding enthalpies and the Qzz values for negative Qzz aro-
matics. This observation is in striking contrast to what has been reported
and mentioned above for positive Qzz aromatics binding anions. The authors
do indicate that the Morokuma-Kitaura decomposition calculations show that
the major contribution to binding in the case of negative Qzz aromatics with
anions is the energy due to polarizability of the aromatic ring. They also
mention that it may be not the polarizability of the aromatic ring density
that is responsible for the binding but rather it may be the polarizability of
Fig. 15 Schematic representations of the geometries for Cl– complexes with TCBm
the aromatics substituents. The work concludes indicating that even electron
rich, negative Qzz aromatics should be considered experimentally tenable for
anion-π interactions. We believe that this statement should be taken with
serious caution.
Johnson, Hay et al. [45] have refined the nature of the interactions be-
tween electron deficient arenes and halide anions. In particular, they have
performed calculations at the MP2/aug-cc-pVDZ level of theory of 1 : 1 com-
plex formed between 1,2,4,5-tetracyanobenzene (TCB) and F– , Cl– , Br– an-
ions. Four geometries were evaluated for each halide. The well-known anion-
aryl centroid (A) (termed non-covalent anion-π complex in this study), two
charge transfer (CT) complexes in which the halide is positioned above
a C – H bond (B) or above a C – CN bond (C), and a C – H hydrogen bond
complex (D) similar to the one already discussed by Mascal in his seminal
study of the interaction of triazine with halide anions [17]. The theoretical
analysis of TCB halide complexes revealed the first example in which non-
covalent anion-aryl centroid complexes are not stable for Cl– and Br– . These
halides do interact with the π system, but the interaction involves CT, char-
acterized by a high second-order stabilization energy ∆E(2) . The resulting
optimized geometries of the CT complexes locate the anion over the periph-
ery ring rather than over the center of the ring.
Since as we have presented above prior computational studies on Cl– and
–
Br have focused almost exclusively on the anion-aryl centroid complex, there
has been no investigation to determine the existence of such CT complexes
for other arenes. Consequently, the authors expand the electronic structure
calculations to halide complexes with triazine, hexafluorobenzene and 1,3,5-
152 P. Ballester
Fig. 16 Calculated EPS at the HF-6-31G level for TCB. The range of energy values is
–39 (red) to +39 (blue) kcal mol–1
tricyanobenzene. The results with F– and triazine are fully consistent with
earlier work [17]. Although the anion-aryl centroid and hydrogen-bonding
geometries have been previously identified as minima for Cl– and Br– [17],
none of the previous studies report the existence of an “attack” geometry
for either Cl– or Br– only for F– . Two geometries were located for hexafluo-
robenzene, the anion-aryl centroid which is not a stable point for F– and the
“attack” geometry that is the global minimum for F– but was not located as
a stable point for Cl– or Br– . These results are in complete agreement with
prior theoretical studies [18, 19]. Results obtained for 1,3,5-tricyanobenzene
are similar to those obtained for TCB. The anion-aryl centroid geometry is
only stable for Br– . The CT complex that locates the anion above a C – H bond
is the global minimum for the three halides. The hydrogen bond geometry is
also a stable structure for all the halides. All anion-aryl centroid complexes
have a low extend of charge transfer. Alternate geometries do have a high
extend of charge transfer. It is worth noting that for the interaction of F–
with triazine in the “attack” geometry, the computed values for the param-
eters used to gauge the extent of charge transfer, indicate the formation of
a strong σ bond. In conclusion, triazine, TCB and 1,3,5-tricyanobenzene form
stable off-center CT complexes with Cl– and Br– . Except for the CT complex
of Br– with triazine, the CT complexes are more stable than the anion-aryl
centroid complexes. The differences in energies, however, are very small
(<1 kcal mol–1 ), indicating a relative flat potential surface for positioning the
halide above the arene plane. In fact, the EPS of TCB shows a widespread
distribution of positive electrostatic potential values all over the π-system.
3
Experimental Evidence of the Anion-π Interaction
3.1
Solution and Related Crystallographic Studies
To the best of our knowledge, the first report suggesting the experimen-
tal existence of attractive interactions between anionic negative charges and
π-electron aromatic systems was published by Schneider et al. in 1993 [9].
The 1 H NMR measurement of the association constant between a simple
Fig. 17 Schematic representation of the complex formed with the cleft-like para-sulfonato
receptor and diphenylamine placing the ionic part above the aromatic guest center
diphenylmethane organic host bearing negatively charged para-sulfonato

groups and an electroneutral diphenylamine show in aqueous solution
weak but attractive interactions that were quantified to reach approximately
0.5 kcal mol–1 per X– /arene unit7 .
The small complexation induced shift observed for the proton Hp
(0.09 ppm) is interpreted by the authors as the existence of a displacement of
the aryl rings in the complex. The authors state “such a slight displacement
will replace repulsions between permanent negative charges of the π-cloud
and of the sulfonato substituent by attractions, as the π-cloud can still be po-
larized by the charge of the sulfonato even if the charge is not exactly above
the arene center”. In other words, although anion-π interactions are claimed
to be responsible for the stability of the complex, the proposed geometry does
not coincide with the one mainly investigated by the theoretical calculations
that locates the anion above the π-aromatic systems. The nature of the attrac-
tion is primarily assigned to the interaction between the anion and the dipole
that it induces in the arene (polarization). While the theoretical studies have
revealed that this “nonelectrostatic” term maybe important, it seems that it’s
not the defining feature of the anion-π interactions. In this and another re-
port [46], Schneider claims that the anion-π interaction can have almost the
same size as the cation-π interaction.
In 2004, an experimental study on the solid-state and solution behavior
of halide complexes with a series of highly electron-deficient arenes (Fig. 18)
also suggests that the complex geometry investigated theoretically, wherein
the anion is located along the principal axes of the π-system, may not be
operative in these highly electron deficient arene systems [47].
In this study, the recognition of halide anions X– (X = Cl, Br, I) is clearly
established by the isolation and X-ray structure determination of a series of
well-defined complexes containing the halide salt and the admixed aromatic
7 This study will suggest anion-arene interactions involving aromatics with negative Qzz values.
A note in reference [44] states based: “on the systems studied and figures presented in the publica-
tion it appears that the binding energy was probably due to arene-arene face to face interactions”.
154 P. Ballester
Fig. 18 Molecular structures of the neutral organic π-acceptors investigated in the

study [47]
compounds in different stoichiometries, as well as by the spectral assign-

ment of diagnostic charge transfer absorption bands to the formation of the
anion-π complex. Intense colorations were observed upon the addition of
the anions as the corresponding tetraalkylammonium salts to acetonitrile
solutions containing the aromatic π acceptor, i.e., tetracyanopyrazine TCP
(calculated Qzz = 18.53 B) [48]. A new absorption band appears and grows
with increasing halide concentration. Close inspection of the new absorption
band reveals that it consists of two Gaussian components. The stability con-
Fig. 19 Mulliken dependence of the energy of the low-energy absorption band (νCT ) in
the TCP/Hal– complexes and the oxidation potential of the anion
Table 11 Solid-state characteristics of the halide complexes obtained by slow diffusion of

hexanes into 1 : 1 mixtures of neutral organic π-acceptors with the corresponding halides
as tetraalkylammonium salts in CH3 CN/CH2 Cl2
Molar ration Counterion X– · · · C [Å] a
TCP/Br– 3:2 Et4 N+ 3.16

4:1 Pr4 N+ 3.15
TCP/I– 2:1 Et4 N+ 3.52
1:1 Bu4 N+ 3.49 b
TCP/Cl– 4:1 Bu4 N+ 3.07
o-CA/Br– 1:1 Pr4 N+ 2.93
a The X– · · · C distance in closest contacts; note that the sums of the van der Waals radii
are 3.45 Å (Cl– · · · C), 3.55 Å (Br– · · · C), and 3.65 Å (I– · · · C)
b The average of the distances to the two or three neighbouring acceptors is given
stant values calculated for the complexes formed in acetonitrile with TCP and
Br– as different tetraalkylammonium salts are in the range of 7–9 M–1 . Fur-
thermore, a job plot indicates a 1 : 1 stoichiometry for the TCP/Br– complex
formed in solution. A clear correlation between the energy of the low-energy
band and the oxidation potential of the anion was also observed and used to
establish a charge-transfer character for the complexes.
Furthermore, the study of the interaction of Br– with the series of neu-
tral organic π-acceptor reveals that the increase of acceptor strength of the
aromatic (characterized by a positive shift of the reduction potential) is ac-
companied by a bathochromic shift of the new absorption band (from 355 to
465 nm). This observation further confirms the charge-transfer (CT) charac-
ter of these complexes.
The isolation and X-ray characterization of single crystals containing the
halide salt and the aromatic compound allows the identification in the solid-
state of the non-covalent anion-π complexes. The anion is located in the space
Fig. 20 Local packing of the solid-state structures of the Br– complexes with TCP. Left:
Bromide used as tetraethylammonium salt. Right: Bromide used as tetrabutylammonium
salt. Only molecules that are in short contact (< sum of vdw radii) with respect to the Br–
atom are shown. In the case of the [(TCP)3(Br– )2 ](Et4 N+ )2 complex, the two different Br–
ions located within the asymmetric unit are represented
156 P. Ballester
close to the electron deficient aromatic ring explaining the observed elec-
tronic (charge-transfer) transitions. The bromide ion lies approximately 3 Å
over the periphery of the aromatic ring. The observed complex geometry dif-
fers from those derived from quantum-mechanical calculations that usually
show the anion placed along an axis perpendicular to the aromatic ring and
passing through its centroid. The existence of multiple halide-π interactions
becomes evident either directly in the asymmetric unit or after the packing
of the lattice. The complex stoichiometries that can be derived from the X-ray
molecular formula of the asymmetric unit i.e., [(TCP)3 (Br– )2 ](Et4 N+ )2 (1.5 :
1), [(TCP)4 Br– ]Pr4 N+ (4 : 1) and [(TCP)I– ]Bu4 N+ (1 : 1), [(TCP)2 I– ]Et4 N+
(2 : 1), as well as, the number of contacts established between the anion and
the aromatic rings seems to be controlled not only by the anion itself but by
the size of its countercation, which should influence the three-dimensional
solid structure.
Johnson, Hay et al. [45] have used 1,2,4,5-tetracyanobenzene (TCB) to
gain further structural information of the anion-π interaction. TCP, 18-
crown-6 and alkali halide were dissolved in acetonitrile (KBr) or 9 : 1
dichloromethane:acetonitrile (KI, NaI) mixture. The purpose of the crown
ether is to enhance the solubility of the salt. It is important to note that the
solution of the crown ether and TCB is colorless, however, the addition of the
halide salt produces a color change consistent with the formation of CT com-
plexes and in complete agreement with Kochi’s previous observation using
TCP instead. Slow evaporation at room temperature yielded single crystals
suitable for X-ray analysis. Despite packing differences, the local environment
about each halide is remarkably similar in all crystals. As shown in Fig. 21,
four TCB molecules surround the anion at interaction distances. Three dif-
ferent orientations can be distinguished for the anion: above the arene plane
nearest to a C atom bearing a CN group (a and b), above the arene plane near-
est to a C atom bearing a H atom (c), and nearly within the plane of the arene,
forming a C – H hydrogen bond (d). This latter orientation was not observed
in the study of Kochi. In conclusion, Kochi’s and Johnson’s results clearly
Fig. 21 Local packing of four TCB molecules around the anion. The closest contact to each
arene is indicated by a dotted black line
show that the anions are not located over the center of the arene (anion-aryl
centroid geometry) in the solid state structures of these complexes. As dis-
cussed above, theoretical studies indicate that TCB and 1,3,5-tricyanobenzene
are capable of forming stable off-center CT complexes with Cl– and Br– .
In 2003, Hoffmann et al. [49] introduced an uncharged host for the selec-
tive binding of Cl– as the tetrabutylammonium salt in CDCl3 : DMSO 95 : 5
– /K – = 105). This host features a triazine-trione platform with three short
(KCl Br
arms that are conformationally preorganized through the introduction of
methyl groups. The three arms are equipped with p-nitrophenylsulfonamide
groups capable of hydrogen-bonding to the anion. In this study, they also used
an analogous receptor with slightly longer side chains in order to evaluate
how the relative position of the anionic host with respect to the triazine-trione
platform (anion-π interaction) could affect the stability of the complex.
The binding parameters of both hosts with chloride were determined
using isothermal titration calorimetry. The microcalorimetry experiments re-
Fig. 22 Molecular structures of the receptors based on the triazine-trione platform hav-
ing three arms of different length terminated with p-nitrosulfonamide hydrogen-bonding
groups
Table 12 Thermodynamic binding parameters determined using ITC for the complexation
of tetrabutylammonium chloride by the tris-sulfonamides in CHCl3 at 298 K
Host long arm Host short arm
K/L mol–1 81 000 ± 10 000 155 000 ± 11 000

∆G/kcal mol–1 –6.7±0.1 –7.1±0.1
∆H/kcal mol–1 –4.7±0.1 –2.2±0.1
– T∆S/kcal mol–1 –2.1±0.1 –4.8±0.1
158 P. Ballester
vealed that the host with the longer arms binds chloride slightly stronger that
the one with the shorter arms (∆∆G ≈ 0.4 kcal mol–1 ).
There are, however, more important differences in the binding processes
of the hosts to the tetrabutylammonium chloride. Although both complex-
ation processes of chloride by the hosts are entropically favored, due to the
release upon binding of ordered solvent molecules from the surface of the
chloride and the host to the bulk, the entropic change (– T∆S) for the host
with longer arms is far less negative. This result may be the consequence of
a higher reduction in the rotational entropy of the host with the long arms on
complex formation. In terms of enthalpy, the binding of chloride by the host
with short arms is clearly less exothermic. In order to explain the observed
result, the authors speculate about the relationship between complex geom-
etry and enthalpy. The more flexible host could adopt a conformation that is
more optimal for coordination of chloride giving rise to higher exothermic-
ity. Furthermore, they conclude that from the measured enthalpy values of
complexation a potential electrostatic attraction of chloride by the triazine-
trione platform in the receptor with the short arms, an anion-π interaction,
can be discarded. We believe that the thermodynamic parameters measured
for the chloride complexation with these systems are highly contaminated
with solvent effects. If the number of solvent molecules released to the bulk
on complexation is different for each host, then the comparison of the data is
completely inappropriate. In terms of free energy of binding, the experimen-
tal results indicate that both chloride complexes formed with the two triden-
tate p-nitrophenylsulfonamide hosts have approximately the same binding
energy and the potential electrostatic attraction of chloride by the triazine-
trione (cyanuric) platform is not reflected.
In 2005, Frontera, Saczewski, Deyà et al. [50] performed a combined
crystallographic and computational study of the anion-π interactions with
Fig. 23 Minimized structures (Maestro, MMFFF) of the chloride complexes formed with
the tridentate p-nitrophenylsulfonamide having long and short arms. The distance of the
chloride atoms to the centroid of the cyanuric acid ring is shown in each case
Fig. 24 Synthesis of the halide salts of the 2-ethyleneamine derivatives of thio- and dithio-
cyanuric acid
cyanuric acids [50]. These authors interpreted the experimental results pre-
viously obtained by Hoffmann et al. making use of MIPp calculations. The
MIPp value for the interaction of cyanuric acid with a chloride anion placed
2.5 Å apart from the ring centroid is almost the same to that computed at
a distance of 3.5 Å (≈–16 kcal mol–1 ). Consequently, a likely explanation for
the similarity of free energies of complexation measured experimentally with
long or short arms tridentate p-nitrophenylsulfonamide hosts is that the ef-
fect of the cyanuric platform on the anion binding is similar and does not
allow to differentiate one with respect to the other.
To obtain experimental evidences of the ability of cyanuric acid to interact
favorable with anions, Frontera, Saczewski, Deyà et al. [50] synthesized
several derivatives of thiocyanuric and dithiocyanuric acid with a flexible
2-ethyleneamine arm attached to the s-triazine ring. The one-step synthesis
of these compounds requires a 1–4 h reflux in aqueous 12% HX (X = Cl, Br,
Fig. 25 Top view and side view of the local packing of the chloride in the crystal structure
of the hydrochloride salt of the 2-ethyleneamine thiocyanuric acid derivative
160 P. Ballester
Fig. 26 Top view and side view of the local packing of the chloride in the crystal structure
of the hydrochloride salt of the 2-ethyleneamine dithiocyanuric acid derivative
Fig. 27 Top view and side view of the local packing of the bromide in the crystal structure
Fig. 28 Top view and side view of the local packing of the iodide in the crystal structure
I) acid. After cooling the reaction mixture to ambient temperature, colorless

crystals suitable for X-ray analysis of the corresponding hydrohalide salts of
the amine were obtained.
The X-ray crystal structures of all the hydrohalide salts except the hydro-
bromide of the thiocyanuric acid derivative revealed the existence of anion-π
interactions. The organic cation, which has several hydrogen-bond donor and
acceptor functionalities, displays an electrostatic interaction between the N
atom of the ammonium moiety and an anion (X– ) located nearly above the
center of the electron-deficient ring. The packing of the lattice reveals that the
anion (X– ) placed on top of the aromatic rings also interacts electrostatically
with at least one ammonium group of an adjacent cyanuric acid derivative.
Additionally, one of the NH groups of the s-triazine ring of another adjacent
organic cation binds the anion (X– ) through a hydrogen-bond interaction.
Finally, two other molecules of the organic cation are also surrounding the
anion. One of them has the 2-ethyleneamine arm positioned at short contact
of the anion (distance < sum of vdW radii).
The distance between the ring centroid and the chloride anion is 0.1 Å
longer in the dithione than in the monothione derivative. In conclusion, the
X-ray structures of the halide salts of mono- and dithiocyanuric acid deriva-
tives having an ethylenenammonium arm attached to one of the nitrogen
atoms of the s-triazine do show the existence of anion-π accompanied by salt-
bridge and hydrogen-bonding interactions as a robust structural motif of the
solid-state packing.
In an attempt to probe the efficacy of the non-covalent anion-π interac-
tion, Johnson et al. [51] have prepared two sulfonamide-derived receptors
(A and B in Fig. 29). The design of the receptor is based on a convergent
two-point recognition motif utilizing both a hydrogen bond and an aromatic
ring that can be involved in the formation of anion-π interactions. The pre-
pared receptors differ in the electronic properties of one of the rings of the
biphenylamino moiety due to the substitution with five fluorine atoms. The
quadrupole moment of such an aromatic ring should be highly positive while
the simple phenyl should have a negative quadrupole moment.
Receptor B has a pentafluoro substituted aromatic ring that is electron-
deficient and more appropriate to engage in an attractive anion-π interaction.
Fig. 29 Molecular structures of the receptors used by Johnson et al. to evaluate the anion-
π interaction in solution
162 P. Ballester
Fig. 30 CPK representation of the single-crystal X-ray structure of the pentafluorophenyl

receptor. The receptor adopts in the solid-state a conformation that is adequate to interact
simultaneously through hydrogen-bonding (sulfonamide NH) and anion-π interaction
with an halide located on top of the pentafluorophenyl ring
Table 13 Ka (M–1 ) for the sulfonamide receptors with Cl– , Br– and I–
Anion Phenyl receptor (A) Pentafluorophenyl receptor (B)
Cl– <1 30 ± 3
Br– <1 20 ± 2
I– <1 34 ± 6
The tetraethylammonium salts of each anion were used
The stability constants of the complex formed with receptor B and a series
of anions (Cl– , Br– , I– ) were evaluated in CDCl3 using NMR titration tech-
niques. Receptor B binds all the screened halides with a measurable but mod-
est association constant. The association constants measured were 20 M–1 for
Fig. 31 X-ray crystal structure of the complex formed between the pentafluorophenyl
receptor B and tetra-n-butylammonium bromide. The sulfonamide receptor and the or-
ganic cation are shown in stick representation while the bromide is represented with van
der Waals surface
Br– , 30 M–1 for Cl– and 34 M–1 for I– . On the other hand, the association
constant for receptor A lacking the electron-deficient aromatic ring required
for the anion-π interaction, were too small to be determined using the same
titration methodology. The enhanced affinity exhibited by receptor B over re-
ceptor A with the array of halides is attributed to the existence of anion-π
interaction in the anion complex.
The experiments presented by Johnson et al. support further studies on
the possible use of the anion-π interaction as an emerging no covalent inter-
action of the selective targeting of anions in solution. The difference in the
reported stability constants (∆∆G) can be used to estimate a value for the
anion-π interaction in the range of approximately –2 kcal mol–1 . This value
almost doubles the value estimated for the cation-π interaction using pro-
tein engineering [37] (–2.75 kcal mol–1 for a cavity lined by three π-systems
– 2.75/3 = 0.9 kcal mol–1 ) and supramolecular chemical model systems [35]
(–2.4 kcal mol–1 for a receptor with four π-systems – 2.4/4 = 0.6 kcal mol–1 ).
Probably, the anion-π interaction is over-evaluated when using the chem-
ical model based on sulfonamides. Johnson et al. performed the 1 H NMR
titrations experiments in CHCl3 , and the literature indicates that tetraalky-
lammonium salts form tight ion pairs with small anions, i.e., Cl– in CHCl3
solutions [34, 52]. Consequently, the sulfonamide receptors bind the tight
ion-pair and the observed enhancement of the association constants for the
pentafluorophenyl receptors could be partially caused due to the existence of
electrostatic interactions between the fluorine atoms and the tetraalkylam-
monium salt. The X-ray crystal structure of the pentafluorophenyl receptor
and tetra-n-butylammonium bromide, in which the ion-pairing in the solid
state is held responsible to force the sulfonamide NH away from the preferred
conformation, shows close contacts between the fluorine atoms and the or-
ganic cation (Fig. 21).
3.2
Further Crystallographic Evidence of Anion-π Interactions
Examples of crystal structures of supramolecular complexes in which an-

ions are nested in the interior of aromatic cavities have been used as ev-
idence for the existence of an attractive anion-π interaction. In 2004, the
first crystallographic evidence of anion recognition by aromatic receptors was
described by Meyer et al. [53], and Gamez, Reedijk et al. [54]. While inves-
tigating the structural and magnetic properties of metal complexes with the
hexakis(pyridine-2-yl)-[1,3,5]triazine-2,4,6-triamine ligand (L), Meyer syn-
thesized a copper (II) chloride complex that showed chloride-triazine bind-
ing with geometrical parameters almost exactly to those calculated com-
putationally for the 1,3,5-triazine· · ·Cl– complex. The cationic moiety of
[L2 (CuCl)3 )]3+ consists of two ligands L that are stacked in a parallel fashion
and held together by three copper(II) ions. The four pyridine-N groups and
164 P. Ballester
Fig. 32 Left: molecular structure of the ligand (L). Right: side view and top view of the
copper(II) chloride complex, few solvent molecules, some hydrogen and chloride atoms
are omitted for clarity. The atoms Cl(8) and Cl(5) are shown in CPK representation. Spe-
cial emphasis is placed on the location of these two atoms in relation to the triazine ring
of the cationic moiety of the complex [L2 (CuCl)3]3+
an apical chloride constitute a square pyramidal coordination of each metal

center. Cl atoms of Cl2 CH2 are placed between the paddles of the structure.
Both triazine rings are arranged in almost perfect face-to-face arrangement.
The most interesting structural feature related to the anion-π interaction is
the position of the chloride atoms 8 and 5 belonging to [CuCl4 ]2– ions. Cl(8)
is located above one of the triazine rings in an anion-aryl centroid geometry.
The distance between the centroid of the ring and the anion is 3.17 Å while
the angle Cl– · · ·centroid axis to the plane of the ring is 87◦ , that is, the chlo-
ride is almost located on the C3 axis above the ring. The Cl(5) chloride shows
a similar geometry.
Fig. 33 Side view and top view of a fragment of the crystal structure of the Cl– complex
reported by Meyer. The six hydrogen-bonding interactions with the aryl C – H groups of
the Cu(II) coordinated pyridine rings are indicated. C – H· · ·Cl– angles ≥150◦ and C to
Cl– distances range from 3.9 to 4.3 Å
Fig. 34 Left: Molecular structure of azadendritz. Right: side views and top view of
the crystal structure of the Cu(II) complex of azadendritz reported by Gamez, Reedijk
et al. [54]. Solvent molecules and hydrogen atoms are omitted for clarity. The structure
shows the four pentacoordinate Cu(II) ions coordinated two apical chloride atoms (scaled
ball-and-stick representation) and two encapsulated chloride anions (CPK representa-
tion)
The authors state that a “Cl– · · ·triazine complex due to electrostatic anion-
π interaction is present in the complex”. Johnson, Hay et al. [45] have recently
noticed that the role of the C – H groups of deficient arenes as potent hydro-
gen bonds should not be overlooked, in fact the structure of Meyer nicely
illustrates this observation. The Cl(8) anion positioned above the center of
the melamine is interacting with the C – H groups of the Cu(II)-coordinated
pyridines. The hydrogen bonds formed may play a dominant role in deter-
mining the position of the anion within this cavity.
Almost at the same time, Gamez, Reedijk et al. [54] described the first co-
ordination compound of the ligand azadendritz. The supramolecular Cu(II)
complex shows intramolecular π–π interactions between two 1,3,5-triazine
rings.
Similar to the structure reported by Meyer, four pyridine-N groups and
an apical chloride constitute a square pyramidal coordination of each metal
center. In this case, however, the chloride atoms are not located on top
of the 1,3,5-triazine rings, which are perfectly stacked, instead the anion
is nestled against four aromatic pyridyl residues. In Meyer’s structure, this
same position was occupied by Cl atoms of Cl2 CH2 molecules. Each en-
capsulated chloride is in close contact with the four pyridine rings with
centroid· · ·Cl– distances ranging from 3.5 to 3.7 Å. It has to be noted that
in both examples the electron-poor character of the pyridine moieties is
probably enhanced by their coordination to the Cu(II) metals. In addition
to the π interaction with the pyridine rings the authors noted that the
chloride ions are in close proximity to the triazine rings, suggesting the
existence of some additional electrostatic interactions with the electron de-
ficient triazine moiety. The angles of the Cl– · · ·centroid axis to the plane
of the different pyridine rings ranges from 74◦ to 82◦ . This result en-
couraged the authors to investigate further the anion · · · triazine interaction
166 P. Ballester
using other triazine-based ligands and anions [55, 56]. Thus, the reaction
of Zn(NO3 ) · 6H2 O or Cu(NO3 ) · 3H2 O with the ligand dipicatriz in acetoni-
trile afforded mono- or trinuclear coordination complexes depending on the
metal-to-ligand ratio used during the crystallization process [56]. In par-
ticular, the trinuclear Zn(II) complex shows anion-π interactions between
two nitrate anions coordinated to two different zinc atoms and the same tri-
azine ring, one in each face of the aromatic ring. On the other hand, the
corresponding trinuclear Cu(II) complex exhibits an even more remarkable
double anion-π interaction between the triazine ring and two nitrate anions.
In this case, the two nitrate anions are now uncoordinated. One of the ni-
trates interacts via a somewhat longer distances, which is most likely due
to the fact that the triazine is already involved in a nitrate-centroid contact.
These examples constitute one of the first reports on anion-π-anion inter-
action. It is also worth noting that the observed spatial arrangement of the
nitrate anion with respect to the triazine ring is distinct to the one pro-
posed by Kim et al. [32] based on theoretical calculations. Gamez, Reedijk
et al. [56] performed ab initio calculations demonstrating that the interaction
energies calculated for this peculiar positioning of the nitrate is compara-
ble to the energy of the binding geometry proposed by Kim et al. in his
theoretical studies of the interaction of triazine with nitrate [32]. In the
model proposed by Kim et al., the anion is parallel to the triazine ring and
the oxygen atoms are located on top of the electropositive carbon atoms
of the ring.
Related to the previous example Gamez, Reedijk et al. [55] also found
crystallographic evidences of another geometry for the nitrate-π interac-
tion in the tetranuclear complex [Cu4 (dpatta)(NO3 )4 ](NO3 )4 obtained under
hydrothermal conditions from the reaction of copper(II) nitrate and the
ligand dpatta [55]. Each two triazine units are perfectly π–π stacked but
staggered with a centroid-to-centroid distance of 3.45 Å. Both triazine rings
Fig. 35 Left: Molecular structure of dipicatriz. Right: side views of the crystal structures
of the Zn(II) and the Cu(II) trinuclear complexes of dipicatriz. The structures show the
three coordinate Zn(II) and Cu(II) ions in scaled ball-and-stick representation and the
distances between the centroids of the triazine ring and the closest oxygen atoms of the
interacting nitrate anions
are alternatively involved in an interaction with a lattice nitrate but with

a different geometry. A DFT analysis was performed with a model complex
triazine · · · nitrate to analyze the anion-π interaction. In particular, the pref-
erence for the coordination of the nitrate ion observed in the X-ray structure
with one oxygen pointing to one nitrogen of the triazine ring. A DFT-B3LYP
geometry optimization indicates that a local minimum exists with a geometry
very similar to that observed in the X-ray crystal structure. The plane of the
nitrate is not completely parallel to the triazine ring and one of the oxygen
atoms points in the direction of one of the N atoms of the ring. This calcu-
lated structure is 0.7 kcal mol–1 more stable that Kim’s proposed geometry.
Furthermore, the dpatta Cu(II) complex represents one of the first crystallo-
graphic evidences of anion-π–π interactions (nitrate-triazine-triazine). The
computed energy for the optimized nitrate-triazine-triazine complex is only
0.46 kcal mol–1 higher than the complex including one ring. Due to this low
energy difference, it is difficult to draw any conclusions about the stabiliza-
tion of the ternary complex induced by the extra π–π interaction. Limitations
in the functionals available to describe the π–π interactions together with
unconsidered coordination forces may account for the impossibility to repro-
duce this experimental observation.
Fig. 36 Molecular structure of dpatta. Stick representation of the crystal structure of

the tetranuclear complex [Cu4 (dpatta)(NO3)4 ](NO3 )4 complex. Hydrogen atoms, non-
coordinated nitrate anions and lattice molecules are omitted for clarity. In box: partial
stick view of the structure showing the anion-π–π interactions
168 P. Ballester
Other beautiful examples of anion-π interactions in which the aromatic

system is a N-containing arene have been described in recent years [57–59].
For example, Dunbar et al. [60] undertook a tandem crystallographic and
computational investigation of the effect of the anion-π interaction on the
preferred structural motifs of Ag(I) complexes with tetrazine and pyridazine-
based ligands. The investigated anions are polynuclear like PF6 – , AsF6 – ,
SbF6 – , and BF4 – . The ligands used are 3,6-bis(2 -pyridyl)-1,2,4,5-tetrazine
(bptz) and 3,6-bis(2 -pyridyl)-1,2-pyridazine (bppn). The MEP of the free
ligands indicated that the bptz tetrazine ring has a higher π-acidic char-
acter compared to the bppn pyridazine ring. Consequently, the bptz ring
is more likely to participate in anion-π interactions. This preference is in
complete agreement with the observed structural motifs of the complexes
of Ag(I)-bptz, regardless of the anion. In the Ag(I)-bptz compounds, multi-
ple and shorter anion-π interactions are established between the anion and
the central tetrazine rings, whereas in the Ag(I)-bppn complexes intermole-
cular interactions are maximized at the expenses of anion-π interactions. The
solid-state evidences are also supported by DFT calculations.
The preceding theoretical evaluation of the anion-π interaction, especially
the results of Johnson, Hay et al. [45], together with the experimental ev-
idences derived from the work of Kochi et al. [47], seem to suggest that
the anion-aryl centroid complex geometry may not be the more common
geometry in the interaction of halides with π-deficient electron systems. The
Cambridge Structural Database (CSD) is a convenient and reliable store-
house for geometrical information. Furthermore, it is well established the
utility of the solid-state structure of small molecules in analyzing the geo-
metric parameters for intermolecular interactions. In order to confirm the
experimental existence of the anion-π interaction in a general way, a sur-
Fig. 37 Molecular structure of the ligands bptz and bppn. Middle: portion of the pack-
ing diagram of [Ag2 (bptz)3 ][SbF6 ]2 depicting the anion and cation arrangement. Right:
portion of the grid-type structure of [Ag2 (bppn)4 ][PF6 ]4 depicting π–π and anion-π
interactions
vey of the CSD has been carried out by several authors [10, 19, 45]. It is
clear that the result of a search of the CSD will be highly dependent on
the search criteria used to store a hit. Johnson, Hay et al. [45] carried
out a search for halide anions located within 4.0 Å of the centroid of six-
membered ring π systems yielded 600 examples. In most cases, however,
the π-system was either positively charged or bonded directly to a posi-
tively charged atom. If only charge-neutral π systems are considered, a much
smaller set of 19 different structures is retrieved. Within these structures,
30 halide arene complexes were found. They performed an analysis of sev-
eral distances of the complexes, i.e., distance anion-to-centroid, distance
anion-to-plane, etc., concluding that 84% of the anions in the data set are
closer to the ring carbons than to the centroid. Thus, the available struc-
tural data of the CSD indicates that the CT-binding motif of a halide in-
teracting with a π systems is more prevalent that the anion-aryl centroid
motif. The authors explained that the incongruence of their results with
prior analysis of the CSD—which reported evidences of a marked prefer-
ence of charge-neutral atoms bearing lone pairs to position themselves over
the center of pentafluoroarenes [19] and trinitrobenzene derivatives [61]—is
due to the use of a misleading search criterion. Finally, based on the sta-
tistical analysis of the hits obtained for a search criteria, which is that the
electronegative atoms must be 4 Å of the pentafluorophenyl centroid, they
established a complete absence of any preferred location over the π sys-
tem. The use of solid-state structures to analyze the geometric parameters
of weak intermolecular forces such as anion-π interactions, has a general
drawback, that is, other intermolecular interactions that are stronger, i.e.,
charge–charge interactions may determine the final crystal packing arrange-
ment. Consequently, the geometry observed in the crystal for the weak inter-
action could be not the preferred one but the one resulting of the balance of
intermolecular forces that controls the packing of the lattice. Probably, this
is one of the causes that produces the complete absence of a preferred loca-
tion of the anion over the π system of perfluorobenzenes. However, several
searches carried out in the CSD do show that many of the geometries the-
oretically calculated for the interaction of anions with π systems are in fact
present in the solid state. Consequently, this type of non-covalent interaction
is worth to be seriously investigated by the practitioners of supramolecular
chemistry.
A recent account by Matile et al. [62] can be considered as the first ex-
perimental example of the use of anion-π interactions for the design of
a synthetic anion channel. Matile and co-workers report the design, syn-
thesis, and evaluation of π-acidic, shape-persistent oligo-(p-phenylene)-N,N-
naphthalenediimide (O-NDI) rods that can transport anions across lipid
bilayer membranes with a rare selectivity Cl– > F– > Br– > I– and a sub-
stantial anomalous mole fraction effect. DFT calculations revealed a global
quadrupole moment Qzz =+ 19.4 B for a model NDI. By comparison with
170 P. Ballester
Fig. 38 a The concept of the anion-π slide in lipid bilayer; b MEP for model NDI; red:
electron-rich, blue: electron-poor
rigid p-oligophenyl rods, the authors deduced that the alignment of three
NDI acceptors separated by phenyl spacers would afford rods with the ap-
propriate length for hydrophobic matching with common lipid bilayer mem-
branes. The results obtained in this study are in agreement with opera-
tional dynamic anion-π interactions and the existence of multiple anion-π
sites for transmembrane anion hopping, that is, anion-π slide as shown in
Fig. 38. A final caveat of the authors indicates that further studies are ne-
cessary to corroborate insights on the novel and complex system introduced
in the study.
4
Summary and Outlook
Theoretical studies indicate the existence of a counter-intuitive attractive in-

teraction between anions and π-systems. The strength of the interaction
(binding energies) seems to be drastically reduced by the inclusion of sol-

vents effects in the calculations, although in some cases the interaction is still
favorable. Calculations of F– , Cl– , Br– and NO3 – complexes with electron-
deficient aromatic systems (quadrupole moment Qzz > 0) like, triazine, hex-
afluorobenzene and polycyanobenzenes establish the existence of different
binding geometries. When the halide lies above the plane of the π-system
strongly covalent sigma complexes, weakly covalent donor-π acceptor (CT)
complexes and anion-aryl centroid complexes may be located as stable ge-
ometries. The halides can also form stable hydrogen-bonded complexes with
π-aromatic systems having C – H donors groups. In fact, the presence of
electro-withdrawing substituents increases the acidity of the C – H donor
and strengthens this type of interaction. The most stable binding geom-
etry (global minimum) depends on the anion and the π-system. Also, some
binding geometries which are not even stable points (minimum) for certain
anions could be global minimum for others. In many cases, the off-center
CT complexes are more stable that the anion-aryl centroid binding geometry.
The physical nature of the anion-π interaction has also been investigated in
great detail from a theoretical point of view. The Qzz molecular quadrupole
of aromatics has the adequate spatial orientation to be involved in the for-
mation of non-covalent anion-π complexes in which the halide is located
above the arene centroid (anion-aryl centroid geometry). The two funda-
mental components contributing to the stabilization energy of non-covalent
anion-π or anion-aryl centroid complexes in the gas phase are electrostatic
and anion-induced polarization. The electrostatic component correlates well
with the magnitude of the Qzz of the aromatic ring and the contribution
due to polarization can be considered as almost constant. The contribu-
tions of the dispersion and correlation terms to the total interaction en-
ergy are small, but they are more important in anion-π complexes, than
in cation-π complexes.
In recent years, numerous experimental studies have shown crystallo-
graphic evidences of anions encapsulated in aromatic cavities formed by
electron-deficient arenes. These observations clearly hint to the existence of
an attractive interaction between anions and π-systems. Nevertheless, many
of the reported crystal structures do not show the anion located above the
arene centroid (anion-aryl centroid geometry). On the contrary, the great
majority of the theoretical studies on the anion-π interaction deal with com-
plexes displaying the anion-aryl centroid geometry in the gas phase.
Although several theoretical studies have proposed structures for novel re-
ceptors based on multiple anion-π interactions, to date, and to the best of
our knowledge, the experimental realization of such type of receptors and
their anion-binding properties have not been reported. The few experimental
studies that have attempted the experimental quantification of the anion-
π interaction in solution have produced very different and even opposite
results.
172 P. Ballester
The last paragraph of a chapter like this one asks the writer to predict
if the non-covalent interaction of anions with π-system will be as import-
ant and useful as the nowadays well-established cation-π interaction. We are
cautious in answering this question affirmatively, but we have to confess that
we are currently working in the construction of synthetic receptors for an-
ions that do incorporate electron deficient π-aromatic systems. We hope that
our designs will be valuable for the experimental evaluation of the anion-
π interaction in solution and we will report shortly our findings. Once the
strength of the anion-π interaction in solution is determined, it will be easier
to evaluate its possible use in the construction of selective receptors for an-
ions making use of the directionality properties that we have discussed in the
chapter.
Acknowledgements I want to thank Dr. Antonio Frontera, Dr. David Quiñonero, and Prof.
Pere M. Deyà from the University of the Balearic Islands for sharing with me his interest
and results of their studies on the anion-π interaction. As a consequence, my group is now
involved in trying to quantify experimentally the strength of this new, counterintuitive and
rather unnoticed non-covalent interaction, and I find myself writing a book chapter on the
topic. Generous financial support from MEC (CTQ2005-08989-C01-02/BQU and CSD2006-
0003), ICIQ Foundation, ICREA Foundation and Generalitat de Catalunya (2005SGR00108)
is gratefully acknowledged.
References
1. Hunter CA (2004) Angew Chem Int Ed 43:5310
2. Amendola V, Esteban-Gomez D, Fabbrizzi L, Licchelli M (2006) Acc Chem Res 39:343
3. Bowman-James K (2005) Acc Chem Res 38:671
4. Beer PD, Schmitt P (1997) Curr Opin Chem Biol 1:475
5. Simmons HE, Park CH (1968) J Am Chem Soc 90:2428
8. Schmidtchen FP, Berger M (1997) Chem Rev 97:1609
9. Schneider HJ, Werner F, Blatter T (1993) J Phys Org Chem 6:590
10. Gamez P, Mooibroek T, Teat S, Reedijk J (2007) Acc Chem Res 40:435
11. Egli M, Sarkhel S (2007) Acc Chem Res 40:197
12. Ma JC, Dougherty DA (1997) Chem Rev 97:1303
13. Mascal M (2006) Angew Chem Int Ed 45:2890
14. Gallivan JP, Dougherty DA (1999) Org Lett 1:103
15. Danten Y, Tassaing T, Besnard M (1999) J Phys Chem A 103:3530
16. Alkorta I, Rozas I, Elguero J (1997) J Org Chem 62:4687
17. Mascal M, Armstrong A, Bartberger MD (2002) J Am Chem Soc 124:6274
18. Alkorta I, Rozas I, Elguero J (2002) J Am Chem Soc 124:8593
19. Quiñonero D, Garau C, Rotger C, Frontera A, Ballester P, Costa A, Deyà PM (2002)
20. Caldwell JW, Kollman PA (1995) J Am Chem Soc 117:4177
21. Sunner J, Nishizawa K, Kebarle P (1981) J Phys Chem 85:1814
22. Burgi HB, Dunitz JD, Shefter E (1973) J Am Chem Soc 95:5065
23. Bürgi HB, Lehn JM, Wipff G (1974) J Am Chem Soc 96:1956
24. Bader RFW (1991) Chem Rev 91:893
25. Cheeseman JR, Carroll MT, Bader RFW (1988) Chem Phys Lett 143:450
26. Koch U, Popelier PLA (1995) J Phys Chem 99:9747
27. Cubero E, Luque FJ, Orozco M (1998) Proc Natl Acad Sci USA 95:5976
28. Hernandez B, Luque FJ, Orozco M (1999) J Comput Chem 20:937
29. Luque FJ, Orozco M (1998) J Comput Chem 19:866
30. Williams JH (1993) Acc Chem Res 26:593
31. Garau C, Frontera A, Quinonero D, Ballester P, Costa A, Deya PM (2003) Chem Phys
Chem 4:1344
32. Kim D, Tarakeshwar P, Kim KS (2004) J Phys Chem A 108:1250
33. Jeziorski B, Moszynski R, Szalewicz K (1994) Chem Rev 94:1887
34. Hunter CA, Low CMR, Rotger C, Vinter JG, Zonta C (2002) Proc Natl Acad Sci USA
99:4873
35. Kearney PC, Mizoue LS, Kumpf RA, Forman JE, McCurdy A, Dougherty DA (1993)
J Am Chem Soc 115:9907
36. Hans-Jörg Schneider TB, Patrick Zimmermann (1990) Angew Chem Int Ed Engl
29:1161
37. Ting AY, Shin I, Lucero C, Schultz PG (1998) J Am Chem Soc 120:7135
38. Garau C, Quinonero D, Frontera A, Ballester P, Costa A, Deya PM (2005) J Phys
Chem A 109:9341
39. Schleyer PvR, Maerker C, Dransfeld A, Jiao H, Hommes NJRvE (1996) J Am Chem Soc
118:6317
40. Vahtras O, Almloef J, Feyereisen MW (1993) Chem Phys Lett 213:514
41. Feyereisen M, Fitzgerald G, Komornicki A (1993) Chem Phys Lett 208:359
42. Hermida-Ramon JM, Estevez CM (2007) Chem Eur J 13:4743
43. Klarner F-G, Kahlert B (2003) Acc Chem Res 36:919
44. Clements A, Lewis M (2006) J Phys Chem A 110:12705
45. Berryman OB, Bryantsev VS, Stay DP, Johnson DW, Hay BP (2007) J Am Chem Soc
129:48
46. Schneider H-J, Yatsimirsky AK (2000) Principles and Methods in Supramolecular
Chemistry. Wiley, Chichester
47. Rosokha YS, Lindeman SV, Rosokha SV, Kochi JK (2004) Angew Chem Int Ed 43:
4650
48. Gaussian 03W Version 6.1. DFT B3LYP/6-31G(d,p)
49. Hettche F, Hoffmann RW (2003) New J Chem 27:172
50. Frontera A, Saczewski F, Gdaniec M, Dziemidowicz-Borys E, Kurland A, Deyà PM,
Quiñonero D, Garau C (2005) Chem Eur J 11:6560
51. Berryman OB, Hof F, Hynes MJ, Johnson DW (2006) Chem Commun, p 506
52. Mo H, Wang A, Wilkinson PS, Pochapsky TC (1997) J Am Chem Soc 119:11666
53. Demeshko S, Dechert S, Meyer F (2004) J Am Chem Soc 126:4508
54. de Hoog P, Gamez P, Mutikainen I, Turpeinen U, Reedijk J (2004) Angew Chem Int Ed
43:5815
55. Casellas H, Massera C, Buda F, Gamez P, Reedijk J (2006) New J Chem 30:1561
56. Maheswari PU, Modec B, Pevec A, Kozlevcar B, Massera C, Gamez P, Reedijk J (2006)
Inorg Chem 45:6637
57. Dorn T, Janiak C, Abu-Shandi K (2005) Cryst Eng Comm 7:633
58. Gural’skiy IyA, Solntsev PV, Krautscheid H, Domasevitch KV (2006) Chem Commun,
p 4808
59. Zhou X-P, Zhang X, Lin S-H, Li D (2007) Crystal Growth & Design 7:485
174 P. Ballester
60. Schottel BL, Chifotides HT, Shatruk M, Chouai A, Perez LM, Bacsa J, Dunbar KR
(2006) J Am Chem Soc 128:5895
61. Quinonero D, Garau C, Frontera A, Ballester P, Costa A, Deya PM (2002) Chem Phys
Lett 359:486
62. Gorteau V, Bollot G, Mareda J, Perez-Velasco A, Matile S (2006) J Am Chem Soc
128:14788
Struct Bond (2008) 129: 175–206
DOI 10.1007/430_2008_083
Published online: 12 March 2008
Anion Templates in Synthesis

and Dynamic Combinatorial Libraries
Ramon Vilar
Department of Chemistry, Imperial College London, London SW7 2AZ, UK
r.vilar@imperial.ac.uk
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
2 Templating Effects in Chemistry . . . . . . . . . . . . . . . . . . . . . . . . 177
3 Recent Examples of Anion-Templated Processes . . . . . . . . . . . . . . . 178

3.1 Macrocycles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
3.2 Cages and Capsules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
3.3 Interlocked Species . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
4 Anions as Templates in Dynamic Combinatorial Chemistry . . . . . . . . 191

4.1 Using Metal–Ligand Coordination Bonds . . . . . . . . . . . . . . . . . . . 192
4.2 Using Reversible Covalent Bonds . . . . . . . . . . . . . . . . . . . . . . . . 201
5 Conclusions and Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
Abstract This review presents an overview of the area of anion-templated synthesis of

molecules and supramolecular assemblies. The review is divided into two main sections:
the first part deals with anion-templated systems where the final products are linked by
bonds that are not reversible under the conditions of the experiment. Several recent ex-
amples of macrocycles, cages and interlocked species are presented in this section. The
second part of the chapter, presents a discussion of anion-templation in systems contain-
ing reversible bonds that give rise to dynamic combinatorial libraries (either by formation
of coordination metal–ligand bonds or by reversible covalent bonds).
Keywords Anion template · Dynamic combinatorial library · Molecular recognition ·

Self assembly
1
Introduction
The use of templates in chemistry is nowadays a widespread strategy for

the synthesis of complex molecules and supramolecular assemblies [1–4].
A plethora of species ranging from macrocycles and cages to interlocked
molecules and imprinted polymers have been efficiently synthesised using
this approach. In a reaction where several potential products can be formed,
176 R. Vilar
chemical templates pre-organize the molecular building blocks to favour the

formation of products with specific nuclearity, geometry and overall struc-
ture. The interactions between the building blocks and the corresponding
template are generally of a non-covalent nature (e.g., hydrogen bonding,
electrostatic interactions, coordination bonds or π-π stacking). In order for
a templated process to successfully yield the targeted product, the structural
and electrostatic properties of the template need to be carefully selected.
From a structural point of view, both the size and geometry of the template
have to be considered, while electrostatically the choice is restricted to neu-
tral, positive or negatively charged species. While cations have been widely
used as templates in synthetic chemistry, the role of anion templates did not
start to be exploited until relatively recently. In spite of the initial reservations
regarding the high solvation energies, sizes and pH dependency of anions, the
past few years have shown the great potential these species have as templates
in a wide range of synthetic routes [5–10].
In a templated reaction, once the building blocks have been pre-organised
by the template, they can be linked together by irreversible covalent bonds or
by reversible interactions (either covalent or supramolecular). Initially, most
templated reactions aimed at forming irreversible covalent bonds between
the building blocks being brought together by the template. However, more
recently, a different approach to templated processes has been developed in
which the building blocks are linked together by reversible bonds which can
rapidly form and break. If in this reaction mixture there are several differ-
ent building blocks that can react with each other in a reversible fashion,
then a dynamic equilibrium between all the possible combinations of the mo-
lecular components can be established. This approach, known as dynamic
combinatorial chemistry (DCC), can then lead to the formation of virtual
libraries of compounds – named dynamic combinatorial libraries (DCL) –
that are thermodynamically controlled [11–15]. As with any other chem-
ical equilibrium, different external stimula (e.g., temperature, concentrations,
addition of a template) can modify the equilibrium between the components
of a DCL. Amongst the most efficient ways of modifying the composition
of these virtual combinatorial libraries, is by adding a template which can
shift the equilibrium towards the formation of one particular assembly or
molecule over others. This process is known as amplification.
The concept of dynamic combinatorial chemistry was pioneered by
Sanders [16, 17] and Lehn [15, 18, 19] in the 1990s. Since then, a wide range
of reports have appeared in which different templates have been used to am-
plify specific components of a dynamic library. In most cases, the templates
used are either cationic or neutral species leading to the amplification of the
corresponding receptors for these types of species. Considering that anion
templates have only started to be used recently in synthesis, it is not surpris-
ing that there are only a handful of examples where DCLs are amplified by
anionic species.
Anion Templates in Synthesis and Dynamic Combinatorial Libraries 177
This chapter aims to provide an update on the role of anions as templates.

The review is divided in two main sections: (a) anion-templated synthesis
of assemblies linked together by irreversible bonds (or bonds that are inert
under mild experimental conditions); (b) anion templates in systems where
the bonds linking the components are reversible and lead to anion-controlled
dynamic combinatorial libraries. Since some comprehensive reviews in the
area of anion templation have appeared over the past few years [5–7], this
chapter will mainly focus on papers published recently and will aim to show
the principles of anion templation rather than being a comprehensive account
of the literature. In addition, the scope of the chapter will be restricted to
finite assemblies (molecular or supramolecular) and not polymeric (for a re-
view on molecularly imprinted polymers using anions see Steinke’s chapter
in this volume).
2
Templating Effects in Chemistry
In the 1960s, Busch carried out pioneering work on metal-directed synthe-

ses of macrocycles establishing the concept of chemical template. As defined
by him “a chemical template organizes an assembly of atoms, with respect
to one or more geometric loci, in order to achieve a particular linking of
atoms” [20–22]. This provides an efficient route to prepare a specific molecu-
lar assembly when several others can be potentially formed. Ideally templates
should be removed from the final product once the reaction has reached com-
pletion; however, templates often form an integral part of the final product;
hence, they cannot always be removed from it.
A templated process can be driven thermodynamically or kinetically [23].
In the former case, the template binds more strongly to one of the prod-
ucts present in an equilibrium (i.e., a mixture under thermodynamic control)
shifting the reaction towards the formation of this specific product which is
then obtained in higher yields. As will be discussed later in this review, this is
particularly important in the generation of dynamic combinatorial libraries.
On the other hand, kinetic templates operate under irreversible conditions by
stabilising the transition state leading to the final product.
In a templated process the interactions between the directing group and
the building blocks can be either covalent or non-covalent. The latter can
make use of a wide range of supramolecular interactions, such as hydrogen
bonding, π-π stacking, electrostatic interactions and hydrophobic effects.
From these, hydrogen bonding interactions are particularly useful since they
are relatively strong and, more importantly, are directional. In fact, hydrogen
bonding interactions are employed as the main driving force in many of Na-
ture’s templated processes. One of the most elegant examples of this is the
replication and transcription of nucleic acids.
178 R. Vilar
3
Recent Examples of Anion-Templated Processes
Since the first examples of anion-directed assemblies were reported in the

1990s, a wide range of molecular and supramolecular systems have been suc-
cessfully prepared using anions as templates. This section has been divided by
the type of molecular or supramolecular species formed and, as mentioned
before, it will mainly focus on systems that have been recently reported in
the literature (making reference to previous examples when important for the
sake of clarity and completion).
3.1
Macrocycles
Böhmer has recently shown that the presence of chloride can dictate the
size of macrocyclic poly-urea systems. The reaction between diamine 1 and
diisocyanate 2 yield the macrocyclic species 3 and 4 in a 5:1 ratio (see
Scheme 1) [24].
When this reaction was repeated in the presence of two equivalents of
tetrabutylammonium chloride, the formation of the larger macrocycle 4 was
favoured over the trimeric species 3. In fact, the ratio reverted to 1:5 in
favour of the larger macrocycle. Crystals of macrocycle 4 were obtained from
the crude of the first reaction when grown in the presence of tetrabutyl-
ammonium chloride. The structure (Fig. 1) revealed an interesting host-guest
complex in which two chlorides are bound by the macrocycle (which explains
the need of two equivalents of chloride to favour the formation of the hexam-
eric compound).
Alfonso and Luis have recently reported an example of anion-templated
synthesis of a pseudopeptidic macrocycle. The reaction between diamine 5
and dialdehyde 6 was carried out (see Scheme 2) [25]. 1 H NMR spectroscopy
showed a complex patter of signals together with protons associated with
aldehyde and methoxyamine suggesting the presence of a range of different
macrocyclic and acyclic products. Addition of different anions to the reaction
mixture had little effect to the distribution of products shown by 1 H NMR
spectroscopy. However, when the reaction was repeated in the presence tereph-
thalate, the almost quantitative formation of a single product as revealed by
1 H NMR spectroscopy was observed. This intermediate product has been
postulated to be host-guest complex 7· Terephthalate). This imine-containing

macrocycle was then reduced to the corresponding amine to yield macrocycle
8, which was isolated and fully characterised. In the first step of this reaction,
terephthalate is expected to form hydrogen bonding interactions between its
anionic carboxylates and the amides of the macrocycle. In addition, π-π in-
teractions between the aromatic ring of the terephthalate and that one of the
di-aldehyde are likely to also play a role in the templating process.
Scheme 1 Synthetic procedure for the preparation of macrocycles 3 and 4
Fig. 1 X-ray crystal structure of macrocycle 4 highlighting the two chloride anions (solid
spheres) bound to the macrocycle
180 R. Vilar
Scheme 2 Terephthalate-templated synthesis of macrocycle 8
It is worth noting that this process could in principle lead to dynamic com-
binatorial libraries of products since the type of bond that brings together
the different components in the reaction is a reversible one (namely imine
formation). In fact, the initial mixture of products observed by the authors
has been postulated to be a mixture of different-sized macrocycles and linear
species (from which one of them is amplified upon addition of terephthalate).
More detailed studies would be needed to determine whether this system in-
deed leads to the formation of a DCL of receptors (see Sect. 4 for examples of
anion-directed DCLs).
A widely used approach to the synthesis of large macrocycles and cages (see
Sect. 3.2) is by self-assembly of metal centres and polydentate ligands. Careful
choice of the geometry around the metal and the number and relative position
of the coordinating groups on the bridging ligands can generate large 2D and
3D assemblies from a one-pot reaction. Often, the addition of a template to the
reaction mixture provides a more efficient path for the formation of one spe-
cific assembly. Some recent examples of anion-templated metallo-macrocycles
and -cages are discussed in this and the following sections.
Lippert has shown that the assembly between platinum(II) centres and
purine bases can be controlled by specific anions. In analogy to the hydrogen-
bonded tetrads that guanine bases can form, this group synthesised a metallo-
square in which the purine bases are interconnected by coordination to plat-
inum(II) centres rather than hydrogen bonding interactions (see Fig. 2) [26].
Fig. 2 Schematic representation of: a guanine quartet with a cationic guest and b metallo-
square 9 based on platinum(II) and methylpurine with an anionic guest
Square 9 formed over a period of 5 days by self-assembly process of four

units of [(NH3 )2 Pt(Pur)(H2 O)]2+ (where Pur = 9-methylpurine). Interest-
ingly, in this process the formation of a second species with a triangular
geometry was observed (see Scheme 3). NMR experiments showed that the
triangular metallo-macrocycle 10 is favoured (in a 0.6:1 ratio) to square 9.
However, if the self-assembly process is carried out in the presence of SO4 2– ,
the proportion of the two species changes and the square is favoured over the
triangle in a 2.5:1 ratio. This change in the preference for the square over the
triangle has been attributed to the templating properties of the sulfate anion.
Scheme 3 Reaction scheme for the formation of metallo-triangle 10
Maekawa and Kitagawa have recently reported an interesting anion-

directed approach to the synthesis of bowl-shaped metallo-macrocycles [27].
Initially, they observed that the reaction between [Cu(MeCN)4 ](PF6 ) and 4-
(2-pyridyl)pyrimidine (pprd) under an atmosphere of C2 H4 and in acetone
yielded the coordination polymer {[Cu(pprd)(C2 H4 )](PF6 )}n (see Scheme 4).
182 R. Vilar
Scheme 4 Formation of a coordination polymer from the reaction between [Cu(MeCN)4]

(PF6 ) and 4-(2-pyridyl)pyrimidine (pprd)
When the same reaction was carried out in methanol the macrocyclic species
[Cu4 (pprd)4 (C2 H4 )4 ](PF6 )4 (11) was formed instead. The crystal structure of
this metallo-assembly revealed a bowl-shaped structure with the PF6 – anion
positioned at its centre (see Fig. 3).
Fig. 3 X-ray crystal structures of a the tetra-copper assembly [Cu4 (pprd)4 (C2 H4 )4 ](PF6 )4
(11) and b the tri-copper assembly {[Cu3(pprd)3 (C2 H4 )3 ](ClO4 )3 }3 (12)
Interestingly, the reaction between [Cu(C2 H4 )n ]ClO4 and pprd in Me2 CO

under a C2 H4 atmosphere yielded the trinuclear metallo-macrocycle {[Cu3
(pprd)3 (C2 H4 )3 ](ClO4 )3 }3 (12). Although the reactions to obtain the two
differently-sized macrocycles are not identical, it is plausible to suggest that
the volume of the anions dictates the size of the macrocycle. With PF6 , the
tetranuclear metallo-macrocycle forms while the smaller tetrahedral ClO4 an-
ion directs the formation of the tri-copper macrocycle.
3.2
Cages and Capsules
In the late 1990s we reported one of the first examples of anion-templated

metallo-cage [28–30]. The system was based on the chloride-directed assem-
bly of eight units of amidinothiourea (H-atu, see Scheme 5) and six nickel(II)
centres to yield [Ni6 (atu)8 ⊂Cl]Cl3 (13). The resulting assembly showed to
contain a completely encapsulated chloride interacting via hydrogen bonding
and weak metal-anion interactions with the cage.
Scheme 5 Reaction scheme for the synthesis of hexanickel cage 13
The approach described above was expanded afterwards to incorporate

a second type of metal within the framework yielding the mixed Ni-Pd and
Ni-Pt complexes [Ni4 M2 (atu)8 ⊂Cl]Cl3 (M = Pd, 15; Pt, 16) (see Fig. 4). This
was carried out by first preparing [Ni(atu)2 ] (14) and then reacting it with the
corresponding palladium(II) or platinum(II) salt. As in the case of the hexa-
nickel cage, it was found that the mixed-metal cage would only form in the
presence of the appropriate halide, namely chloride or bromide.
The assembly of the nickel(II)/H-atu cage 13, is accompanied by a dramatic
colour change, from orange to green. Considering that in methanol only chlo-
ride acts as a template for the formation of cage 13, we have recently employed
this anion-templated process to develop a colorimetric chemical sensor for
chlorides [31].
Bidentate pyrazolyl-based ligands have shown to be versatile building
blocks for the synthesis of a plethora of coordination assemblies. Initial work
by McCleverty and Ward in the late 1990s showed that the synthesis of cages
[Co4 (Ln )6 ⊂X](X)7 (where L1 and L2 = bidentate pyrazolyl-pyridine ligands;
184 R. Vilar
Fig. 4 X-ray crystal structure of mixed-metal cage [Ni4 Pt2 (atu)8⊂Cl]Cl3 showing the en-
capsulated chloride anion
X = BF4 – , ClO4 – ; see Figs. 5 and 6) was dependant on the presence of specific
anions [32]. Detailed investigations (by NMR spectroscopy) of the assem-
bly process in solution demonstrated that the tetrahedral BF4 – and ClO4 –
anions indeed act as templating agents for the formation of these metallo-
assemblies [33].
More recently, Ward has explored the effect that the length of bidentate
pyrazolyl-based ligands has on the formation of the metallo-cages [34]. When
Fig. 5 A selection of bidentate pyrazolyl-based ligands employed for the synthesis of

metallo assemblies
Fig. 6 X-ray crystal structure of [Co4 (L1 )6 ⊂(ClO4 )](ClO4)7 highlighting the encapsulated
tetrahedral anion (in space-fill representation)
L3 was used as ligand, the expected [Co4 (L3 )6 ⊂X](X)7 (X = BF4 – , ClO4 – ,
PF6 – , I– ) cages were formed, but in this case the anion did not seem to play
a determinant role in defining the geometry of the final assembly. As indi-
cated by the authors, the resulting cage is sufficiently large to leave gaps in the
centre of the faces through which the encapsulated anion can easily exchange
with the external anions.
186 R. Vilar
Amouri has reported the anion-directed synthesis of a series of coordi-

natively unsaturated metallo-cages with general formula [Co2 (L4 )4 (RCN)2 ⊂
(BF4 )](BF4 )3 (R = Me, Et, Ph) [35]. The X-ray crystal structures of some of
these assemblies (see Fig. 7 for an example) have revealed the presence of an
encapsulated BF4 – anion which interacts with the coordinatively unsaturated
cobalt(II) centres. Interestingly when analogous reactions were performed in
the presence of other anions (such as Cl and NO3 ) different metal-organic
assemblies were formed [36].
Fig. 7 X-ray crystal structure of [Co2(L4 )4 (MeCN)2 ⊂(BF4 )](BF4 )3 showing the encapsu-
lated anion and its interactions with the metal centres
3.3
Interlocked Species
Molecular interlocked systems such as catenanes and rotaxanes can also be

prepared using anion-directed approaches. Although anionic templates did
not make their way into this area until relatively recently, nowadays there are
several examples that demonstrate the utility of this approach for the synthe-
sis of this topologically interesting species.
The first examples of interlocked species synthesized by anion tem-
plated approaches were the pseudorotaxanes and rotaxanes reported in the
late 1990s by Stoddart and Vögtle. Stoddart reported that by mixing four
equivalents of [NH2 (CH2 Ph)2 ][PF6 ] with one equivalent of the macrocycle
tetrakis-p-phenylene[68]crown-20, the quadruply-stranded pseudorotaxane
(17) formed (see Fig. 8) [37]. This assembly was structurally characterized re-
vealing the presence of a PF6 – anion at its center forming multiple C–H· · ·F
hydrogen bonds with the hydroquinone methine and the benzylic methylene
hydrogen atoms.
Fig. 8 Schematic representation of quadruply-stranded pseudorotaxane 17
Using a different approach, Vögtle successfully showed that organic an-

ions can induce the formation of rotaxanes [38–41]. In this approach a strong
host-guest complex between a tetralactam macrocycle and a phenolate anion
is first formed (see Scheme 6). In this assembly the phenolate anion is posi-
tioned at the center of the ring to further react with a second component (e.g.,
an alkyl bromide or acyl chloride) yielding a rotaxane. The negatively charged
phenolic functionality can be located either at the stopper component or at
the axle precursor providing a range of different possibilities for the synthesis
of the interlocked species.
More recently, Beer has developed a series of synthetic procedures for the
halide-templated syntheses of pseudorotaxanes, rotaxanes and catenanes. This
methodology is based on combining the recognition of halides by a hydrogen-
bonding host (e.g., a macrocycle) with ion-pairing between the corresponding
halide and a cationic species (see Fig. 9). These two interactions allow for the
interpenetration of the two molecular components to yield the interlocked
molecules. Some examples of the macrocycles and axels employed to generate
the corresponding pseudorotaxanes are shown in Fig. 10 [38–41].
A similar approach was later employed to prepare rotaxanes by “clipping”
(via ruthenium-catalysed olefin metathesis) the acyclic hydrogen bonding
188 R. Vilar
Scheme 6 Schematic representation of the anion templated synthesis of rotaxanes based

on tetralactam macrocycles and a phenolate anions
Fig. 9 Schematic representation of the anion-templated synthesis of interlocked species

based on combining halides recognition by a hydrogen-bonding host with ion-pairing
species 18 with the corresponding axel 19 in the presence of chloride (see

Scheme 7) [42]. The resulting rotaxane (20) has been structurally characterised
showing the strong binding between the interlocked species and the templating
chloride (see Fig. 11).
Using the same general strategy described above, Beer reported the first
anion-templated synthesis of catenanes [43, 44]. Mixing macrocycle 21 with
Fig. 10 Selection of macrocycles and axels employed to synthesise pseudorotaxanes and

rotaxanes
Scheme 7 Reaction scheme for the chloride-templated synthesis of rotaxane 20

190 R. Vilar
Fig. 11 X-ray crystal structure of rotaxane 20 showing the chloride anion bound to the
interlocked species
Scheme 8 Reaction scheme for the chloride-templated synthesis of catenane 23
Fig. 12 X-ray crystal structure of catenane 23 showing the chloride anion bound to the
interlocked species
the acyclic hydrogen-bonding molecule 22 in the presence of chloride as

a template, led to the formation of [2]catenane 23 (see Scheme 8). As for ro-
taxane 20, the crystal structure of this interlocked species revealed the strong
interaction between the templating chloride and the two macrocyclic compo-
nents of the catenane (Fig. 12).
For a comprehensive review of this area, see the recent reviews published
by Beer [7, 9, 10].
4
Anions as Templates in Dynamic Combinatorial Chemistry
One of the seminal papers in defining the concept of dynamic combinatorial

chemistry was published by Lehn in the 1990s [45] In this work it was shown
Scheme 9 Chloride-templated synthesis of circular helicate 24

192 R. Vilar
that the assembly of iron(II) salts and a tris-bipy ligand (L5 ) is a dynamic
process that can yield a range of different metal helicates. Which assembly
is formed, is highly dependent on the nature of the counter-anions present
in solution. With FeCl2 , the pentanuclear circular helicate [Fe5 (L5 )5 Cl]9+ (24)
was formed in high yields (see Scheme 9), while a mixture of the penta-
and hexa-nuclear helicates were obtained in the presence of bromide (and
only the hexa-nuclear helicate [Fe6 (L5 )6 (SO4 )]10+ (25) was formed with sul-
fate). Further studies by the same authors demonstrated that in the reaction
with FeCl2 (and also in the analogous one with NiCl2 ) a linear helicate is
formed first (i.e., the kinetic product) which progressively converts into the
thermodynamic circular helicate product [Fe5 (L5 )5 Cl]9+ [46] Structural char-
acterization of 25 confirmed the assembly to be a circular double helix with
a chloride ion located in the central cavity.
Although this anion-directed system was one of the first examples of DCCs
reported in the literature, there are in fact very few known systems to date
where negatively charged species are used to amplify the formation of a spe-
cific assembly from a dynamic combinatorial library. Such examples will be
reviewed in Sects. 4.1 and 4.2.
4.1
Using Metal–Ligand Coordination Bonds
Dunbar has elegantly demonstrated the use of anionic templates for the syn-
theses of a range of nickel(II) and zinc(II) metalla-cyclophanes using 3,6-
bis(2-pyridyl)-1,2,4,5-tetrazine (bptz) as bridging ligand (see Scheme 10) [47,
48]. Anions such as BF4 – and ClO4 – induce the formation of the tetra-metallic
square assemblies [{M4 (bptz)4 (CH3 CN)8 }X](X)7 , (M = ZnII , NiII ; X = BF4 – ,
ClO4 – ), while the larger octahedral anion SbF6 – templates the formation
of the molecular pentagon [{Ni5 (bptz)5 (CH3 CN)10 }SbF6 ](SbF6 )9 . The X-ray
crystal structures of these species (see Figs. 13 and 14) have shown that in
both the squares and pentagon one anion is encapsulated at the centre of the
corresponding metalla-cyclophane (displaying anion-π interactions between
the O and F atoms of the anions and the tetrazine rings of bptz).
Further studies by the same authors showed that the molecular pentagon
can be easily converted into the corresponding molecular square in the pres-
ence of excess BF4 – and ClO4 – . The conversion of the molecular pentagon to
the square is also observed upon addition of iodide (which due to its large size
and polarizability it can adopt the directionality of a tetrahedral anion).
In contrast to the above, the conversion of the nickel square to the corres-
ponding pentagon upon addition of excess SbF6 – is not readily observed (only
partial conversion takes place). This apparent higher stability of the molecu-
lar square in comparison to the molecular pentagon has been attributed to
more strain in the pentagon. This is supported by the X-ray crystal structure
of the metallo-pentagon in which the btpz ligands are considerably bent.
Scheme 10 Reaction scheme showing the anion-directed synthesis of metallo-macrocycles

and their interconversion
Fig. 13 X-ray crystal structure of metallo-square [{Ni4 (bptz)4 (CH3 CN)8 }BF4 ](BF4 )7
More recently, Fujita has reported an interesting example of anion-

controlled dynamic equilibrium of palladium-containing assemblies [49].
The reaction of ligands L6 and L7 with palladium(II) was investigated under
different experimental conditions, namely in different solvents, at differ-
ent concentrations of the ligand and in the presence of different counter-
anions.
194 R. Vilar
Fig. 14 X-ray crystal structure of metallo-square [{Ni5 (bptz)5 (CH3 CN)10 }SbF6 ](SbF6 )9
In this study it was shown that the nature of the resulting metallo-
assemblies was highly dependant on the conditions employed. For example,
the reaction between L6 and [Pd(en)(NO3 )2 ] in DMSO was found to yield two
macrocycles (26 and 27) in roughly 60:40 proportions when the concentra-
tion of the ligand was 5 mM. By reducing the concentration of the ligand to
1 mM, the equilibrium was shifted to the simpler [2+2] assembly 26 which
was present in nearly 90%.
Increasing the concentration of ligand to 20 mM or above, yielded yet an-
other product, metallo-assembly 28 which at 500 mM concentration of ligand
was practically the only product observed by 1 H NMR spectroscopy.
In the same paper, the reactions between these two ligands and “naked”
palladium(II) cations were also discussed. Interestingly, the structure and nu-
clearity of the resulting assemblies was shown to depend on the counteranion
of the palladium salt used. Thus, the reaction between Pd(NO3 )2 and ligand
196 R. Vilar
Fig. 15 X-ray crystal structure of metallo-tetrahedron 29
L6 yielded mainly the tetrahedral assembly 29 (see Fig. 15). An analogous as-
sembly was obtained in the presence of BF4 – . However, when the reaction
was carried out in the presence of triflate a double-walled triangle (30) was
obtained as the major product (Scheme 11).
Scheme 11 Schematic representation of double-walled metallo-triangle
When carrying out a similar reaction with ligand L7 (which is longer than
L6 )the structures of the resulting metallo-assemblies were again found to be
anion-dependant. In this case, dynamic equilibrium between the two assem-
blies 31 and 32 was observed. With nitrate as the counter-anion, a roughly 1:1
mixture of the two assemblies was observed; however, with triflate the main
Scheme 12 Schematic representation of the equilibrium between the double-walled

metallo-triangle 31 and metallo-tetrahedron 32
product formed was the double-walled triangle. In contrast in the presence of

the aromatic p-tosylate anion, the formation of the M4 L8 assembly was favored.
A similar approach to that reported by Fujita for the generation of dynamic
combinatorial libraries of metallo-assemblies has been developed in our own
group [50]. In contrast to most of the bipyridyl-based ligands employed for
the synthesis of metallo-macrocycles reported so far, we were interested in
using bipyridyl ligands containing spacers with hydrogen bonding function-
alities such as ligands L8 and L9 . It was rationalized that having hydrogen
bonding donor groups would aid in the interaction with potential anionic
guests/templates.
The reactions between each of these two ligands and [Pd(dppp)(OTf)2 ]

(dppp = 1,3-bis(diphenylphosphino)propane) were studied aiming at pro-
198 R. Vilar
Scheme 13 Reaction scheme for the synthesis of metallo-macrocycles 33 and 34
ducing cyclic metallo-assemblies (see Scheme 13). The combination of these

ligands and the metal complex (in 1:1 ratios) could in principle give a range
of different cyclic and acyclic materials. Since palladium-pyridine bonds are
relatively labile, it was expected that a dynamic equilibrium of different as-
semblies would be established.
Crystals were grown from the corresponding reaction mixtures. The X-ray
crystal structures obtained revealed that the crystallised products correspond
to [2+2] assemblies with general formula [Pd(dppp)(L)]2 (OTf)4 (L = L8 , 33;
L = L9 , 34) and a “bowl-type” structure (see Fig. 16).
Fig. 16 X-ray crystal structure of assembly 34 showing the triflate anion at the centre of
the bowl-type assembly
Although the solid-state structures obtained for these systems indicate the
formation of the [2+2] assembly, 1 H NMR studies showed that more than one
species co-existed in solution. As indicated above, in principle a 1:1 mixture
of the bis-pyridyl ligands and cis-[Pd(dppp)]2+ centres could yield macrocy-
cles of different sizes or a range of different acyclic products. Several 1 H NMR
spectroscopic and ESI-mass spectrometric studies were carried out (using
[Pd(dppp)(OTf)2 ] and L8 ) to establish the behaviour of the system in solu-
tion. These studies suggested that there is an equilibrium between two species
which have been assigned to the [2+2] and a [3+3] metallo-assemblies (see
Scheme 14).
Scheme 14 Reaction scheme showing the equilibrium between the [2+2] and [3+3]
metallo-macrocycles
Interestingly, this equilibrium can be shifted by modify the experimental

conditions such as the solvent, concentration of ligand and temperature. Fur-
thermore, the equilibrium between the different metallo-assemblies present
in solution can also be shifted by the presence of different anions. For ex-
ample, addition of several equivalents of H2 PO4 – to a DMSO solution con-
taining a mixture of [2+2] and [3+3] assemblies, shifted the equilibrium to
the formation of only the [2+2] assembly. Surprisingly, addition of HSO4 –
to the 1:1 mixture in DMSO of [Pd(dppp)(OTf)2 ] and L8 did not modify the
equilibrium between the [2+2] and [3+3].
Williams has reported another type of system in which a dynamic combi-
natorial library of metal complexes is generated by metal-ligand interactions
and controlled by anionic templates [51]. In this work it was shown that
chloride can modify the distribution of products in an equilibrated solution
containing cobalt(II) salts and 2,2 -bipyridyl ligands (bipy or the chiral ligand
L10 – see Scheme 15). More specifically, mixing Co(NO3 )2 with bipy and L10
generated a library of complexes with general formula [Co(bipy)X (L10 )3–X ]2+ .
This was shown by electrospray mass spectrometry which revealed that
all the possible combinations of products were indeed present in solution:
200 R. Vilar
Scheme 15 Reaction scheme showing the different possible complexes that can be formed
by mixing Co(NO3 )2 with L10 and bipy
[Co(L10 )3 ]2+ (80%), [Co(bipy)(L10 )2 ]2+ (100%), [Co(bipy)2 (L10 )]2+ (91%)
and [Co(bipy)3 ]2+ (11%). Furthermore, 1 H NMR spectroscopy indicated that
for each of the complexes – except for [Co(bipy)3 ]2+ which gave an enan-
tiomeric pair – the corresponding ∆- and Λ-diastereomers were present.
The equilibrium of the above mixture was shown to change upon add-
ition of CF3 COOH. Addition of the acid led to protonation of the amines
groups on L8 inducing diasteroselectivity and, as a consequence, some of the
complexes initially present in the mixture disappeared. Interestingly, when
DCl rather than CF3 COOH was added to the reaction mixture, only the two
homoleptic complexes {Cl2 ⊂∆-[Co(L10 H2 )3 ]6+ } and [Co(bipy)3 ]2+ could be
detected suggesting that chloride acts as a template amplifying the formation
of these species. These observations have been rationalised on the grounds
that coulombic repulsion between the protonated amines is minimized by the
presence of chloride.
A similar approach to the one discussed above has been recently reported
by Rice [52, 53]. In these investigations it was shown that nitrate can mod-
ify the distribution of products in a mixture of cobalt(II) and two different
N,N chelating ligands. First, the reaction between Co(ClO4 )2 and ligand L11
was studied showing that a triple helicate with formula [Co2 (L11 )3 ](ClO4 )4
Scheme 16 Reaction scheme showing the different possible complexes that can be formed
by mixing Co(ClO4)2 with L11 and L12
formed (see Scheme 16). The ligands around the cobalt(II) centres generate
“pockets” of the right size to bind perchlorate anions via hydrogen bonding
(see Fig. 17).
Fig. 17 X-ray crystal structure of the triple helicate [Co2 (L11 )3 ](ClO4 )4 showing the per-
chlorate anions bound to the “pockets” formed by the three ligands
The authors then studied the possibility of anion exchange in this complex.
Thus, upon addition of two equivalents of [Bu4 N][NO3 ] to [Co2 (L11 )3 ](ClO4 )4
a new host-guest complex with formula [Co2 (L11 )3 ](ClO4 )2 (NO3 )2 was ob-
tained. The X-ray crystal structure of this mixed-anion helicate showed that
the perchlorates initially bound to the binding pockets of the helicate, had
been replaced by nitrates.
Having established the basic host-guest chemistry between the helicate
and the two anions, the authors then investigated the reaction between
[Co(ClO4 )2 ]·6H2 O, L11 and L12 (see Scheme 16 for the chemical structure of
the ligands) in a 2 to 1.5 to 1.5 ratio. This reaction resulted in the forma-
tion of four complexes: [Co2 (L11 )3 ]4+ , [Co2 (L11 )2 (L12 )]4+ , [Co2 (L11 ) (L12 )2 ]4+
and [Co2 (L12 )3 ]4+ , with a 1:3:3:1 statistical distribution. This was confirmed
by both 1 H NMR spectroscopy and ES mass spectrometry. Interestingly,
upon addition of KNO3 to this mixture a dramatic change in product dis-
tribution was observed. The two homoleptic complexes [Co2 (L11 )3 ]4+ and
[Co2 (L12 )3 ]4+ were found to be the main components of the mixture (with
only 5% of the heteroleptic compounds being present). This behaviour has
been attributed to the strong binding of nitrate (which acts as a template) to
the anion-binding pockets present in [Co2 (L11 )3 ]4+ .
4.2
Using Reversible Covalent Bonds
A large number of dynamic combinatorial libraries reported in the literature

make use of reversible covalent chemistry to generate the required dynamic
equilibria [12]. In spite of this, there are very few examples of anion-directed
202 R. Vilar
DCLs where this type of reversible bond-formation is used to generate the

virtual library of receptors. One of these examples has been reported by
Otto and Kubic who developed a DCL of anion receptors [54]. The library
is based on disulfide exchange reactions between a dimeric cyclic peptide
(35) – where the two peptidic cycles are linked by a disulfide group – and
a range of different thiol-substituted spacers a–f (see Scheme 17). Mixing of
all these components yields a range of dimeric cyclic peptides in different
proportions.
Scheme 17 Components of a DCL of dimeric cyclic peptides linked by disulfide bonds
Interestingly, addition of K2 SO4 or KI to the DCL amplified the formation

of three receptors (35a, 35b and 35c), while addition of other anions such
as chloride or fluoride did not shift the equilibrium. Two of these receptors
(35b and 35c) were amplified more than 35a; these were then chosen for fur-
ther studies and therefore prepared and isolated using a second generation
biased library. Isothermal titration calorimetry studies revealed the binding
constants between the selected dimeric-receptors and sulfate or iodide to be
around 106 (in a mixture of MeCN/H2 O). This example nicely demonstrates
how dynamic combinatorial chemistry can be employed to improve the selec-
tivity and binding properties of a specific receptor; in this case this has been
done by simply modifying the length and flexibility of the spacer that joins
the two macrocycles.
A different type of reversible covalent bond has been employed by Sessler
to generate anion-directed DCLs. More specifically, the synthesis of new
bipyrrole-based macrocyclic receptors has been achieved by reacting diamine

36 with diformylbipyrrole (37) in acidic media (see Scheme 18) [55]. Interest-
ingly, depending on the acid used (namely HCl, HBr, CH3 CO2 H, CF3 CO2 H,
H3 PO4 , H2 SO4 or HNO3 ) different distributions of oligomeric species and
macrocycles were obtained. While several products were formed with most
acids, in the presence of sulfuric acid the [2+2] macrocycle 38·2H2 SO4
formed in nearly quantitative yield.
Scheme 18 Synthesis of macrocycle 38·2H2 SO4 by reacting 36 and 37 in the presence of

H2 SO4
The anion-free macrocycle 38 was isolated upon addition of triethylamine

to 38·2H2 SO4 . This macrocycle was structurally characterised and its anion-
binding properties studied, revealing high association constants between the
receptor and HSO4 – (1:1; Ka = 63 500±3000 M–1 ) and H2 PO4 – (2:1; Ka1 =
191 000±15 400 M–1 and Ka2 = 60 200±6000 M–1 ). Interestingly, when 38 was
allowed to stand in acetonitrile for 5 days in the presence of HSO4 – or H2 PO4 –
(as tetrabutylammonium salts) a rearrangement of the poly-imine compound
took place. More specifically, the [2+2] macrocycle 38 expanded into the
[3+3] macrocycle 39 (see Fig. 18) quantitatively in the presence of H2 PO4 –
(and in 47% yield in the presence of HSO4 – ). More recently, the same authors
have published a comprehensive study in which they show a similar be-
haviour for other related di-amino and di-aldehyde building blocks [56, 57].
The nature of the anions (from the corresponding acid) present in solution,
dictates the size of the macrocycle formed. Under specific circumstances,
some of the macrocycles undergo anion-induced ring-expansion.
204 R. Vilar
Fig. 18 Schematic representation of the [3+3] macrocycle 39
The results discussed above, strongly suggest the presence of a dynamic

equilibrium between various macrocyclic species in solution. Upon addition
of the appropriate anionic template, amplification of one of them is then
observed.
5
Conclusions and Outlook
The first examples of anion-templated processes were reported in the early

1990s. Since then, this area of supramolecular chemistry has grown steadily
showing the important role that anionic templates can play in directing
the synthesis of specific molecules and supramolecular assemblies. A wide
range of organic and metal-organic macrocycles have now been prepared by
anion-templated processes. Similarly, the syntheses of molecules with com-
plex topologies (such as those of pseudorotaxanes, rotaxanes and catenanes)
can now be achieved by anion-directed assembly of simple building blocks.
Another area where anion templates have had an important impact is in the
synthesis of coordination cages and capsules. More recently, the use of anions
as templates in dynamic combinatorial libraries has been realized showing
that structurally challenging anion-receptors can be developed using this

dynamic approach. This promises to be an area of important future develop-
ments with the potential to generate receptors that can not be easily achieved
using more “classical” synthetic methodologies.
References
1. Schalley CA, Voegtle F, Doetz KH (eds) (2005) Templates in Chemistry II. Top Curr
Chem, vol 249
2. Schalley CA, Voegtle F, Doetz KH (eds) (2005) Templates in Chemistry I. Top Curr
Chem, vol 248
3. Vilar R (2004) Struct Bond 111:85
4. Diederich F, Stang PJ (eds) (2000) Templated Organic Synthesis. Wiley VCH, Wein-
heim
5. Gimeno N, Vilar R (2006) Coord Chem Rev 250:3161
6. Vilar R (2003) Angew Chem Int Ed 42:1460
7. Beer PD, Sambrook MR, Curiel D (2006) Chem Commun, p 2105
8. Beer PD, Wong WWH (2005) Macrocyclic Chem, p 105
9. Lankshear MD, Beer PD (2006) Coord Chem Rev 250:3142
10. Lankshear MD, Beer PD (2007) Acc Chem Res 40:657
11. Corbett PT, Leclaire J, Vial L, West KR, Wietor J-L, Sanders JKM, Otto S (2006) Chem
Rev 106:3652
12. Rowan SJ, Cantrill SJ, Cousins GRL, Sanders JKM, Stoddart JF (2002) Angew Chem
Int Ed 41:899
13. Otto S (2003) Curr Opin Drug Discov Dev 6:509
14. Lehn J-M (2007) Chem Soc Rev 36:151
15. Lehn J-M, Eliseev AV (2001) Science 291:2331
16. Otto S, Furlan RLE, Sanders JKM (2000) J Am Chem Soc 122:12063
17. Rowan SJ, Lukeman PS, Reynolds DJ, Sanders JKM (1998) New J Chem 22:1015
18. Eliseev AV, Lehn JM (1999) Curr Topics Microbiol Immunol 243:159
19. Lehn J-M (1999) Chem Eur J 5:2455
20. Melson GA, Busch DH (1964) J Am Chem Soc 86:4834
21. Thompson MC, Busch DH (1964) J Am Chem Soc 86:3651
22. Busch DH (1992) J Inclusion Phenom 12:389
23. Anderson S, Anderson HL, Sanders JKM (1993) Acc Chem Res 26:469
24. Meshcheryakov D, Boehmer V, Bolte M, Hubscher-Bruder V, Arnaud-Neu F, Hersch-
bach H, Van Dorsselaer A, Thondorf I, Moegelin W (2006) Angew Chem Int Ed
45:1648
25. Bru M, Alfonso I, Burguete MI, Luis SV (2006) Angew Chem Int Ed 45:6155
26. Roitzsch M, Lippert B (2006) Angew Chem Int Ed 45:147
27. Maekawa M, Konaka H, Minematsu T, Kuroda-Sowa T, Munakata M, Kitagawa S
(2007) Chem Commun, p 5179
28. Vilar R, Mingos DMP, White AJP, Williams DJ (1998) Angew Chem Int Ed 37:1258
29. Vilar R, Mingos DMP, White AJP, Williams DJ (1999) Chem Commun, p 229
30. Cheng S-T, Doxiadi E, Vilar R, White AJP, Williams DJ (2001) J Chem Soc Dalton
Trans, p 2239
31. Diaz P, Mingos DMP, Vilar R, White AJP, Williams DJ (2004) Inorg Chem 43:7597
206 R. Vilar
32. Fleming JS, Mann KLV, Carraz C-A, Psillakis E, Jeffery JC, McCleverty JA, Ward MD
(1998) Angew Chem Int Ed 37:1279
33. Paul RL, Bell ZR, Jeffery JC, McCleverty JA, Ward MD (2002) Proc Nat Acad Sci USA
99:4883
34. Paul RL, Argent SP, Jeffery JC, Harding LP, Lynam JM, Ward MD (2004) Dalton Trans,
p 3453
35. Amouri H, Mimassi L, Rager MN, Mann BE, Guyard-Duhayon C, Raehm L (2005)
36. Amouri H, Desmarets C, Bettoschi A, Rager MN, Boubekeur K, Rabu P, Drillon M
(2007) Chem Eur J 13:5401
37. Fyfe MCT, Glink PT, Menzer S, Stoddart JF, White AJP, Williams DJ (1997) Angew
Chem Int Ed Engl 36:2068
38. Hubner GM, Glaser J, Seel C, Vogtle F (1999) Angew Chem Int Ed 38:383
39. Reuter C, Schmieder R, Vogtle F (2000) Pure Appl Chem 72:2233
40. Reuter C, Wienand W, Hubner GM, Seel C, Vogtle F (1999) Chem Eur J 5:2692
41. Seel C, Vogtle F (2000) Chem Eur J 6:21
42. Wisner JA, Beer PD, Drew MGB, Sambrook MR (2002) J Am Chem Soc 124:12469
43. Sambrook MR, Beer PD, Wisner JA, Paul RL, Cowley AR (2004) J Am Chem Soc
126:15364
44. Ng K-Y, Cowley AR, Beer PD (2006) Chem Commun, p 3676
45. Hasenknopf B, Lehn J-M, Kneisel BO, Baum G, Fenske D (1996) Angew Chem Int Ed
Engl 35:1838
46. Hasenknopf B, Lehn J-M, Boumediene N, Dupont-Gervais A, Van Dorsselaer A,
Kneisel B, Fenske D (1997) J Am Chem Soc 119:10956
47. Campos-Fernandez CS, Schottel BL, Chifotides HT, Bera JK, Bacsa J, Koomen JM, Rus-
sell DH, Dunbar KR (2005) J Am Chem Soc 127:12909
48. Campos-Fernandez CS, Clerac R, Dunbar KR (1999) Angew Chem Int Ed 38:3477
49. Chand DK, Biradha K, Kawano M, Sakamoto S, Yamaguchi K, Fujita M (2006) Chem
Asian J 1:82
50. Diaz P, Tovilla JA, Ballester P, Benet-Buchholz J, Vilar R (2007) Dalton Trans, p 3516
51. Telfer SG, Yang X-J, Williams AF (2004) Dalton Trans, p 699
52. Harding LP, Jeffery JC, Riis-Johannessen T, Rice CR, Zeng Z (2004) Dalton Trans,
p 2396
53. Harding LP, Jeffery JC, Riis-Johannessen T, Rice CR, Zeng Z (2004) Chem Commun,
p 654
54. Otto S, Kubik S (2003) J Am Chem Soc 125:7804
55. Katayev EA, Pantos GD, Reshetova MD, Khrustalev VN, Lynch VM, Ustynyuk YA,
Sessler JL (2005) Angew Chem Int Ed 44:7386
56. Katayev EA, Boev NV, Khrustalev VN, Ustynyuk YA, Tananaev IG, Sessler JL (2007)
J Org Chem 72:2886
57. Katayev EA, Sessler JL, Khrustalev VN, Ustynyuk YA (2007) J Org Chem 72:7244
Struct Bond (2008) 129: 207–248
DOI 10.1007/430_2008_084
Published online: 25 April 2008
Molecularly Imprinted Polymers

Using Anions as Templates
Sally L. Ewen · Joachim H. G. Steinke (u)
Department of Chemistry, Imperial College London, South Kensington Campus,
London SW7 2AZ, UK
j.steinke@imperial.ac.uk
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
1.1 Context . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
2 Molecularly Imprinted Polymers . . . . . . . . . . . . . . . . . . . . . . . 209

2.1 The Concept . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
2.2 Inspiration, Incorporation and Assimilation . . . . . . . . . . . . . . . . . 210
2.3 A Brief “Developmental” History . . . . . . . . . . . . . . . . . . . . . . . . 211
2.4 Scope and Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
2.5 Formats . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
2.6 Design Criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
3 Molecular Imprinting Using Anionic Templates . . . . . . . . . . . . . . . 222

3.1 Anionic Phosphate Derivatives . . . . . . . . . . . . . . . . . . . . . . . . . 225
3.2 Carboxylates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
3.3 Anionic Sulfate Derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
3.4 Other Anions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
3.5 Concluding Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
Abstract Molecularly imprinted polymers (MIPs) are a class of solid-phase artificial re-
ceptors that are prepared using templates during the polymer network forming step; these
are subsequently removed to generate the selective receptor sites. The ease of synthesis,
the possibility of nanomolar binding constants and high levels of molecular discrimina-
tion, as well as environmental stability and ability to be reused, have led to a dramatic
boost in research interest in MIPs. One particularly promising area of study is the use
of anionic templates in the synthesis of MIPs and the targeting of substrates that carry
biologically important anionic functionalities. Benefiting from concurrent developments
in supramolecular receptor design and synthesis, it has become clear that MIPs for anion
recognition will impact biological screening, diagnosis, point-of-care devices (including
online sensing) and read-out.
Keywords Anionic templates · Molecularly imprinted polymers · Self-assembly ·

Biomimicry · Synthetic receptors
Abbreviations
FIP Functional group imprinted polymer
ISFET Ion-sensitive field effect transistor
208 S.L. Ewen · J.H.G. Steinke
ITC Isothermal titration calorimetry

ITO Indium tin oxide
MIP Molecularly imprinted polymer
MPA Methylphosphonic acid
NIP Non-molecularly imprinted polymer
NMR Nuclear magnetic resonance
ODS Octadecylsiloxane
PMP Pinacolyl methylphosphonate
SDS Sodium dodecyl sulfate
SPE Solid phase extraction
TSA Transition state analogue
1
Introduction
Molecular templates are involved universally in relaying structure and func-

tion into both biological and synthetic molecular assemblies. Information on
shape, size and electronic spatial configuration of a template, i.e. the molecu-
lar state of a molecule, is used to guide and control the formation, structure
and function of new self-assembled molecular entities. The template either
becomes part of the new structure or is separated from it and reused, or in-
volved in other templated processes. Messenger RNA in the polymerase chain
reaction functions as a reusable template for new complementary polynu-
cleotide strands [1]. Diblock copolymers can be used as templates for min-
eralisation of non-trivial inorganic phosphate structures [2, 3]. Patterned sur-
faces of phosphate groups template the mineralisation and resulting complex
morphology of inorganic salts such as calcium carbonate [4]. In liquid crys-
tals the addition of dopants can cause a switch to a different liquid crystalline
phase [5]. The formation of a molecular capsule can be triggered through the
presence of the correct solvent which occupies the interior [6]. Micelles have
been used for many years as templating agents in the formation of zeolites [7].
In this review we will be discussing synthetic polymer receptors formed as
crosslinked networks in the presence of molecular templates. The focus will
be on anionic templates and the performance of the resulting solid-phase re-
ceptor structures in terms of chemical selectivity and molecular recognition.
1.1
Context
Anion templation as a means of producing molecularly imprinted polymers

has to deal with the issues that one encounters designing and synthesising
molecular receptors for anions in general [8–10]. The increasing understand-
ing of the factors involved in anion recognition, such as how to design recep-
tor sites and ligand arrays that are appropriate for the larger and more diffuse
Molecularly Imprinted Polymers Using Anions as Templates 209
charge distribution in anions, is one reason for the recent surge in interest in
preparing MIPs for anion recognition. Another reason is the impact of pro-
teomics and the challenges and opportunities that accompany it with regard
to sensing and detection of charged species (simple anions as well as carboxy-
late, phosphate and sulfate derivatives) in the interrogation and analysis of
complex biological systems. A third reason relates to advances in the ability of
chemists to apply supramolecular and self-assembly approaches to the design
of complex structures and materials, from which MIP synthesis has already
begun to benefit.
2
Molecularly Imprinted Polymers
2.1
The Concept
The synthesis of a molecularly imprinted polymer is at first sight a straight-

forward affair. A template molecule, which can essentially be freely chosen, is
mixed together with one or several polymerisable receptor molecules (“func-
tional monomers”) [11–14] (Fig. 1). The latter, so-called binding sites, are
selected on the basis of preferably strong and directed interactions with
the template molecule to maximise the molecular fidelity of the imprint-
ing process (covalent and non-covalent interactions are possible). This step
Fig. 1 Pictorial representation of molecular imprinting, showing a formation of pre-

polymerisation complex; b polymerisation; c removal of template molecule to generate
molecularly imprinted recognition site
takes place once template and binding sites have self-assembled, typically
in solution, and crosslinking molecules have been added which produce
a crosslinked polymer network once polymerisation has been initiated, in
most instances thermally or photochemically although electrochemical or re-
dox initiation has also been demonstrated successfully. The template is then
removed from the polymer, most commonly through liquid extraction, leav-
ing behind a polymer matrix made up of supramolecular receptor sites which
have been formed throughout the polymer network as a result of the pre-
organisation of the binding site/template complex [15]. These receptor sites
are often referred to as molecularly imprinted cavities as a polymer matrix
has formed around the template, becoming part of the receptor structure.
The complementarity of shape and electron density distribution transferred
into the polymer matrix has obvious analogies with the substrate recognition
sites of enzymes or antibodies where the spatial organisation of the func-
tional groups of amino acid side chains define catalytic sites and substrate
selectivity [12]. As the ability of conformational adaptation to the substrate
is a key feature of enzyme recognition and catalysis, so is a certain level
of conformational freedom in a MIP [16]. Regarding the molecular recogni-
tion mechanism operational in MIPs, a number of recent studies illustrate
the decade-old controversy between the imprint mechanism (cavity gener-
ated by template) and the association mechanism (trapped templates act as
binding/nucleation sites) [17, 18].
2.2
Inspiration, Incorporation and Assimilation
The concept of molecularly imprinted polymers is strikingly simple and

generic. Very often MIPs are referred to as synthetic enzyme or antibody
mimics due to the analogy of having spatially defined functional groups
positioned inside a molecular pocket. Once this simplistic view took hold,
most MIP research revolved around identifying the level to which MIPs can
mimic the structure and function of protein receptors and catalyst active
sites [11–14]. However, just as inspirational as the likening of MIPs to an-
tibodies and enzymes has been, as incomplete is the appreciation of their
potential under conditions where proteins are mostly unsuitable or inade-
quate; examples are their use in organic solvents and in variable and ex-
treme temperature environments. In recent years, MIPs have been discussed
more generally as a special type and format of receptor design and struc-
ture, within the wider chemical world of supramolecular chemistry [19–21].
Consequently more of the advances in self-assembly and molecular recog-
nition have been explored and incorporated into MIP design and synthe-
sis. This trend will be even more beneficial to the development of MIPs
in years to come, during which many concepts and approaches will have
been assimilated, causing a step change in MIP performance. Some instruc-
tive examples of this progress include stoichiometric non-covalent binding

sites (vide infra Sects. 2.3 and 3), high-throughput and parallel synthe-
sis experimentation [22–24], integration of sensing modules into MIPs [25,
26], new polymerisation methods [27], modelling approaches of templat-
ing complexes [28, 29] and single site MIPs as dendrimer [30] or micro-
gel [31]. Future activities will include: improvements of recognition in wa-
ter [32]; the use of (dynamic) combinatorial chemistry [10, 33] and thermo-
dynamically controlled approaches for MIP synthesis; the use of MIPs as
the sensing component for medical diagnostics [34] and environmental an-
alysis [35], also as polymeric drugs, but, more importantly, as drug deliv-
ery vehicles (including molecular tags and labels) [36]; the application of
MIPs as identification and separation tools for biological screening [37];
the preparation and processing of MIPs with smaller feature sizes and
more complex shapes [38, 39]; the successful use of more challenging tem-
plate molecules (proteins, complex drug molecules). On the whole, MIPs
are relatively easily prepared multidentate receptors, in contrast with small
molecule receptor analogues, and have shown nanomolar binding affini-
ties [40]. Their solid-state nature is in itself a useful format for screening
strategies, with further benefits arising from long-term stability, recyclability
and modularity.
2.3
A Brief “Developmental” History
Polyakov et al. were the first to report the effect of generating selectivity in
a polymer matrix through addition of a template molecule (for a more de-
tailed account see [14]). In 1931 they prepared silica particles from sodium
silicate and (NH4 )2 CO3 in water, adding benzene, toluene or xylene in the
process. After prolonged drying and exhaustive extraction with hot water, the
polymer showed higher capacity for the additive (template) than for struc-
turally related compounds. A contemporary debate ensued about the origin
of antibody selectivity in the immune system with contributions from Breinl
and Haurowitz [41], Mudd [42] and Linus Pauling [43]. The latter favoured
the view that antibodies are generated in the presence of an intruding anti-
gen which would determine the antibody conformation (as he attempted to
demonstrate in 1942 [14]). Similar work to that of Polyakov was reported in
1949 by Dickey, this time using alkyl orange dyes as the templates. The im-
printed silica gel showed pronounced selectivity for the template, which was
present throughout the silica network-forming step [44]. Curti et al. showed
the first examples of enantioselectivity with both mandelic acid and camphor
sulfonic acid (also in silica) and their application as stationary phases for
chromatography (Fig. 2) [45, 46].
A particularly imaginative example was contributed by Patrikeev et al. in
1960, in which they resorted to bacteria as templates. Curiously the imprinted
Fig. 2 A selection of important anionic template molecules and their corresponding bind-
ing sites reflecting the development of MIPs
silica promoted certain bacterial growth over that of control samples im-
printed with other varieties [47]. In a related study either the levo or dextro
form of the same bacillus (Bacillus cereus var. mycoides) was incorporated
into a silica gel matrix showing discrimination for L- and D-linanool vapour
respectively [48]. The same group provided perhaps the earliest example of
a catalytic MIP templating amino acid condensation reactions, though rates
were enhanced modestly by a factor of two and experimental details are
scarce [49]. Two decades later, in 1972, Wulff et al. [50] and Klotz et al. were
independently pioneering the use of organic polymers as tailor-made recep-
tors. Wulff et al. were the first to harness the versatility of radical vinyl chem-
istry as polymerisation methodology and exemplified the covalent imprinting
approach (Fig. 2), whereas Klotz and coworkers harnessed the benefits of
ring-opening polymerisation combined with reversible disulfide crosslinking
employing non-covalent interactions [51]. In the following years Wulff et al.
elaborated the covalent approach with the addition of non-covalent (predom-
inantly electrostatic) interactions [52, 53]. A method to imprint on a surface
(“surface imprinting”) was first developed by Sagiv et al. in 1979 using sil-
ica particles as surfaces with polymerisable siloxane as surface modifiers [54].
This methodology is transferable to other surfaces as long as the template
absorbs onto the chosen surface [55]. The work in the group of Mosbach
in 1981 revolutionised molecular imprinting through the non-covalent ap-
proach [56] (Fig. 2). The impact derived from the much simpler synthetic
route for making MIPs. Rather than having to prepare templates connected
to polymerisable binding sites via reversible covalent bonds, non-covalent in-
teractions allow the self-assembly of template and binding sites in solution,
affording a pre-polymerisation complex prior to vinyl monomer network for-
mation. In further development, Mosbach et al. demonstrated that MIPs with
high selectivity could be obtained even without the necessity to invoke cova-
lent or ionic bonds [57]. With time, more and more groups have taken up the
non-covalent approach and today it is the most widely used methodology to
prepare MIPs [14].
However, comparative studies of covalent and non-covalent imprinting are
rare, and even those reported lack the confidence in which optimised condi-
tions have been employed for both approaches, so that clear conclusions from
which to select the better approach could not be drawn [58, 59].
Although the notion of combining the advantages of covalent bonds during
the imprinting step with those of non-covalent interactions for rebinding was
present in earlier work by Wulff et al. [50, 52, 60, 61], the first example of using
exclusively covalent interactions for the imprinting step and noncovalent ones
for rebinding was executed by Sellergren and Andersson in 1990 [62]. Poly-
merisable groups were linked by ester bonds, though the target molecule,
p-aminophenylalanine ethyl ester, was necessarily different to the template.
This changed completely in 1994, when Whitcombe et al. significantly refined
the semi-covalent approach, outlining a concept based on a small covalent
fragment as space holder (“sacrificial spacer approach”) [63]. With this strat-
egy, template and a binding site are covalently linked with a cleavable spacer
which is designed to reveal complementary functional groups upon removal,
closely following the geometry of the covalent juncture for improved com-
plementarity [64]. The semi-covalent approach is an attempt to synergise the
advantages of the covalent methodology (strict control of functional group
location, more uniform distribution) with that of the non-covalent one (re-
duced kinetic restriction upon rebinding).
As a “natural” coalescence of the covalent and the original non-covalent
approach, a stoichiometric non-covalent approach was developed [65]. If the
non-covalent interactions are strong enough to produce association constants
of at least 103 M–1 (or preferably higher) the equilibrium will lie well on the
side of the template-functional monomer complex. On the way to stoichio-
metric non-covalent interactions (discussed in detail in Sect. 3) a representa-
tive example of carboxylic acid templates and various N-base binding sites is
found in Kempe et al. (1993) [66]. Polymerisable amidines were first used for
stoichiometric imprinting of a transition-state analogue (TSA) by Wulff et al.
in 1997 (Fig. 2) [67]. Other noteworthy developments were valine-derived
binding sites for dipeptide imprinting by Yano et al. in the same year [68]
and bidentate dioxoborolanyl recognition sites for carboxylates exploited by
Lübke et al. in 2000 [69]. In 1994 Sellergren was already successful in comple-
menting the standard MAA binding site monomer with pentamidine, offering
association constants close to those typically considered to be required for
stoichiometric imprinting strategies (Fig. 2) [70].
With time and improved synthetic protocols, larger templates (fullerenes,
dendrimers, nanoparticles, colloids, micelles, lipid bilayers, self-assembled
block copolymers, oligonucleotides, DNA and proteins) have been im-
printed [14] and the choice of matrices has expanded to liquid crystal
polysiloxanes, carbon networks, zeolites, layered aluminophosphates and
colloidal crystals, though organic polymer networks remain the dominant
imprint casting medium [14].
Further important developments related to templates emerged in the mid-
and late 1990’s. In 1999, Sreenivasan et al. demonstrated that it is possible
to imprint with two different template molecules simultaneously [71]. A year
later, Rachkov and Minoura introduced the epitope imprinting methodology
in which a structurally unique 3-amino acid fragment of a peptide chain was
shown to be a sufficient template to produce MIPs that are selective for the
entire nonapeptide [72]. Another year later, Sellergren et al. used a template
analogue rather than the target molecule itself for generating MIPs [73], fur-
ther suggesting that complex molecules can be imprinted successfully on the
basis of a molecular information-rich substructure or derivative. Also, ma-
nipulation of the binding site chemistry has widened the opportunities for
MIP applications. Useful synthetic post-polymerisation transformations were
elaborated on disulfide templates by Mukawa et al. in 2002, as their reduction
was found to yield thiols as binding sites (semi-covalent approach), which

can also be oxidised up to sulfonic acids, changing the chemical nature of
the non-covalent binding site in a controlled manner [74]. The first MIPs
designed for influencing molecular reactivity came from the laboratories of
Shea et al. in 1978 [75], closely followed by Damen and Neckers in 1980 [76].
Cycloadditions leading to cyclopropane and cyclobutane dicarboxylic acids
respectively were performed, with significant regio- and diastereoselectivity
in the latter case. Also in 1980, Belokon et al. organised amino acid tem-
plates via Schiff base formation inside a MIP cavity, and showed that upon
deprotonation the amino acid carbanion retained its original configuration
unlike the same reaction in solution [77]. A few years later, in 1987, Sarhan
et al. showed the possibility of stereospecific reversal of configuration rather
than retention with mandelic acid [78]. Andersson et al. achieved the cata-
lytic deprotonation of bound amino acids (vide supra Belokon et al.) electing
a pyridoxal-coenzyme analogue as a template in a non-covalent imprinting
system [79]. In this particularly eventful year Wulff et al. were the first to
carry out an asymmetric reaction in a chiral cavity designed to generate op-
tically active amino acids, which was achieved with an ee of 36% and used
glycine as a template [80]. The observed chiral induction is solely the re-
sult of enantioselective induction by the chiral cavity. The above examples
give a taste of the potential of MIP cavities as nanoreactor sites. MIPs that
are catalytically active (apart from some early work in silica matrices [81])
and which thereby influence reaction outcomes, were only embarked upon in
1987, by Leonhardt et al. [82]. Initially the product of a reaction was pursued,
and not the better mimicry of using a transition state analogue. It produced
esterolytic activity with a 2–3 fold rate enhancement and substrate selectivity
of MIPs. Two years later however Robinson et al. prepared a MIP imprinted
with a TSA. Poly(vinylimidazole) and p-nitrophenylphosphonate TSAs were
coordinated to a CoII ion and the resulting template assembly was crosslinked
with dibromobutane. Despite the more rational design of the template only
a small enhancement of 1.6-fold was observed [83]. As the development of
catalytic MIPs is strongly intertwined with the evolution of stoichiometric
non-covalent binding sites as oxyanionic receptor sites, we will return to this
connection in one of the later sections (Sect. 3).
2.4
Scope and Limitations
Discussions about the current and general scope and limitations associated
with MIPs are ongoing. For a while, a good number of publications were of
two minds in their assessment of the potential of MIPs, where small advances
in performance were hailed as stepping stones to rival, if not excel, anti-
body and enzyme methods. This was followed by a healthy sobering period.
Over the last ten years a much more realistic assessment of the potential of
MIPs has emerged, accompanied by a substantial increase in the number of

researchers involved in MIP science and technology, creating a rise in publi-
cation numbers. Strengths of MIPs include:
• Robustness/longevity
– MIPs have been shown to keep their molecular recognition perform-
ance over months and even years; this is a stark contrast to enzymes
and antibodies. Depending on the chosen polymer matrix, MIPs are
chemically and physically robust and not limited to an aqueous envi-
ronment or well-controlled physiological conditions (pH, temperature,
solvent). MIPs can be recycled and reused [84].
• Solid-state format
– Most MIPs are solid-phase which makes them amenable to automated
processes requiring only extraction and reloading cycles, with sub-
strate isolation being a simple filtration step [85].
• Relative ease of synthesis
– The concept of molecular imprinting is simple and elegant, which is
also reflected in the relatively straightforward manner in which they
can be synthesised [14]. This is certainly true if one compares MIP syn-
thesis with other strategies available to produce synthetic receptors.
Many of the latter require a large number of synthetic steps, rather than
self-assembly, to achieve the spatial positioning of functional groups
identified as being key for mimicking enzyme catalysis or antibody
performance.
• Relative ease of design
– MIPs rely entirely on self-assembly (cf. Pauling’s theory of antibody
formation) and present a modular system, which also allows MIPs
to be designed more easily in contrast to approaches in which the
geometric requirements of the receptor molecule have to be carefully
calculated or estimated during the design phase [86] (though the orig-
inally chosen template may not be the best “template” for the desired
performance) [87, 88].
• Unrestricted choice of substrate
– As long as it is possible, either in solution or in bulk, to self-assemble
template with binding sites, and as long as the polymerisation chem-
istry is compatible with such a complex, the choice of template
molecule would appear to be unrestricted [14].
• Can be made to give a response (e.g. QCM, fluorescence).
– Through the addition of multi-functional binding sites, additional
function for sensing, or some other form of responsive behaviour,
can be incorporated without having to redesign template or receptor
sites [25, 26, 34, 35].
Limitations of MIPs:
• There are some common limitations associated with MIPs regardless of
the chosen template, binding site, solvent, etc.:
– Binding sites are heterogeneous (polyclonal rather than monoclonal
as in enzymes and monoclonal antibodies) as a consequence of the
statistical nature of the polymer network forming process, leading to
undesirable band broadening in chromatographic separations and to
a distribution of different activities of catalytic sites [89].
– Similarly not all template molecules may be accessible within the poly-
mer matrix, or will require very long extraction times leading to prob-
lems of template leaching especially when employing MIPs for trace
analysis [35].
– Diffusion to and from the recognition sites within the MIP matrix is
typically slow, depending on the solid-phase format and phase struc-
ture. High pore volume and interconnectivity can reduce the diffusion
problem but generating such a hierarchical pore structure may not
be compatible with the requirement of forming a strongly associated
template complex. Thin film, surface imprinting and membrane ap-
proaches are alternative means of addressing diffusion problems but
are in many cases accompanied by lower selectivity due to interface
effects or changes in the crosslinking stoichiometry, leading to a reduc-
tion of the molecular fidelity of the imprint [14].
– For applications such as enantiopolishing or chiral separations, MIPs
are said to offer low capacity. This is true when comparing MIPs with
sorbents that rely on an interaction with a surface or a surface modi-
fied with a chiral selector (e.g. Pirkle phases). On the other hand, MIPs
are at least competitive if one compares their atom economy with that
of enzymes or antibodies adding up the molar mass of crosslinker and
receptor sites per template (between ∼5–100 kDa depending on syn-
thesis recipe) [14].
– Dealing with polar and particularly water soluble template molecules
and recognition in aqueous milieu have been major issues for MIPs
for many years. Progress across the field however indicates that generic
strategies are available to overcome this limitation [34].
• Major issues regarding deficits in technology for anion MIPs will be dis-
cussed in more detail in Sects. 3.1–3.4.
2.5
Formats
Molecularly imprinted polymers are implicated for a wide range of applica-

tions. Their solid-phase nature necessitates that the synthesis step includes
the processing step. As is the case with any type of covalently crosslinked ma-
terial, once polymerisation has taken place the molecular structure is fixed
and cannot be reconfigured. The most common formats encountered for
MIPs are:
• Irregular particles
– The most popular format prepared by grinding the bulk-polymerised
MIP into smaller particles. For chromatographic applications this is
usually followed by sizing through sieving [90].
• Monoliths
– Upon crosslinking, the polymerisation mixture retains the shape of the
reaction vessel. Separating the inner wall of the reaction vessel from
the polymer reveals a single piece of MIP, a monolith. The monoliths
could be used for MIP applications that have been shown to be use-
ful as a direct means of packing chromatographic columns (HPLC,
CEC) [90, 91]. If the monolith is exposed to large variations in sol-
vent polarity without space constraints it will slowly disintegrate into
smaller particles due to the high density of crosslinks in connection
with a glass transition temperature (Tg ) that is usually well above room
temperature.
• Beads
– This third format is particularly useful to pack chromatography columns
and generally to manipulate MIPs in automated processes [90]. As it re-
quires a two- or multi-phase solvent system during polymerisation to
impart the spherical shape, until recently it has proven quite challeng-
ing to obtain narrow disperse bead sizes with high yield at the desired
size. Protocols are now available that offer generic solutions for beading
a wide range of MIP formulations [92]. Beads are particularly desir-
able for catalytic MIP applications as selectivity improves (compared to
monolith synthesis) and catalytic activity increases due to faster mass
transfer [93].
• Films
– Especially for sensing (UV-Vis, luminescence, QCM), film formats, i.e.
surfaces coated with a MIP layer, are attractive as a simple means of
device fabrication. This does not differ greatly from the monolith for-
mat, and film thicknesses can be controlled through an appropriately
shaped substrate or through spin coating. The changes in solvent con-
centration caused by evaporation of solvent and precipitation of the
MIP during the film forming processes will alter the stoichiometry of
the template/binding site complex set in the starting solution [94].
• Membranes
– Free standing films of MIPs have been prepared in various ways.
Mechanical robustness has been imparted through lower crosslink
densities or monomers generating lower Tg polymer backbones com-

pared to monoliths. Reducing crosslink density or rigidity generally
reduces selectivity and long-term stability of MIPs. Membrane com-
posites offer a means of accommodating brittle MIP particles into an
overall flexible matrix. Phase inversion techniques with polar high-
performance polymers have also been successfully applied to form
MIP membranes [95].
• (Hydro)gels
– Instead of high crosslink levels and thus polymer networks with high
Tg , MIP gels and hydrogels can be produced in the same way as mono-
liths, by selecting monomers leading to lower Tg polymer backbones
and/or by choosing polar monomers with high affinity to water and
protic solvents, so that the solvent causes the polymer network to swell
and to become more flexible, i.e. gel-like. Interesting applications are
antibody mimics with MIPs becoming single site catalysts [31] and,
more recently, controlled drug delivery [96].
• Single molecule species
– Zimmerman et al. were the first to have synthesised a dendrimer
imprinted with a porphyrin derivative using the sacrificial spacer
methodology: a single cavity within a single polymer confinement
structure [30].
• Spun fibres
– Electrospinning has been applied to MIP synthesis employing mem-
brane formation techniques with the help of a support polymer,
promising higher surface area, faster diffusion and more efficient
washing out of the template molecule [97].
2.6
Design Criteria
MIPs are complex synthetic receptors prepared by self-assembly, followed

by network formation. They are by their very nature amorphous solids, and
those are inherently difficult to analyse at the molecular level. Most ana-
lyses of MIPs are indirect, assessing the fidelity of the molecular recognition
events through binding assays of various kinds. Nuclear magnetic resonance
(NMR) studies on the self-assembled complex formed prior to polymerisa-
tion as well as molecular modelling have been shown to be useful in arriving
at a more rational design of MIPs [98], though pre-polymerisation analysis
and modelling [99–102] are still in stages too early to extrapolate trend pat-
terns and reach more general conclusions on reaction conditions, binding
strength and template/binding site complex formation. The vastness of the
parameter space involved when synthesising MIPs has prompted the use of
parallel synthesis combined with statistical data analysis to more rapidly ana-
lyse data sets and identify trends. These activities will mature and, with time,
will have wider impact on our ability to design MIPs for any given template
molecule with tunable thermodynamic and kinetic performance parameters.
Although there are always exceptions, the following design criteria have been
selected as being useful and reliable guides that help one arrive at a reason-
able starting point for developing a new MIP:
• Polymerisation temperature
– Lower temperatures improve selectivity and capacity. The exothermic
nature of radical vinyl polymerisation compounded by the gelling of
the imprint mixture is likely to cause higher internal temperatures than
those used to control the environment externally.
• Binding site design
– Statistical copolymerisation kinetics with the crosslinker lessens the
heterogeneity of receptor sites.
– Too much conformational freedom between binding site and polymer
backbone reduces selectivity.
– Strong, ideally close to stoichiometric ratios of binding interactions
improve the quality of the imprint.
– Faster equilibration kinetics increase performance in chromatography-
based application.
• Crosslinker
– There is an optimum level of crosslinking with typical values to be
around 70–95 mol %. Conformational flexibility of the crosslinker has
to be balanced with the overall level of crosslinks. The chemical na-
ture of the crosslinker determines the useful solvent range for a MIP.
The size of the crosslinker ideally matches the dimensions of the tem-
plate, to avoid conformational frustrations within the polymer matrix
as a result of a structural misfit, which can lead to a loss in capacity
and/or selectivity.
• Solvent
– The best choice is a solvent that maximises template binding site in-
teractions while producing a highly porous MIP matrix to minimise
diffusion limitations.
• Template
– Imprinted cavities can be viewed as multi-point receptor sites plus
the added shape of the cavity as an additional recognition feature.
A template offering many points for binding and where stoichiomet-
ric binding sites for the exhibited functional groups exist offers the
best conditions for high selectivity. On the other hand, large templates
with many binding positions become more difficult to imprint with as

the differences between related molecules become relatively smaller.
This is also the case for templates with only one functional group,
where the shape of the cavity becomes a more important contribu-
tor. Larger templates pose the problem of being difficult to extract
from the polymer as they more readily become entrapped. Degrad-
able templates or template fragments (epitopes) have been shown to be
helpful in overcoming this limitation. Conformationally dynamic and
demanding templates such as proteins are a particular case in point.
Presenting proteins in native form on a degradable nanocrystal sur-
face, for example, is an ingenious way to address both shape stability
of the template and access to the imprinted cavity.
• Template to binding site stoichiometry
– A good starting point is a ratio that reflects the number of sites avail-
able on the template for interaction with the chosen polymerisable
receptor. For covalently bound templates or stoichiometrically coordi-
nating binding sites the ratio is obvious.
• Crosslinker to self-assembled complex stoichiometry
– The minimum number of “atoms” that are required to form the desired
binding pocket and typical stoichiometries that give high selectivity
are 20 : 1 in equimolar terms. Larger ratios can lead to enhanced selec-
tivity but this is traded against a loss in capacity.
• Covalent versus non-covalent imprinting
– The covalent approach is synthetically more labour-intensive but pro-
duces MIPs that have better defined affinity distributions. It is still
undecided as to which situation one of these two strategies will lead to
the higher performing MIP.
The polymerisation chemistry and other network forming processes applied
to MIP generation (e.g. membrane phase inversion) lack a thermodynamic
means of controlling the formation of receptor sites to render them less poly-
clonal. Some means of increased control such as living [103] and thermody-
namically controllable polymerisation chemistry [27, 104, 105] (in analogy to
dynamic combinatorial library synthesis [106]), are at a very early stage. Im-
proved binding site design, especially in water or buffered system, would ben-
efit from hydrophobic effects for recognition cf. enzyme pocket; for this, how-
ever, one needs to incorporate the necessary conformational/environmental
change associated with it. Also, the area of modelling for selecting optimised
stoichiometries for MIP manufacture would become a more powerful means
of guiding the synthesis, with more precisely defined interactions that gener-
ate an imprint with increased 3D/shape fidelity.
3
Molecular Imprinting Using Anionic Templates
One of the most attractive features of molecular imprinting is its applicabil-

ity to such a diversity of analytes. Thus the literature boasts MIPs made using
a striking array of templates, which range from small molecules (<100 Da) to
biological macromolecules (several kDa), and include sugars, steroids, dyes,
herbicides, pharmaceuticals and biochemicals amongst others (Table 1). Nev-
ertheless, upon inspection it is apparent that anionic templates have seldom
been considered throughout the history of imprinting. Indeed, up until the
last 7 or 8 years, there has been a surprising paucity of MIPs that are designed
either to recognise “naked” anions (such as oxyanions, halides or cyanide)
or to exploit the predominating anionic character of a particular analyte. In-
directly, however, anionic templates have played a role as early as the 1940s
up to the present time. For example, Dickey prepared silica gels in the pres-
ence of methyl orange (p-dimethylamino-azobenzenesulfonic acid) and then
later on using a sulfonamide analogue as template [44, 107]. In a subsequent
communication it was disclosed that the two MIPs exhibited very similar se-
lectivity for their target molecule, indicating that the negative charge on the
p-substituent was nonessential to the recognition mechanism [107]. Curti
et al. imprinted in silica using enantiomers of mandelic acid and camphor-
sulfonic acid. While the conditions for their MIP preparation (pH 4) suggest
that the sulfonic acid, and to some extent the carboxylic acid, were disso-
ciated throughout the imprinting process, the aforesaid functional groups
were not deliberately targeted in the MIP synthesis [45]. In a more recent
example, Rajkumar et al. generated a MIP for fructosyl valine recognition,
using boronic acid-functionalised monomers to covalently interact with the
diol moieties [108]. The carboxyl group, which would have been fully dissoci-
ated during the rebinding event (pH 11.4), was not targeted by any functional
monomer.
The relative scarcity of anion-imprinted polymers is especially remark-
able given the prevalence of anionic molecules of importance. It is widely
recognised that the design of anion hosts is a challenging task, especially in
contrast to cation receptor design. The main reason for this is the diffuse
nature of anions. Since Anions are significantly larger than the equivalent
isoelectronic cations, they have a lower charge to radius ratio; consequently,
electrostatic binding interactions are markedly diminished. For the same rea-
son, anions are generally more polarisable than cations, and it is therefore
more difficult to develop specific binding sites for their recognition.
With regard to molecular imprinting, a significant hindrance to the devel-
opment of anion-recognition MIPs has been the fact that anionic species are
very often incompatible with apolar media. The molecular imprinting process
involves the formation of a pre-polymerisation complex between the template
molecule and functional monomers. This is typically based upon H-bonding
Table 1 Examples of molecules that have been used as templates for molecular imprinting
Template molecule Selected Refs.
Cyclobarbital [162, 163]
Naproxen [164–166]
(S)-Propranolol [167]
L-Menthol [27]
Glucose [168, 169]
Cholesterol [64, 170, 171]
Nicotine [172, 173]
Bisphenol A [174]
and electrostatic interactions (though it might also include dipole–dipole

attractions or metal ion coordination). In general, such interactions are max-
imised when a non-competitive, apolar solvent is employed during the im-
printing step, thereby affording a well-defined binding site. Similarly, the use
of such a solvent for subsequent rebinding of the target molecule tends to pro-
mote specific binding as well as minimise levels of undesirable nonspecific
interactions. The last 20 years have witnessed notable advances pertaining to
the compatibility of molecular imprinting with polar/aqueous media. While
much of that is outside of the scope of this review, one particular issue –
the influence of functional monomers used – is of prime importance to our
discussion.
Traditionally, the majority of MIPs have been made using a limited reper-
toire of simple, commercially available functional monomers (for example,
methacrylic acid, methacrylamide, 1-vinylimidazole), in accordance with the
non-covalent imprinting strategy. In this regard, Simon and Spivak recently
investigated a selection of commercially available functional monomers
deemed to be particularly suitable for oxyanionic templates (in particular,
for carboxylates and phosphonates) [109]. They prepared a series of MIPs,
incorporating various functional monomers including pyridine-derivatives,
aliphatic amines, and 1-vinylimidazole. A chiral template molecule, t-Boc-L-
phenylalanine, was employed to study enantioselective discrimination. The
MIP particles were slurry-packed into HPLC columns, and capacity factors
and separation factors of the L- and D-enantiomers were calculated to provide
values for binding affinity and enantioselectivity. Interestingly, the authors
observed that, while the highest binding affinity was exhibited by the more
basic aliphatic amine derivatives, the best selectivity was obtained by func-
tional monomers bearing aromatic amine groups. Directional binding, as
afforded by these latter functional monomers, is evidently vital for the pro-
duction of selective binding sites.
However, while the use of simple, commercially available functional
monomers is relatively undemanding from a synthetic chemistry perspective,
the drawback is that such monomers generally tend to afford only relatively
weak binding interactions. This is all the more evident in aqueous/polar me-
dia. As a consequence, it is usually necessary to imprint using a large excess
of functional monomer relative to the amount of template molecule, in order
to push the equilibrium and ensure a high degree of complexation. Unfor-
tunately, this approach results in undesirable binding site heterogeneity and
subsequent non-specific rebinding of the analyte.
The introduction of a novel imprinting methodology, which has attracted
considerable attention within the field, serves to explain much of the re-
cent progress with anion-imprinted polymers. Termed “stoichiometric non-
covalent imprinting”, this approach uses carefully selected (often custom-
made) functional monomers which bind with relatively high affinity to par-
ticular motifs on the template molecule [110]. Since the vast majority of
functional monomer is associated with the template molecule, it is possible

to use an equimolar mixture of template and functional monomer during im-
printing. In theory, this not only reduces non-specific interactions but also
affords binding sites that are more defined, since the functional monomers
are more precisely arranged by the template molecule in a definite spatial
orientation within the imprint cavity. Furthermore, the strong binding inter-
actions afforded by these functional monomers means they are often strong
enough to work in solvent systems more suited to anionic molecules, as well
as apolar systems. Wulff et al. were the first to propose the concept of stoi-
chiometric non-covalent imprinting [111] (this can be thought of as a logical
extension to the covalent imprinting strategy, also pioneered by Wulff). They
reported the use of custom-made polymerisable amidinium species for tar-
geting oxyanions, in line with the prevalence of guanidinium-based arginine
residues in oxyanion binding sites in biological systems. More recently, other
authors have promoted functional monomers based upon urea [112, 113],
thiourea [114] and guanidinium groups [115, 116].
Regarding the scope of this review, in the following subsections, the use of
molecular imprinting for the recognition of selected classes of anions will be
discussed. In defining “anions”, we refer not only to small “typical” anions,
such as nitrate, phosphate, sulfate, etc., but also to larger, increasingly more
complex substrates, which bear an anionic moiety. Moreover, we shall limit
our discussion to MIPs in which the anionic part of the template molecule is
deliberately targeted or used to advantage. In this way, we disregard any re-
ports of molecular imprinting where the template molecule might happen to
feature an anionic component (for example, the C-terminus of peptides) but
where this is not deliberately accounted for in the design of the functional
binding site.
3.1
Anionic Phosphate Derivatives
Phosphate derivatives are ubiquitous in biological systems, rendering them

a very attractive target for anion recognition chemistry. For example, many
coenzymes and metabolic intermediates are esters of phosphoric and pyro-
phosphoric acid; the nucleic acids DNA and RNA are based upon a phosphate
diester backbone; and phospholipids, the major component of all biologi-
cal membranes, contain a phosphate diester moiety. Protein phosphorylation
is a reversible post-translational modification involving the kinase-catalysed
transfer of a phosphate group from ATP to a protein (typically to serine, thre-
onine or tyrosine residues). This highly important process can be thought of
as a means of “switching on” proteins, and underlies the regulation of virtu-
ally all cellular events. In addition, many man-made molecules of interest are
phosphate derivatives. This includes environmental pollutants and degrada-
tion products of chemical warfare agents.
One of the earliest examples of a polymer imprinted with a phosphate

derivative came from Turkewitsch et al., who constructed a MIP for the
selective recognition of cAMP in aqueous media (Fig. 3) [117]. The custom-
designed functional monomer 1 was used to target the nucleotide via both
aryl stacking and electrostatic interactions between the positively-charged
nitrogen and the phosphate diester anion (Fig. 4). Rather cleverly, this func-
Fig. 3 Schematic representation of cAMP inside MIP, showing possible binding interac-
tions
Fig. 4 Functional monomer custom-designed for cAMP-imprinted polymer

tional monomer also served as a transducer element: binding of cAMP ef-

fected fluorescence quenching within the imprinted polymer, enabling it to be
used as a chemosensor. Further recognition points within the imprinted cav-
ity were afforded by the use of a large excess of HEMA, which was assumed
to bind to the purine moiety via a network of H-bonds. The fluorescent na-
ture of the MIP was used to calculate an association constant, Ka , for binding
of cAMP by the MIP in aqueous media: this was determined to be in the order
of 105 M–1 . With regard to selectivity, the MIP was shown to discriminate be-
tween the sodium salts of cAMP and the potentially competitive analogue,
cGMP, thus highlighting the role of the adenosine moiety in the recognition
process. From the standpoint of our discussion, an investigation into selectiv-
ity over other oxyanions would also have been of interest.
A steadily increasing number of publications report the use of phosphate-
derived template molecules to prepare MIPs for solid-phase extraction (SPE).
SPE involves the selective pre-concentration of trace-level compounds from
complex mixtures. The high selectivities and affinities obtainable by molecu-
lar imprinting render these polymers ideal for use as SPE sorbents.
For example, Möller et al. prepared a MIP for the selective isolation of
diphenyl phosphate (a flame retardant hydrolysis product) from urine [118].
The MIP was prepared using an excess of 2-vinylpyridine as the functional
monomer, with ditolyl phosphate as the template molecule (imprinting with
a structural analogue of the target molecule eliminates complications caused
by possible leaching of the template during subsequent rebinding analysis).
In this study, the influence of matrix components normally present in hu-
man urine (urea, NaCl) was investigated. While urea did not compete with
diphenyl phosphate for the basic binding sites, it was observed that NaCl af-
fected both the recovery and repeatability of diphenyl phosphate binding.
This was explained by the assumed complexation between Na+ ions and the
anionic analyte, which would render the molecule more neutral and thereby
hinder ionic interaction with the polymer.
The last decade has seen considerable interest in the application of mo-
lecular imprinting to nerve agent detection. The general objective is to de-
velop a sensitive and specific portable sensor device that boasts faster re-
sponse times than existing nerve agent analysis techniques (gas chromatog-
raphy, HPLC, immunoassay). In this regard, Jenkins et al. used pinacolyl
methylphosphonate (PMP), the hydrolysis product of Soman, as a template
in the preparation of an imprinted polymeric luminescent sensor [119, 120].
PMP reversibly bound to ligand-coordinated Eu3+ , which was incorporated
into the recognition site, and this binding event corresponded to the ap-
pearance of a narrow luminescence band in the 610 nm region of the Eu3+
spectrum. The presence of organophosphorous pesticides and herbicides was
determined not to interfere with PMP recognition (although none of the com-
pounds screened bore an anionic moiety).
Fig. 5 Schematic representation of MPA-imprinted polymer
Zhou et al. developed a potentiometric chemosensor for the detection

of methylphosphonic acid (MPA), another degradation product of nerve
agents [121]. In the presence of MPA, an octadecylsiloxane (ODS) thin
layer was covalently bound to the surface of an indium tin oxide (ITO)
transducer (Fig. 5). Subsequent removal of the physisorbed MPA afforded
what the authors claim to be imprinted recognition sites, with functional
and steric complementarity. It was demonstrated that rebinding of MPA
by the “imprinted” film generated a greater potentiometric response than
binding of homologous substrates (ethylphosphonic acid, propylphosphonic
acid, tert-butylphosphonic acid). However, binding of MPA by the con-
trol, non-imprinted film, afforded a greater potentiometric response than
binding of MPA homologues by the “imprinted” film. It is thus prob-
able that recognition in this case is governed by a size-exclusion phe-
nomenon. Further experiments should be carried out to eliminate this
ambiguity.
Prathish et al. also generated a potentiometric sensor for MPA detection,
although using an altogether different methodology [122]. MIP particles, pre-
pared via the copolymerisation of EDMA and either methacrylic acid or
4-vinylpyridine in the presence of MPA (which was afterwards removed using
1:1 (v/v) methanol:acetic acid) were dispersed in 2-nitrophenyloctyl ether
(plasticizer) and polyvinyl chloride, and this solution was used to cast a mem-
brane of thickness ∼0.45 mm. Potentiometric analysis of ground water sam-
ples, which were spiked with MPA and adjusted to pH 10 in the presence of
Tris buffer, yielded a greater response by the MIPs than by the corresponding
NIPs, thus indicating an imprinting effect. However, the high level of bind-
ing by the NIPs suggests that a substantial degree of nonspecific, presumably
hydrophobic interactions is involved in, or even dominates, the recognition
event. It should also be noted that since the amount of template recovered
from the imprinted polymers was not reported, the possibility of residual
template “bleeding” from the MIPs and affecting the responses cannot be
wholly discounted.
Imprinting with phosphorylated/phosphonylated template molecules has

been particularly valuable in the synthesis of catalytically-active enzyme
mimicking MIPs. In this approach, the tetrahedral geometry of the phos-
phate/phosphonate ester functional group is likened to the tetrahedral transi-
tion state of carbonate or ester hydrolysis reactions and the template molecule
is thus referred to as a TSA. The recognition sites in the resulting im-
printed polymer are designed to stabilise the high energy transition state
of the hydrolysis reaction, thereby leading to enhanced rates of reaction.
This strategy, pioneered by Mosbach [83], emulates a method of produc-
ing catalytically-active polyclonal antibodies, viz. immunisation with stable
TSAs [123]; however, in contrast to polyclonal antibodies, MIPs are consider-
ably more robust and cheaper to produce.
The work of Kawanami et al., who produced a MIP to catalyse p-nitro-
phenyl acetate hydrolysis, serves to illustrate the methodology (Fig. 6) [124].
p-Nitrophenyl phosphate, employed as a template molecule, was observed
to form an insoluble complex when mixed in solution with a 10-fold excess
of the chosen functional monomer, 1-vinylimidazole; this was then copoly-
merised with a 20-fold excess of divinylbenzene. Following polymerisation,
Fig. 6 Schematic representation of: a molecular imprinting with template (TSA);

b imprinted recognition site following template removal; c,d catalysis
the template molecule was removed by means of continuous extraction with

acetonitrile, phosphate buffer and methanol, affording an insoluble macro-
porous polymer. MIP-catalysed hydrolysis of p-nitrophenyl acetate was ob-
served to be 85 times faster than uncatalysed hydrolysis. This was two times
faster than the catalytic activity of a NIP, thus indicating that catalysis oc-
curs within the imprinted sites. Further evidence for this was provided by
the fact that addition of the template molecule, p-nitrophenyl phosphate,
clearly inhibits hydrolytic activity. However, the authors conceded that the
catalytic performance of the MIP does not yet compare with that of cata-
lytic antibodies for the same hydrolysis reaction. Although no explanation
has been given for this shortcoming, the use of an excess of functional
monomer (cf. stoichiometric non-covalent imprinting strategy), which would
have afforded a heterogeneity of binding sites encompassing a broad spec-
trum of affinities and selectivities, is likely to have been one contributing
factor.
Also using a phosphate-derived TSA template, Wulff et al. have demon-
strated significant progress in the synthesis of catalytic MIPs modelled on
carboxypeptidase A [115]. The active site of this hydrolytic enzyme features
two guanidinium groups (arginine residues) and a Zn2+ cation. One of the
guanidinium moieties is thought to bind the oxyanion formed during hydro-
lysis, while the metal ion promotes catalysis via polarisation of the substrate;
the second guanidinium and a hydrophobic pocket influence substrate speci-
ficity. Following this, the authors prepared a MIP with a phosphate diester
template molecule employed as a TSA of carbonate hydrolysis (Fig. 7). A func-
tional monomer was custom-designed, incorporating an amidinium group
as an H-bond donor (cf. guanidinium groups) and a triamine to coordinate
Zn2+ . This monomer bound to the template molecule through stoichiometric
non-covalent interaction.
Fig. 7 Schematic representation of: a molecular imprinting with template (TSA);

b catalysis of diphenylcarbonate hydrolysis
The catalytic performance of the MIP and appropriate controls were in-
vestigated with the hydrolysis of diphenylcarbonate. Reaction kinetics, fol-
lowed by HPLC analysis of aliquots, were calculated as pseudo first-order
rate constants. Rather encouragingly, the imprinted polymer showed typical
Michaelis-Menten kinetics, in line with natural enzymes. The catalytic activ-
ity of the MIP (expressed as kcat /kuncat , the ratio of turnover number of the
catalysed reaction to turnover number in the absence of catalyst) was calcu-
lated to be 6900. This is markedly higher than has been reported for catalytic
antibodies for carbonate hydrolysis (kcat /kuncat = 810), clearly demonstrating
the great potential of MIPs for use as artificial enzymes.
All of the imprinted polymers discussed above involve phosphory-
lated/phosphonylated organic molecules, where the recognition process relies
equally upon binding to, and accommodation of, both the organic backbone
and the anionic functional group of the template. In fact, we are aware of only
one report of a MIP designed to target inorganic phosphate, H2 PO4 – . In this
one example, Kugimiya et al. present a series of MIPs prepared using thiourea-
derived functional monomers [114]. The thiourea group is known to be a good
host for anions, particularly for the phosphate anion, to which the N – H moi-
eties bind through a set of coplanar H-bonds. The authors’ initial experiments
indicated that 1-allyl-2-thiourea is particularly suitable for use in aqueous
media: this was in direct contrast to a related functional monomer, N-methyl-
N -(4-vinylphenyl)-thiourea, which showed almost no binding ability under
the conditions studied. Such results were explained by the greater hydrophilic-
ity of 1-allyl-2-thiourea. Thus a polymer was imprinted with phenylphosphonic
acid as the template molecule, using EGDMA as the crosslinker, acetonitrile as
the porogen and four equivalents (to template molecule) of 1-allyl-2-thiourea
as the functional monomer. Incidentally, Kugimiya et al. omit explaination as
to why a phosphonic acid template molecule was chosen for a MIP designed for
NaH2 PO4 recognition. Presumably they had anticipated that the difference in
geometry between these two functional groups would be negligible.
To evaluate the ability of the MIP to selectively bind NaH2 PO4 , MIP and
NIP particles were stirred in distilled water with a mixture of various anions
(NaH2 PO4 , KNO3 , NaCl, Na2 SO4 , CH3 COONa). After 1 h, the supernatant
was removed and analysed by ion chromatography. Under these conditions,
both the MIP and the NIP preferentially bound phosphate over the other an-
ions measured, and the MIP bound 58% more template than the NIP, thus
demonstrating a clear imprinting effect and selectivity for the target anion.
Unfortunately, due to the insufficient level of information provided, it is diffi-
cult to fully appreciate the data. For example, while the amount of each anion
bound is presented as % binding activity, it is unclear how this relates to the
estimated number of available binding sites. Furthermore, it was belatedly
discovered that CH3 COO– could not actually be detected via ion chromatog-
raphy, and the influence of this potentially-competitive anion was therefore
not determined.
3.2
Carboxylates
The prevalence of the carboxylate moiety in both biogenic and man-made

molecules of interest makes this functional group a popular target for an-
ion host chemistry. Needless to say, carboxylates are a major constituent of
proteins, peptides and amino acids, and the expansion of proteomics begets
increasing requirements for means of specific detection of such biomolecules.
Other relevant examples of carboxylates include fatty acids, while many small
molecule di- and tricarboxylates are implicated in key metabolic pathways
such as the citric acid cycle (e.g. citrate, succinate, fumarate and malonate).
Carboxylated anthropogenic molecules include trichloroacetic acids, anionic
surfactants and β-lactam antibiotics.
While a small number of early examples of molecular imprinting with
carboxylate-bearing templates exist (e.g. Curti et al. imprinting with mandelic
acid [45]) in many of these cases the carboxylate moiety was not deliber-
ately accounted for. In 1982, Sarhan and Wulff imprinted with D-glyceric acid,
using monomers bearing a boronic acid (for covalent interactions) and an
amino functionality [125]. In 1990, Andersson and Mosbach prepared MIPs
using N-protected amino acid derivatives as templates with a 4-fold excess
of methacrylic acid as the functional monomer, demonstrating that imprint-
ing (and enantiomeric resolution) is possible in the absence of both covalent
and ionic interactions (binding was predominantly due to H-bond interac-
tions) [126]. The imprinting of sialic acid reported in 1996 by Kugimiya et al.
is worthy of mention as it used a combination of covalent (boronate esterifi-
cation) and ionic interactions for MIP formation, illustrating that this is pos-
sible in aqueous media [127]. This brief synopsis illustrates the initial under-
representation of carboxylate-targeted MIPs. The situation began to change
with a few groups exploring the use of polymerisable nitrogen bases as po-
tential functional monomers for (non-stoichiometric) non-covalent binding
of carboxylic acids [66, 128, 129]. Today, a number of groups focus their ef-
forts on constructing MIPs which deliberately target the carboxylate moiety.
In particular, various tailor-made functional monomers have been reported,
many adapted from naturally-occurring binding motifs, which are designed
to bind to the template with affinity sufficient to incorporate them stoichio-
metrically into the polymer matrix.
In one of the earliest examples of stoichiometric non-covalent imprinting,
Lübke et al. used a combination of custom-designed functional monomers
to imprint with ampicillin, a β-lactam antibiotic which bears both an
amine and a carboxylate functional group [69]. Rather elegantly, a poly-
merisable electron-deficient quinone 2 was used to interact with the amine
(via formation of an n–π complex), while the carboxylate was targeted
using a bis(boronate-amide)-functionalised monomer 3 (Figs. 8–10). This
latter species, adapted from synthetic receptors described by Hughes and
Fig. 8 Functional monomer custom-designed to bind the carboxylate of ampicillin
Fig. 9 Functional monomer custom-designed to bind the amine of ampicillin
Smith [130], exploits internal Lewis acid coordination between the boron and
the carbonyl oxygen, which serves to polarise the amide, thereby increasing
affinity for the oxyanion. Imprinting was carried out by allowing stoichiomet-
ric amounts of the monomers and the tetrabutylammonium salt of ampicillin
to complex in DMSO over 24 h, prior to incorporation into a highly crosslinked
polymer. Subsequent extraction with acetonitrile afforded a 92% recovery of
the template. The resulting MIP showed good potential, performing well in
aqueous media with optimal binding at pH 8 (where the carboxylate is present
as the anion) and with selectivity over structural analogues.
Zhang et al. reported the design and four-step synthesis of a novel
guanidinium-functionalised monomer which was built upon a vinylan-
thracene scaffold to effect fluorescence [116]. NMR titration experiments
in deuteriomethanol indicated the formation of a 1 : 1 complex with ac-
etate, with estimated association constants of the order of 105 M–1 . Although
synthesis of this particular monomer does demand more expertise than
monomers typically used for imprinting – with somewhat uninspiring re-
coveries reported by the authors – it nonetheless appears to be a promising
candidate for stoichiometric non-covalent imprinting with carboxylates.
Hall et al. are amongst the most prolific authors in the field of car-
boxylate imprinting. Making use of expertise in the field of supramolecular
Fig. 10 Schematic representation of ampicillin inside MIP, showing possible binding in-
teractions
chemistry, viz. Hamilton et al. [131, 132], they have developed a series of
novel monotopic and ditopic functional monomers based upon a urea core
(Fig. 11) [112, 133]. These synthetically accessible host species can usually be
prepared in a one-step procedure, either from 1-isopropenyl-4-isopropyl-2-
isocyanate and the corresponding amine or by reaction of 4-vinylaniline with
the corresponding isocyanate. Ureas bind to oxyanions via donation of two
H-bonds, and it was shown that varying the substitution of the monomers
– and hence modifying the acidity of the urea core – can result in quite
dramatic changes in affinity for the model substrate, tetrabutylammonium
benzoate. For example, NMR titration experiments in DMSO-d6 afforded an
association constant of 1322 ± 48 M–1 for 4 where R1 = R2 = H, while the
corresponding value for 4 where R1 = R2 = CF3 was 8820 ± 1600 M–1 . The
authors have explored a number of potential applications for these urea-based
monomers. These include a class-selective MIP for β-lactam antibiotics, for
which it was hypothesised that the mono-urea functionalised binding site
directly targets the β-lactam carboxylate moiety while providing sufficient
Fig. 11 Functional monomers based upon urea core, designed by Hall et al.
space to accommodate pendant substituents [88, 134]. In another example,

a nitrophenyl-functionalised analogue was used to prepare a chromogenic
MIP for the enantioselective recognition of Z-glutamate and the related drug,
methotrexate; this monolithic polymer exhibited a change in colour intensity,
visible to the naked eye, upon binding the target analytes in basic, aque-
ous media [135]. Along similar lines, Schmitzer and Gomy have used NMR
titrations, isothermal titration calorimetry (ITC) and molecular modelling to
design, synthesise and screen a library of functional monomers that are also
based upon the urea motif [113]. Out of 16 candidates, the two monomers 5
and 6 (Fig. 12) demonstrated particularly high affinity for the model sub-
strate, bis(tetrabutylammonium)-N-Z-L-glutamate. Association constants be-
tween this latter species and both 5 and 6 were calculated to be 3370 ±
260 M–1 and 4000 ± 400 M–1 respectively (in DMSO-d6 ). These monomers
should be promising receptors for future stoichiometric incorporation into an
imprinted polymer.
While most of the aforementioned MIPs were constructed using func-
tionalised, highly crosslinked organic polymers, other workers have demon-
strated that inorganic matrices, assembled by a surface sol–gel technique,
are also valid candidates for anion receptor chemistry [136, 137]. With this
Fig. 12 Functional monomers based upon a urea core, designed by Schmitzer and Gomy
approach, molecularly imprinted ultrathin TiO2 films were prepared by first

allowing titanium butoxide to covalently complex (as confirmed by FT-IR
measurements) with a carboxylated template molecule in a mixture of toluene
and ethanol (Fig. 13). Water was subsequently added, and the complex was
allowed to chemisorb onto a hydroxylated solid substrate, such as the gate
surface of an ion-sensitive field effect transistor (ISFET) device [138]. Follow-
Fig. 13 Hydrolysis of Ti(IV)-carboxylate template and subsequent rebinding

ing complete hydrolysis, aqueous ammonia was used to remove the template,
affording a sol–gel film with imprinted recognition sites. It was suggested
that rebinding of the template molecule occurs via a combination of covalent
bonding, H-bonding, metal coordination, and hydrophobic interactions with
Ti–O moieties. Such molecularly imprinted films are particularly suited to use
as transducer elements in chemosensor devices.
This method was demonstrated to afford molecularly imprinted TiO2
films selective for azobenzene carboxylic acids [136, 139], chloroaromatic
acids [138], chiral carboxylic acids including amino acid derivatives [140, 141]
and anthracenecarboxylic acids [139]. The methodology was also adapted
for use with water-soluble di- and tri-peptides [142]. In this latter case, even
though NaOH solution was used as an alternative to aqueous ammonia, com-
plete template removal proved impracticable, with 30–40% of the template
trapped in the film: nonetheless, the TiO2 films were said to show repro-
ducible binding for guest molecules. More recently, TiO2 films were also used
for imprinting thiolate- and phosphonate-functionalised templates [143].
The inherent simplicity of metal oxide sol–gel films, in particular the
lack of requirement for carefully chosen functional monomers, makes this
approach an attractive alternative to the use of highly crosslinked organic
polymers. The technique is optimised for – indeed, necessitates – an anionic
functional group on the template, and results in imprinted films with a no-
table degree of regioselectivity, structural selectivity and enantioselectivity.
To date no studies have been published demonstrating the degree of selectiv-
ity between substrates with differing anionic functional groups.
Finally, we wish to mention an interesting and somewhat more unusual
example of molecular imprinting, albeit it is debatable as to whether this
particular application should rightly be included within a discussion on
anionic templates. D’Souza et al. copolymerised 6-methacrylamidohexanoic
acid and divinylbenzene in the presence of calcite (calcium carbonate poly-
morph) crystals and, after exhaustive removal of the template, showed that
the imprinted polymer promoted nucleation of calcite from an aqueous su-
persaturated calcium carbonate solution [144, 145]. Using several methods of
analysis, it was demonstrated that no template remained in the MIP after the
wash steps; subsequent crystal-formation was therefore attributed to hetero-
geneous nucleation as opposed to seeding by residual crystals. Regarding the
functional monomer used, though the authors do not explicitly state their
reasoning behind choosing 6-methacrylamidohexanoic acid, they report that
the use of other functional monomers (acrylic or methacrylic acids) in place
of 6-methacrylamidohexanoic acid afforded MIPs with inferior nucleating
abilities. It was suggested that the presence of the flexible spacer allowed the
acidic head group of the monomer to more easily match the spacing of ions
on the surface of the crystal. More recently, Egan et al. applied this same
methodology to imprinting with calcium oxalate crystals, using the resulting
MIP to model nucleation of renal calculi from artificial and real urine [146].
3.3
Anionic Sulfate Derivatives
While perhaps not as ubiquitous as phosphate derivatives and carboxylates,

sulfate derivatives are nonetheless a frequently encountered class of oxyan-
ions and are therefore another attractive target for anion receptor chemistry.
Important examples of such molecules include sulfated sugars and anionic
surfactants.
Intriguingly, there are actually relatively few accounts of imprinting with
sulfates/sulfonates and, with the exception of aforementioned historical re-
ports where the anionic component was either not directly targeted [45] or
found to be unnecessary for the recognition event [107], most examples come
from the last 3–4 years. That said, looking at the binding motifs that tar-
get sulfate derivatives in biological systems and in small molecule receptor
chemistry, it is apparent that many of the functional monomers designed for
phosphate and carboxylate recognition could eventually be adapted for use
with sulfates/sulfonates.
In 2004, Caro et al. imprinted a polymer with 1-naphthalene sulfonic acid,
using a 4-fold excess of 4-vinylpyridine as functional monomer (cf. stoichio-
metric non-covalent imprinting), EGDMA as crosslinker and AIBN as free
radical initiator [147]. A 4 : 1 v/v mixture of methanol and water was used
as the porogen, due to the high polarity of the template, and it was assumed
that molecular recognition would occur via a combination of hydrophobic
effect and ionic interactions. Following extraction of the template, the poly-
mer particles were packed into a polyethylene cartridge, and a mixture of
different polar compounds (including eight naphthalene sulfonates) was per-
colated through at pH 2.3. All compounds, except oxamyl and methomyl,
were retained on the MIP in the loading step. The MIP was then washed
with MeOH to remove non-selectively bound species. Somewhat unexpect-
edly, all eight of the naphthalene sulfonates were retained by the imprinted
polymer in spite of the wash step; these included various mono- and di-
sulfonated species, some of which were hydroxyl- or amine-substituted, as
well as the original template, 1-naphthalene sulfonic acid. However, all other
polar compounds (phenol, nitrophenols, bentazone and 4-chloro-2-methyl-
phenoxy acetic acid) were eluted.
The explanation offered by the authors is that the MIP was cross-reactive
for naphthalene sulfonates. However, at pH 2.3, only the naphthalene sul-
fonates were anionic (pKa 0.5–0.6). It is quite plausible that in this example,
molecular recognition was predominantly an anion-exchange mechanism,
and that this overshadowed the influence of any molecularly imprinted cav-
ities. In this way, the MeOH wash step could be seen to have disrupted
non-covalent, non-ionic interactions, leaving behind the anionic naphthalene
sulfonates, which were retained much more strongly via ionic interactions
with the protonated pyridyl moieties. Unfortunately, this is always a poten-
tial hazard with constructing such hosts: there is often a fine balance between
obtaining high affinity binding interactions while still retaining the selectivity
afforded by the act of imprinting.
The recent work of Albano et al. illustrates a particularly innovative ap-
plication of molecular imprinting [148]. In the presence of sodium dodecyl
sulfate (SDS), a thin layer of polypyrrole was electrodeposited onto the sur-
face of a quartz crystal. Removal of the SDS by washing with water afforded
a molecularly imprinted sensor element which could be used to monitor lev-
els of pollutant surfactants in river water. Thus, the piezoelectric quartz sensor
was contained within a flow cell so that the MIP-coated side of the crystal re-
mained in contact with analyte solution while the electrodes of the sensor were
connected to an oscillator circuit. A significant drop in the quartz crystal os-
cillation frequency occurred when the sensor was in contact with SDS. This is
consistent with an increase in mass on the surface of the sensor, suggesting that
SDS molecules had bound to the layer of imprinted polypyrrole. A reference
polypyrrole-coated sensor, prepared in parallel to the MIP-sensor but in the
absence of SDS, showed only a minimal response to the target analyte.
The influence of pH on sensor performance was investigated: MIP-coated
sensors exhibiting the highest response and sensitivity were those buffered
at pH 9 during the polymerisation step; likewise, an alkaline measurement
solution (pH 8) was found to afford the best results. It is evident that the
anionic form of SDS, rather than the free acid, is required for optimal bind-
ing to the pyrrole moieties. With regard to selectivity, the sensor was also
applied to solutions of both sodium dodecanoate (structurally identical to
SDS except bearing a carboxylate group in place of the sulfate ester) and
sodium dodecylbenzenesulfonate. While the sensor showed low sensitivity to
the carboxylate-functionalised analyte, a response equal to that afforded by
SDS was attained with the benzenesulfonate analogue. Nonetheless, the degree
to which molecular recognition by the MIP was governed by ionic strength, as
opposed to a definite molecular imprinting effect, is as yet unclear.
Recently, Sineriz et al. reported the first example of a MIP imprinted with
a sulfated sugar [149]. Sulfated sugars, such as heparan sulfate and chon-
droitin sulfate, are a class of highly abundant biological molecules with im-
portant intercellular roles. Their activities are directly related to patterns and
degree of sulfation. Understanding their biological functions therefore neces-
sitates structure elucidation, yet antibodies developed for this purpose have
generally been found to be insufficiently selective.
In an initial study, various amine-functionalised monomers (Fig. 14) were
incorporated into polymers templated around glucose-6-O-sulfate. Relative
binding strengths of the resulting MIPs were determined by equilibration
with glucose-6-O-sulfate solution followed by HPLC analysis of the super-
natant. In DMSO, it was found that the MIP made using quaternary amine-
functionalised monomer 7 exhibited high affinity towards the analyte but no
selectivity (the MIP and corresponding NIP performing equally). In contrast,
Fig. 14 Functional monomers screened by Sineriz et al.
the MIP made using a primary amine (8) was observed to bind up to 80% of
a given concentration of glucose-6-O-sulfate in DMSO, while the correspond-
ing NIP showed negligible binding. Consistent with the aforementioned work
of Simon and Spivak [109], it was apparent that the directional nature of H-
bonds from the primary amine, coupled with the inherent pre-organisation
of the imprinted binding site, were crucial for selectivity. Further experi-
ments showed that the MIP prepared using the primary amine-functionalised
monomer preferentially bound glucose-6-O-sulfate over other sulfated sac-
charides (galactose-6-sulfate, glucose-3-sulfate and N-acetyl-glucosamine-6-
sulfate). The degree of selectivity over other anionic functional groups was
not investigated.
3.4
Other Anions
While relatively few MIPs have been made to recognise the oxyanionic
functional groups phosphate/phosphonate, carboxylate and sulfate/sulfonate,
MIPs designed to target inorganic anions are even more scarce. The following
paragraphs list examples of this mode of imprinting.
Fujiwara and co-workers attempted to surface imprint pyridine-bearing
microspheres using tetravalent ferrocyanide anions as the template [150]. Ad-
sorption studies indicated that, while the imprinted microspheres did show
greater affinity for the template anion in comparison to non-imprinted mi-
crospheres, they lacked selectivity and bound all other polyvalent anions
tested. This suggests that, although the pyridine groups were apparently pre-
organised by the imprinting process, negligible selectivity for the target an-
ions may be due to the absence of imprinted cavities on the surface of the
microspheres.
In 1988, Dong et al. reported a chloride-ion selective electrode, prepared
via electropolymerisation of pyrrole in the presence of LiCl [151]. The resul-
tant polymer film was observed to give a Nernstian response to chloride, with
a limit of detection of 3.5 × 10–5 M (said to be comparable to that of most
other contemporary chloride-selective electrodes). Later, the same methodol-
ogy was adapted to produce a chloride-selective microsensor, used to detect
chloride in serum [152]. Both the ion-selective electrode and the microsensor
showed poor selectivity for chloride over other anions. For example, poten-
tial selectivity coefficients of the electrode were of the order 10–1 for other
halides, IO4 – and ClO4 – , and >1 for NO3 – , HCOO– [151]. It was postulated
that selectivity was related to ionic radius.
A rare example of electropolymerisation employed for the formation of
an anion selective crosslinked polymer matrix was introduced by Kamata
et al. [153, 154]. Reductive coupling between trifunctional p-cyanopyridinium
crosslinkers (Fig. 15) in water containing the required counteranion led to
stable polymer networks which showed thermodynamic and kinetic anion
selectivity. A size exclusion effect was observed whereby only the imprint
halogen (and any smaller halogen anion) was electrochemically recognised.
Diffusion of the counterion was reduced in the case of the same template an-
ion MIP, but increased with the size of the templating anion. Experiments
clearly demonstrating anion selectivity have yet to be carried out.
Fig. 15 Trifunctional p-cyanopyridinium crosslinker used to prepare halide-imprinted

polymer
In the mid-1990s, the molecular imprinting principle was adopted for the
preparation of a nitrate-selective electrode [155]. Ion-selective electrodes,
i.e. sensors which convert the activity of a specific ion into an electrical
potential, have widespread application in biochemical and biophysical an-
alysis; however, their major limitation is poor selectivity leading to interfer-
ence from other ions. Hutchins and Bachas electropolymerised pyrrole onto
a glass-carbon electrode in the presence of an aqueous solution of NaNO3 .
The polymer-coated electrode was demonstrated to detect aqueous nitrate,
giving a near-Nernstian response, with response times ranging from 24 to
<6 s. Recognition appeared to be based on size-exclusion phenomena. There-
fore, while the polymer was able to discriminate over traditional interferents
whose radii is larger than that of NO3 – , such as ClO4 – and I– , the much
smaller SCN– still effected a response. It was hypothesised that the hydropho-
bicity of the polypyrrole films induced the superimposition of a Hofmeister-
type selectivity (i.e. based upon lipophilicity) upon the nitrate imprinting
selectivity. Thus lipophilic anions not large enough to be sterically hindered
from the film, such as thiocyanate, still interfered with nitrate recognition.
Much more recently, this technology has been translated into the production
of nanolitre-scale electrochemical cells [156].
3.5
Concluding Comments
Tailoring molecular imprinting for anion recognition is a recent development

in a relatively young art. To date, the large majority of anion-imprinted poly-
mers have been designed to recognise organic molecules which bear one or
more oxyanionic functional group, and we see few examples where the tar-
get analyte is a simple, inorganic anion. A plausible explanation for this bias
may be that there is simply less demand for such receptors: after all, simple
anions can be resolved with relative ease using ion chromatography. Yet this
contrasts sharply with the rest of the synthetic receptor scene, where the focus
is very much upon developing host molecules for anions such as phosphate,
sulfate, acetate and the halides [10, 157, 158]. As an alternative explanation,
it has been suggested that small and simple species are much poorer “im-
printogens” than larger, multi-functionalised molecules [159], and that it is
generally quite difficult to create a well-defined recognition site when less
chemical information is available during the imprinting step. For example,
Li and Wu reported an inability to imprint with formate, acetate and pro-
pionate [160]. Following computer simulation studies, they suggested that
the crosslinker (EDMA) had been unable to form adequate cavities for such
small, noncomplex molecules. The extent to which this applies to the use of
other crosslinkers has not been investigated.
It is apparent that the current emphasis is towards MIPs that can dis-
criminate by both the anionic moiety and the organic residue of a substrate.
Progress in this direction is likely to be particularly beneficial to the recent
efforts to imprint with proteins [20, 161], or indeed to any application where
the target species is adorned with multiple anionic functionalities. To sepa-
rately target each functional group on a complex template molecule should
naturally give rise to better defined binding sites with greater affinity and
selectivity for the target species. Equally beneficial will be progress towards
compatibility with aqueous media and pH independent binding.
An interesting topic of discussion concerns the different levels of selectiv-
ity required by anion-recognition MIPs; such requirements are dictated by the
intended final application of the imprinted polymer. To date, most MIPs have
been designed to be specific for one particular molecule, and in this regard
the aim has been to optimise imprinting technology so as to make binding
sites as uniform and “monoclonal” as possible. Still, there are a number of in-
stances where imprinted polymers benefit from a degree of cross-selectivity.
This is illustrated by the aforementioned MIP from Hall et al., which recog-
nises compounds bearing the anionic structural motif particular to β-lactam
antibiotics [134]. However, while recognition of a particular (anionic) sub-

structure is one notion of cross-selectivity, a possible extension of this would
be to develop imprinted polymers that are specific for one particular (an-
ionic) functional group, irrespective of the type, class, substructure or size of
the rest of the molecule (Fig. 16). In a way, this could be thought of as “turn-
ing the technology on its head”. Whereas with classic molecular imprinting
the intention is to eradicate all substrate promiscuity, this novel mode of
imprinting would afford binding sites that selectively recognise a particular
functional group present on the substrate, while discrimination by compound
type, class, substructure or size is deliberately suppressed. One might theorise
that such “functional group imprinted polymers” (“FIPs”) could be prepared
using an adaptation of the “epitope approach” to imprinting, whereby a sub-
structure of the target molecule(s) – in this case the functional group –
is used to imprint. Certainly the very act of imprinting, where functional
binding sites are pre-organised into an exact spatial and geometric arrange-
ment, lends itself (conceptually, at least) to discrimination between even
near-isosteric functional groups.
While molecular imprinting, both in general and with regard to anionic
templates, has seen significant development since the first, exploratory ex-
periments of Polyakov and Dickey, there is still vast room for improve-
ment. The technology will continue to benefit from concomitant advances
in a breadth of disciplines, from supramolecular chemistry to polymer sci-
ence to organic synthesis. And yet, the possibilities of molecular imprinting
are manifest. We hope that this review will serve to promote the potential of
Fig. 16 Pictorial representation of functional group imprinted polymer (FIP) concept

MIPs, as an extremely versatile class of potent synthetic receptors, where the

benefits for anion recognitions are beginning to emerge.
Acknowledgements Support for S.L.E. through a Chemical Biology Centre studentship at

Imperial College London is gratefully acknowledged.
References
1. Mullis KB (1994) Angew Chem Int Edit 33:1209
2. Cheng YJ, Gutmann JS (2006) J Am Chem Soc 128:4658
3. Smarsly B, Antonietti M (2006) Eur J Inorg Chem: 1111
4. Antonietti M, Breulmann M, Goltner CG, Colfen H, Wong KKW, Walsh D, Mann S
(1998) Chem Eur J 4:2493
5. Shibaev V, Bobrovsky A, Boiko N (2003) Prog Polym Sci 28:729
6. Conn MM, Rebek J (1997) Chem Rev 97:1647
7. Tanev PT, Pinnavaia TJ (1995) Science 267:865
8. Blondeau P, Segura M, Perez-Fernandez R, de Mendoza J (2007) Chem Soc Rev 36:198
9. Hay BP, Firman TK, Moyer BA (2005) J Am Chem Soc 127:1810
10. Kubik S, Reyheller C, Stuwe S (2005) J Incl Phenom Macrocycl Chem 52:137
11. Steinke J, Sherrington DC, Dunkin IR (1995) Adv Polym Sci 123:81
12. Wulff G (1995) Angew Chem Int Edit 34:1812
13. Zhang HQ, Ye L, Mosbach K (2006) J Mol Recognit 19:248
14. Alexander C, Andersson HS, Andersson LI, Ansell RJ, Kirsch N, Nicholls IA, O’Ma-
hony J, Whitcombe MJ (2006) J Mol Recognit 19:106
15. Andersson HS, Nicholls IA (1997) Bioorganic Chem 25:203
16. Wulff G (2002) Chem Rev 102:1
17. Maier WF, Benmustapha W (1997) Catal Lett 46:137
18. Davis ME, Katz A, Ahmad WR (1996) Chem Mat 8:1820
19. Liu ZS, Zheng C, Yan C, Ga RY (2007) Electrophoresis 28:127
20. Bossi A, Bonini F, Turner APF, Piletsky SA (2007) Biosens Bioelectron 22:1131
21. Qiao FX, Sun HW, Yan HY, Row KH (2006) Chromatographia 64:625
22. Dirion B, Cobb Z, Schillinger E, Andersson LI, Sellergren B (2003) J Am Chem Soc
125:15101
23. Lanza F, Sellergren B (2004) Macromol Rapid Commun 25:59
24. Martin-Esteban A, Tadeo JL (2006) Comb Chem High Throughput Screen 9:747
25. Southard GE, Van Houten KA, Ott EW, Murray GM (2007) Macromolecules 581:202
26. Rachkov A, McNiven S, El’skaya A, Yano K, Karube I (2000) Anal Chim Acta 405:23
27. Patel A, Fouace S, Steinke JHG (2004) Anal Chim Acta 504:53
28. Karim K, Breton F, Rouillon R, Piletska EV, Guerreiro A, Chianella I, Piletsky SA
(2005) Adv Drug Deliv Rev 57:1795
29. Wei ST, Jakusch M, Mizaikoff B (2006) Anal Chim Acta 578:50
30. Zimmerman SC, Wendland MS, Rakow NA, Zharov I, Suslick KS (2002) Nature
418:399
31. Wulff G, Chong BO, Kolb U (2006) Angew Chem Int Edit 45:2955
32. Wang XJ, Xu ZL, Yang ZG, Bing NC (2007) Prog Chem 19:805
33. Batra D, Shea KJ (2003) Curr Opin Chem Biol 7:434
34. Piletsky SA, Turner NW, Laitenberger P (2006) Med Eng Phys 28:971
35. Caro E, Marce RM, Borrull F, Cormack PAG, Sherrington DC (2006) Trends Anal
Chem 25:143
36. Sellergren B, Allender CJ (2005) Adv Drug Deliv Rev 57:1733

37. Theodoridis GA, Papadoyannis LN (2006) Curr Pharm Anal 2:385
38. Voicu R, Faid K, Farah AA, Bensebaa F, Barjovanu R, Py C, Tao Y (2007) Langmuir
23:5452
39. Belmont AS, Sokuler M, Haupt K, Gheber LA (2007) Appl Phys Lett 90:3
40. Berglund J, Nicholls IA, Lindbladh C, Mosbach K (1996) Bioorg Med Chem Lett
6:2237
41. Breinl F, Haurowitz F (1930) Z Physiol Chem 192:45
42. Mudd S (1932) J Immunol 23:423
43. Pauling L (1940) J Am Chem Soc 62:2643
44. Dickey FH (1949) Proc Nat Acad Sci USA 35:227
45. Curti R, Colombo U (1952) J Amer Chem Soc 74:3961
46. Curti R, Colombo U, Clerici F (1952) Gazz Chim Ital 82:491
47. Patrikeev VV, Balandin AA, Klabunovskii EI, Mardashev YS, Maksimova GI (1960)
Dokl Akad Nauk SSSR 132:850
48. Patrikeev VV, Smirnova ZS, Maksimova GI (1962) Dokl Akad Nauk SSSR 146:707
49. Patrikeev VV, Kozarenko TD, Balandin AA (1962) Izvest Akad Nauk SSSR, Ser
Khimi, p 170
50. Wulff G, Sarhan A (1972) Angew Chem Int Edit 11:341
51. Takagishi T, Klotz IM (1972) Biopolymers 11:483
52. Wulff G, Sarhan A, Zabrocki K (1973) Tetrahedron Lett, p 4329
53. Wulff G, Sarhan A, Gimpel J, Lohmar E (1974) Chem Ber 107:3364
54. Sagiv J (1979) Isr J Chem 18:346
55. Rubinstein I, Steinberg S, Tor Y, Shanzer A, Sagiv J (1988) Nature 332:426
56. Arshady R, Mosbach K (1981) Macromol Chem Phys 182:687
57. Andersson L, Sellergren B, Mosbach K (1984) Tetrahedron Lett 25:5211
58. Kugimiya A, Matsui J, Takeuchi T (1997) Mater Sci Eng C 4:263
59. Piletsky SA, Piletskaya EV, Panasyuk TL, El’skaya AV, Levi R, Karube I, Wulff G
(1998) Macromolecules 31:2137
60. Wulff G, Sarhan A, Gimpel J, Lohmar E (1974) Chem Ber 107:3364
61. Lauer M, Boehnke H, Grotstollen R, Salehnia M, Wulff G (1985) Chem Ber 118:246
62. Sellergren B, Andersson L (1990) J Org Chem 55:3381
63. Whitcombe MJ, Rodriguez ME, Vulfson EN (1994) Sep Biotech III 158:565
64. Whitcombe MJ, Rodriguez ME, Villar P, Vulfson EN (1995) J Am Chem Soc 117:7105
65. Wulff G, Knorr K (2001) Bioseparations 10:257
66. Kempe M, Fischer L, Mosbach K (1993) J Mol Recogn 6:25
67. Wulff G, Gross T, Schonfeld R (1997) Angew Chem Int Edit 36:1962
68. Yano K, Nakagiri T, Takeuchi T, Matsui J, Ikebukuro K, Karube I (1997) Anal Chim
Acta 357:91
69. Luebke C, Luebke M, Whitcombe MJ, Vulfson EN (2000) Macromolecules 33:5098
70. Sellergren B (1994) Anal Chem 66:1578
71. Sreenivasan K, Sivakumar R (1999) J Appl Polym Sci 71:1823
72. Rachkov A, Minoura N (2000) J Chromatogr A 889:111
73. Quaglia M, Chenon K, Hall AJ, De Lorenzi E, Sellergren B (2001) J Am Chem Soc
123:2146
74. Mukawa T, Goto T, Takeuchi T (2002) Analyst 127:1407
75. Shea KJ, Thompson E (1978) J Org Chem 43:4253
76. Damen J, Neckers DC (1980) J Am Chem Soc 102:3265
77. Belokon YN, Tararov VI, Savel’eva TF, Vitt SV, Bakhmutov VI, Belikov VM (1980)
Makromol Chem 181:89
78. Sarhan A, El-Zahab MA (1987) Makromol Chem Rapid Commun 8:555

79. Andersson LI, Mosbach K (1989) Makromol Chem Rapid Commun 10:491
80. Wulff G, Vietmeier J (1989) Makromol Chem 190:1727
81. Wulff G (2002) Nanoporous Mater III 141:35
82. Leonhardt A, Mosbach K (1987) React Polym 6:285
83. Robinson DK, Mosbach K (1989) J Chem Soc Chem Commun, p 969
84. Owens PK, Karlsson L, Lutz ESM, Andersson LI (1999) Trends Anal Chem 18:146
85. Turiel E, Tadeo JL, Martin-Esteban A (2007) Anal Chem 79:3099
86. Mayes AG, Whitcombe MJ (2005) Adv Drug Deliv Rev 57:1742
87. Caro E, Marce RM, Cormack PAG, Sherrington DC, Borrull F (2005) Anal Chim Acta
552:81
88. Urraca JL, Moreno-Bondi MAC, Hall AJ, Sellergren B (2007) Anal Chem 79:695
89. Toth B, Pap T, Horvath V, Horvai G (2007) Anal Chim Acta 591:17
90. Liu HY, Row KH, Yan GL (2005) Chromatographia 61:429
91. Lanza F, Hall AJ, Sellergren B, Bereczki A, Horvai G, Bayoudh S, Cormack PAG,
Sherrington DC (2001) Anal Chim Acta 435:91
92. Yoshimatsu K, Reimhult K, Krozer A, Mosbach K, Sode K, Ye L (2007) Anal Chim
Acta 584:112
93. Strikovsky AG, Kasper D, Grun M, Green BS, Hradil J, Wulff G (2000) J Am Chem
Soc 122:6295
94. Schmidt RH, Mosbach K, Haupt K (2004) Adv Mater 16:719
95. Ulbricht M (2004) J Chrom B 804:113
96. Siemoneit U, Schmitt C, Alvarez-Lorenzo C, Luzardo A, Otero-Espinar F, Concheiro A,
Blanco-Mendez J (2006) Int J Pharm 312:66
97. Chronakis IS, Milosevic B, Frenot A, Ye L (2006) Macromolecules 39:357
98. Svenson J, Ning Z, Fohrman U, Nicholls IA (2005) Anal Lett 38:57
99. Liu Y, Wang F, Tan TW, Lei M (2007) Anal Chim Acta 581:137
100. Dineiro Y, Menendez MI, Blanco-Lopez MC, Lobo-Castanon MJ, Miranda-Ordieres AJ,
Tunon-Blanco P (2006) Biosens Bioelectron 22:364
101. Nantasenamat C, Naenna T, Ayudhya CIN, Prachayasittikul V (2005) J Comput Aided
Mol Des 19:509
102. Dong WG, Yan M, Zhang ML, Liu Z, Li YM (2005) Anal Chim Acta 542:186
103. Vaughan AD, Sizemore SP, Byrne ME (2007) Polymer 48:74
104. Buchmeiser MR (2001) J Chrom A 918:233
105. Enholm EJ, Allais F, Martin RT, Mohamed R (2006) Macromolecules 39:7859
106. Corbett PT, Leclaire J, Vial L, West KR, Wietor JL, Sanders JKM, Otto S (2006) Chem
Rev 106:3652
107. Bernhard SA (1952) J Amer Chem Soc 74:4946
108. Rajkumar R, Warsinke A, Moehwald H, Scheller FW, Katterle M (2007) Biosens Bio-
electr 22:3318
109. Simon RL, Spivak DA (2004) J Chrom B 804:203
110. Wulff G, Knorr K (2002) Bioseparation 10:257
111. Wulff G, Schonfeld R (1998) Adv Mater 10:957
112. Hall AJ, Manesiotis P, Emgenbroich M, Quaglia M, De Lorenzi E, Sellergren B (2005)
J Org Chem 70:1732
113. Gomy C, Schmitzer AR (2006) J Org Chem 71:3121
114. Kugimiya A, Takei H (2006) Anal Chim Acta 564:179
115. Liu JQ, Wulff G (2004) Angew Chem Int Edit 43:1287
116. Zhang H, Verboom W, Reinhoudt DN (2001) Tetrahedron Lett 42:4413
117. Turkewitsch P, Wandelt B, Darling GD, Powell WS (1998) Anal Chem 70:2025
118. Moller K, Nilsson U, Crescenzi C (2004) J Chromatogr B 811:171

119. Jenkins AL, Uy OM, Murray GM (1997) Anal Commun 34:221
120. Jenkins AL, Uy OM, Murray GM (1999) Anal Chem 71:373
121. Zhou YX, Yu B, Shiu E, Levon K (2004) Anal Chem 76:2689
122. Prathish KP, Prasad K, Rao TP, Suryanarayana MVS (2007) Talanta 71:1976
123. Lerner RA, Benkovic SJ, Schultz PG (1991) Science 252:659
124. Kawanami Y, Yunoki T, Nakamura A, Fujii K, Umano K, Yamauchi H, Masuda K
(1999) J Mol Catal A 145:107
125. Sarhan A, Wulff G (1982) Makromol Chem 183:85
126. Andersson LI, Mosbach K (1990) J Chrom 516:313
127. Kugimiya A, Takeuchi T, Matsui J, Ikebukuro K, Yano K, Karube I (1996) Anal Lett
29:1099
128. Steinke JHG, Dunkin IR, Sherrington DC (1999) Trends Anal Chem 18:159
129. Spivak D, Shea KJ (1999) J Org Chem 64:4627
130. Hughes MP, Smith BD (1997) J Org Chem 62:4492
131. Fan E, Van Arman SA, Kincaid S, Hamilton AD (1993) J Am Chem Soc 115:369
132. Linton BR, Goodman MS, Fan E, Van Arman SA, Hamilton AD (2001) J Org Chem
66:7313
133. Hall AJ, Achilli L, Manesiotis P, Quaglia M, De Lorenzi E, Sellergren B (2003) J Org
Chem 68:9132
134. Urraca JL, Hall AJ, Moreno-Bondi MC, Sellergren B (2006) Angew Chem Int Edit
45:5158
135. Manesiotis P, Hall AJ, Emgenbroich M, Quaglia M, De Lorenzi E, Sellergren B (2004)
Chem Commun, p 2278
136. Lee SW, Ichinose I, Kunitake T (1998) Langmuir 14:2857
137. Kunitake T, Lee SW (2004) Anal Chim Acta 504:1
138. Lahav M, Kharitonov AB, Katz O, Kunitake T, Willner I (2001) Anal Chem 73:720
139. Lee SW, Yang DH, Kunitake T (2005) Sens Actuators B 104:35
140. Lahav M, Kharitonov AB, Willner I (2001) Chem Eur J 7:3992
141. Lee SW, Ichinose I, Kunitake T (2002) Chem Lett, p 678
142. Ichinose I, Kikuchi T, Lee SW, Kunitake T (2002) Chem Lett, p 104
143. Pogorelova SP, Kharitonov AB, Willner I, Sukenik CN, Pizem H, Bayer T (2004) Anal
Chim Acta 504:113
144. D’Souza SM, Alexander C, Carr SW, Waller AM, Whitcombe MJ, Vulfson EN (1999)
Nature 398:312
145. D’Souza SM, Alexander C, Whitcombe MJ, Waller AM, Vulfson EN (2001) Polym Int
50:429
146. Egan TJ, Rodgers AL, Siele T (2004) J Biol Inorg Chem 9:195
147. Caro E, Marce RM, Cormack PAG, Sherrington DC, Borrull F (2004) J Chrom A 1047:175
148. Albano DR, Sevilla F (2007) Sens Actuators B 121:129
149. Sineriz F, Ikeda Y, Petit E, Bultel L, Haupt K, Kovensky J, Papy-Garcia D (2007)
Tetrahedron 63:1857
150. Fujiwara I, Maeda M, Takagi M (2003) Anal Sci 19:617
151. Dong S, Sun Z, Lu Z (1988) Analyst 113:1525
152. Dong SJ, Che GL (1991) Talanta 38:111
153. Kamata K, Suzuki T, Kawai T, Iyoda T (1999) J Electroanal Chem 473:145
154. Kamata K, Kawai T, Iyoda T (2001) Langmuir 17:155
155. Hutchins RS, Bachas LG (1995) Anal Chem 67:1654
156. Lenihan JS, Ball JC, Gavalas VG, Lumpp JK, Hines J, Daunert S, Bachas LG (2007)
Anal Bioanal Chem 387:259
157. Schmidtchen FP (2005) Anion Sensing (Topics in Current Chemistry) 255:1

158. Katayev EA, Ustynyuk YA, Sessler JL (2006) Coord Chem Rev 250:3004
159. Sun BW, Yang ML, Li YZ, Chang WB (2003) Acta Chim Sin 61:878
160. Wu LQ, Li YZ (2004) Anal Chim Acta 517:145
161. Turner NW, Jeans CW, Brain KR, Allender CJ, Hlady V, Britt DW (2006) Biotech Prog
22:1474
162. Tanabe K, Takeuchi T, Matsui J, Ikebukuro K, Yano K, Karube I (1995) J Chem Soc
Chem Commun, p 2303
163. Kubo H, Yoshioka N, Takeuchi T (2005) Org Lett 7:359
164. Kempe M, Mosbach K (1994) J Chrom A 664:276
165. Caro E, Marce RM, Cormack PAG, Sherrington DC, Borrull F (2004) J Chrom B
813:137
166. Haginaka J, Sanbe H (2001) J Chrom A 913:141
167. Andersson LI (1996) Anal Chem 68:111
168. Oral E, Peppas NA (2006) J Biomed Mater Res A 78A:205
169. Seong H, Lee HB, Park K (2002) J Biomater Sci – Polym Edit 13:637
170. Asanuma H, Kakazu M, Shibata M, Hishiya T, Komiyama M (1997) Chem Commun,
p 1971
171. Hwang CC, Lee WC (2002) J Chrom A 962:69
172. Matsui J, Takeuchi T (1997) Anal Commun 34:199
173. Zander A, Findlay P, Renner T, Sellergren B, Swietlow A (1998) Anal Chem 70:3304
174. Sanbe H, Hosoya K, Haginaka J (2003) Anal Sci 19:715
Subject Index
Adenosine triphosphate (ATP) 96 Cages 182

Alkaline phosphatase 102 Calix[4]arenes 90
Aluminophosphates 214 –, receptors 52
Amides 2 Calix[4]diquinone receptor 77
Amidinothiourea 183 Calix[6]furan 28
Amidoferrocenylalkylthiol (AFAT)-gold Calix[n]pyrroles 26–28
64 cAMP 97, 103, 226
7-Amino-4-trifluorocumarin 111 Capsules 182
6-Aminopicolinic acid 11 Carbazole-based receptors 29
Ammonium-containing receptors 39 Carboxylates 232
Ampicillin 232 Carboxypeptidase A 230
Anionic templates 175, 207 Catechol 32
Anion-π interactions 127, 152 Catenanes 78, 186
Anthracene 13, 99 Cd(II) 85
Anthracenecarboxylic acids 237 CENOTE phosphorylation 97
Anticrown chemistry 130 Charged receptors 33
Aryl bisamide 77 Chemosensors, artificial 98
ATPase 96 Chloride-selective microsensor 240
Azadendritz 165 Cholapod anion receptors 17
Azaferrocenophane 71 Chondroitin sulfate 239
Azobenzene carboxylic acids 237 Cobalt (III) dithiocarbamate cryptands
60
Bidentate pyrazolyl-based ligands 183 Cobaltocenium/cobaltocene 48
Binaphthalene 15 Coenzyme A (CoA) 96
Biomimicry 207 Complexation 1
3,6-Bis(2 -pyridyl)-1,2-pyridazine (bppn) Coordination polymer, bowl-shaped
168 metallo-macrocycles 182
3,6-Bis(2-pyridyl)-1,2,4,5-tetrazine (bptz) Covalent bonds, reversible 201
192 Cryptates 81
1,2-Bis(4-nitrophenyl) urea 12 Crystal structures 1
1,1 -Bis(alkyl-N-amido)ferrocene, gold Crystallographic studies 152
electrodes 63 p-Cyanopyridinium crosslinker 241
Bis-bipyridinephenylphosphine oxide 87 Cyanuric acid (CNA) 147
Bis-cobaltocenium receptor 48 Cyclen-Cd complex,
Bis-copper(II) cryptates 81 7-amino-4-trifluorocumarin 111
Bis-phosphinylferrocene 77 Cyclic peptides 202
Bis-terpyridine Iridium(III) 76 Cyclobarbital 223
Boroxine (BOX) 147 Cyclotriveratrylene amides 51
Bowl-type structure 198 Cytokine receptor 117
250 Subject Index
Dendrimers, redox anion sensors 60 Hexapeptide receptors 11

Deoxythymidine 106 Host/anion “sandwich” 7
2,5-Diamidopyrrole skeleton 23 Host–guest systems 95, 127
1,8-Diamino-3,6-dichlorocarbazole 29 Hydrogel, supramolecular 119
Diazabicycloalkanes 129 Hydrogel fibers, hydrophobic micro-
5,5 -Dicarboxamido-dipyrrolylmethanes domains, phosphate derivatives 120
24 Hydrogen bond donors 2
2,5-Dicarboxamidopyrrole skeleton 21 Hydrogen bonding 1
Diindolylquinoxalines receptors 30 Hydroxy (OH) donors, neutral receptors
Diphenyl phosphate 227 32
Diphenylcarbonate hydrolysis 230 Hyper-phosphorylation 117
Diphenylurea 15
Dipicatriz 166 Imidazolium-based receptors 34
Dithiocarbamates, redox anion sensing Indium tin oxide (ITO) 228
59 Indole-based receptors 29
Dithiocyanuric acid, 2-ethyleneamine Indolocarbazoles sensors 29
derivatives 159 Interlocked species 186
DNA 96, 98, 225 Ion-sensitive field effect transistor (ISFET)
Dodecanethiol 88 device 236
Dynamic combinatorial chemistry (DCC) IP3 97, 102
176, 191 Iridium(III) polypyridine thiourea,
Dynamic combinatorial library 175 cyclometallated 76
Isophthalamide receptors 2
Electrochemical sensing 47 1-Isopropenyl-4-isopropyl-2-isocyanate
ERK2 (extracellular signal-regulated 234
kinase 2) 97
Eu(III) 86 Kemp’s triacidic imides 103
Excimers 108 Kinase-catalysed transfer 225
EYPC lipid bilayers 4
Lanthanide(III) complexes 86
Ferrocene 48
Ferrocene-boronic acid 51 Macrocycles 178
[3,3]Ferrocenophane 72 Macrocyclic amide receptors 6
Flavin adenine dinucleotide (FAD) 96 Metal complexes, anion sensing 46
Fluorescent sensing 95 Metal–ligand coordination bonds 192
Fluoride-ferrocenium 51 Metalla-cyclophanes 192
Metallocene optical anion sensors 68
Glucose-6-O-sulfate 239 Metallocene redox anion sensors 48
Guanidinium-based receptors 37, 102, Metallocene-Lewis acid anion receptors
225 53
Metallodithiocarbamate macrocyclic
H donor, aromatic, neutral receptors 21 receptors 59
Halides 2 Metallo-squares 193
–, π aromatic systems 132 Metallo-tetrahedron 196
Helicates 191 Metallo-triangle 196
Heparan sulfate 239 6-Methacrylamidohexanoic acid 237
Hexafluorobenzene 132 Methotrexate 235
Hexakis(pyridine-2-yl)-[1,3,5]triazine- Methyl orange (p-dimethylamino-
2,4,6-triamine 163 azobenzenesulfonic acid) 222
Hexanickel cage 183 Methylphosphonic acid (MPA) 228
Subject Index 251
Methylpurine 181 Phosphoprotein 95

Micelles 208 Photo-induced electron transfer (PET) 98
MIPs 207 Pinacolyl methylphosphonate (PMP),
Mitogen-activated protein kinase (MAPK) Soman 227
97 Platinum(II), methylpurine 181
Molecular interaction potential with Point-of-care devices 207
polarization (MIPp) 139 Poly(vinylimidazole) 215
Molecular recognition 95, 175 Polyamine, cyclic/acyclic 98
Molecularly imprinted polymers (MIPs) Polyammonium-based anion receptors 39
207, 209 Polyaza receptors 73
Multisite phosphorylation 117 Poly-guanidinium 102
Myoinositol-1,4,5-triphosphate 97, 102 Polynucleotides 208
Polypyridyl anion receptors 56
Naphthalene 15 Polypyrrole 239
Naphthalene sulfonates 238 Poly-urea 178
Nerve agent detection, MIPs 227 Prodigiosin mimics 23
Neutral receptors 2 Protein phosphorylation 225
Nicotinamide adenine dinucleotide (NAD) Protein surfaces, phosphorylated 112
96 Pseudopeptidic macrocycle 178
p-Nitrophenyl acetate hydrolysis 229 Pseudorotaxanes 66, 186
p-Nitrophenylphosphonate 215 Pt(II) 90
Nitrophenylurea 82 Pyridine-2,6-dicarboxamide “caps” 7
Nucleoside phosphates (nucleotides) 96 Pyridinium-based receptors 34
Pyromellitamide 4
Octadecylsiloxane (ODS) 228 Pyrophosphate 99
Octamethylcalix[4]pyrrole 26 Pyrrole-based receptors 21
Oligo-(p-phenylene)-N,N-naphthalene-
diimide (O-NDI) 169 Quinoxaline phenanthroline 59
Online sensing 207
Optical anion sensors 68 Read-out 207
Organometallic receptors 45 Redox anion-sensing systems, surface
Organophosphorous pesticides/herbicides confined 63
227 Resorcin arene 104
Resorcinol 32
PDGFR-β 117 Rhenium(I)tricarbonylchloride 77
Pentafluorophenyl receptor 162 RNA 96, 225
Pentamethyl-amidoferrocene dendrimers Rotaxanes 67, 78, 186
63 Ruthenium(II)bipyridylcalix[4]diquinone
Perfluoroaromatic compounds 135 receptor 73
Pesticides/herbicides, organophosphorous Ruthenium(II) polypyridyl 72
227 Ruthenium-(II)-tris-(2,2 -bipyridiyl),
PET sensors 12 Ru(bpy)3 101
Phenylbutyrate 27
Phosphate 95 Sacrificial spacer approach 214
Phosphate anion recognition, coordination Sapphyrin 21
chemistry 104 Self-assembly 175, 207
Phosphate derivatives 225 Semi-wet sensor array, phosphate
–, ratiometric detection 109 derivatives, high throughput sensing
Phosphate recognition, electrostatic 119
interactions/hydrogen bonds 98 Sialic acid 232
252 Subject Index
Solid-phase extraction (SPE), MIPs 227 Transition metal dithiocarbamates, redox

Soman 227 anion sensing, receptors 59
Spermine/spermidine 98 Transition metal polypyridyl anion
Squaramide-containing macrocyclic receptors 56
receptor, fluorescent 40 Transition metal-based receptors,
Squaramido-based receptors 5 redox-active 47
Staphylococcus nuclease 102 Transition metals, anion-binding groups
Steroids 33 79
Sulfate derivatives 238 Tren skeleton 2
Sulfate in water 40 Triazine 133
Sulfonamide receptors 162 Tricarboxylate anions 37
Sulfonamide-based receptors 2 Trichloroacetic acids 232
Supramolecular chemistry 1, 127 Trifluoro-1,3,5-triazine 133
Surface confined systems, optical anion Tris-[(2-pyridyl)methyl]amine 82
sensing 88 Tris-amides 2
Surface imprinting 213 Tris-sulfonamides 2
Symmetry-adapted perturbation theory Tweezers 149
(SAPT) 143
syn–syn conformation 3 UDP-Gal 97, 111
Synthetic receptors 207 Urea 225, 234
Urea-based receptors 12
Tb(III) 86 Urea-crown ether functionalised ferrocene
Templating effects 177 53
Terephthalate 178
Tetracyanobenzene (TCB) 151, 156 Vesicle membranes, transport of chloride
Tetrakis(imidazolium) macrocyclic ions 18
receptors 35 Vinylanthracene 233
Tetrakis-p-phenylene[68]crown-20 186
Tetralactam macrocycles 188 Zeolites 208, 214
Thiocyanuric acid, 2-ethyleneamine Zinc(II)-dipicolylamine (Zn-Dpa) type
derivatives 159 artificial chemosensors 106
Thiourea 225 Zn-cyclen type artificial chemosensors
Thiourea-based receptors 12 105

Recognition of Anions 13ThePoet05

Загружено:

Сведения о документе

Исходное описание:

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Recognition of Anions 13ThePoet05

Загружено:

Авторское право:

Доступные форматы

http://www.nd-warez.

Contemporary Metal Boron Chemistry I Non-Covalent Multi-Porphyrin Assemblies

Organometallic and Coordination Chemistry Semiconductor Nanocrystals

Halogen Bonding Magnetic Functions Beyond

High Energy Density Materials Intermolecular Forces and Clusters II

Ferro- and Antiferroelectricity Intermolecular Forces and Clusters I

Single-Molecule Magnets Principles and Applications

Volume Editor: Ramón Vilar

Springer WWW home page: springer.com

ISBN 978-3-540-79091-4 e-ISBN 978-3-540-79092-1

Structure and Bonding ISSN 0081-5993

Library of Congress Control Number: 2008923960

You will ﬁnd information about the

A large number of biologically relevant species are negatively charged, there-

London, February 2008 Ramón Vilar

An Introduction to Anion Receptors Based on Organic Frameworks

Metal-Based Anion Receptor Systems

Recent Progress of Phosphate Derivatives Recognition

Anions and π-Aromatic Systems. Do They Interact Attractively?

Anion Templates in Synthesis and Dynamic Combinatorial Libraries

Molecularly Imprinted Polymers Using Anions as Templates

Subject Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249

An Introduction to Anion Receptors

3 Aromatic NH Donor Containing Neutral Receptors . . . . . . . . . . . . . 21

4 Hydroxy (OH) Donors in Neutral Receptors . . . . . . . . . . . . . . . . . 32

Keywords Anion recognition · Complexation · Crystal structures · Hydrogen bonding ·

The development of new anion receptors based on organic frameworks con-

systems functioning as effective and selective anion receptors. This review

Fig. 1 X-ray crystal structure of a bromide complex of 2a

Almost contemporaneously, B.D. Smith and co-workers reported the use of

Thordarson et al. have studied the aggregation of pyromellitamide 7 and

Fig. 2 X-ray crystal structures of 4 (left) and 6 (right)

vealed that compound 7 bound a range of anions with 2 : 1 anion/receptor

Prohens and co-workers have synthesized compounds 8a and 8b, simple

A.P. Davis and co-workers have designed a number of acyclic receptors

Macrocyclic receptors often possess a higher degree of selectivity than acyclic

Chmielewski and Jurczak have reported a series of extended tetra-amide

12a to the 20-membered macrocycle 12b results in a 30-fold increase in the

2,6-dicarboxamide macrocycles (12a–12c), a result rationalized in terms of

Scheme 1 Intramolecular hydrogen-bonding vs. anion binding in compound 13a

Bowman-James and co-workers have designed polyamide cryptand-type

The authors have also synthesized 16, a tricyclic cryptand-like receptor,

Kubik and co-workers have developed a series of highly effective anion

The formation of the 2 : 1 complexes observed in 17 led the authors to

Stability constants were determined for compound 19 by UV-Vis spec-

Yoon and co-workers have reported a series of mono- and bis-func-

UV-Vis spectroscopy titration experiments in acetonitrile solution were

in DMSO-d6 /0.5% water resulted in high stability constants for acetate

Fig. 6 X-ray crystal structure of a carbonate complex of 32 reproduced by permission of

In 2000 Lee and Hong synthesized tris-thiourea macrocycles 33a and

Gale and co-workers have developed a number of receptors based on

Fig. 7 X-ray crystal structure of 37b benzoate complex

Gale, Smith and co-workers have developed prodigiosin mimics based on

hanced selectivity for both dihydrogen phosphate and benzoate compared to

Elucidation of the solution phase anion complexation properties of recep-

The meso-substituted derivatives showed lower afﬁnities for anion com-

(compounds 45 and 46) [58]. UV-Vis spectroscopic titrations in acetoni-

In 1996, Sessler and co-workers reported the anion complexation proper-

One approach to dramatically increase the afﬁnity of calixpyrroles for

Recently, Lee and co-workers have shown that a binol-strapped calix[4]pyr-

that receptor (S)-49 shows selectivity for (S)-2-phenylbutyrate over (R)-

Association constants were determined by Cram extraction methods,