Complexation and electrochemical sensing of anions by amide-substituted ferrocenyl ligands

Article abstract of DOI:10.1016/S0022-328X(01)00935-4

New amide-containing ferrocenyl ligands, L_1-5, were prepared and the voltammetric and ¹H NMR investigations of anion binding were carried out in organic media. The electrochemical recognition ability of L_1-5 towards F^-, HSO₄^-, H₂PO₄^- and ATP^2- is based on the synergy between H-bonding to amide protons in anion complexation to reduced, neutral ligands and ion-pairing interactions developed with the oxidized, cationic form of the ligands. The strength of the anion-ligand interactions depends on the number of ferrocene centers and amide groups in the receptor, and on the accessibility of the binding sites. Clear two-wave cyclic voltammetry features allowed the amperometric titration of H₂PO₄^- and ATP^2- by ligands L₄ and L₅ built from a cyclotriveratrylene structural unit, and containing a combination of three ferrocene centers with three (L₄) or six (L₅) amide groups.

Full text of DOI:10.1016/S0022-328X(01)00935-4

Journal of Organometallic Chemistry 637–639 (2001) 356–363
www.elsevier.com/locate/jorganchem
Complexation and electrochemical sensing of anions by
amide-substituted ferrocenyl ligands
Olivier Reynes a, Frederic Maillard a, Jean-Claude Moutet a,*, Guy Royal a,
Eric Saint-Aman a, Gabriella Stanciu a, Jean-Pierre Dutasta b, Isabelle Gosse b,
Jean-Christophe Mulatier b
a Laboratoire d’Electrochimie Organique et de Photochimie Re´dox, UMR CNRS 5630, Uni6ersite´ Joseph Fourier Grenoble I, BP 53,
38041 Grenoble Cedex 9, France
b
´
Laboratoire de Ste´re´ochimie et Interactions Mole´culaires, UMR CNRS 5532, Ecole Normale Supe´rieure de Lyon, 46 Alle´e d’Italie,
F-69364 Lyon Cedex 07, France
Received 10 January 2001; received in revised form 15 March 2001; accepted 19 March 2001
Abstract
New amide-containing ferrocenyl ligands, L1–5, were prepared and the voltammetric and 1H NMR investigations of anion
binding were carried out in organic media. The electrochemical recognition ability of L1–5 towards F−, HSO4−, H2PO4− and
ATP2− is based on the synergy between H-bonding to amide protons in anion complexation to reduced, neutral ligands and
ion-pairing interactions developed with the oxidized, cationic form of the ligands. The strength of the anion–ligand interactions
depends on the number of ferrocene centers and amide groups in the receptor, and on the accessibility of the binding sites. Clear
two-wave cyclic voltammetry features allowed the amperometric titration of H2PO−4 and ATP2− by ligands L4 and L5 built from
a cyclotriveratrylene structural unit, and containing a combination of three ferrocene centers with three (L4) or six (L5) amide
groups. © 2001 Elsevier Science B.V. All rights reserved.
Keywords: Ferrocene; Redox-active ligands; Anion complexation; Electrochemical recognition; Cyclotriveratrylene
containing the ferrocene entity are of particular interest,
as they offer the possibility to modulate the anion–re-
ceptor interactions according to the redox state; e.g.
electrostatic interactions can be switched on by the
oxidation of ferrocene to ferricinium. The driving force
responsible for the anion-binding ability of these recep-
tors in their neutral ferrocene form is usually the H-
bonding interactions between the guest anion and
H-bond donor groups appended to the receptor, such
as amide [1,2,8] or pyrrole [9] groups. The presence of
additional H-bond acceptor groups like amine, pyridine
and bipyridine [10–13] in the binding sites enhances the
interaction between the receptor and dihydrogenphos-
phate or hydrogensulphate anions. Another strategy
based on hard acid–hard base interaction has been
developed for sensing fluoride anion with boronic
derivatives of ferrocene [14]. Taking advantage of the
dendritric effect, ferrocenyl dendrimers exhibiting an-
ionic electrochemical recognition properties have also
been reported [15,16]. The dendritic effect found in the
1. Introduction
The design of molecular receptors having the ability
to selectively bind and sense cationic, anionic, or even
neutral guests via a macroscopic physical response is an
area of intense activity [1,2]. From the first reports in
the late 1970s on the synthesis and complexation of
metallocene-containing macrocyclic ligands [3,4] capa-
ble of electrochemical detection of a metal cation [5],
ferrocene has largely proved to be a simple and remark-
ably robust building block for the construction of re-
dox-responsive receptors [2]. In this context, the design
of reagents for anionic sensing is rapidly gaining inter-
est [6,7], stimulated by the importance of small inor-
ganic anions in biochemical processes and from an
environmental point of view. Electrochemical sensors
* Corresponding author. Tel.: +33-4-76514-600; fax: +33-4-
76514-267.
E-mail address: jean-claude.moutet@ujf-grenoble.fr (J.-C. Moutet).
0022-328X/01/$ - see front matter © 2001 Elsevier Science B.V. All rights reserved.
PII: S0022-328X(01)00935-4

O. Reynes et al. / Journal of Organometallic Chemistry 637–639 (2001) 356–363
357
electrochemical recognition of small oxo-anions has
been demonstrated using poly(amidoferrocene) and
poly(aminoferrocinium). Sensing properties of the latter
are enhanced through the synergy between X−···H−N
hydrogen bonding, electrostatic attraction in the oxi-
dized state and shape selectivity [15]. Silicon-based fer-
rocenyl dendrimers containing SiꢀNH group are able to
sense oxo-anions potentiometrically and amperometri-
cally in homogeneous solution, and at the elec-
trode ꢀ solution interface after the immobilization of
dendrimer onto the electrode surface [16].
In this paper, we present a series of new ferrocenyl
ionophores, L1–5 (Chart 1), designed for anionic sens-
ing. They are based on the combination of ferrocene as
the signaling unit and secondary amide groups as H-
bond donors. L1–5 differ by: (i) the number of ferrocene
centers; (ii) the number of amide groups; and (iii) in the
preorganization of the binding site.
at least 4 days before use. TBAP (Fluka) was dried
under vacuum at 80 °C for 3 days. Tetraethylammo-
nium fluoride, tetra-n-butylammonium hydrogensul-
phate and dihydrogenphosphate were of the highest
purity commercially available and were used without
further purification. The di(tetra-n-butylammonium)
salt of adenosine-5%-triphosphate (ATP2−) was ob-
tained from its corresponding disodium salt by ion
exchange on Amberlite IRC50 in the n-Bu4N+ form.
1H NMR experiments were conducted at 22 °C on a
Brucker AC250 spectrometer using the solvent deu-
terium signal as an internal reference. UV–vis spectra
were obtained using a Cary 1 spectrophotometer.
2.2. Redox ligands
Ligands L1–5 were synthesized by the reaction of
1-chlorocarbonylferrocene [17] or 1,1%-bis-chlorocar-
bonylferrocene [18] with the desired primary amine, i.e.
4-methoxyaniline (L1, L2), 1,2-bis(2-aminoethoxy)-
benzene (L3), (9)-3,8,13-triamino-2,7,12-trimethoxy-
10,15-dihydro-5H-tribenzo[a,d,g]cyclononene [19] (L4)
and (9)-2,7,12-tris(2-aminoethyl carbamoylmethoxy)-
3,8,13 - trimethoxy - 10,15 - dihydro - 5H - tribenzo[a,d,g]-
cyclononene [20] (L5), in dried CH2Cl2 containing
freshly distilled triethylamine. For L1–3 the stoichiomet-
ric mixture was stirred overnight at r.t. under an argon
atmosphere, then filtered and evaporated to dryness in
vacuo. For the synthesis of L3, solutions of 1,1%-bis-
chlorocarbonylferrocene (0.5 mmol in 50 ml of CH2Cl2)
and 1,2-bis(2-aminoethoxy)benzene (0.5 mmol in 50 ml
of CH2Cl2) were added dropwise simultaneously over 4
h in 200 ml of CH2Cl2 containing 1 mmol of triethyl-
amine. The mixture was stirred for 24 h. The crude
product was extracted with CH2Cl2, the organic phase
washed with H2O, and the solvent removed under
vacuum. The resulting orange–brown solid was purified
2. Experimental
2.1. Instrumentation, sol6ents and reagents
The electrochemical equipment has been described in
Ref. [13]. Experiments were conducted in a three-elec-
trode cell under an argon atmosphere and at room
temperature (r.t.). The working electrode was a plat-
inum or a carbon disc (5 and 3 mm in diameter,
respectively) polished with 1 mm diamond paste. The
Ag/10 mM AgNO3 in CH3CN+0.1 M tetra-n-buty-
lammonium perchlorate (TBAP) system was used as a
reference electrode. The redox potential of the regular
ferrocene/ferricinium (Fc/Fc+) couple is 0.07 V under
our experimental conditions. Acetonitrile (Rathburn,
HPLC grade S) was used as received. Dichloromethane
was dried over neutral aluminum oxide (activity I) for
Chart 1. Redox ligands used in this study.

358
O. Reynes et al. / Journal of Organometallic Chemistry 637–639 (2001) 356–363
by thin layer chromatography on neutral alumina
plates eluted with CH2Cl2, to yield L1 (75%), L2 (38%)
and L3 (55%) as orange solids. Cyclotriveratrylene-
based hosts L4,5 were prepared from their correspond-
ing amine precursors, four equivalents of 1-chloro
carbonylferrocene and an excess of triethylamine in
CH2Cl2. The mixture was stirred overnight at r.t. under
an argon atmosphere, and the solvent was removed
under vacuum. Crude L4 was dissolved in CH2Cl2 and
the organic phase was washed with water, dried over
Na2SO4 and the solvent removed under vacuum to give
L4 almost quantitatively, which was used without fur-
ther purification. Crude L5 was dissolved in a minimum
amount of CH2Cl2. Diethylether was added until a
precipitate was formed. After one night stirring the
solid orange compound was recovered by filtration to
give L5 (77%).
1H NMR titration curves were obtained by monitoring
variations of HꢀN and HꢀCp chemical shifts in 5–10
mM solutions of redox ligands in CD3CN or CD2Cl2
on the addition of anions in the 0.1–10 molar equiva-
lent range. Association constants K between the re-
duced neutral ligands
L
and anions A− were
determined by considering the formation of 1:1 LA−
complexes, according to:
L+A− ? LA−
(1)
with K=[LA−]/([L][A−]).
Assuming a fast exchange at the NMR time scale, the
1:1 binding isotherm can be expressed as Eq. (2)
Dl=([L]t+[A−]t+K−1
−(([L]t+[A−]t+K−1)2−4[L]t[A−]t)1/2
)
Dlmax/(2[L]t)
(2)
1
L1. FABMS; m/z (positive mode): 435 [M+H]+; H
NMR (CDCl3,): l 3.80 (s, 3H, CH3ꢀO), 4.24 (s, 5H,
HꢀCp), 4.40 (m, 2H, HbꢀCp), 4.75 (m, 2H, HaꢀCp),
6.89 (d, 2H, HꢀPh), 7.29 (s, 1H, NH), 7.49 (d, 2H,
HꢀPh); UV–vis: u (nm) (m, M−1 cm−1) (d–d band;
CH3CN): 444 (2020).
where [ ]t denotes the total concentration, and Dlmax the
maximal downfield shift of the resonance of the consid-
ered proton. Using a non-linear regression method [21],
the titration curves Dl versus [A−]t were fitted to the
1:1 binding isotherm giving consistent results in all
cases.
1
L2. FABMS; m/z (positive mode): 484 [M+H]+; H
NMR (CDCl3): l 3.76 (s, 6H, CH3ꢀO), 4.43 (m, 4H,
HbꢀCp), 4.62 (m, 4H, HaꢀCp), 6.88 (d, 4H, HꢀPh), 7.67
(d, 4H, HꢀPh), 8.69 (s, 2H, NH); UV–vis: u (nm) (m,
3. Results and discussion
M
−1 cm−1) (d–d band; CH3CN): 433 (2030).
1
L3. FABMS; m/z (positive mode): 435 [M+H]+; H
3.1. Association of neutral receptors with F−, H2PO−4 ,
NMR (CDCl3): l 3.56–3.62 (m, 4H, ꢀNHꢀCH2), 4.29
(t, 4H, ꢀCH2ꢀOꢀ), 4.48 (m, 4H HbꢀCp), 4.53 (m, 4H,
HaꢀCp), 6.80 (t, 2H, NH), 7.08 (m, 4H, HꢀPhꢀ);
UV–vis: u (nm) (m, M−1 cm−1) (d–d band; CH3CN):
355 (1960).
ATP2− and HSO−4
The association constants K were determined at 295
K in millimolar solutions of ligands in CD2Cl2 or
CD3CN, using a standard H NMR titration and mon-
1
L4. FABMS; m/z (positive mode): 1042 [M+H]+;
1H NMR (CDCl3): l 3.67 (d, 3H, PhꢀCH2ꢀPhꢀ), 3.95
(s, 9H, ꢀOꢀCH3), 4.18 (s, 15H, HꢀCp), 4.35 (m, 6H,
HꢀCp), 4.71 (m, 6H, HꢀCp), 4.81 (d, 3H,
PhꢀCH2ꢀPhꢀ), 7.03 (s, 3H, HꢀPh), 8.01 (s, 3H, HꢀPh),
8.50 (s, 3H, NH); UV–vis: u (nm) (m, M−1 cm−1) (d–d
band; CH3CN): 444 (12 550).
itoring the shifts of amide or cyclopentadienyl proton
resonances on the addition of increasing amounts of a
given anion (see Section 2). Results were analyzed using
a non-linear regression method, and by assuming a 1:1
binding isotherm (Table 1). The 1:1 binding isotherm
gave consistent fits for all the experimental data. The
formation of complexes with higher stoichiometry, e.g.
2:1 anion–ligand ratio, can be reasonably excluded
because of the establishment of strong repulsive electro-
static interactions between the two anions in close
proximity.
L5. FABMS; m/z (positive mode): 1345 [M+H]+;
1H NMR (CDCl3): l 3.48 (m, 15H, NꢀCH2CH2ꢀN and
PhꢀCH2ꢀPh), 3.79 (s, 9H, ꢀOꢀCH3), 4.13 (s, 15H,
HꢀCp), 4.29 (m, 6H, HꢀCp), 4.48 (m, 6H,
OꢀCH2ꢀCOꢀ), 4.63 (m, 9H, HꢀCp and PhꢀCH2ꢀPh),
6.60 (br t, 3H, NH), 6.77 (s, 3H, HꢀPh), 6.82 (s, 3H,
HꢀPh), 7.36 (br t, 3H, NH); UV–vis: u (nm) (m,
Addition of anions to solutions of L1, L2, L3 and L5
1
caused significant perturbations in their H NMR spec-
tra. A typical example (L5+ATP2−) is shown in Fig.
1. However, no changes were observed in the L4 spec-
trum when either fluoride or dihydrogenphosphate an-
ion was used. In contrast, the ꢀCOꢀNH and HaꢀCp
resonances in L1, L2, L3 and L5 are shifted downfield,
showing the formation of complexes between these
neutral ligands and the surveyed anions.
M
−1 cm−1) (d–d band; CH3CN): 439 (7580).
2.3. Titration procedure and determination of
association constants
1
Electrochemical and H NMR titrations were carried
out by adding small volumes of concentrated stock
solutions of the considered anion to ligand solutions.
It is noteworthy that the strongest perturbations were
observed for the NꢀH resonances on complexation with

O. Reynes et al. / Journal of Organometallic Chemistry 637–639 (2001) 356–363
359
Table 1
Association constants (K, M−1) between L1–5 and the surveyed anionsa
b
L1
L2
L3
L4
L5
d
d
f
F− c
20 (0.64)
230 (0.48)
110 (0.16)
69 (0.48)
89 (1.80)
H2PO−4 e
ATP2− e
HSO−4 e
40 (2.00)
65 (2.98)
25 (1.56)
f
f
f
63 (2.13)
f
f
f
f
10 (0.52)
a Determined from 1H NMR data (see the experimental section), at 295 K in CD2Cl2 unless otherwise noted; in parentheses is the maximum
downfield shift (Dlmax, ppm) for H-amide or HaꢀCp resonances.
b In CD3CN.
c Dlmax for the HaꢀCp proton.
d Too weak to be determined (no significant change in the NMR spectrum).
e Dlmax for the H-amide proton.
f Not studied.
H2PO−4 . For example, the amide protons of L1 and L5
are shifted (Dlmax=2.0 and 1.8 ppm) much more than
the cyclopentadienyl protons (Dlmax=0.23 and B0.01
ppm, respectively, for the HaꢀCp protons). This shows
that the complexation of the dihydrogenphosphate an-
ion proceeds mainly via H-bonding interactions with
the amide protons. In addition, the two resonances for
the two non-equivalent amide protons in L5 are shifted
downfield by the same order of magnitude in the pres-
ence of H2PO−4 . This observation indicates that they
are both involved in the complexation process. A simi-
point is further highlighted by the behavior of L4 and
L5 built on the rigid cyclotriveratrylene structural unit.
Ligand L4 contains three amidoferrocene moieties di-
rectly bound to the aromatic rings, giving a very rigid
structure, which does not allow a favorable organiza-
tion of the three NH groups. Consequently, L4 does not
present any significant complexation ability in its re-
duced form towards all the surveyed anions. In con-
trast, fluoride and dihydrogenphosphate anions interact
with L5, in which the length of the chains linking the
ferrocene groups to the cyclotriveratrylene unit was
increased by the introduction of additional amide
groups, favoring the spatial organization of the binding
sites.
lar association constant was found for L5+ATP2−
but Dlmax for the NꢀH resonance was higher than with
H2PO−4 , because of the dianionic form of ATP2−
,
.
In contrast, on complexation of L1–3 with F− the
strongest downfield shifts were observed for the HaꢀCp
resonances. For example, with L3 Dlmax was 0.16 ppm
for the a-cyclopentadienyl protons, but was less than
0.01 ppm for the amide protons. Obviously complexa-
tion of L1–3 with F− involves significant interaction
between the hard base F− and the partial positive
charge on the iron(II) center in the ferrocene unit, in
addition to a weak N−H···F− hydrogen bonding.
Complexation of L5 with F− follows the same binding
mode, but the amide–fluoride interaction appeared
much stronger (Dlmax=5.51 ppm for the amide proton
signal) than the FcꢀF− interaction (Dlmax=0.48 ppm
for the HaꢀCp protons).
From all these observations, it can be concluded that
the strength of the interactions between these neutral
ligands and the surveyed anions depends on the number
of binding sites contained in the ligands, and on the
steric constraints. From L1 to L2, and L4 to L5 the
number of amide groups increases, leading to an en-
hancement of the hydrogen bonding interactions. How-
ever, complexation of F− and H2PO−4 is significantly
weaker with L3 than with the less rigid ligand L2, both
of which contain a 1,1%-bis(amidoferrocene) unit. Acces-
sibility to the binding sites is thus a limiting factor for
the strength of the anion–receptor interaction. This
Fig. 1. 1H NMR spectra of L5 in CD2Cl2. (A) Free L5; (B) ATP2−
L5=1; (C) amide proton shift vs. ATP2−/L5.
/

360
O. Reynes et al. / Journal of Organometallic Chemistry 637–639 (2001) 356–363
values of the complexes formed between L2 and chlo-
ride or carboxylate anions were markedly higher in
acetone than in dichloromethane [22]. This was at-
tributed to a competition between the solvent and the
receptor for binding and solvation of these anions.
3.2. Electrochemical recognition of anions
The electrochemical behavior of the ferrocene deriva-
tives L1–5 has been investigated by cyclic voltammetry
(CV) in CH3CN and CH2Cl2 containing 0.1 M tetra-n-
butylammonium perchlorate (TBAP). The CV curves
for L1–5 exhibit the regular wave corresponding to the
reversible ferrocene/ferricinium (Fc/Fc+) redox couple
(see, e.g. curve 1 Figs. 2 and 3). Compared with the
simple unsubstituted ferrocene, E1/2s for L1–5 are
shifted towards more positive potentials (Table 2) ow-
ing to the substitution of the Cp rings with electron-
withdrawing carboxamide groups. It is noteworthy that
oxidation of the three Fc centers in L4 or L5 occur at
the same potential in each ligand, leading to the obser-
vation of a single three-electron oxidation wave. This
suggests that the three ferrocene groups are not elec-
tronically coupled in L4 and L5. Moreover, owing to
different solvation effects in CH3CN and CH2Cl2 small
shifts in E1/2s ranging from −40 mV (L4) to +80 mV
(L2) have been observed depending on the solvent used.
The CV behavior of L1–5 is changed in the presence
of F−, HSO4−, H2PO4− and ATP2−, according to the
receptor–anion couple considered. From a general
point of view, two different types of CV behavior for
the Fc center in L1–5 have been observed on the addi-
tion of increasing amounts of a given anion: (i) a
gradual negative shift in the Fc/Fc+ redox potential
(one-wave behavior); or (ii) the growth at a less positive
potential of a new redox wave (two-wave behavior) at
the expense of the original wave for the free ligand.
Negative shifts in the E1/2 values on the addition of a
guest anion are owing to ion-pairing interactions be-
tween the anion and the oxidized, positively charged
ligand leading to the stabilization of ferricinium moi-
eties in L1–5. The two-wave behavior is linked to a very
strong increase in the anion–ligand interaction follow-
ing the oxidation of the ligand. This point will be
discussed further. During redox cycling, ion-pairing
associations are accompanied by adsorption phenom-
ena owing to the weak solubility of the ion pairs in
CH3CN and CH2Cl2 electrolyte. This is revealed in the
CV curves by a loss of reversibility of the Fc/Fc+ redox
wave, accompanied by the rise of a broad shoulder at
the negative foot of the oxidation peak, or by the
appearance of a sharp negative stripping peak corre-
sponding to the electroreductive desorption of ion pairs
adsorbed on the electrode surface in the oxidation step.
Table 2 summarizes the potential shifts measured in
the presence of one molar equivalent of anions. Ac-
Fig. 2. CVs in CH3CN+TBAP 0.1 M at a Pt disc (5 mm in diameter)
electrode. (A) Curve 1: free L3; curve 2: F−/L3=5; curve 3: F−/L3=
10. (B) Curve 1: free L4; curve 2: F−/L4=1.5; curve 3: F−/L4=2.
Scan rate: 0.1 V s−1
.
Fig. 3. (A,B) CVs in CH3CN+TBAP 0.1 M at w=0.1 V s−1. (A)
Curve 1: free L4; curve 2: ATP2−/L4=0.6; curve 3: ATP2−/L4=1;
Pt disc (5 mm in diameter) electrode. (B) Curve 1: free L5; curve 2:
H2PO24−/L5=1; curve 3: H2PO4−/L5=2.2; carbon disc (3 mm in
diameter) electrode. (C,D) Oxidation peak current (Icpa) vs. anion L
ratio for the complexed ligand: (C) L4+ATP2−; (D) L5+H2PO42−
.
Further, the same association constant value was
found for L5+H2PO−4 in CD3CN and in CD2Cl2 (90
M
−1), suggesting that the polarity of the solvent does
not markedly affect the strength of the H-bond. It
should be noted however that the stability constant

O. Reynes et al. / Journal of Organometallic Chemistry 637–639 (2001) 356–363
361
cording to the electrochemical features, i.e. a one- or
two-wave CV behavior, potential shifts correspond ei-
ther to the difference between positive peak potentials
(DEpa), or half-wave potentials (DE1/2) for free and
complexed ligands.
higher F−/L ratios (Fig. 2B, curve 3). This stripping
peak is owing to the reduction and desorption of ion
pairs that precipitate onto the electrode surface during
the positive scan. Obviously L +F− species are more
3+
1–3
soluble than those formed with L+1–3. However, the CV
wave becomes more and more irreversible in the pres-
ence of a large excess of fluoride anions, which favors
the formation of weakly soluble L+F− complex spe-
cies and inhibits the dissolution of the ion-pair coating
on reduction. On the whole, the CV curves for L4 and
L5+F− present a ill-behaved two-wave behavior, with
poorly separated ‘free’ and ‘complexed’ positive peaks
and adsorption phenomena. These features preclude
any accurate amperometric sensing of F− with L4 and
L5.
3.2.1. Electrochemical response to F−
The electrochemical response for L1–5 in CH3CN
electrolyte is affected by the addition of increasing
amounts of fluoride anion. Typical CV curves for L3
and L4 are presented in Fig. 2.
With L1–3, a shoulder is seen at the negative foot of
the oxidation peak, along with a decrease in the re-
versibility of the original Fc/Fc+ wave and with a
significant increase in the positive peak current (Fig.
2A, curve 2). The electrochemical response becomes
fully irreversible and the shoulder reaches full develop-
ment when more than two molar equivalents of F−
have been added (Fig. 2A, curve 3). It should be noted
that the original oxidation peak is always seen in the
CV curves, even in the presence of a large excess of
fluoride anion. This behavior is characteristic of an EC
mechanism with product adsorption [23], suggesting
that oxidized, positively charged L+1–3 ligands form with
F− insoluble ion pairs that remain strongly adsorbed
onto the electrode surface and cannot be reduced in the
reverse scan. Thus, the absence of reduction peak in the
presence of an excess of fluoride anions can be ex-
3.2.2. Electrochemical response to H2PO−4 and ATP2−
Ion-pairing interactions and adsorption phenomena
also mainly govern the electrochemical response in the
CH3CN electrolyte of L1–5 in the presence of dihydro-
genphosphate anions. For L1–3 the CV behavior is close
to that observed with F−, and is characterized by the
growth of a shoulder at the negative foot of the oxida-
tion peak and by a loss of reversibility in the Fc/Fc+
redox wave. The larger potential shifts were obtained
with L1 and L2 (Table 2), and maximum perturbations
were observed at anion–ligand molar ratios greater
than 2 (L=L2), 3 (L=L1) and 5 (L=L3). These
observations show that the interaction between H2PO4−
and the rigid L3 ligand is lower than with L1 and L2.
plained by the very poor solubility of both the L+
+
F− ion pair and the L+F− complex.
The electrochemical response of L4 (Fig. 2B) and L5
to F− is different from that for L1–3, but can also be
explained in terms of ion-pairing association between
the ferricinium cation and the fluoride anion. The addi-
tion of a slight excess of F− (e.g. 1.5 molar equivalent)
induces the rise of a new negative peak at a less positive
potential than the original reduction wave (Fig. 2B,
curve 2). This reduction peak becomes sharper at
The same trend has been found with L1–3 and ATP2−
.
The CV curves seem to present a two-wave behavior,
especially in the case of L1 and L3. However, oxidation
peaks appeared broad and ill defined, which did not
allow the determination of E1/2s for the complexed
ligands.
In contrast, a clear two-wave behavior was obtained
when considering the L4,5+H2PO−4 or ATP2− systems.
Table 2
Electrochemical dataa for free L1–5, and in the presence of one molar equivalent of guest anions
Free ligand
E1/2 (V)
F−
H2PO−4
DE (V)
CH3CN
ATP2−
DE (V) c
CH3CN
HSO−4
DE (V) b
CH3CN
DE (V) b
CH3CN
−0.005
CH3CN
CH2Cl2
CH2Cl2
CH2Cl2
CH2Cl2
CH2Cl2
L1
L2
L3
L4
L5
0.28
0.45
0.45
0.33
0.28
0.35
0.53
0.50
−0.22
−0.09
−0.23
−0.06
−0.09
−0.31
−0.21 b
−0.21 b
−0.01 b
−0.17 b
−0.15
−0.14
−0.12
−0.16
−0.16
−0.16
−0.01
−0.29
−0.09 b
−0.26
−0.005
−0.02
d
d
d
e
d
e
c
e
e
0.29
−0.14
−0.17
−0.23 c
−0.23
−0.02
d
d
c
d
d
d
a Versus Ag–AgNO3 10 mM+TBAP 0.1 M in CH3CN at w=0.1 V s−1; A−=F−, H2PO4−, ATP2−, HSO−4
b DE=Epa ([A−]"0)−Epa([A−]=0); one-wave behavior.
c DE=E1/2 ([A−]"0)−E1/2([A−]=0); two-wave behavior.
d Not studied.
.
e No significant variation in the electrochemical response.

362
O. Reynes et al. / Journal of Organometallic Chemistry 637–639 (2001) 356–363
Typical CV curves are shown in Fig. 3A and B. A
well-behaved redox peak system grows at E1/2=0.19 V
for L4+H2PO−4 , 0.17 V for L4+ATP2−, 0.11 V for
L5+H2PO−4 , and 0.12 V for L5+ATP2− (Table 2).
This wave reaches full development at anion–L molar
ratios close to 2 in the presence of the dihydrogenphos-
phate anion (Fig. 3D). Maximum perturbation of the
CV curves for L4 and L5 is obtained with 1 and 1.6
equivalent of the ATP2− dianion added, respectively.
This last result is not surprising, taking into account
that Fc centers are less sterically hindered in L5 than in
L4. In these experimental conditions, the original wave
for the free Fc/Fc+ redox couple vanishes and the CV
curves present the sole electrochemical response of Fc+
-anion species. Ion pairs are adsorbed onto the elec-
trode surface during the anodic scan, and dissolve in
the reverse cathodic scan. Adsorption–desorption phe-
nomena are responsible for a peak-to-peak separation
greater than the 60 mV theoretical value for a fully
reversible system. However, in sub-stoichiometry mix-
tures the L4,5+H2PO−4 and ATP2− redox systems ap-
pear fully reversible, with DEp$0.1 V. This
electrochemical behavior allows amperometric titration
curve to be drawn by considering, e.g. the intensity of
the positive peak Icpa corresponding to the complexed
ligand versus the anion–ligand molar ratio (Fig. 3C
and D). Icpa increases linearly with the amount of anion
added to the solution, to reach a constant value. The
two-wave behavior can be attributed to a large increase
in the association constant (e.g. by several orders of
magnitude) [24] on oxidation, owing to the establish-
CH2Cl2 electrolyte, the CV behavior of L1–5 is quantita-
tively the same as in CH3CN. However, in most cases
potential shifts are significantly higher in CH2Cl2 (Table
2), suggesting an enhancement of anion–ligand interac-
tions. For example, DE1/2 for L4+H2PO−4 is −230 mV
in CH2Cl2 electrolyte, against −140 mV in CH3CN
solution. We have seen from 1H NMR experiments that
the polarity of the solvent has a weak influence on
H-bond interactions between the reduced neutral lig-
ands and anions. Thus, it can be assumed that the
solvent polarity affects mainly the ion-pairing associa-
tions between the oxidized ligands and anions, as elec-
trostatic interactions are reinforced from a polar to a
non-polar solvent. In addition, adsorption phenomena
owing to the coating of ion pair species onto the
electrode surface are stronger in the CH2Cl2 electrolyte.
Thus, from an electrochemical recognition point of
view two opposing effects occur in the CH2Cl2 elec-
trolyte: (i) a favorable effect with an increase in oxi-
dized ligand–anion interactions; and (ii) a negative
effect owing to an increased adsorption phenomena
which leads to more complex voltammograms.
3.3. Discussion
Comparison between results obtained with L1–5 in
the presence of various anions highlights the main
parameters that govern anion sensing properties of
ferrocene–amide receptors in organic electrolytes. The
number of binding sites and ferrocene groups in the
receptor, and its topology determine the electrochemi-
cal recognition behavior. Interactions between the sur-
veyed anions and receptors L1–5 in their reduced
neutral state are relatively weak, with K lower than 250
M−1 for all the host–guest systems studied (Table 1).
1H NMR experiments have shown that complexation is
mainly owing to H-bonding of anions to amide pro-
tons. In the case of the fluoride anion, additional hard
base–hard acid interactions with the iron center in the
ferrocene group could also be involved. Strong electro-
static interactions are switched on upon the oxidation
of ferrocene to ferricinium, especially with L4 and L5,
which contain three ferrocene groups and show a re-
markable two-wave behavior in the presence of H2PO4−
and ATP2− anions. It is noteworthy that even when the
association constant between the neutral receptor and
ment of very strong electrostatic interactions between
H2PO−4 , ATP2− and the triply charged L
species.
3+
4,5
3.2.3. Electrochemical response to HSO−4
The CV behavior of L1–5 in the CH3CN electrolyte
changed slightly in the presence of hydrogensulphate
anions (Table 2). Small negative shifts (around 5 mV)
in the E1/2s of the Fc/Fc+ center were observed with L1
and L2 (one-wave behavior). Further, no significant
variation in E1/2s could be measured for L3, L4 and L5,
even in the presence of a large excess or HSO−4 . How-
ever, the formation of ion pairs between this anion and
the oxidized ligands, and their adsorption onto the
electrode surface was evidenced by a clear sharpening
of the Fc+“Fc reduction wave, especially in the case
of ligands L4 and L5 that contain three ferrocene
groups.
1
the anion is too weak to be accessible by H NMR
studies (e.g. L4+H2PO−4 ), the redox ligand is able to
recognize the guest anion electrochemically. It can thus
be assumed that the oxidation of ferrocene in the
vicinity of amide groups increases the acidity of the
amide protons and makes hydrogen bonding stronger
[25], leading to a synergy between ion pairing and
H-bond interactions. Thus, the recognition behavior
can be explained in terms of H-bonding complexation,
and the magnitude of the electrochemical sensing is
3.2.4. Sol6ent effect on the electrochemical recognition
beha6ior
As the electrochemical recognition properties of L1–5
towards anions depend on both the hydrogen bonding
complexation and ion-pairing interactions, the polarity
of the solvent could play an important role in the
recognition ability of these ligands. In the less polar

O. Reynes et al. / Journal of Organometallic Chemistry 637–639 (2001) 356–363
363
essentially dependent on the strength of ion-pairing
interactions. It is not surprising that the best results
have been obtained with receptors L4 and L5, which
contain three and six amide binding sites, respectively,
and three ferrocene groups in close proximity. Al-
though the anion–ligand interactions in the reduced
state are stronger with L5 than with L4, their recogni-
tion properties towards F−, H2PO4− and ATP2− are
very close.
cenyl ligands containing one (L1) or two (L2, L3) amide
groups, but cannot be used in a simple way to sense
anions. Clear CV features, such as a well-behaved
two-wave behavior allowing an accurate amperometric
titration, require that the redox receptor contain several
binding sites and ferrocene groups in close proximity.
This goal was achieved with ligands L4 and L5 based on
the cyclotriveratrylene structural unit.
The shape of the receptor also has an important
effect on the complexation and recognition properties.
For example, the association of F− and H2PO4− with
neutral L2 is stronger than with L3, while both contain
one ferrocene unit and two amide groups. This is likely
to be because of the open shape of the binding site in
L3. However, both ligands exhibit very close electro-
chemical recognition properties, although maximal per-
turbations in the CV curves are reached at higher
anion–ligand ratios with L3 than with L2. In the case of
L4 and L5, which bear three ferrocene centers in close
proximity, repulsive electrostatic interactions and steric
hindrance do not allow the binding of three anions per
receptor molecule. This is clearly seen in the case of
H2PO−4 , where maximal CV perturbations are reached
at anion–ligand molar ratios around 2. Further, only
one dianionic ATP2− can be bound by ion-pairing
association to L43+. With L35+ this ratio is slightly
higher (around 1.6), because of its less sterically hin-
dered structure.
Finally, it should be emphasized that the electro-
chemical recognition properties of L4 and L5 towards
the dihydrogenphosphate anion are as important as
that reported earlier for dendrimers containing at least
nine amido–ferrocene moieties [15]. This comparison
shows that the cyclotriveratrylene unit provides an ap-
propriate structural effect in the construction of fer-
rocene-containing receptors able to sense dihydro-
genphosphate and ATP anions in organic electrolytes.
It has been shown that cyclotriveratrylene platforms
with cationic Ru(arene) [26] or CpFe(arene) [27] pen-
dant arms are good candidates for anion recognition.
The new hosts L4,5 thus benefit from the preorganiza-
tion of the cyclotriveratrylene platform with the three
metal centers and NH amide binding sites.
References
[1] L. Fabbrizzi, A. Poggi, Chem. Soc. Rev. 24 (1995) 197.
[2] P.D. Beer, P.A. Gale, Z. Chen, Adv. Phys. Org. Chem. 31 (1998)
1.
[3] G. Oepen, F. Vo¨gtle, Liebiegs Ann. Chem. (1979) 1094.
[4] A.P. Bell, C.D. Hall, J. Chem. Soc. Chem. Commun. (1980) 163.
[5] T. Saji, Chem. Lett. (1986) 275.
[6] P.D. Beer, Assoc. Chem. Res. 31 (1998) 71.
[7] P.D. Beer, J. Cadman, Coord. Chem. Rev. 205 (2000) 131.
[8] K. Kavallieratos, S. Hwang, R.H. Crabtree, Inorg. Chem. 38
(1999) 5184.
[9] M. Sherer, J.L. Sessler, A. Gebauer, V. Lynch, J. Chem. Soc.
Chem. Commun. (1998) 85.
[10] Z. Chen, A.R. Graydon, P.D. Beer, J. Chem. Soc. Faraday
Trans. 92 (1996) 97.
[11] A.R. Graydon, P.D. Beer, A.O.M. Johnson, D.K. Smith, Inorg.
Chem. 36 (1997) 2112.
[12] J.D. Carr, L. Lambert, D.E. Hibbs, M.B. Hursthouse, K.M.
Malik, J.H.R. Tucker, J. Chem. Soc. Chem. Commun. (1997)
1649.
[13] M. Buda, A. Ion, J.-C. Moutet, E. Saint-Aman, R. Ziessel, J.
Electroanal. Chem. 469 (1999) 132.
[14] H. Yamamoto, A. Ori, K. Ueda, C. Dusemund, S. Shinkai, J.
Chem. Soc. Chem. Commun. (1995) 333.
[15] C. Valerio, J.L. Fillout, J. Ruiz, J. Guitard, J.-C. Blais, D.
Astruc, J. Am. Chem. Soc. 119 (1997) 2588.
[16] C.M. Casado, I. Cuadrado, B. Alonso, M. Moran, J. Losada, J.
Electroanal. Chem. 463 (1999) 87.
[17] H.-J. Lorkowski, R. Pannier, Z.A. Wende, J. Prakt. Chem. 4
(1967) 141.
[18] J.-C. Moutet, E. Saint-Aman, M. Ungureanu, T. Visan, J.
Electroanal. Chem. 410 (1996) 79.
[19] C. Garcia, J. Maltheˆte, A. Collet, Bull. Soc. Chim. Fr. 130
(1993) 93.
[20] G. Ve´riot, J.-P. Dutasta, G. Matouzenko, A. Collet, Tetrahe-
dron 51 (1995) 389.
[21] KALEIDAGRAPH, Version 3.0.5, Synergy Software (PCS Inc.),
Reading, developed by Abelbeck Software Incorporation.
[22] P.D. Beer, M. Shade, J. Chem. Soc. Chem. Commun. (1997)
2377.
[23] R.H. Wopshall, I. Shain, Anal. Chem. 39 (1967) 1514.
[24] S.R. Miller, D.A. Gutowski, Z. Chen, G.W. Gokel, L.
Echegoyen, A.E. Kaifer, Anal. Chem. 60 (1988) 2021.
[25] J.D. Carr, S.J. Coles, M.B. Hursthouse, M.E. Light, J.H.R.
Tucker, J. Westwood, Angew. Chem. Int. Ed. 39 (2000) 3296.
[26] J.L. Atwood, K.T. Holman, J.W. Steed, J. Chem. Soc. Chem.
Commun. (1996) 1401.
4. Conclusion
This study emphasizes the required parameters for an
efficient electrochemical recognition of small anions
with ferrocenyl ligands. Significant CV perturbations in
the presence of anions were found with mono-ferro-
[27] K.T. Holman, G.W. Orr, J.W. Steed, J.L. Atwood, J. Chem.
Soc. Chem. Commun. (1998) 2109.