A new calix[4]arene-bis(crown ether) derivative displaying an improved caesium over sodium selectivity: Molecular dynamics and experimental investigation of alkali-metal ion complexation

View Article Online / Journal Homepage / Table of Contents for this issue

A new calix[4]arene-bis(crown ether) derivative displaying an

improved caesium over sodium selectivity: molecular dynamics

and experimental investigation of alkali-metal ion complexation

Véronique Lamare,*^aJean-François Dozol,^aSaowarux Fuangswasdi,^bFrançoise Arnaud-Neu,^b

Pierre Thuéry,^cMartine Nierlich,^cZouhair Asfari^dand Jacques Vicens^d

^aCEA CADARACHE, DESD/SEP, F-13108 St-Paul Lez Durance, Cedex, France

^bECPM, Laboratoire de Chimie Physique, UMR 7512 du CNRS, 1 rue Blaise Pascal,

F-67008 Strasbourg, France

^cCEA SACLAY, DSM/SCM, F-91191 Gif-sur-Yvette, France

^dECPM, Laboratoire de Chimie Analytique et Minérale, UMR 7512 du CNRS,

1 rue Blaise Pascal, F-67008 Strasbourg, France

Received (in Cambridge) 27th August 1998, Accepted 26th November 1998

In the light of computational studies suggesting enhanced cation-binding abilities for various benzo derivatives

of calix[4]arene-bis-crown-6 macrocycles, the synthesis and properties of one of these ligands, BC6B2, have been

investigated. In this compound, di(1,2-phenylene)-annelated glycolic chains link the 1,3- and 2,4-phenyl rings of

the calixarene, thereby forcing it into the 1,3-alternate conformation. The detailed structures of BC6B2ؒCH₃NO₂

1 and Cs₂(NO₃)₂BC6B2ؒCHCl₃ؒH₂O 2 have been established by X-ray crystallography. Studies of coordination and

solvent extraction of the alkali-metal ions have been used to compare this ligand with a number of related systems

known previously.

formed crown ether loops contain six oxygen atoms, to be suit-

Introduction

able ligands for the removal of caesium ions from both acidic

and basic nuclear wastes highly concentrated in sodium

ions.^7,15–17These ligands, designated by the acronym BC6 (for

bis-crown-6; followed, as shown later, by other letters indi-

cating the nature and number of any substituents within the

crown ether loops), are a primary subject of our continuing

studies of selective complexation of alkali-metal cations

through solution equilibrium measurements^18,19and single-

crystal X-ray diﬀraction structure determinations.^20–24Much

work has also been done on dialkyloxycalix[4]arene mono-

crown derivatives, involving synthesis, X-ray crystal structures,

and study of the calixarene conformation or crown size on

the complexation and extracting properties towards alkali

cations.^25–28

Earlier computational studies of both mono- and bis-crown

derivatives of calix[4]arene, considered in the gas phase and

solution or at immiscible solvent interfaces, were devoted to the

truly glycolic systems with simple ethylene spacers between the

ligating oxygen atoms. The conclusions drawn from these works

were that for larger cations, size complementarity of the

polyether loop and the cation was a dominant factor but that

for smaller cations, such as sodium, solvation eﬀects were

more important, since there was not enough aﬃnity with the

ligand to overcome the high desolvation energy upon

complexation.^29–34A recent publication has also shown that

the calculated Cs^ϩ/Na^ϩextraction selectivity in chloroform

decreases as the water content in the organic phase increases.³⁵

The possible eﬀects of modiﬁcation of the spacer groups were

not considered. This and in particular the introduction of aro-

matic spacers, has in fact been the focus of our computational

work and we were encouraged by the fact that the very ﬁrst

experimental studies of a calix[4]bis-crown-6 bearing a single

aromatic spacer substituent within each crown loop showed this

ligand to extract caesium more eﬃciently than its unsubstituted

analogue and to do so with a higher Cs^ϩ/Na^ϩselectivity.¹⁶

Our initial simulations involved the application of molecular

As reﬂected in a burgeoning literature, studies of chemical sep-

aration techniques and the design and synthesis of new extrac-

tion reagents have gained worldwide impetus in recent years.^1–3

In part, this impetus results from greater concerns for the

environment, energy savings and recycling at the industrial

level. It is also a result of the rise of the ﬁeld of “supramolecu-

lar chemistry”, which has provided a variety of new reagents

with the capacity to substantially improve the selectivity and

eﬃciency of many separation techniques.⁴Solvent extraction is

one of the more prominent of these techniques and in the ﬁeld

of nuclear waste treatment it is of special interest for the

selective elimination of caesium (as the radionuclides ¹³⁵Cs

and ¹³⁷Cs).⁵The element may be present as Cs^ϩcation in trace

amounts in diﬀerent types of aqueous solutions: solutions gen-

erated from spent fuel reprocessing are acidic and contain either

ﬁssion products or, as in the case of evaporator concentrates,

high concentrations of sodium nitrate.⁶Other high-salinity

liquid wastes are basic and may contain signiﬁcant amounts of

potassium nitrate along with the sodium salt, as in the case of

solutions stored at the Hanford depository.⁷The achievement

of a very high selectivity for caesium over the other alkali-metal

cations, especially sodium but to some extent potassium, is a

major goal of eﬀorts to use extraction to diminish the volume

of radioactive waste ultimately to be disposed of in geological

repositories.²

The rich coordination chemistry of the calixarenes includes

a strong aﬃnity of particular derivatives for the alkali metal

cations and a high lipophilicity for the complexes.^8–10Our own

earlier work, for example, has demonstrated in addition that

materials given the trivial name “calixbiscrowns” can be re-

markably selective in their binding and transport of the alkali

metal cations.^11–14This family of receptors, in which calix[4]-

arene is forced into the “1,3-alternate” conformation by glycolic

chain links between the phenolic oxygen atoms of the 1,3 and

2,4 phenyl rings, have been found, in the cases where the newly

J. Chem. Soc., Perkin Trans. 2, 1999, 271–284

271

View Article Online

Fig. 1 1,3-Alt-calix[4]arene-bis-crowns modelled by MD and TI/MD simulations.

dynamics (MD) calculations in a vacuum and in water as well

as free-energy-perturbation calculations (MD/FEP). They were

applied to several BC6 derivatives in which the central ethyl-

ene link of the crown loops was replaced by 1,2-, 1,3- or 1,4-

phenylene units, and the qualitative criterion used to assess the

ligands was the complementarity of the crown unit and the

cation as indicated by the MD simulations in a vacuum (intrin-

sic complementarity) and in a water “box”.^36,37The possible

enhancement of the desired properties, as revealed for the 1,2-

phenylene species, encouraged us to perform further simu-

lations for BC6 derivatives with additional 1,2-phenylene units

in the ring and the promising results obtained for the molecules

bearing two or three aromatic spacers within the crown units

(Fig. 1; BC6B2 and BC6B3) led us to undertake their synthesis

and practical evaluation.

Thus, we report herein the results of MD simulations, in

a vacuum and in water, for calix[4]arene bis-crown derivatives

used to calibrate the force ﬁeld (validation of the charge set)

and then the predicted properties for BC6B2 and BC6B3.

Additional calculations of free-energy derivatives with respect

to the non-bonded parameters for the cations and of cation

mutations within the calixarene–cation complex have been used

to provide a qualitative insight into the origins of ligand select-

ivity. As an experimental evaluation of the validity of the pre-

dictions, we describe the synthesis of the calixcrown BC6B2

which has been characterised through determination of crystal

structure, and in quantitative equilibrium studies of alkali-

metal ion extraction and complexation.

method was chosen involving MNDO scaled charges⁴²which

enabled the calculation of point charges without the necessity

of dividing the calixarene into several parts. Validation of this

charge set towards more polar MNDO/ESP charges has been

recently performed on 1,3-alt-dioctyloxycalix[4]arene-C6B2, a

monocrown analogue of BC6B2, for which X-ray crystal struc-

tures of CsPic and KClO₄–H₂O complexes were determined.⁴³

To ﬁnally evaluate the ligands, the well known concept of host–

guest complementarity,⁴⁴which involves assessment of the

steric and electrostatic ﬁt of host and guest, was used.

The intended use of the calixcrown ligands being studied was

in liquid–liquid extraction experiments where the cation to be

extracted is initially present in an aqueous phase and the ligand

in an immiscible organic phase. A criterion of complementarity

is thus the ability of the calixcrown to provide a coordination

environment similar to that of the hydration shell of the cation.

Part of this environment (number of donor atoms and their

statistical distance to the cation) can be easily modelled by the

summation of the radial distribution functions (rdfs) of the

oxygen atom donor sites, rdfO_C, collected during MD simu-

lations, and their comparison with the rdf of the water oxygen

atoms, rdfO_W, around the “free” cation under similar simulation

conditions.³⁹Fig. 2 shows some examples of the rdf curves

computed from MD simulations in vacuum. As an eﬀective

summary of the curves obtained for all the systems being

studied, values of 〈d(M^ϩ–O_C)〉, average of the 5 to 7 single d(M^ϩ–

O_C) distances occurring during each MD run, are given in Table

1. This parameter provides a simple and useful interpretation of

the rdf curves, although some information such as the ﬂuctu-

ations in single d(M^ϩ–O_C) distances is lost.

From simulations in a vacuum in the absence of a counter-

anion, it seems that BC6N (where N denotes a 2,3-disubstituted

naphthalene unit), BC6B2 (Fig. 3) and BC6B3 provide an excel-

lent complementarity for Cs^ϩ. BC6 is slightly too large for a

perfect ﬁt and BC5 is too small, though it does provide a good

ﬁt for Rb^ϩand K^ϩ. Calix[4]arene-crown-6 ligands bearing 1,2-

phenylene groups show appropriate rdfO_Ccurves for Rb^ϩbut

the peak maximum is not quite ideal and 〈d(Rb^ϩ–O_C)〉 is too

large, with a signiﬁcant dispersion of the single d(M^ϩ–O_C)

values. These discrepancies are greater for K^ϩand there is even

poorer complementarity for Na^ϩ, for which rdfO_Ccurves show

one peak with an integral ranging between 3 and 4 atoms, with

the additional O_Catoms located in the second coordination

shell.

Results and discussion

Molecular modelling

Molecular dynamics simulations. MD simulations were per-

formed on the alkali-metal ion complexes in a vacuum and in an

explicit aqueous phase to evaluate the intrinsic stability of the

complexes, the host–guest complementarity and the inﬂuence

of a water-molecule environment upon these properties. These

calculations are very sensitive to the non-bonded interaction

parameters of the force-ﬁeld equation, particularly the atomic

charge sets on the calixarene and the van der Waals parameters

of the cations. The van der Waals parameters of the alkali-

metal cations have been calibrated by Aqvist to reproduce the

free energies of hydration³⁸and these constitute a widely-used

set^{29–37,39,40}but the choice of atomic charges is not straight-

forward for molecules as large as the calixarenes, for which it is

conventional to break the molecule up into small components

and calculate the charges either from ab initio electrostatic

potentials or through cruder procedures. Of several charge sets

used so far in calixcrown studies or in those of 1,2-phenylene-

substituted crown ethers,⁴¹none could fully describe the series

of calixarenes involved in the present work. Thus, a simple

In water, most of the complexes built without counter-ions

are hydrated (Table 2), though adoption of a central position

can render Na^ϩinaccessible to solvent, as is also the case for K^ϩ

in BC6B3. Coordination of a water molecule also improves the

complementarity between Cs^ϩand the substituted BC6 ligands.

The same is true for Rb^ϩand K^ϩwith BC5.

Introduction of a counter anion (e.g., NO₃^Ϫ) into the simu-

lation system gives rise to a major electrostatic interaction

272

J. Chem. Soc., Perkin Trans. 2, 1999, 271–284

View Article Online

Table 1 Distances 〈d(M^ϩ–O_C)〉 and 〈d(C–O_C)〉 (for empty crowns) during 500 ps of MD run in a vacuum or 100 ps in water (average of the single time

averages)

b

BC5

BC6B3

BC6B2

BC6N

BC6

d(M^ϩ–O_w)^a

d(M^ϩ–O_w)_exp

c

〈d(M^ϩ–O_C)〉_vac

Cs^ϩ

Rb^ϩ

K^ϩ

3.01(11)

2.95(14)

2.93(17)

3.09(48)^d

3.18(13)

3.12(21)

3.09(30)

3.06(57)

3.15(9)

3.14(17)

3.04(9)

3.11(73)

3.22(10)

3.16(14)

3.14(19)

3.26(81)

3.30(14)

3.24(22)

3.32(34)

3.45(81)^d

3.2

2.9

2.8

2.5

3.13(7)

2.88

2.79(8)

3.35(6)

Na^ϩ

CsNO₃

RbNO₃

KNO₃

3.05(5)

2.98(5)

2.94(5)

2.83(14)

3.36(9)

3.33(4)

3.42(10)

3.38(20)

3.38(61)

3.15(1.03)

3.47(13)

3.48(25)

3.46(28)

3.47(35)

3.32(19)

3.27(19)

3.23(29)

3.26(30)

3.27(37)

3.14(78)

NaNO₃

e

〈d(M^ϩ–O_C)〉_wat

Cs^ϩ

Rb^ϩ

K^ϩ

3.00(4)

2.90(5)

2.88(3)

3.07(36)

3.22(9)

3.23(8)

3.06(35)

3.02(62)

3.19(3)

3.21(3)

3.29(10)

3.40(23)

3.66(81)

3.21(91)

3.28(23)

3.13(14)

3.02(70)

3.17(19)

3.21(14)

3.64(70)^f

Na^ϩ

g

〈d(C–O_C)〉_vac

Empty crown^h

2.87(36)

2.96(38)

3.04(35)

3.16(47)

3.24(39)

3.26(50)

3.25(41)

3.36(40)

3.37(44)

3.50(50)

Free calixareneⁱ

^ad(M^ϩ–O_w), distance cation–O_waterat the maximum of the ﬁrst hydration peak of the rdf collected during 20 ps of MD run for the free cation in

water. ^bRef. 62. ^c〈d(M^ϩ–O_C)〉 (Å): average of the 5 to 6 individual d(M^ϩ–O) time averages during the MD sampling at 300 K. ( ) ﬂuctuation. ^dAverage

calculated on 100 ps, on a longer sampling Na^ϩoscillates between both crowns. ^eAverages calculated on stabilized structures if hydration occurs

during the MD. ^fAverage calculated over 80 ps, decomplexation occurs at 85 ps. ^gC: mass center of the oxygen atoms of the crown. 〈d(C–O_C)〉 (Å):

average of the 5 to 6 d(C–O) distances. ^hAverage on the 8 simulations performed at 300 K on the mononuclear complex with and without counter-

ion. An additional sampling for BC6N of 1 ns at 500 K gives 〈d(C–O_C)〉 = 3.22(37) for the empty crown. ⁱSampling time at 300 K: 500 ps for BC5 and

BC6B3, 1 ns for BC6B2 and BC6. At 500 K: 1 ns for BC6N. In bold characters: closest values compared to d(M^ϩ–O_w).^a

Fig. 2 Radial distribution function rdfO_Cof the crown oxygen atoms around the cation during 500 ps of MD in vacuum at 300 K.

which tends to draw the cations out of the mean plane of the

crown ether loop and provides evidence of the existence of

various metastable conﬁgurations. Simulations with CsNO₃

and calixcrown-6 ligands (Fig. 4) showed a slight discrimin-

ation between BC6B3, BC6B2 and BC6N, since the best ﬁt was

then obtained with BC6B2 (on the rdf curve, the peak maxi-

mum is set at 3.2 Å for BC6B2 (narrow) and BC6B3 (enlarged),

to be compared to 3.3 Å for BC6N and BC6), but structures are

J. Chem. Soc., Perkin Trans. 2, 1999, 271–284

273

View Article Online

Table 2 Cation hydration in complexes with calix[4]arene-bis-crowns

a

nO_w

BC5

BC6B3

BC6B2

BC6N

BC6

Cs^ϩ

Rb^ϩ

K^ϩ

1.0

0.5

1.0

1.3

0

1.5

1.7

1.3

0

0.8

1.0

b

1.1

1.0

3.0

0

Na^ϩ

0

^aNumber of water oxygen atoms in the ﬁrst hydration shell of the

cation complexed by the calixcrown during the MD run in a water box

(rdf collected from 15 to 100 ps of MD run excepted for BC6B3–Cs^ϩ

(hydration at 40 ps) and BC6–K^ϩ(stabilization at 3.0 O_wsince 60 ps,

average of 2.0 O_wfrom 15 to 100 ps)). ^bDecomplexed.

Fig. 4 Minimized CsNO₃complexes with calix[4]arene-bis-crown-6 in

vacuo. (H atoms not represented).

total decomplexation of sodium is always observed during MD

simulations in water on binuclear sodium nitrate complexes

with BC6B3, BC6B2 and BC6N. The situation is less clear

starting with mononuclear NaNO₃complexes, since, in one

simulation with BC6B2, such a hydration occurred, although

with BC6B3 the simulation was stopped at 150 ps of MD run

without any contact between the cation and the water phase.

So if complete dehydration of the sodium cation is energetic-

ally at a disadvantage (see free energy calculations in water),

the Na^ϩcomplexation with partial dehydration appears to be

possible but obstructed by the presence of hydrophobic benzo

groups on the crown. The decrease of aﬃnity towards sodium

in this calixarene family must be related to this hydrophobic

unfavorable environment more than to worse Na^ϩ–calixarene

complementarity.

The crown size can be represented by the 〈d(C–O_C)〉 par-

ameter, mean distance of the oxygen atoms of the crown to

their mass center. This value was calculated on a MD sampling

of at least 500 ps in a vacuum, for both crowns in the free ligand,

and, for the empty crown, in the mononuclear alkali complexes

(Table 1). The free calixarenes need diﬀerent sampling times at

300 K to obtain the convergence of 〈d(C–O_C)〉 towards the same

value for both crowns. Indeed, 500 ps at 300 K are suﬃcient for

BC5, BC6B3 and BC6B2, 1 ns at 300 K is needed for BC6, but

convergence was not obtained for BC6N at 4 ns of MD run at

300 K, and one more sampling of 1 ns at 500 K was necessary

to obtain the value. This diﬃculty in sampling BC6N crowns

Fig. 3 Minimized BC6B2–Na^ϩ, K^ϩand Cs^ϩstructures in a vacuum

(H atoms not represented).

rather close to those obtained without counter-ion for these

ligands. On the contrary, structures simulated in a vacuum with

NaNO₃give rise to a diﬀerent location of the sodium ion: what-

ever the calixarene, this small cation is shifted to the upper part

of the crown, interacting with the nitrate and four oxygen

atoms of the crown (Fig. 5). This location leaves enough room

for a water molecule in the crown. In a previous study, taking

advantage of a BC6(NaNO₃)₂(H₂O)₂X-ray structure, we have

described how sodium complexation was probably favored by

one of the water molecules of the cation hydration shell,

leading to a partial dehydration of the sodium upon complex-

ation.²⁴This behavior has been checked on several MD simu-

lations with BC6, and is independent of the charge sets tested.

The presence of hydrophobic aryl groups on the complexation

site is a handicap for the action of the water molecule. Indeed,

274

J. Chem. Soc., Perkin Trans. 2, 1999, 271–284

View Article Online

Fig. 6 Time evolution of the free energy derivative with respect to the

van der Waals parameter R* for Cs^ϩand Rb^ϩcations complexed in the

benzocrown calixarenes (simulations in vacuo and in water). (——)

BC6B2 complex, (---) BC6B3 complex, (–ؒ–ؒ) BC6N complex, (ؒؒؒؒ) free

cation in water, * partial decomplexation.

pling time for 200 ps in a vacuum and 100 ps in water. Their

convergence is slower for the complex cations than for their

hydrated forms (see Experimental section). The van der Waals

parameters of Rb^ϩand Cs^ϩconverge very slowly and, in these

cases, it seems necessary to sample for at least 50 ps to achieve

full convergence whether in a vacuum or in water (Fig. 6). Some

FEDs did not converge due either to partial decomplexation

(BC6B2–Rb^ϩin water, where, after 45 ps, the cation is partially

out of the crown, bound by one water molecule within the

crown) or to great sensitivity to changes in OCCO dihedral

angles (BC6B2–Cs^ϩin a vacuum). The protocol 11(2 ϩ 50) was

then chosen for mutations in a vacuum but the necessity to keep

reasonable calculation times led us to choose as the standard

for mutations in water, the protocol 11(2 ϩ 20). It can be

argued that this sampling is suﬃcient, as it has already been

shown, here and elsewhere,⁴⁵that full convergence of the free

energy derivatives is not necessary to gain satisfactorily accur-

ate results for the free energy calculations for cation mutation.

Nevertheless, the sampling time evolution of the FEDs with

respect to the cation parameters shows that, as in the case of

BC6B2, it will be diﬃcult to get reliable quantitative results for

the mutation of larger cations in either a vacuum or water.

These FED calculations do not allow us to conclude which

calixcrown-6 has the best aﬃnity towards Cs^ϩor Rb^ϩ, but, in

any case, the optimum ion for these macrocycles has a size

between Rb^ϩand Cs^ϩ, as it has been previously found that

the optimum cation size for 18-crown-6 is slightly smaller

than K^ϩ.⁴⁶

Selectivity calculations were performed with the thermo-

dynamic cycle method where the intrinsic preference of a ligand

for a cation is indicated by the value of ∆G_{4 vac}. Values for the

present systems were calculated with the protocol 11(2 ϩ 50)

and show the usual preference for the smallest ions (∆G_{4 vac}> 0,

Table 3). The diﬀerence ∆G_{3 wat}Ϫ ∆G_{4 vac}can give an indication

of possible selectivity in extraction from an aqueous phase to

an apolar organic phase, if there are no major diﬀerences in

cation solvation energy in the complexes, as has been found in

simulations of BC6 alkali-metal ion complexes in chloro-

form.^30,33The predicted selectivity for BC6, BC6N and BC6B2

is much the same, with Cs^ϩ> Rb^ϩ> K^ϩ> Na^ϩ, though BC6B2

would appear to oﬀer least discrimination. For BC6B3, the

Fig. 5 Minimized BC6B2–NaNO₃structure in vacuo (orthogonal

views, H atoms not represented).

has been previously observed with our former charge set,²³and

appears to be a property of BC6N, compared to other calix-

crowns: the rigidiﬁcation of the top of the crown increases the

energetic barrier to sample the free dihedrals.

When one of the crowns is complexed, an inﬂuence of the

complexed crown on the organization of the free crown is

observed for BC6B3, BC6N and BC6, whose 〈d(C–O_C)〉 and

associated ﬂuctuation decrease. Finally, the size of the com-

plexed crown, which changes with respect to the cation and the

presence of the counter-ion, is generally smaller than the size of

the free crown due to the converging position of the oxygen

atoms towards the cation.

Thus, our overall conclusions from the present MD simu-

lations are that all the calixcrown-6 ligands, especially those bear-

ing aromatic substituents, show a good complementarity for Cs^ϩ.

The adjunction of those aromatic substituents decreases the

crown size and ameliorates the aﬃnity for Rb^ϩ, although it

remains far from the perfect ﬁt obtained with BC5. But none of

the ligands studied shows a signiﬁcant complementarity for Na^ϩ.

Relative discrimination of these calix[4]arene-crown-6 ligands

towards Na^ϩcomplexation seems mostly driven by the possibil-

ity of solvation of the cation in the complex.

Free energy calculations

The three free energy derivatives were calculated on alkali-

metal cations in calixcrown complexes, as a function of sam-

J. Chem. Soc., Perkin Trans. 2, 1999, 271–284

275

View Article Online

a

Table 3 ∆G₄

calculated from MD/TI in vacuo, and ∆G₃

calcu-

calc

OH

O

vac

Cl

OH

lated from MD/TI in water

K₂CO₃/MeCN

Na^ϩ→K^ϩ

K^ϩ→Rb^ϩ

Rb^ϩ→Cs^ϩ

∆G_{4 vac}/kJ mol^Ϫ1

BC6

34.5(1.0)

31.8(1.3)

38.4(5)

37.1(1)

38.5(5)

73.4(1)

14.0(0.5)

13.7(0)

15.2(0)

18.1(0)

20.4(1)

22.7(0)

26.3(6)

28.0(2)

30.2(8)

35.7(0)^b

44.0(0)^b

32.8(1)

BC6N

BC6B2

BC6B3

BC5

OH

1

Br

K₂CO₃

Br

∆G_{3 calc}/kJ mol^Ϫ1

^aAverages of two independent simulations. ^bIn bold characters, select-

ivity for rubidium (∆G_{3 calc}Ϫ ∆G_{4 vac}< 0).

O

TsCl/Et₃N

CH₂Cl₂

selectivity is predicted, on the basis of the mutation criterion, to

change such that there is the greatest aﬃnity for Rb^ϩ(as is the

case, though more markedly, for BC5) but this result is inconsis-

tent with estimates of the host–guest complementarity which

indicate that the preference still should be for Cs^ϩ.

OTs

OH

OTs

OH

3

2

Calculation of ∆G₄

in water could not be fully achieved

wat

due to cation decomplexation during most of the mutations.

One could notice that ∆G_{4 wat}values between Rb^ϩand Cs^ϩwere

O

close to ∆G₄

even when these cations were hydrated in the

vac

calixarene. A diﬀerence of about 4 kJ mol^Ϫ1between ∆G₄

wat

O

and ∆G_{4 vac}was found for mutations between K^ϩand Rb^ϩ, but

the ∆G₄values most sensitive to the hydration of the com-

wat

OH

HO

O

plex were found for the mutations between Na^ϩand K^ϩ,

ranging from 40 kJ mol^Ϫ1for the non-hydrated ion (close to

∆G_{4 vac}value) to about 76 kJ mol^Ϫ1when the cation is in contact

with a water molecule, leading to a strong decrease in calculated

selectivity with respect to Na^ϩ. These simulations conﬁrm that

the hydration of the sodium ion in the calixcrown-6 is a key

feature of selectivity between Na^ϩand other alkali cations.

OH

K₂CO₃

MeCN

O

+

3

O

BC6B2

Experimental investigations

Scheme 1

Synthesis of BC6B2

The synthesis of the new ligand BC6B2 is presented in Scheme

1. It follows procedures which are now well established from the

preparation of a number of calixcrowns.¹⁶The synthesis is

remarkable in all cases in that the formation of two macro-

cyclic, polyether units on one calixarene base occurs in accept-

able yields without the need for special reaction conditions

such as the use of high dilution. It is possible that cation tem-

plate eﬀects are of some inﬂuence in the reactions but since the

reaction mixture is at all times heterogeneous, it is extremely

diﬃcult to be sure of any mechanistic details.

with the other one (O7–O12). The location of this solvent mole-

cule is comparable to that in the BC6 and BC6N cases: the

C(methyl) ؒ ؒ ؒ O(ether) distances range from 3.16(1) to 3.49(1)

(mean value 3.4(1)) Å and the C(methyl) atom is 1.29(1) Å from

the mean plane deﬁned by the six oxygen atoms O1–O6 (deﬁned

within 0.305(6) Å). The crown conformations cannot be dir-

ectly compared to those of BC6 and BC6N since two torsion

O–C–C–O angles are constrained to the zero value by the

phenyl rings. These conformations can be described by the

sequence of O–C–C–O torsion angles, which is g^Ϫ0g^ϩ0g^ϩfor

O1–O6 and a0g^Ϫ0a for O7–O12 (g = gauche angle, a = anti

angle). Moreover, not all the C–O–C–C torsion angles are anti

ones: some of them are gauche angles or intermediate values:

the departure from 180Њ is particularly evident near O8 and O11

whose lone pairs point outwards. On the contrary, the oxygen

atoms’ lone pairs in the crown, including the solvent molecule,

are all directed towards the crown centre as in the caesium

complexes.^20–23

Doubly crowned calix[4]arene BC6B2 was fully characterized

1

by H-NMR, FAB spectrometry and elemental analysis. The

1,3-alternate conformation was deduced from its ¹H-NMR

spectra, which showed singlets at 3.74 ppm for the Ar–CH₂–Ar

methylene protons in the macrorings of BC6B2. The simplicity

of the polyether chain methylene resonances in particular indi-

cates that the chains must be conformationally mobile on the

NMR timescale.

The molecular unit of 2 is shown in Fig. 8. The two caesium

ions are included in the crown ether moieties, as in the com-

plexes described previously.^20–23A particular feature of the

present structure concerns the counter-ions. Whereas in the

previous structures, the nitrate ions were always bonded to

the caesium ion in a bidentate mode (with the exception of a

monodentate one²¹), one of the nitrate ions in the present case

is not bonded and the other one bridges the caesium ions of two

neighboring molecules (bidentate for one, monodentate for the

other) thus giving rise to polymeric inﬁnite chains. The form-

ation of dimers has already been observed with thiocyanate

counter-ions.²⁰The caesium ions are bonded to the six oxygen

atoms of each crown, with distances ranging from 3.094(3) to

3.366(3) (mean value 3.26(11)) Å for Cs1 and from 2.995(4) to

Crystal structures of BC6B2ؒCH₃NO₂1 and Cs₂(NO₃)₂BC6B2ؒ

CHCl₃ؒH₂O 2

An ORTEPII view of the molecular unit of 1, which does not

possess any symmetry element, is given in Fig. 7. We have

recently reported the crystal structures of solvent adducts of

BC6 and BC6N.^22,23The presence of complexed solvents has

been shown to have an inﬂuence on the crown conformation:²³

polar solvents such as acetonitrile and nitromethane were

shown to interact with the crown ether oxygen atoms via weak

C–H ؒ ؒ ؒ O hydrogen bonds with their methyl group pointing

towards the crown center. The present structure illustrates by

itself such an eﬀect since a nitromethane molecule is associated

with one of the two crowns (corresponding to O1–O6) and not

276

J. Chem. Soc., Perkin Trans. 2, 1999, 271–284

View Article Online

Fig. 8 X-Ray crystal structure of Cs₂(NO₃)₂BC6B2ؒCHCl₃ؒH₂O 2.

Fig. 7 X-Ray crystal structure of BC6B2ؒCH₃NO₂1. Hydrogen atoms

omitted for clarity. Shorter intermolecular contacts (Å): C65 ؒ ؒ ؒ O1

3.38(1), C65 ؒ ؒ ؒ O2 3.46(1), C65 ؒ ؒ ؒ O3 3.49(1), C65 ؒ ؒ ؒ O4 3.16(1),

C65 ؒ ؒ ؒ O5 3.36(1), C65 ؒ ؒ ؒ O6 3.48(1).

Hydrogen atoms, non-coordinating nitrate ion and solvent molecules

omitted for clarity. Symmetry operations: i = 1/2 Ϫ x, y Ϫ 1/2, 3/2 Ϫ z;

ii = 1/2 Ϫ x, 1/2 ϩ y, 3/2 Ϫ z. Selected bond lengths (Å): Cs1–O1

3.366(3), Cs1–O2 3.305(4), Cs1–O3 3.094(3), Cs1–O4 3.299(3), Cs1–O5

3.336(3), Cs1–O6 3.137(3), Cs1–O13 3.363(4), Cs1–O14 3.208(4),

Cs1 ؒ ؒ ؒ C47 3.600(5), Cs1 ؒ ؒ ؒ C48 3.367(5), Cs1 ؒ ؒ ؒ C49 3.602(5),

Cs1 ؒ ؒ ؒ C61 3.695(5), Cs1 ؒ ؒ ؒ C62 3.416(5), Cs1 ؒ ؒ ؒ C63 3.619(5), Cs2–

O7 3.229(3), Cs2–O8 3.297(4), Cs2–O9 2.995(4), Cs2–O10 3.278(4),

Cs2–O11 3.349(4), Cs2–O12 3.088(3), Cs2–O15ⁱ3.186(4), Cs2 ؒ ؒ ؒ C39

3.627(5), Cs2 ؒ ؒ ؒ C40 3.436(5), Cs2 ؒ ؒ ؒ C41 3.615(5), Cs2 ؒ ؒ ؒ C54

3.680(5), Cs2 ؒ ؒ ؒ C55 3.458(5), Cs2 ؒ ؒ ؒ C56 3.679(5).

3.349(4) (mean value 3.21(14)) Å for Cs2. These values are in

agreement with those in the complexes previously described;

they are slightly lower than those determined for BC6 com-

plexes with bidentate nitrate ions (3.4(1) Å), and closer to those

obtained with a monodentate nitrate ion (3.18(6) Å) or in the

case of BC6B or BC6N complexes (3.3(2) Å). This is in qualita-

tive agreement with the crown sizes deduced from molecular

modelling (see Table 1 above). The distances between caesium

ions and the oxygen atoms’ mean planes are 0.83(2) and 0.72(2)

Å for Cs1 and Cs2 respectively. Both distances are in the usual

range; however, the caesium ion is bonded to only one nitrate

oxygen atom, hence possesses a less bulky counter-ion, and thus

seems to be slightly nearer to the crown. Due to the peculiar

binding mode of the counter-ion, the distances between cae-

sium ions and the nitrate oxygen atoms lie a little above the

usual range: 3.363(4) and 3.208(4) Å for Cs1, 3.186(4) Å for

Cs2, to be compared to mean values of 3.08(6)–3.2(1) Å in

the previous cases (3.15(1) Å for the monodentate nitrate ion).

The monodentate bond length appears slightly shorter than the

bidentate means. The shorter Cs ؒ ؒ ؒ C distances, as usual, are

with the three terminal atoms of the phenolic rings, with mean

values of 3.5(1) Å for Cs1, 3.6(1) Å for Cs2 (3.5(2) to 3.6(2) Å in

the previous cases). The conformations of the two crowns are

identical: the O–C–C–O angles give the sequence g^Ϫ0g^Ϫ0g^ϩ,

analogous to the one observed in the solvent-containing crown

in 1, and two C–O–C–C angles deviate from the ideal anti value

(81 and 94Њ). Once more, it is found that a caesium ion or a

solvent molecule (nitromethane or acetonitrile) organizes the

crowns in very similar ways, whereas empty crowns, due to

their high ﬂexibility, may display completely diﬀerent conform-

ations.²³The conformation observed here is roughly the same as

those found in the caesium and potassium complexes of the

mono-crown derivative: g^Ϫ0α0g^ϩ, with α = 29Њ for Cs^ϩand

g^Ϫ0g^ϩ0g^ϩfor K^ϩand, in both cases, deviations of some C–O–

C–C angles from the ideal value.⁴³The crystal structures of the

complexes with the unsubstituted crown derivative have already

shown that caesium and potassium ions induce the same crown

conformation.²⁴

Complexation studies

Preliminary studies of alkali picrate complexation by BC6B2

were monitored with ¹H-NMR. After a reaction period of

approximately 72 hours between solid alkali metal picrates (Li^ϩ,

Na^ϩ, K^ϩ, Rb^ϩand Cs^ϩ) and a chloroform solution of BC6B2,

the ratios of metal to ligand in solution, as estimated from

integration of the picrate proton resonances versus those for the

methylenic protons Ar–CH₂–Ar, were 0.4:1 (Li^ϩ), 0.1:1 (Na^ϩ),

0.4:1 (K^ϩ), 1.3:1 (Rb^ϩ), and 1.0:1 (Cs^ϩ). These ratios must be

determined partly by diﬀerences in the lattice energy of the

solid picrates as well as by diﬀerences in complex ion stability,

and may possibly be aﬀected by the fact that lithium and

sodium picrates are hydrated, whereas those of potassium,

rubidium and caesium are not, though no special care was

taken to dry CDCl₃used. All spectra in fact showed (weak)

water resonances of similar intensity, but one can deduce from

these values a preference for Rb^ϩand Cs^ϩover the other alkali

metal ions.

J. Chem. Soc., Perkin Trans. 2, 1999, 271–284

277

View Article Online

b

Table 4 Interaction of alkali metal ions with BC6B2:percentage extraction E^afrom water into dichloromethane; stability constants log β_xyand

thermodynamic parameters^cin acetonitrile; association constants log K_aand position of the maximum absorption λ_max^dof the separated ion pair in

tetrahydrofuran, with, in parentheses, the shift vs. the position of the maximum absorption of the tight ion pair. (Mean values of n ≥ 2 independent

determinations, with the standard deviation on the mean, σ_{n Ϫ 1}

Li^ϩNa^ϩ

0.9 0.2 0.9 0.1

)

K^ϩ

11.3 0.1

5.22 0.05

29.8 0.3

10.7 0.7

Rb^ϩ

Cs^ϩ

56.0 0.1

6.3 0.1

35.9 0.6

E

40.4 0.3

5.5 0.2

31

22.7 0.2

Log β₁₁

2.18 0.03

≤1

≤5.70

Ϫ∆G₁₁

Ϫ∆H₁₁

T∆S₁₁

Log β₂₁

Ϫ∆G₂₁

Ϫ∆H₂₁

T∆S₂₁

Log K_a

λ_max(∆λ_max

—

1

e

29

7

1

2

—

19

1

8

1

8.05 0.07

45.9 0.4

23.8 0.6

9.4 0.1

53.6 0.6

46.1 0.5

10.1 0.2

58

54

4

1

3

4

—

22

4.0 0.1

375(17)

1

8

1

f

4.91 0.09

378(19)

5.75 0.04

377(16)

)

—

xϩ

^aT = 20 ЊC. ^bT = 25 ЊC, I = 0.01 M (Et₄NClO₄); corresponding to the general equilibrium: xM^ϩϩ yL

M_xL_y.

^cT = 25 ЊC, ∆G, ∆H and T∆S in

kJ mol^{Ϫ1 d}λ_maxin nm. ^eNot measurable. ^fNo spectral changes observed.

.

Extraction data (Table 4) show that with BC6B2, as with

all the bis-crown-6 derivatives previously studied,¹⁸there is an

increase in percentage extraction (E) along the alkali metal

series. It can be noted that the extraction level of the larger

cations is less aﬀected by the substitution of the crown linkage

than the smaller ones. For instance, the extraction of Cs^ϩis

only slightly decreased when one and two benzo groups are

attached to the crown moieties, whereas the extraction of Na^ϩ

is strongly decreased. These variations lead to an increase in the

Cs^ϩ/Na^ϩextraction selectivity S = E_Cs/E_Nafrom 14 to 29 and

62, for BC6, BC6B and BC6B2 respectively.

ϩ

Complexation studies of BC6B2 in a homogeneous medium

were limited because of the low solubility of this ligand. How-

ever stability constants could be determined in acetonitrile and

were compared to those obtained previously for BC6, BC6B

and BC6N.¹⁸These results had shown that the substitution of

the crown-6 parts of BC6 by a 1,2-phenylene or a naphtho

group led to practically no change in the complexing properties

towards the large cations K^ϩ, Rb^ϩand Cs^ϩ, and a small but

systematic decrease of the complex stability with the smaller

cations Li^ϩand Na^ϩ. This led, as already observed in extrac-

tion, to an increase of the Cs^ϩ/Na^ϩselectivity from BC6 to

BC6B and BC6N,¹⁸which has been interpreted in terms of loss

of ﬂexibility of the crown chain, the substituents rigidifying the

crown and preventing optimal interactions between the binding

sites and the smaller cations. The introduction of a second 1,2-

phenylene group like in BC6B2 leads to important modiﬁ-

cations in both the nature and the stability of the complexes

formed. With K^ϩ, Rb^ϩand Cs^ϩ, the experimental spectro-

photometric data present ﬁrstly a series of spectra passing

through an isosbestic point until C_M/C_L≈ 1 (for Cs^ϩand Rb^ϩ)

and C_M/C_L≈ 2 (for K^ϩ), followed by a second set of spectra

with increasing intensities for higher values of C_M/C_L. Fig. 9

illustrates the case of Cs^ϩ. These spectral changes clearly show

the presence of three distinct absorbing species (including

the ligand) and have been interpreted by the formation of a

binuclear species as well as the mononuclear complex (Table 4).

The binuclear complex of Cs^ϩhas been isolated in the solid

state and its structure is reported in this paper. The formation

of binuclear complexes had already been shown with Cs^ϩin

solution by NMR²¹and in the solid state for the three ligands

BC6, BC6B and BC6N.^14,20,21,23Complexation of Li^ϩleads to

a small but regular decrease in the intensities of the spectrum

of the ligand, as was previously observed with the other bis-

crown-6 derivatives and all alkali metal ions,¹⁸indicating the

formation of a mononuclear complex only. With Na^ϩ, no spec-

tral changes could be observed even in the presence of a large

excess of metal ion.

Fig. 9 Changes in the UV absorption spectrum of BC6B2 (C_L=

8.2 × 10^Ϫ5M) upon addition of CsNO₃in acetonitrile: a) 0 ≤ C_M/

C_L≤ 0.9 b) 1.1 ≤ C_M/C_L≤ 14.3; spectrophotometric cells of 1 cm

pathlength.

large cations K^ϩ, Rb^ϩand Cs^ϩ.¹⁸The introduction of two 1,2-

phenylene groups on each crown unit of BC6 results in a

decrease in the stability of the sodium complex and an increase

in the stability of the complexes with the larger cations K^ϩ, Rb^ϩ

and Cs^ϩ(e.g. ∆log β₁₁= 1.4 in the case of Cs^ϩ), whereas the

stability of the Li^ϩcomplex surprisingly remains almost the

same. The Cs^ϩ/Na^ϩselectivity increases from BC6 to BC6B,

BC6N and BC6B2, although the selectivity of the latter ligand

5

ϩ

(S = β_Cs/β_Naу 2 × 10 ) cannot be strictly evaluated because

of the very weak stability of the Na^ϩcomplex, which could only

be estimated to be lower than 1 log unit.

The stability of the 1:1 complexes of BC6B2 roughly con-

ﬁrms the trends observed in extraction, i.e. an aﬃnity for the

The substitution of the crown parts of bis-crown-6 deriv-

278

J. Chem. Soc., Perkin Trans. 2, 1999, 271–284

View Article Online

atives by one aryl group had been shown to lead to important

changes in the stabilization of the complexes.¹⁸In particular,

it induces a strong decrease in the enthalpy of complexation

(Ϫ∆H₁₁) and a strong increase in the corresponding entropy

(T∆S₁₁). This had been interpreted by the rigidiﬁcation of the

molecule, which weakens the interactions between the binding

sites and the cations, and counteracts the loss of degrees of

freedom upon complexation. The substitution of the crown

moieties by two 1,2-phenylene groups also leads to some

changes in the stabilization of the complexes. As seen from the

calorimetric data in Table 4, the stabilization of the K^ϩ, Rb^ϩ

and Cs^ϩcomplexes of BC6B2 in acetonitrile is both enthalpy

and entropy driven (Ϫ∆H₁₁> 0 and T∆S₁₁> 0). This contrasts

with the stabilization of their homologs with BC6 which was

mostly enthalpy controlled.¹⁸However, in agreement with what

was observed with BC6, Ϫ∆H₁₁values decrease in the series

from Cs^ϩ, to Rb^ϩand K^ϩ, whereas T∆S_cvalues increase. As for

the unsubstituted ligand, the enthalpy variations can be inter-

preted in terms of size adequacy between the cation and the

crown loop, the crown size being too large to allow eﬃcient

interaction with the smaller cations. The entropy changes are

related to the stronger desolvation of these cations upon com-

plexation and to a smaller loss of conformational freedom of

the complexed crown moiety.

The higher stability of BC6B2 complexes over BC6 ones is

entirely due to the more favourable entropy terms (Table 4).

This can be explained by the greater rigidity of BC6B2 and its

smaller loss of freedom upon complexation. The rigidity is also

reﬂected in the enthalpy changes. Although the Ϫ∆H₁₁value

for the complexation of Cs^ϩ, whose size is well adapted to the

crown, is similar to that observed with BC6, it is lower with

the smaller and less suited Rb^ϩand K^ϩcations, indicating the

diﬃculty encountered by the di-substituted crown part in

accommodating these cations.

Comparison between BC6B2 complexes and their homologs

with BC6B and BC6N bearing only one aryl group is less

straightforward (Fig. 10). Stabilization of BC6B and BC6N

complexes has been shown to be mostly entropy driven

(Ϫ∆H₁₁> 0 and T∆S₁₁> 0, T∆S₁₁being dominant).¹⁸The extra

stability of Rb^ϩand Cs^ϩcomplexes with BC6B2 is now due to a

more favorable enthalpy term overcoming a less favorable

entropy term. Diﬀerent factors may account for this behavior:

the higher Ϫ∆H₁₁value may reﬂect a better preorganization or

complementarity of the bis(1,2-phenylene)-crown derivative as

well as the possibility of strong π-interactions with the cation

through the two benzo rings.⁴⁶The lower solvation of the di-

substituted ligand may also be associated with a higher

enthalpy contribution. All these eﬀects must be predominant

over the eﬀect of decreased basicity of the oxygen donor sites

due to the presence of the benzo rings.⁴⁷The smaller entropy

term is consistent with the lower solvation of the ligand and its

lower rigidity with respect to BC6B and BC6N.

40

30

20

10

0

Rb⁺

–10

40

BC6

BC6B

Cs⁺

BC6N

BC6B2

30

20

10

0

–10

BC6

BC6B

BC6N

BC6B2

Fig. 10 Thermodynamic parameters for the complexation of Rb^ϩand

Cs^ϩby calix[4]arene-bis-crown-6 in acetonitrile.

T∆S₁₁.

∆G₁₁, ᮀ ∆H₁₁, ᭿

constants log K_acorresponding to the formation of the separ-

ated ion pairs from the tight ion pairs are also given in Table 4.

These values do not display any systematic trend. The introduc-

tion of two 1,2-phenylene groups on the crown-6 parts can lead

to either a decrease in log K_aas in the case of Rb^ϩ, to an

increase as in the case of Cs^ϩor to practically no change as for

K^ϩ. However, the optical shifts, ∆λ, experienced by the maxi-

mum absorption of the tight ion pair seems to be more inform-

ative: they are higher for Rb^ϩand Cs^ϩwith BC6B2 than with

the other ligands.⁴⁸In this case, the maximum absorption of the

separated ion pair approaches the value of 380 nm, character-

istic of the fully separated ion pair, suggesting a better shielding

of these cations by BC6B2 than by the other bis-crown-6

derivatives.

The binuclear complexes found with BC6B2 and the three

cations K^ϩ, Rb^ϩand Cs^ϩare less stable than the corresponding

mononuclear complexes. The lack of positive cooperative eﬀect

in their formation is essentially due to unfavorable entropy

terms. The enthalpy of complexation corresponding to the

stepwise formation of the binuclear complex ∆H₂= ∆H₂₁Ϫ

∆H₁₁and the corresponding entropy terms T∆S₂= T∆S₂₁Ϫ

T∆S₁₁can be calculated from the overall values given in Table 4.

It can be seen that, whereas Ϫ∆H₂and Ϫ∆H₁₁values are of

the same order of magnitude, T∆S₂values are very low when

compared to T∆S₁₁. For instance, T∆S₂is found to be 3 kJ

mol^Ϫ1instead of 19 kJ mol^Ϫ1for T∆S₁₁, for the K^ϩcomplex.

This may show a loss of freedom of the system upon com-

plexation of the second cation, resulting from its organization

and rigidiﬁcation.

Discussion

In this study, there is an overall agreement between observ-

ations from MD simulations in a vacuum about crown size,

crown rigidity, host–guest complementarity, and experimental

data for the bigger cations. BC6 appears too large for Cs^ϩ, but

ﬂexible enough to optimize interactions on condition of a slight

energetic loss in the entropic term upon complexation. The

rigidiﬁcation occurring in the top of the crown of BC6N

decreases dramatically the ﬂexibility of the crown, which is still

too large, but is now unable to ﬁt the cation well, which causes

the loss in the enthalpic term. On the contrary, the gain in the

entropic term shows a good preorganization of the crown due

to the loss of ﬂexibility and probably (but not illustrated in

simulations) a weak solvation of the crown before complex-

ation.

The separation of the tight ion pairs formed by alkali picrates

in a medium of low relative permittivity gives another picture

of the complexing ability of a given ligand. The association

The BC6B2–Cs^ϩcomplex is comparable to BC6–Cs^ϩfor the

J. Chem. Soc., Perkin Trans. 2, 1999, 271–284

279

View Article Online

enthalpic term, which underlines the good size complementar-

ity. The entropic term is intermediate, stabilizing the complex

but to a lesser extent than BC6N, and shows that the actual

rigidiﬁcation of the crown depends more on the location of the

aromatic spacers than on their number. Thus, one can expect

for BC6B3 the qualities of BC6N (maximum of rigidiﬁcation

and minimum of solvation) without its drawback (smaller

crown size). Nevertheless, the tail observed on the rdfO_Ccurves

could be a signal of too much rigidiﬁcation, leading to diﬃcul-

ties for the oxygen atoms in arranging around the cation.

Moreover, the crown size could this time be too small, which

would lead to a selectivity for rubidium instead of caesium,

as suggested by mutation calculations. So, this compound

could have the same behavior as BC5, which is selective for

Rb^ϩand K^ϩin extraction experiments. Here, the prediction

limits of our model are reached, as there remains a doubt on

the real selectivity of this new calixarene which has not been

evaluated yet.⁴⁹

Complex solvation has been taken into account with MD

simulations in water. This solvent can be supposed not to be

convenient for the study of these highly hydrophobic ligands

because, although the structures obtained through simulations

in vacuum may be compared with those obtained by X-ray

diﬀraction, the hydrophobicity of the ligands means that the

structural and energetic results obtained from simulations in an

aqueous phase cannot be directly compared with experimental

data for extraction. Water is present at every step of the extrac-

tion process and incomplete dehydration of the cation during

complexation can give rise to a hydrated complex ion product.

Such species have been proposed in order to rationalize extrac-

tion isotherms for K^ϩ, Rb^ϩand Cs^ϩcomplexes of calix-

crowns.⁵⁰No evidence was obtained for similar behavior

involving Na^ϩbut in fact the crystal structure of the binuclear

NaNO₃complex of BC6 obtained from undried salt and

methanol–acetonitrile solution shows that a hydrated species

can form.²⁴The main problem involved in MD simulations in

water is the impossibility of simulating alkali complexes with

counter-ions, with the exception of sodium, due to the high

dissociating power of this solvent. Nevertheless, MD results

from simulations in an explicit water phase show the possible

hydration in a facial position of bigger alkali cations not

coordinated with a counter-ion, which had been conﬁrmed by

an X-ray structure of the KClO₄–H₂O complex of 1,3-alt-

dioctyloxycalix[4]arene C6B2, a monocrown analogue of

BC6B2, whose perchlorate anion is dissociated from potas-

sium.⁴³In the case of sodium ion, MD simulations with nitrate

counter-ion suggest that the selectivity in the alkali series,

unfavorable to sodium, is due to the diﬃculty of sodium com-

plexation without co-complexation of a water molecule, which

is in agreement with the meanings of mutation calculations.

The greater Cs^ϩ/Na^ϩselectivity of calixcrown-6 bearing 1,2-

benzo groups with respect to the unsubstituted one, is expected

to arise from the diﬃculty of complexation of Na^ϩ–H₂O species

in a more hydrophobic complexation site.

provided by the optimization of the carbamoylmethylphos-

phine oxide (CMPO) ligands used in the “TRUEX” process for

the extraction of trivalent lanthanides and actinides from nitric

acidic solutions,⁵²explained either by a weaker solvation of aryl

substituted extractants, which are weaker bases than alkyl sub-

stituted ones and compete less with nitric acid,⁵³or by an

increase of the donor properties of the phenyl group as the

oxygen atom is electron-depleted by charge transfer to the

cation.⁵⁴Such properties are not in the scope of current mol-

ecular modelling studies and should be investigated through

quantum mechanics.

Conclusion

Molecular modelling, based on force ﬁeld methods, has been

used as a suitable method to explain the experimental Cs^ϩ/Na^ϩ

selectivity of a family of 1,3-alt-calix[4]arene-bis-crown-6,

using an appropriate choice of force ﬁeld parameters, mainly

point charges. Calculations for new ligands belonging to this

family before their synthesis, and focusing on host–guest com-

plementarity, have shown that their aﬃnity towards alkali

cations was comparable to those of known compounds, with

a possible inversion of the Cs^ϩ/Rb^ϩselectivity for the tri-

benzocrown-6 derivative not yet tested. In fact, the newly tested

dibenzocrown-6 presents an enhanced Cs^ϩ/Na^ϩselectivity

due to a poorer aﬃnity for sodium cation. This feature is not

the result of complementarity diﬀerences, and has been inter-

preted by the greater diﬃculty for the sodium cation to be

complexed without total dehydration in a more hydrophobic

complexation site.

Experimental

Simulation protocols

All calculations were carried out on SGI workstations

(INDIGO 2 R8000 or Origin 200 R10000) with the AMBER

4.1 software,⁵⁵using as a force ﬁeld, the all-atom parameters

and the representation of the potential energy given in eqn. (1).⁵⁶

E_pot= ͚ K_r(r Ϫ r_eq)²ϩ ͚K_θ(θ Ϫ θ_eq)²ϩ

bonds

angles

12

6

V

R*

͚

ⁿ(1 ϩ cos(nφ Ϫ η)) ϩ ͚ͫε_ijͩͩ ͪ

Ϫ

ͩ ͪͪͬ

ͩ ͪ ͪͬ

ϩ

dihedrals

i < j

2

R_ij

R*

R_ij

12

10

q_iq_j

R*

(1)

͚ͫ ͬϩ ͚ ͫε_ijͩͩ ͪ

Ϫ

i < j

H-bonds

εR_ij

R_ij

The calixarenes were model built on the graphic workstation

with the InsightII software⁵⁷from previous optimized struc-

tures then submitted to a ﬁrst minimization without atomic

charges with the presence of the caesium cation in one crown.

The atomic charges on the calixarene were calculated with

the MNDO semiempirical method without geometry optimiz-

ation. The calculation was performed on the whole calixarene

and the crude point charges thus determined were then aver-

aged for equivalent atoms and scaled up with a 1.26 scaling

factor to allow a nice ﬁt with 6-31G*/ESP values commonly

used for crown-ether moieties.⁴²These point charges take into

account diﬀerent conformational environments (average on the

complexed and the empty crown conformations) and no arti-

fact is introduced by cutting the calixarene cavity. When pos-

sible, charges on equivalent atoms were kept constant through

the series, but some signiﬁcant diﬀerences were found due to the

presence of benzo substituents on the crown (Fig. 11).

Studies of homogeneous phase complexation have shown

that the introduction of a 1,2-benzo spacer at the head of the

crown link has little inﬂuence on the binding of Cs^ϩbut

strongly diminishes the interaction with Na^ϩ. This eﬀect is also

reﬂected in the results for extraction of the alkali-metal picrates

by the calixcrown ligands into dichloromethane. Enhancement

of the extraction of Cs^ϩunder some experimental conditions,

where the aqueous phase contains nitric acid, must be attrib-

uted to factors not quantiﬁed within our simulations and is

possibly related to the increase of extractant hydrophobicity

and the presence of acid.^17,51Indeed, extraction isotherms have

highlighted the formation of cation–nitric acid complexes for

Cs^ϩand Na^ϩwith the monocrown 1,3-alt-diisopropoxycalix-

[4]arene-crown-6.⁵⁰It is known, in fact, that the substitution of

alkyl by aryl groups within an extractant can improve the eﬃ-

ciency of the extractant in acidic media, a good example being

The cation parameters came from Aqvist³⁸and were adapted

to the AMBER force ﬁeld (TIP3P water model and periodic

boundary conditions). The 1–4 non bonded contributions were

scaled down by a factor of 0.5.

In a vacuum, the alkali complexes were submitted to minimiz-

ation with 50 steepest descent iterations followed by 1950 con-

280

J. Chem. Soc., Perkin Trans. 2, 1999, 271–284

View Article Online

Table 5 Evolution of calculated ∆G₃with respect to the protocol of cations mutation

Exp.

21(1 ϩ 4)^a

105 ps

11(5 ϩ 5)

110 ps

11(2 ϩ 20)

240 ps

11(20 ϩ 20)

440 ps

11(2 ϩ 40)

460 ps

∆G₃/kJ mol^Ϫ1

(Aqvist)³⁸

Na^ϩ→K^ϩ

Rb^ϩ→Cs^ϩ

73.7

32.3

72.2

33.3

73.4

33.0

73.3

32.9

72.6

32.8

73.4

33.1

^a21(1 ϩ 4): 21 windows (δλ = 0.05) with 1 ps of equilibration and 4 ps of data collection by MD at 300 K. 20 ps of MD correspond to about 1 CPU

hour on the SG R8000 workstation.

PBC and an isothermal/isobaric ensemble of 300 K and 1 atm

through coupling to temperature and pressure baths. All the

starting structures were complexes minimized at the end of the

MD run in a vacuum, and immersed in a TIP3P cubic box,

removing water molecules within 2 Å of the solute (about 1200–

1600 water molecules around the solute). These systems were

energy minimized and submitted to at least 100 ps of molecular

dynamics simulation (100 ps of MD in water took approxi-

mately 15 hours on one CPU of the R10000 computer).

Free energy calculations were performed with the thermo-

dynamic integration procedure (TI).⁵⁹The free energy deriv-

atives were calculated in a vacuum and in water as a function of

sampling time (one cumulative calculation each 5 ps) as con-

vergence tests in order to check our protocol for mutation

calculations and to improve the accuracy of the selectivity

calculations without changing the force ﬁeld parameters.^45,60

Previously, the MD/FEP method has been used in standard cal-

culations with the protocol 11(1 ϩ 4) ps.^29,37The equilibration

BC6

BC6B2

BC6B3

BC6N

and data collection samplings were ﬁxed and, in the case of

convergence problems between two independent mutations, the

width of the window, δλ, was changed, leading to mutation

lengths varying from 55 ps (11 windows, δλ = 0.1) to 505 ps (101

windows, δλ = 0.01). Based on the hypothesis that the free

energy is not a sensitive function of λ, it has been argued that,

to obtain a good ratio of sampling length to computational

accuracy, it is better to use the molecular dynamics/thermo-

dynamic integration (MD/TI) procedure with a relatively large

δλ and to sample longer in each window in order to minimize

the equilibration step and improve sampling for the free energy

calculation.⁶¹The calculation of free energy derivatives for a

single λ value (here, λ = 1) then provides the necessary sampling

length at each λ to obtain convergence of each parameter

involved in the free energy calculations. For cations, the non-

bonded parameters considered are the point charge q, the van

der Waals radius R* and the well depth ε.

The free energy derivatives (FED) for hydrated alkali-metal

cations were calculated during 180 ps of MD in water. The

evolution of the derivatives as a function of time indicates that

a period of 20 ps is necessary to reach convergence for the

three derivatives. To discern the impact of the simulation

protocol, we calculated ∆G₃in water for Na^ϩ→K^ϩand Rb^ϩ→

Cs^ϩmutations (Table 5). The worst results were obtained with

the 21(1 ϩ 4) protocol. For the same length of time, the

protocol 11(5 ϩ 5) gives a ∆G₃value closer to the experimental

reference. In the simulations with 11 windows, no improvement

results in increasing the sampling length beyond 20 ps. The

Rb^ϩ→Cs^ϩmutation overestimates the experimental value by

~0.7 kJ mol^Ϫ1, while the Na^ϩ→K^ϩmutation underestimates the

target value by ~0.4 kJ mol^Ϫ1, an inaccuracy which cannot be

attributed to the inﬂuence of sampling length, as is shown by

the unsatisfactory result for the 11(20 ϩ 20) protocol.

C1

C2

C3

H3

C4

H4

C5

H5

O

C6

H6

O1

C7

C8

H8

C9

H9

O2

C10

C11

H11

C12

H12

C13

H13

C14

H14

0.157

Ϫ0.140

Ϫ0.013

0.076

Ϫ0.087

0.080

Ϫ0.164

0.000

Ϫ0.360

0.200

0.150

Ϫ0.119

Ϫ0.020

0.071

Ϫ0.082

0.075

0.143

0.001

Ϫ0.346

0.220

0.000

Ϫ0.346

0.120

Ϫ0.100

0.087

Ϫ0.074

0.080

0.150

Ϫ0.119

Ϫ0.020

0.071

Ϫ0.082

0.075

0.143

0.001

Ϫ0.346

0.220

0.000

Ϫ0.346

0.103

Ϫ0.083

0.091

Ϫ0.071

0.085

Ϫ0.275

0.103

Ϫ0.083

0.091

Ϫ0.071

0.085

0.150

Ϫ0.119

Ϫ0.020

0.071

Ϫ0.082

0.075

0.143

0.001

Ϫ0.346

0.203

0.000

Ϫ0.400

Ϫ0.387

Ϫ0.346

0.124

Ϫ0.089

0.088

Ϫ0.042

Ϫ0.051

0.077

Ϫ0.069

0.077

Fig. 11 Point charges used in simulations.

jugate gradient iterations. Then MD simulations were per-

formed at constant temperature (300 K) and energy for 500 ps

with a 1.0 fs time step, the relative permittivity set at 1.0 and a

10 Å residue-based cut-oﬀ. One conformation was saved on

each picosecond calculation and the trajectories, constituted

of the collection of these conformations, were visualized on a

graphics screen by the MD/DRAW software and analyzed by

the MDS software.⁵⁸The ﬁrst ten picoseconds, corresponding

to the system thermal equilibration, were not taken into

account in the structural and energetic analysis and averages

were calculated over 490 ps of MD.

For selectivity calculations, the thermodynamic cycle method

was considered to be as in Scheme 2⁵⁹where, assuming that

∆G₁

calix ϩ M^ϩ

∆G₃

calix:M^ϩ

∆G₄

∆G₂

In water, the molecular dynamics simulations used a time

step of 2.0 fs, and a 12 Å cut-oﬀ. The SHAKE procedure was

used to constrain bonds involving hydrogen atoms. We used

calix ϩ N^ϩ

calix:N^ϩ

Scheme 2

J. Chem. Soc., Perkin Trans. 2, 1999, 271–284

281

View Article Online

the system is reversible, the relative binding selectivity is com-

puted by the diﬀerence: ∆∆G = ∆G₁Ϫ ∆G₂= ∆G₃Ϫ ∆G₄. ∆G₃

is calculated in water for the hydrated cations with TI and the

protocol 11(2 ϩ 20) ps. The cations were previously equilibrated

in a 403 TIP3P water box by a 20 ps MD run.

calix[4]arene (0.425 g, 1 mmol), K₂CO₃(1.382 g, 10 mmol)

and acetonitrile (100 ml) was stirred at room temp. for 4 h.

Then, the ditosylate 3 (0.643 g, 1 mmol) was added and the

mixture was allowed to reﬂux for 7 days. The same quantities of

K₂CO₃and ditosylate 3 were then added to the mixture which

was reﬂuxed for 7 additional days. The solvents were evapor-

ated on a rotatory evaporator and the residue was solubilized in

CH₂Cl₂and neutralized to pH 1 by addition of 1 M HCl. The

organic layer was dried over Na₂SO₄, ﬁltered and evaporated.

The crude product was chromatographed using 90:10 CH₂Cl₂–

acetone as eluent to give BC6B2 (0.850 g, 83%) as a white solid,

For ∆G₄in a vacuum, the cation M^ϩis mutated into the

cation N^ϩby dividing the calculation into 11 intermediate states

(windows) deﬁned by a coupling factor λ (λ = 1 initial state,

λ = 0 ﬁnal state). In a vacuum, each window was sampled by 52

ps of MD (2 ps equilibration and 50 ps data collection) sum-

marized in the text by 11(2 ϩ 50) ps, or 21(2 ϩ 50) ps in case

of convergence problems (mutations between Na^ϩand K^ϩ).

The value reported is the average value of the two mutations

M^ϩ→N^ϩand N^ϩ→M^ϩ. The starting point of each mutation

was a structure equilibrated after 100 ps of MD run at 300 K.

1

mp 142–143 ЊC. H-NMR δ 7.14–7.08 (m, 8H), 6.96–6.89 (m,

8H), 6.97–6.81 (m, 8H), 6.72 (d, J = 7.0 Hz, 8H), 6.58 (t, J = 7.0

Hz, 4H), 4.39 (s, 8H), 3.74 (s, 8H), 3.49 (m, 16H). ϪMS(FAB^ϩ);

m/z: 1021.5 (100%). Anal. Calc. for C₆₄H₆₀O₁₂: C, 75.28; H,

5.92. Found: C, 75.18; H, 5.88%.

Synthesis

X-Ray crystallography

Starting materials for synthesis. The solvents and all com-

pounds were commercial reagents and were used without fur-

ther puriﬁcation.

BC6B2ؒCH₃NO₂1.† Preparation of single crystals. Beautiful

colorless single crystals suitable for X-ray crystallography

were obtained by slow evaporation of a solution of BC6B2 in

nitromethane.

Crystal data. O₁₄N₁C₆₅H₆₃, M_r= 1082.23, crystallizes in the

monoclinic space group P2₁/n, a = 12.260(3), b = 16.998(3),

c = 27.624(7) Å, β = 102.50(2)Њ, U = 5620(4) Å³, Z = 4, D_c=

1.279 g cm^Ϫ3, µ = 0.838 cm^Ϫ1, F(000) = 2288.

Analytical procedures. The melting points (mps) were taken

on a Büchi 500 apparatus in capillaries sealed under nitrogen.

Silica gel columns were prepared with Kieselgel Merck (N^o

11567). Elemental analyses were performed at the Service de

Microanalyse of the Institut de Chimie de Strasbourg. The

¹H-NMR spectra were recorded at 200 MHz on a Bruker

SY200 spectrometer. The FAB mass spectra were obtained on a

VG-Analytical ZAB HF apparatus.

Data collection, structure determination and reﬁnement. A

parallelepipedic single crystal (dimensions 0.64 × 0.56 × 0.32

mm) was sealed in a glass capillary. The cell constants were

obtained from the least-squares reﬁnement of the setting angles

of 25 reﬂections in the range 8 < θ < 12Њ. The data were

collected with an Enraf-Nonius CAD4 diﬀractometer using

graphite monochromated MoKα radiation (0.71073 Å) in the

range 1 < θ < 20Њ, at room temperature, in the ω/θ scan mode.

2848 reﬂections with I > 3σ(I), out of 5243 unique reﬂections

measured, were used after Lorentz-polarization correction. The

intensity decay was not signiﬁcant and no absorption correc-

tion was done. The main part of the structure was solved by

direct methods with SHELXS-86,⁶³and the remaining atoms

were found from subsequent Fourier diﬀerences. Oxygen atoms

and the atoms of the solvent molecule were reﬁned anisotropic-

ally. Hydrogen atoms were included at calculated positions and

constrained to ride on their parent carbon atoms (C–H 0.95 Å,

isotropic displacement parameter B 6 Å²). Analytical scattering

factors for neutral atoms were corrected for ∆f Ј and ∆f Љ.

Reﬁnement of 401 parameters was done by full-matrix least-

squares on F and led to a ﬁnal conventional R1 value of 0.072

(R1_w= 0.091, w = 1/σ(F)²). The goodness of ﬁt was S = 3.7 and

2-(2-Hydroxyethoxy)phenol (1). A mixture of 1,2-dihydroxy-

benzene (33.03 g, 300 mmol), 2-chloroethanol (24.15 g, 300

mmol), K₂CO₃(20.75 g, 150 mmol) and acetonitrile (750 ml)

was reﬂuxed for 48 h. The mixture was allowed to cool to room

temp., ﬁltered and the ﬁltrate concentrated by evaporation

under reduced pressure. The residue was chromatographed on a

silica column using 80:20 CHCl₃–acetone as eluent to aﬀord 1

(18.50 g, 40%) as a white solid, mp 102–103 ЊC. ¹H-NMR

δ 6.99–6.78 (m, 4H), 4.16–4.11 (m, 2H), 4.03–3.98 (m, 2H), 2.79

(s large, 2H). Anal. Calc. for C₈H₁₀O₃: C, 62.33; H, 6.54. Found:

C, 62.21; H, 6.59%.

1,2-Bis(2-ethoxyphenyloxy)ethane (2). A mixture of 1 (7.71 g,

50 mmol), K₂CO₃(69.12 g, 500 mmol), 1,2-dibromoethane

(5.64 g, 30 mmol) and acetonitrile (750 ml) was reﬂuxed for

4 days. After cooling to room temp., solvents were removed

by evaporation under reduced pressure. The residue was mixed

with CH₂Cl₂and water and neutralized with 1 M HCl. The

organic layer was dried over Na₂SO₄, ﬁltered and evaporated on

a rotary evaporator. The crude product was chromatographed

using 85:15 CH₂Cl₂–acetone as eluent to give 2 (4.20 g, 50%) as

a white solid, mp 76–77 ЊC. ¹H-NMR δ 6.97 (s, 8H), 4.40–4.34

(m, 6H), 4.14–4.10 (m, 4H), 3.89–3.86 (m, 4H). Anal. Calc. for

C₁₈H₂₂O₆: C, 64.66; H, 6.63. Found: C, 64.48; H, 6.52%.

the highest and lowest residual densities 0.56 and Ϫ0.09 e Å^Ϫ3

All calculations were performed on a VAX 4000-200 computer

with the Enraf-Nonius MolEN system.⁶⁴The molecular draw-

ing was done with ORTEPII.⁶⁵

.

Cs₂(NO₃)₂BC6B2ؒCHCl₃ؒH₂O 2. Preparation of single crys-

tals. BC6B2 (0.2 mmol) was reacted with CsNO₃(large

excess: 2.7 mmol) in CHCl₃–CH₃CN (1:1, 10 ml) at 65 ЊC for

two days. The remaining CsNO₃was removed by ﬁltration

and the solution was allowed to slowly evaporate, leading

to beautiful colorless single crystals suitable for X-ray

crystallography.

Ditosylate of 1,2-bis(2-ethoxyphenyloxy)ethane (3). A mix-

ture of 2 (2.08 g, 6 mmol), toluene-p-sulfonyl chloride (2.35 g,

12 mmol) and CH₂Cl₂(70 ml) was cooled at 0 ЊC. Then, NEt₃

(2.34 g, 22 mmol) was added dropwise and the reaction mixture

was stirred for 48 h. It was then neutralized to pH 1 by addition

of 1 M HCl, dried over Na₂SO₄, ﬁltered and evaporated on a

rotary evaporator. The crude product was chromatographed

using 95:5 CH₂Cl₂–acetone as eluent to give 3 (2.42 g, 63%) as a

white solid, mp 95–96 ЊC. ¹H-NMR δ 7.77 (d, J = 8.5 Hz, 4H),

7.28 (d, J = 8.5 Hz, 4H), 6.97–6.81 (m, 8H), 4.31–4.27 (m, 8H),

4.20–4.16 (m, 4H), 2.40 (s, 6H). Anal. Calc. for C₃₂H₃₄O₁₀S₂: C,

59.80; H, 5.33. Found: C, 59.71; H, 5.27%.

Crystal data. Cs₂Cl₃O₁₉N₂C₆₅H₆₃, M_r= 1548.34, crystallizes

in the monoclinic space group P2₁/n, a = 13.800(1), b =

† Full crystallographic details for 1 and 2, excluding structure factor

tables, have been deposited at the Cambridge Crystallographic Data

Centre (CCDC). For details of the deposition scheme, see ‘Instructions

for Authors’, J. Chem. Soc., Perkin Trans. 2, available via the RSC web

page (http://ww w .rsc.org/authors). Any request to the CCDC for this

material should quote the full literature citation and the reference

number 188/151.

Calix[4]arene-bis(dibenzo)crown-6 (BC6B2). A mixture of

282

J. Chem. Soc., Perkin Trans. 2, 1999, 271–284

View Article Online

24.636(4), c = 18.735(1) Å, β = 102.20(1)Њ, U = 6226(2) Å³,

Z = 4, D_c= 1.652 g cm^Ϫ3, µ = 1.377 cm^Ϫ1, F(000) = 3120.

Suprapur) and CsI (Janssen Chimica). ∆H_xyvalues were reﬁned

using the program Sirko,⁷⁰the stability constants obtained by

spectrophotometry being taken as constant. The corresponding

entropies of complexation ∆S_xywere then derived from the

expression ∆G_xy= ∆H_xyϪ T∆S_xy.

Data collection, structure determination and reﬁnement. A

parallelepipedic single crystal (dimensions 0.25 × 0.20 × 0.20

mm) was sealed in a glass capillary. The cell constants were

determined and the data collected with a Nonius Kappa-CCD

diﬀractometer using graphite monochromated MoKα radi-

ation (0.71073 Å) in the range 1 < θ < 24.6Њ, at 173 K, with 1Њ

steps for the φ scan, 60 second frame intervals and a 28 mm

crystal to detector distance. The data were processed with the

HKL package (DENZO, XDisplayF, Scalepack).⁶⁶No absorp-

tion correction was done. 6474 unique reﬂections were used.

The main part of the structure was solved by direct methods

with SHELXS-86,⁶³and the remaining atoms were found from

subsequent Fourier diﬀerences. All non hydrogen atoms were

reﬁned anisotropically. Hydrogen atoms were included at calcu-

lated positions and constrained to ride on their parent carbon

atoms with a displacement factor equal to 1.2 times that of the

parent atom. Analytical scattering factors for neutral atoms

were corrected for ∆f Ј and ∆f Љ. Reﬁnement of 820 parameters

was done by full-matrix least-squares on F²and led to ﬁnal

conventional values R1 = 0.037 (all reﬂections, 0.030 for 5641

“observed” (I > 2σ(I)) ones) and wR2 = 0.107 (all reﬂections,

Association constants in THF

Spectrophotometric titrations in tetrahydrofuran (THF) have

been performed according to Smid’s procedure⁷²as described

earlier.⁴⁸Increasing amounts of ligand solution in THF (SDS,

Analytical Grade) were added to 1 ml of metal picrate in THF,

directly in the spectrophotometric cell. After each addition of

ligand the spectra were recorded between 300 and 500 nm on

a Shimadzu UV-2101-PC spectrophotometer using supracil

cuvettes of 0.5 cm path length and thermoregulated at 25 ЊC.

The experimental spectra were treated by the program Sirko⁷⁰

as follows: the molar absorptivities and the association con-

stants K_aof the species formed, corresponding to the following

equilibria: M^ϩPic^Ϫϩ L

MLPic, were ﬁrst reﬁned simul-

taneously from the absorbances at ten signiﬁcant wavelengths

where the greatest changes occurred. Then, the molar absorp-

tivities at several wavelengths, chosen every nm around the

expected maximum, were reﬁned in order to determine the

position of the maximum absorption of the complexes. The

association constants obtained from the ﬁrst reﬁnement were

taken as constant during this treatment.

2

0.088 for “observed” ones), with w = 1/[σ²(F_o) ϩ (0.1302P)²ϩ

71.6189P], where P = (F_o²ϩ 2F_c²)/3. The goodness of ﬁt was

S = 0.54 and the highest and lowest residual densities were 0.63

and Ϫ0.53 e Å^Ϫ3. All calculations were performed on a Silicon

Graphics Indy station with the SHELXTL system.⁶⁷The

molecular drawing was done with ORTEPII.⁶⁵

Acknowledgements

V. L. thanks COGEMA for ﬁnancial support. S. F. thanks the

French Government and SFERE (Société Française d’Expor-

tation des Ressources Educatives) for a grant.

Picrate extraction experiments

The percentages of alkali picrates extracted from water into

dichloromethane (E) have been determined at 20 ЊC according

to the Pedersen’s procedure.⁶⁸The experimental details and the

preparation of the metal picrates have already been reported.⁶⁹

References

1 Chemical Separations with Liquid Membranes, R. A Bartsch and

J. D. Way Eds., ACS Symposium Series 642, American Chemical

Society, 1996.

2 New Separation Chemistry Techniques for Radioactive Waste and

Other Speciﬁc Applications, L. Cecille, M. Casarci and L. Pietrelli

Eds., Elsevier Applied Science, London and New York, 1991.

3 Future Industrial Prospects of Membrane Processes, L. Cecille and

J.-C. Toussaint Eds., Elsevier Applied Science, London and New

York, 1989.

4 Supramolecular Chemistry, J.-M. Lehn Ed., VCH, Weinheim, 1995.

5 B. A. Moyer and Y. Sun, in Ion Exchange and Solvent Extraction,

J. A. Marinsky and Y. Marcus Eds., Marcel Dekker Inc., 1997, Vol.

13, Chap. 6, pp. 295–391.

6 M. Bourgeois, in Techniques de l’Ingénieur, Traité Mécanique et

Chaleur, B 3650.

7 T. J. Haverlock, P. V. Bonnesen, R. A. Sachleben and B. A. Moyer,

Radiochim. Acta, 1997, 76, 103

8 C. D. Gutsche, Calixarenes, Royal Society of Chemistry, Cam-

bridge, 1989.

9 Calixarenes: a Versatile Class of Macrocyclic Compounds, J. Vicens

and V. Böhmer Eds., Kluwer Academic Publishers, Dordrecht, 1991.

10 V. Böhmer, Angew. Chem., Int. Ed. Engl., 1995, 34, 713.

11 Z. Asfari, S. Wenger and J. Vicens, Supramol. Sci., 1994, 1, 103.

12 Z. Asfari, S. Wenger and J. Vicens, in Calixarenes’ 50^thAnniversary,

J. Vicens, Z. Asfari and J. M. Harrowﬁeld Eds., Kluwer Academic

Publishers, 1995, pp. 137-148. Reprinted from J. Inclusion Phenom.

Mol. Recognit. Chem., 1994, 19, 137.

Stability constant in acetonitrile

The stability constants, expressed as the concentration ratios

β_xy= [M_xL_y^xϩ]/[M^ϩ]^x[L]^yand corresponding to the overall

equilibrium: xM^ϩϩ yL

M_xL_yhave been determined in

xϩ

acetonitrile (SDS, purex for Analyses), BC6B2 not being sol-

uble enough in methanol. The measurements were performed

by UV-absorption spectrophotometry. The procedure was iden-

tical to that already described.¹⁸As the equilibria were reached

rapidly, titration of the ligand (C_L= 2–8 × 10^Ϫ5M) by the metal

ion solution was performed directly in the spectrophotometric

cell. The ﬁnal ratio R = C_M/C_Lof the concentration of metal to

the concentration of ligand was adjusted in order to obtain the

maximum spectral changes. In the case of very weak complex-

ation, as for Li^ϩand Na^ϩcations, values of R higher than 900

were used. The spectra were recorded between 250 and 300 nm,

at 25 ЊC, with a Shimadzu UV-2101-PC spectrophotometer

after each addition of metal salt. As far as possible, the ionic

strength was maintained at 0.01 M in all solutions by the use of

Et₄NClO₄. The metal salts were the following perchlorates:

NaClO₄(Fluka, purum), KClO₄(Prolabo, Normapur), RbClO₄

(Sigma) and caesium nitrate: CsNO₃(Merck). All these salts

were dried under vacuum for 12 h before use. The spectral

changes recorded have been interpreted by the program Sirko.⁷⁰

13 Z. Asfari, S. Wenger and J. Vicens, Pure Appl. Chem. 1995, 67, 1037.

14 Z. Asfari, M. Nierlich, P. Thuéry, V. Lamare, J.-F. Dozol, M. Leroy

and J. Vicens, Anal. Quím. Int. Ed., 1996, 92, 260.

15 C. Hill, J.-F. Dozol, V. Lamare, H. Rouquette, S. Eymard, B.

Tournois, J. Vicens, Z. Asfari, C. Bressot, R. Ungaro and A.

Casnati, in Calixarenes’ 50^thAnniversary, J. Vicens, Z. Asfari and

J. M. Harrowﬁeld Eds., Kluwer Academic Publishers, 1995, pp. 399–

408. Reprinted from J. Inclusion Phenom. Mol. Recognit. Chem.,

1994, 19, 399.

16 Z. Asfari, C. Bressot, J. Vicens, C. Hill, J.-F. Dozol, H. Rouquette,

S. Eymard, V. Lamare and B. Tournois, Anal. Chem., 1995, 67,

3133.

Thermodynamic parameters in acetonitrile

The enthalpies of complexation ∆H_xyhave been determined by

calorimetric titrations using a Tronac isoperibolic calorimeter.

The procedure has been reported in detail elsewhere.⁷¹Solu-

tions of the metal salts were titrated into solutions of the ligand

(2.5 × 10^Ϫ4M ≤ C_L≤ 10^Ϫ3M). For solubility reasons, the metal

iodides were used instead of the perchlorates used for spectro-

photometric measurements: KI (Prolabo, p.a.), RbI (Merck,

17 Z. Asfari, C. Bressot, J. Vicens, C. Hill, J.-F. Dozol, H. Rouquette,

S. Eymard, V. Lamare and B. Tournois, in Chemical Separations

J. Chem. Soc., Perkin Trans. 2, 1999, 271–284

283

View Article Online

with Liquid Membranes, R. A. Bartsch and J. D. Way Eds., ACS

Symposium Series 642, American Chemical Society, 1996, pp. 376–

390.

spectra of Cs^ϩ, Rb^ϩand K^ϩcomplexes of BC6N do not show any

modiﬁcation in the naphthalene region with respect to the free

BC6N.²³

18 F. Arnaud-Neu, Z. Asfari, B. Souley and J. Vicens, New J. Chem.,

1996, 20, 453.

19 A. D’Aprano, J. Vicens, Z. Asfari, M. Salomon and M. Iammarino,

J. Solution Chem., 1996, 25, 955.

47 R. M. Izatt, J. S. Bradshaw, S. A. Nielsen, J. D. Lamb, J. J.

Christensen and D. Sen, Chem. Rev., 1985, 85, 271.

48 F. Arnaud-Neu, Z. Asfari, B. Souley and J. Vicens, Anal. Quím.

Int. Ed., 1997, 93, 404.

20 P. Thuéry, M. Nierlich, C. Bressot, V. Lamare, J.-F. Dozol, Z. Asfari

and J. Vicens, J. Inclusion Phenom. Mol. Recognit. Chem., 1996, 23,

305.

21 Z. Asfari, C. Naumann, J. Vicens, M. Nierlich, P. Thuéry, C.

Bressot, V. Lamare and J.-F. Dozol, New J. Chem., 1996, 20, 1183.

22 P. Thuéry, M. Nierlich, Z. Asfari and J. Vicens, J. Inclusion Phenom.

Mol. Recognit. Chem., 1997, 27, 169.

23 P. Thuéry, M. Nierlich, J. C. Bryan, V. Lamare, J.-F. Dozol, Z. Asfari

and J. Vicens, J. Chem. Soc., Dalton Trans., 1997, 4191 (and

references therein).

24 P. Thuéry, M. Nierlich, V. Lamare, J.-F. Dozol, Z. Asfari and

J. Vicens, Supramol. Chem., 1997, 8, 319.

25 R. Ungaro, A. Casnati, F. Ugozzoli, A. Pochini, J.-F. Dozol, C. Hill

and H. Rouquette, Angew. Chem., Int. Ed. Engl., 1994, 33, 1506.

26 F. Ugozzoli, O. Ori, A. Casnati, A. Pochini, R. Ungaro and D. N.

Reinhoudt, Supramol. Chem., 1995, 5, 179.

27 A. Casnati, A. Pochini, R. Ungaro, F. Ugozzoli, F. Arnaud, S.

Fanni, M.-J. Schwing, R. J. M. Egberink, F. de Jong and D. N.

Reinhoudt, J. Am. Chem Soc., 1995, 117, 2767.

49 J.-F. Dozol, V. Lamare, C. Bressot, R. Ungaro, A. Casnati, J. Vicens

and Z. Asfari, Fr. Pat. 97 02490.

50 C. Sorel, PhD Thesis, Université Paris VI, France, 1996.

51 J.-F. Dozol, N. Simon, V. Lamare, H. Rouquette, S. Eymard,

B. Tournois and D. De Marc, Poster presented at 10^thSymposium

on Separation Science and Technology for Energy Applications,

Gatlinburg, Tennessee, October 20–24, 1997 (accepted in Sep. Sci.

Technol.).

52 B. F. Myasoedov, M. K. Chmutova, N. E. Kochetkova, O. E. Koiro,

G. A. Pribylova, N. P. Nesterova, T. Ya. Medved and M. I.

Kabachnik, Solvent Extr. Ion Exch., 1986, 4, 61.

53 R. Chiarizia and E. P. Horwitz, Solvent Extr. Ion Exch., 1992, 10,

101.

54 M. N. Litvina, M. K. Chmutova, B. F. Myasoedov and M. I.

Kabachnik, Radiochemistry, 1996, 38, 494.

55 D. A. Pearlman, D. A. Case, J. A. Caldwell, W. S. Ross, T. E.

Cheatham III, D. M. Ferguson, G. L. Seibel, U. C. Singh, P. Weiner

and P. A. Kollman, AMBER 4.1, University of California, San

Francisco, 1995.

28 F. Arnaud-Neu, N. Deutsch, S. Fanni, M. J. Schwing-Weill, A.

Casnati and R. Ungaro, Gazetta Chim. Ital., 1997, 127, 693.

29 G. Wipﬀ and M. Lauterbach, Supramol. Chem., 1995, 6, 187.

30 A.VarnekandG.Wipﬀ,J.Mol.Struct.(THEOCHEM),1996,363,67.

31 G. Wipﬀ, E. Engler, P. Guilbaud, M. Lauterbach, L. Troxler and

A. Varnek, New J. Chem., 1996, 20, 403.

32 M. Lauterbach and G. Wipﬀ, in Physical Supramolecular Chemistry,

NATO ASI series, L. Echegoyen and A. Kaifer Eds., Kluwer,

Dordrecht, 1996, pp. 1–38.

56 S. J. Weiner, P. A. Kollman, D. T. Nguyen and D. A. Case,

J. Comput. Chem., 1986, 7, 230.

57 Insight II, Biosym Technologies, 1993.

58 E. Engler, G. Wipﬀ, MDS. Université Louis Pasteur, Strasbourg,

France, 1992; E. Engler and G. Wipﬀ, in Crystallography of

Supramolecular Compounds, G. Tsoucaris Ed., Kluwer, Dordrecht,

1996, pp. 471–476.

59 P. A. Kollman, Chem. Rev., 1993, 93, 2395.

60 D. A. Pearlman, J. Comput. Chem., 1994, 15, 105.

61 T. P. Straatsma and H. J. C. Berendsen, J. Chem. Phys., 1988, 89,

5876.

33 A. Varnek and G. Wipﬀ, J. Comput. Chem., 1996, 17, 1520.

34 M. Lauterbach, E. Engler, N. Muzet, L. Troxler and G. Wipﬀ,

J. Phys. Chem. B, 1998, 102, 245.

62 Y. Marcus, Chem. Rev., 1988, 88, 1475.

35 M. Lauterbach, G. Wipﬀ, A. Mark and W. van Gunsteren, Gazetta

Chim. Ital., 1997, 127, 699.

36 C. Bressot, PhD Thesis, Université Louis Pasteur de Strasbourg,

France, 1995.

63 G. M. Sheldrick, SHELXS-86: Program for the Solution of Crystal

Structures, University of Göttingen, 1985.

64 MolEN: An Interactive Structure Solution Procedure, Enraf-Nonius,

Delft, 1990.

37 V. Lamare, C. Bressot, J.-F. Dozol, J. Vicens, Z. Asfari, R. Ungaro

and A. Casnati, Sep. Sci. Technol., 1997, 32, 175.

38 J. Aqvist, J. Phys. Chem., 1990, 94, 8021.

39 S. Miyamoto and P. A. Kollman, J. Am. Chem. Soc., 1992, 114,

3668.

40 T. Kowall and A. Geiger, J. Phys. Chem., 1994, 98, 6216.

41 P. D. J. Grootenhuis and P. A. Kollman, J. Am. Chem. Soc., 1989,

111, 2152.

65 C. K. Johnson, ORTEPII, Report ORNL-5138, Oak Ridge

National Laboratory, Tennessee, 1976.

66 Z. Otwinowski and W. Minor, Processing of X-ray Diffraction Data

Collected in Oscillation Mode, in Methods in Enzymology, Vol. 276:

Macromolecular Crystallography, Part A, pp. 307–326, C. W. Carter,

Jr. and R. M. Sweet Eds., Academic Press, 1997.

67 SHELXTL, version 5.03, Analytical X-ray Systems, Karlsruhe,

1997.

42 J. Marrone, D. S. Hartsough and K. M. Merz, J. Phys. Chem., 1994,

98, 1341.

43 V. Lamare, J.-F. Dozol, F. Ugozzoli, A. Casnati and R. Ungaro, Eur.

J. Org. Chem., 1998, 1559.

68 C. J. Pedersen, Fed. Proc. Fed. Am. Soc. Expl. Biol., 1968, 27, 1305.

69 F. Arnaud-Neu, M. J. Schwing-Weill, K. Ziat, S. Cremin, S. J. Harris

and M. A. McKervey, New J. Chem., 1991, 15, 33.

70 V. Vetrogon, N. G. Lukyanenko, M. J. Schwing-Weill and F.

Arnaud-Neu, Talanta, 1994, 41, 2105.

71 F. Arnaud-Neu, G. Barrett, S. Fanni, D. Marrs, W. McGregor,

M. A. McKervey, M. J. Schwing-Weill, V. Vetrogon and S. Wechsler,

J. Chem. Soc., Perkin Trans. 2, 1995, 453.

44 D. J. Cram, Angew. Chem., Int. Ed. Engl., 1998, 27, 1009.

45 P. Cieplak, D. A. Pearlman and P. A. Kollman, J. Chem. Phys., 1994,

101, 627.

1

46 H-NMR spectra of Cs^ϩand Rb^ϩcomplexes of BC6B2 show

important modiﬁcations in the region of the protons belonging to

the benzo substituents as compared to the free ligand, which may

be attributed to π–cation interactions. Such modiﬁcations are not

visible in the spectrum of the K^ϩcomplex. On the other hand, the

72 M. Bourgoin, K. H. Wong, J. Y. Hui and J. Smid, J. Am. Chem. Soc.,

1975, 97, 3462.

Paper 8/06714G

284

J. Chem. Soc., Perkin Trans. 2, 1999, 271–284

Article Doi

A new calix[4]arene-bis(crown ether) derivative displaying an improved caesium over sodium selectivity: Molecular dynamics and experimental investigation of alkali-metal ion complexation

DOI: 10.1039/a806714g

Source and publish data:

Authors:

Article abstract of DOI:10.1039/a806714g

Full text of DOI:10.1039/a806714g

Products guided by the article

R&D Labs maybe for 82264-97-9

Relevant to this article

Hot Product