Supramolecular pseudorotaxane polymers from complementary pairs of homoditopic molecules

Article abstract of DOI:10.1021/ja020900a

Self-assembly of supramolecular pseudorotaxane polymers from complementary homoditopic building blocks comprised of bis(dibenzo-24-crown-8) esters derived from the hydroxymethyl crown ether and aliphatic diacid chlorides (CxC, x = number of methylene unit

Full text of DOI:10.1021/ja020900a

Published on Web 03/01/2003
Supramolecular Pseudorotaxane Polymers from
Complementary Pairs of Homoditopic Molecules
Harry W. Gibson,* Nori Yamaguchi, and Jason W. Jones
Contribution from the Department of Chemistry, Virginia Tech, Blacksburg, Virginia 24061
Received July 1, 2002 ; E-mail: hwgibson@vt.edu
Abstract: Self-assembly of supramolecular pseudorotaxane polymers from complementary homoditopic
building blocks comprised of bis(dibenzo-24-crown-8) esters derived from the hydroxymethyl crown ether
and aliphatic diacid chlorides (CxC, x ) number of methylene units in the diacid segment) and 1,10-bis-
[p-(benzylammoniomethyl)phenoxy]alkane bis(hexafluorophosphate)s (AyA, y ) number of methylene units
in the linker) has been studied. 1H NMR spectroscopic studies of bis[(2-dibenzo-24-crown-8)methyl] sebacate
(C8C) with dibenzylammonium hexafluorophosphate (6) showed that the two binding sites of the ditopic
host are equivalent and independent (no positive or negative cooperativity). Likewise the binding sites in
1,10-bis[p-(benzylammoniomethyl)phenoxy]decane bis(hexafluorophosphate) (A10A) were shown to behave
independently with dibenzo-24-crown-8 (1a). Then using 1H NMR spectroscopy on dilute equimolar solutions
(<1 mM) of CxC and AyA association constants were estimated for the formation of the linear (lin-CxC•AyA)
and cyclic (cyc-CxC•AyA) dimers, thus enabling effective molarities to be estimated for the various systems.
Finally 1H NMR spectroscopy was used to semiquantitatively or qualitatively demonstrate the formation of
linear supramolecular polymers lin-[CxC•AyA]n in more concentrated solutions (up to 2.0 M) of the
complementary pairs of CxC and AyA. The sizes of the assemblies (n values) are not as great as the
dilute solution studies predict; this is attributed to the deleterious effect of ionic strength and exo complexation
at high concentrations. However, as expected from the dilute solution results, linear extension is indeed
favored with the longer building blocks, meaning that “monomer” end-to-end distance is a key factor in
reducing the amount of cyclic species that form. Viscosity experiments clearly demonstrate the formation
of large noncovalent polymers lin-[CxC•AyA]n in concentrated solutions. Cohesive film and fiber formation
also indicate that supramolecular polymers of sufficient size to enable entanglement self-assemble in these
solutions.
Introduction
fibers. Such interactions permit reversibility at the molecular
level, affording thermodynamically controlled polymeric su-
Chemists are extending the concept of self-organizing non-
covalent interactions from supramolecular science1 into the field
of material science to produce novel structures with functional
properties. The formation of noncovalently bound crystalline,
“polymeric” solids has a long history,2 but much progress has
been made recently.3
prastructures. This is advantageous from the point of view of
producing polymers with potential for commercial use, since
kinetically induced defects in conventional covalently bonded
polymers are irreversible. Indeed, it has been shown that strong
associations between self-organizing building blocks promote
the construction of well-defined supramolecular polymeric
materials with properties comparable to covalent polymers.4
The self-assembly and study of “noncovalent polymers” in
solution is more recent; the aims are to control and engineer
linear and network supramolecular polymers using the liquid
phase to facilitate the process and ultimately produce films or
(3) Inter alia: Desiraju, G. R. Acc. Chem. Res. 1996, 29, 441-449. Karle, I.
L.; Ranganathan, D.; Haridas, V. J. Am. Chem. Soc. 1997, 119, 2777-
2783. Brunet, P.; Simard, M.; Wuest, J. D. J. Am. Chem. Soc. 1997, 119,
9, 2737-2738. Hajek, F.; Hosseini, M. W.; Graf, E.; De Cian, A.; Fischer,
J. Angew. Chem., Int. Ed. Engl. 1997, 36, 1760-1762. Desiraju, G. R.
Chem. ReV. 1998, 98, 1375-1406. Batten, S. R.; Robson, R. Angew. Chem.,
Int. Ed. 1998, 37, 1461-1494. Marsman, A. W.; Leussink, E. D.; Zwikker,
J. W.; Jenneskens, L. W. Chem. Mater. 1999, 11, 1484-1491. Ozin, G.
A. Chem. Commun. 2000, 419-432. Moulton, B.; Zaworotko, M. J. Chem.
ReV. 2001, 101, 1629-1658. Holman, K. T.; Pivovar, A. M.; Swift, J. A.;
Ward; M. D. Acc. Chem. Res. 2001, 34, 107-118. Ward, M. D. Chem.
ReV. 2001, 101, 1697-1726. Nguyen, T. L.; Fowler, F. W.; Lauher, J. W.
J. Am. Chem. Soc. 2001, 123, 11057-11064.
(4) H-bonded systems: Lehn, J.-M. Makromol. Chem., Macromol. Symp. 1993,
69, 1-17. Rebek, J., Jr. Chem. Commun. 2000, 637-643. Kato, T.;
Mizoshita, N.; Kanie, K. Macromol. Rapid Commun.2001, 22, 797-814.
Meijer E. W.; Brunsveld, L.; Folmer, B. J. B.; Meijer, E. W.; Sijbesma, R.
P. Chem. ReV. 2001, 101, 4071-4098. Berl, V.; Schmutz, M.; Krische,
M. J.; Khoury, R. G.; Lehn, J.-M. Chem. Eur. J. 2002, 8, 1227-1244.
Metal-induced systems: Schubert, U. S.; Hien, O.; Eschbaumer, C.
Macromol. Rapid Commun. 2000, 21, 1156-1161.
(1) Vo¨gtle, F. Supramolecular Chemistry; John Wiley and Sons: New York,
1991. Cram, D. J.; Cram, J. M. Container Molecules and Their Guests;
Royal Society of Chemistry, Cambridge, UK, 1994. Lehn, J.-M. Supramo-
lecular Chemistry; VCH Publishers: New York, 1995. ComprehensiVe
Supramolecular Chemistry; Atwood, J. L., Davies, J. E. D., MacNicol, D.
D., Vo¨gtle, F., exec. Eds.; Pergamon Press: New York, 1996. Moore, J.
S. Curr. Opin. Colloid Interfac. Sci. 1999, 4, 108-116. Caulder, D. L.;
Raymond, K. N. Acc. Chem. Res. 1999, 32, 975-982. Rowan, A. E.;
Elemans, J. A. A. W.; Nolte, R. J. M. Acc. Chem. Res. 1999, 32, 995-
1006. Leininger, S.; Olenyuk, B.; Stang, P. J. Chem. ReV. 2000, 100, 853-
908. Lindoy, L. F.; Atkinson, I. M. Self-Assembly in Supramolecular
Systems; Royal Society of Chemistry: Cambridge, UK, 2000. Steed, J.
W.; Atwood, J. L. Supramolecular Chemistry; J. Wiley and Sons: New
York, 2000. Holliday, B. J.; Mirkin, C. A. Angew. Chem., Int. Ed. 2001,
40, 2022-2043.
(2) Schmidt, G. J. Pure Appl. Chem. 1971, 27, 647-678.
9
3522
J. AM. CHEM. SOC. 2003, 125, 3522-3533
10.1021/ja020900a CCC: $25.00 © 2003 American Chemical Society

Supramolecular Pseudorotaxane Polymers
A R T I C L E S
Scheme 1. Syntheses of Homoditopic Hosts, Bis(dibenzo-24-crown-8) Esters C8C and C2C, and Bis(m-phenylene)-26-crown-8 Ester
MC8MC
Results and Discussion
Pseudorotaxanes, in which linear molecules are threaded
through cyclic species, have been intensely investigated over
the past decade.5 There have been several reports of attempts
to use this structural motif to self-assemble noncovalent
polymers,6 including preliminary accounts from this laboratory.7
In one of our efforts we utilized a relatively simple system with
dibenzylammonium hexafluorophosphate guest and dibenzo-
24-crown-8 (DB24C8) host units in a pair of homoditopic
building blocks.7a For the model monotopic building blocks the
association constant for pseudorotaxane formation is reported
to be desirably high (Ka ) 2.7 × 104 M-1 in chloroform-d at
25 °C).8 Association of homoditopic molecules containing such
complementary units spontaneously leads to reversible chain
extension in 1:1 stoichiometric solutions, forming linear su-
pramolecular polymers based on pseudorotaxane formation, as
we describe in this work.
Our approach seeks to incorporate flexible units into both
building blocks for two reasons: (1) to increase solubility,
particularly in the nonpolar solvents required for these com-
plexations, enabling the necessary high concentrations to be
achieved and (2) to produce amorphous or semicrystalline
supramolecular polymers having better mechanical properties
than fully crystalline solids.
I. Synthesis of Building Blocks. The synthetic methodologies
employed for the homoditopic host molecules are depicted in
Scheme 1. The primary alcohols 27a were esterified to afford
the homoditopic hosts, bis(crown ether)s C2C, C8C,7a and
MC8MC.7a
Syntheses of homoditopic guests 1,10-bis[p-(benzylammo-
niomethyl)phenoxy]alkane bis(hexafluorophosphate)s, diam-
monium salts (AyA), are shown in Scheme 2. The PF6 salts
-
showed improved solubility in organic solvents such as acetone
and acetonitrile compared to Cl- salts. The lack of solubility
of the salts A4A, A10A, and A22A in CDCl3 precluded use of
this solvent for complexation studies; however, all the precursors
are soluble in acetone-d6 and 1:1 acetone-d6:CDCl3.
II. Model Complexation Studies. Since both the host and
guest species are ditopic, the possibility exists in each case that
the binding is statistical, positively cooperative, or anticoop-
erative.9 In these model studies the binding of complementary
monotopic complementary species was examined to determine
which situation obtained.
II. a. Ditopic Guest A10A and Monotopic Host Dibenzo-
24-crown-8 (1a). The C10-linked diammonium salt A10A was
examined with DB24C8 (1a) at 22 °C in acetone-d6 by 1H NMR
spectroscopy. The concentration of A10A was held constant at
10 mM, and the crown concentration was systematically varied.
The extent of complexation, p, of the ammonium moieties to
form pseudorotaxanes was determined by integration of (1) the
NCH2 signals for complexed (4.77 and 4.67 ppm) versus
uncomplexed (4.60 and 4.58 ppm) species and (2) the signal
for Hyc in the complex (d, 6.70 ppm) versus Hyu in the
uncomplexed state (d, 6.99 ppm) observed under slow ex-
change.10 The results are shown in Figure 1a in the form of a
Scatchard plot.9 The linear nature of the plot indicates that the
two binding sites of A10A are independent of each other and
that they behave in a noncooperative mode.9 The association
constants11 K1 and K2 (see Scheme 3a) for formation of the [2]-
and [3]-pseudorotaxanes, therefore, will be in the ratio 2:0.5,
based on statistics.9 The negative of the slope of the Scatchard
plot is the average of K1 and K2 and the intercept is Kave. Taking
the average of the slope and intercept values [1.4 ((0.1) ×
(5) For reviews of pseudorotaxanes and rotaxanes, see: Amabilino, D. B.;
Stoddart, J. F. Chem. ReV. 1995, 95, 2725-2828. Gibson, H. W. in Large
Ring Molecules; Semlyen, J. A., Ed.; John Wiley and Sons: New York,
1996; Chapter 6, pp 191-262. Nepogodiev, S. A.; Stoddart, J. F. Chem.
ReV. 1998, 98, 1959-1976. Raymo, F. M.; Stoddart, J. F. Chem. ReV. 1999,
99, 1643-1664. Sauvage, J.-P., Dietrich-Buchecker, C. O., Eds. Molecular
Catenanes, Rotaxanes and Knots; Wiley-VCH: Weinheim, 1999. Hubin,
T. J.; Busch, D. H. Coord. Chem. ReV. 2000, 200-202, 5-52. Cantrill, S.
J.; Pease, A. R.; Stoddart, J. F. J. Chem. Soc., Daltons Trans. 2000, 3715-
3734. Mahan, E.; Gibson, H. W. in Cyclic Polymers, 2nd ed.; Semlyen, J.
A., Ed.; Kluwer Publishers: Dordrecht, 2000; pp 415-560.
(6) Hirotsu, K.; Higuchi, T.; Fujita, K.; Ueda, T.; Shinoda, A.; Imoto, T.;
Tabushi, I. J. Org. Chem. 1982, 47, 1143-1144. Alvarez-Parrilla, E.;
Cabrer, P. R.; Al-Soufi, W.; Meijide, F.; Nunez, E. R.; Tato, J. V. Angew.
Chem., Int. Ed. 2000, 39, 2856-2858. Sandier, A.; Brown, W.; Mays, H.;
Amiel, C. Langmuir 2000, 16, 1634-1642. Hoshino, T.; Miyauchi, M.;
Kawaguchi, Y.; Yamaguchi, H.; Harada, A. J. Am. Chem. Soc. 2000, 122,
9876-9877. Cantrill, S. J.; Youn, G. J.; Stoddart, J. F. J. Org. Chem. 2001,
66, 6857-6872.
(7) (a) Yamaguchi, N.; Gibson, H. W. Angew. Chem., Int. Ed. 1999, 38, 143-
147. (b) Yamaguchi, N.; Gibson, H. W. Chem. Commun.1999, 789-790.
(c) Yamaguchi, N.; Nagvekar, D. S.; Gibson, H. W. Angew. Chem., Int.
Ed. 1998, 37, 2361-2364.
(8) Ashton, P. R.; Campbell, P. J.; Chrystal, E. J. T.; Glink, P. T.; Menzer, S.;
Philp, D.; Spencer, N.; Stoddart, J. F.; Tasker, P. A.; Williams, D. J. Angew.
Chem. Int. Ed. Engl. 1995, 34, 1865-1869. Ashton, P. R.; Chrystal, E. J.
T.; Glink, P. T.; Menzer, S.; Schiavo, C.; Spencer, N.; Stoddart, J. F.;
Tasker, P. A.; White, A. J. P.; Williams, D. J. Chem. Eur. J. 1996, 2, 709-
728. Fyfe, M. C. T.; Stoddart, J. F. AdV. Supramol. Chem. 1999, 5, 1-53.
Chang, T.; Heiss, A. M.; Cantrill, S. J.; Fyfe, M. C. T.; Pease, A. R.; Rowan,
S. J.; Stoddart, J. F.; Williams Org. Lett. 2000, 2, 2943-2946. Chang, T.;
Heiss, A. M.; Cantrill, S. J.; Fyfe, M. C. T.; Pease, A. R.; Rowan, S. J.;
Stoddart, J. F.; White, A. J. P.; Williams, D. J. Org. Lett. 2000, 2, 2947-
2950. Cantrill, S. J.; Pease, A. R.; Stoddart, J. F. J. Chem. Soc., Dalton
Trans. 2000, 3715-3734. Amirsakis, D. G.; Garcia-Garibay, M. A.; Rowan,
S. J.; Stoddart, J. F.; White, A. J. P.; Williams, D. J. Angew. Chem., Int.
Ed. 2001, 40, 4256-4261.
(9) Freifelder, D. M. Physical Biochemistry; W. H. Freeman and Co.: New
York, 1982; pp 659-660. Marshall, A. G. Biophysical Chemistry; J. Wiley
and Sons: New York, 1978; pp 70-77. Connors, K. A. Binding Constants;
J. Wiley and Sons: New York, 1987; pp 78-86.
(10) See Supporting Information.
(11) Association constants reported here are based on Ka ) [pseudorotaxane]/
([crown]o - [pseudorotaxane])([ammonium salt]o - [pseudorotaxane]); that
is, they are concentration-based and assume that all nonpseudorotaxane
species are “free”. This is the usual method of calculating such association
constants; see ref 5.
9
J. AM. CHEM. SOC. VOL. 125, NO. 12, 2003 3523

A R T I C L E S
Gibson et al.
Scheme 2. Syntheses of Homoditopic Guests: Diammonium Salts AyA, y ) Number of Methylene or Ethyleneoxy (EO) Units in the
Spacer Ra
a Reaction conditions: a. K2CO3/DMF/110 °C. b. C6H5CH2NH2/toluene/110 °C. c. NaBH4/MeOH, 25 °C. d. 1) HCl, 2) NH4PF6.
Scheme 3. Sequential Formation of [2]- and [3]Pseudorotaxanes
from (a) (left) a Homoditopic Guest and a Monotopic Host and (b)
(right) a Homoditopic Host and a Monotopic Guest
uncomplexed (∼4.4 ppm) states and (2) Hyc complexed (d, 6.63
ppm) versus Hxu uncomplexed (m, 7.46 ppm). Again the linear
Scatchard plot (Figure 1b) confirmed the independent, nonco-
operative nature of the two complexation sites of A10A. In this
solvent system from Kave [4.7 ((0.2) × 102 M-1], K1 ) 7.5
((0.3) × 102 M-1 and K2 ) 1.9 ((0.1) × 102 M-1. The average
values from the Scatchard plots may be compared with
association constants11 of 2.0 ((0.1) × 102 M-1 and 4.6
((0.1) × 102 M-1 determined by us for the monotopic system
dibenzylammonium hexafluorophosphate (6) (10 mM) /DB24C8
(1a) (concentration varied from 10 to 30 mM) in acetone-d6
and acetone-d6:chloroform-d (1:1, v:v), respectively. The agree-
Figure 1. Scatchard plots derived from 1H NMR data for complexation of
ment for the two systems in the mixed solvent is within
homoditopic guest A1A0 (total concentration 10 mM) with monotopic host
experimental error, as expected9 if the ammonium sites in A10A
dibenzo-24-crown-8 (1a) (total concentration varied from 10 mM to 0.10
are essentially equivalent to that in dibenzylammonium hexaflu-
orophosphate (6).
M): (a) (top) in acetone-d6 and (b) (bottom) in acetone-d6:choroform-d
(1:1, v:v). p ) fraction of ammonium moieties bound. p was determined
from integration of the signals for uncomplexed and complexed NCH2 at
4.68/4.77 and 4.56/4.59 ppm in (a), and 4.40 and 4.54/4.64 ppm in (b),
respectively. These values in (b) were confirmed using the signal for
complexed Hy at 6.62 ppm versus the signal for uncomplexed Hz at 7.46
ppm. [1a] was determined from the percentage of uncomplexed A10A. Error
bars in p: (0.03 absolute; error bars in p/[DB24C8]: (15% relative.
102 M-1] from the Scatchard plot, K1 was determined to be 2.2
((0.1) × 102 M-1 and K2 ) 55 ( 1 M-1. This system was
similarly examined in acetone-d6/chloroform-d (1/1, v/v) using
(1) NCH2 signals in the complexed (4.54 and 4.64 ppm) and
II. b. Ditopic Guest AEO35A and Monotopic Dibenzo-
24-crown-8 (1a). The H NMR spectra of a series of chloro-
form-d solutions of AEO35A and DB24C8 (1a) (10/10-10/40
1
9
3524 J. AM. CHEM. SOC. VOL. 125, NO. 12, 2003

Supramolecular Pseudorotaxane Polymers
A R T I C L E S
Diammonium salt A10A has an electron-donating alkoxy group
para to the NCH2 moiety; this is expected to decrease the acidity
of A10A relative to dibenzylammonium hexafluorophosphate
(6). On the other hand bis(crown ether) C8C possesses a CH2-
OOC moiety that may increase its basicity relative to dibenzo-
24-crown-8 (1a). [No Hammett constants could be found for
this substituent.] Thus, the pair C8C/6 consists of the more
active host and guest species relative to both 1a/A10A and 1a/
6.
III. Complexation Studies in Dilute Solutions: Depen-
dence of Cyclization on Spacer Length. The formation of
cyclic species14 is a competitive pathway that detracts from the
formation of high molar mass linear supermolecules. Therefore,
we next investigated the relative stabilities of cyclic and linear
dimeric complexes cyc-CxC•AyA and lin-CxC•AyA, respec-
tively, (Scheme 4) based on homoditopic guest molecules A4A
through A22A and host C8C.
1
To identify and assign H NMR signals for these species, a
series of 10 mM solutions of A10A was prepared using
nonstoichiometric amounts of C8C (from 20 to 80 mM), and
the H NMR spectra were recorded (Figure 3). Increasing the
Figure 2. Scatchard plot derived from 1H NMR data for complexation of
homoditopic host C8C (total concentration 2.46 mM) with monotopic guest
dibenzylammonium hexafluorophosphate (6) (total concentration varied from
1.20 to 15.0 mM) in acetone-d6:choroform-d (1:1, v:v). p ) fraction of
crown ether moieties bound. p was determined from integration of the signals
for uncomplexed and complexed γ-protons at 3.73 and 3.49 ppm,
respectively; [6] was determined by integration of the signals for uncom-
plexed and complexed NCH2 signals at 4.46 and 4.65 ppm, respectively.
Error bars in p: (0.03 absolute; error bars in p/[DB24C8]: (15% relative.
1
concentration of C8C with respect to A10A shifts the com-
plexation equilibrium toward the 1:2 complex lin-C8C•-
A10A•C8C, analogous to the [3]pseudorotaxane of Scheme 3a,
emulating the environment of the internal, i.e., complexed,
benzylic N-methylene units in linear dimer lin-C8C•A10A and
extended lin-[C8C•A10A]n. Since the signals at 4.53 and 4.63
ppm became more intense as [C8C] increased while the signals
at 4.45 and 4.74 ppm became less intense, the former (inner)
signals correspond to the linearly linked chains lin-[C8C•A10A]n,
lin-[C8C•A10A]n•C8C and lin-[A10A•C8C]n•A10A), and the
latter (outer) signals, to the cyclic dimer cyc-C8C•A10A. The
integral ratio of the outer to the inner signals decreased from
3.4 at 10/10 mM to 0.29 at 10/80 mM. The chemical shifts of
the complexed NCH2 signals for the linearly linked species lin-
C8C•A10A and lin-[C8C•A10A]n are consistent with the
positions of analogous signals in the model [2]pseudorotaxane
from DB24C8 (1a) and ditopic guest A10A, 4.54 and 4.64 ppm,
as described above. Signals in the aromatic region were used
for corroboration. The signal for Hyu of the diammonium
component of the cyclic dimer cyc-C8C•A10A appears upfield
(6.49 ppm) relative to complexed Hyc of linear species lin-
[A10A•C8C]n•A10A (6.64 ppm); note that the latter is es-
sentially the same position as analogous signals in 1a•A10A
(6.63 ppm, see above). Likewise complexed Hxc of cyclic dimer
cyc-C8C•A10A at 7.00 ppm (a doublet) was distinct over most
of the concentration range examined.
mM at 22 °C) were recorded to study the binding affinity of
AEO35A. From the linear Scatchard plot the average apparent
association constant11 for pseudorotaxane formation (Kave
)
was determined to be 20 ( 3 M-1, yielding K1 ) 32 ( 5 M-1
and K2 ) 8 ( 1 M-1. The 1000-fold lower apparent Ka value
obtained relative to the model system composed of dibenzy-
lammonium hexafluorophosphate (6) and DB24C8 (1a) [2.7 ×
104 M-1 in CDCl3 at 25 °C]8 is undoubtedly a consequence of
competitive intra- and intermolecular hydrogen bonding between
the 34 aliphatic oxygen atoms of the PEO linker segment and
ammonium salt moieties of AEO35A, reducing the true
concentration of the latter species in the free state. The same
effect was noted for solutions in CDCl3:acetone-d6 (1:1, v:v)
[K1 ) 40 ( 8 M-1, K2 ) 10 ( 2 M-1 at 22°C].11
II. c. Ditopic Host C8C with Monotopic Guest Dibenzy-
lammonium Hexafluorophosphate (6). The interaction of C8C
with dibenzylammonium hexafluorophosphate (6) to form [2]-
and [3]pseudorotaxanes (Scheme 3b) was also studied. The
concentration of C8C was maintained at 2.46 mM and that of
6 was varied from 1 to 15 mM in acetone-d6/chloroform-d (1/
1, v/v). Extents of complexation were determined by 1H NMR
by comparison of integrals for complexed (3.48 ppm) vs
uncomplexed (3.72 ppm) γ-protons.10 As can be seen in Figure
2 the Scatchard plot is linear, meaning that the two host sites
To evaluate quantitatively the equilibria leading to cyc-
CxC•AyA and lin-CxC•AyA it is necessary to use concentra-
tions that do not allow the formation of higher linear oligomers.
9
in C8C act independently and equivalently. From the values
(12) Substituent effects for secondary ammonium salts in complexation with
DB24C8 follow the Hammett equation: Ashton, P. R.; Fyfe, M. C. T.;
Hickingbottom, S. K.; Stoddart, J. F.; White, A. J. P.; Williams, D. J. J.
Chem. Soc., Perkin Trans. 2 1998, 2117-2128. We have shown that the
electron-withdrawing carbomethoxy substituent on the crown ether in the
pair 6/7a causes a lower Ka than observed with 1a/6 (ref 13).
(13) Gibson, H. W.; Yamaguchi, N.; Hamilton, L.; Jones, J. W. J. Am. Chem.
Soc. 2002, 124, 4653-4665.
(14) Cyclic dimer formation in other ditopic pseudorotaxane systems: Ashton,
P. R.; Baxter, I.; Cantrill, S. J.; Fyfe, M. C. T.; Glink, P. T.; Stoddart, J.
F.; White, A. J. P.; Williams, D. J. Angew. Chem., Int. Ed. 1998, 37, 1294-
1297. Cyclic dimers in quadruple hydrogen bonded systems: So¨ntjens, S.
H. M.; Sijbesma, R. P.; van Genderen, M. H. P.; Meijer, E. W.
Macromolecules 2001, 34, 3815-3818.
of the slope and intercept the average apparent association
constant11 for pseudorotaxane formation (Kave) was determined
to be 7.4 ((0.6) × 102 M-1, yielding K1 ) 1.2 ((0.1) × 103
M-1 and K2 ) 3.0 ((0.2) × 102 M-1. These values are higher
than those for the model system consisting of dibenzo-24-
crown-8 (1a) and homoditopic guest A10A [K1 ) 7.5 ((0.3)
× 102 M-1 and K2 ) 1.9 ((0.1) × 102 M-1] and Kave is greater
than Ka for 1a•6 [4.6 ((0.1) × 102 M-1] under similar
conditions. The differences are attributed to substituent effects.12
9
J. AM. CHEM. SOC. VOL. 125, NO. 12, 2003 3525

A R T I C L E S
Gibson et al.
Figure 3. 1H NMR spectra of solutions of C8C and A10A at (a) 10/0, (b) 0/10, (c) 10/10, (d) 20/10, (e) 40/10, (f) 80/10 mM/mM (400 MHz, acetone-
d6:chloroform-d (1:1, v:v), 22 °C). The three sets of NCH2 signals are for (1) uncomplexed moieties, i.e., end groups, (denoted by “u”) in A10A, lin-
[A10A•C8C]n•A10A and lin-[C8C•A10A]n, (2) complexed moieties (denoted by “l”) in lin-[C8C•A10A]n, lin-[A10A•C8C]n•A10A and lin-
[C8C•A10A]n•C8C, and 3) complexed moieties in cyc-C8C•A10A (denoted by “c”). As the proportion of C8C increases statistics predict that
lin-[C8C•A4A]n•C8C, n ) 1 will become the predominant species.
Scheme 4. Cartoon Illustration of Construction of Linear and Cyclic Dimers lin-CxC•AyA and cyc-CxC•AyA, and Supramolecular
Assemblies lin-[CxC•AyA]n, lin-[CxC•AyA]n•CxC and lin-[AyA•CxC]n•AyA
Therefore, the 1H NMR spectra of quite dilute (0.2 and 0.4 mM)
equimolar solutions of C8C individually with A4A through
A22A were examined. These revealed four sets of N-CH2
signals (Figure 4a-c) corresponding to (1) uncomplexed
moieties of the ditopic guest molecule AyA (∼4.3 ppm), (2)
complexed moieties in cyclic dimer cyc-CxC•AyA (∼4.45 and
4.75 ppm), (3) complexed (∼4.55 and 4.65 ppm) and (4)
uncomplexed (∼4.4 ppm) moieties in the linear dimer lin-
CxC•AyA, respectively. Figure 4d for the model system from
DB24C8 (1a) and A10A displays signals for the benzylic
protons of the ammonium salt units in A10A (totally uncom-
plexed, u) and the [2]pseudorotaxane lin-C8C•A10A (com-
9
3526 J. AM. CHEM. SOC. VOL. 125, NO. 12, 2003

Supramolecular Pseudorotaxane Polymers
A R T I C L E S
tance of the increasing penalty for cyclization as the end-to-
end distance increases.16,17As we anticipated, Kcyclic for
C8C•A22A was reduced, 4-fold, compared to that for C8C•A4A.
Simultaneously Klinear increased 4-fold.
An important parameter is the effective molarity (EM),16,17
which is the concentration at which extension of linear dimer
lin-CxC•AyA is as likely as cyclization to cyc-CxC•AyA; above
this concentration linear self-assembly of trimers and higher
oligomeric assemblies is favored. As indicated in Table 1 the
EM value for C8C/A22A (0.060 mM) clearly stands out,
showing a 16-fold improvement with respect to that of C8C/
A4A (0.95 mM). These results indicate that the end-to-end
distance of AxA plays a key role in determining the outcome
of the complexation process. We expected this effect would
enhance construction of supramolecular polymers lin-[CxC•AyA]n
using longer building blocks.
We also examined the self-assembly of the shorter bis(crown
ether) host C2C with the C22 linked diammonium guest A22A.
Klinear is higher for this system than for any of the other pairs
(Table 1). The total number of methylene spacer units is 24 for
C2C/A22A, and in terms of Kcyclic it falls between C8C/A10A
(18 CH2) and C8C/A22A (30 CH2). The observed EM is
essentially the same for C2C/A22A and C8C/A22A.
1
Figure 4. Partial H NMR spectra (NCH2 region) of equimolar solutions
of (a) C8C and A4A, (b) C8C and A10A, and (c) C8C and A22A (0.40
mM each), and (d) a solution of DB24C8 (1a) and A10A (0.80/0.40 mM)
at 22 °C (400 MHz, acetone-d6:chloroform-d, 1:1, v:v). The four pairs of
NCH2 signals in (a-c) are for (1) uncomplexed moieties in AyA, (2)
uncomplexed moieties (end groups) of linear dimers lin-C8C•AyA, denoted
by “lu”, (3) complexed moieties in lin-C8C•AyA, (denoted by “lc”) and
complexed moieties in cyc-C8C•AyA (denoted by “cc”). In (d) the upfield
pair of signals is for the NCH2 protons of A10A; the next downfield pair
of signals is for the uncomplexed NCH2 protons of the [2]pseudorotaxane,
denoted by “u”, and the most downfield pair derives from the complexed
NCH2 protons of the pseudorotaxane, denoted by “c”.
The NMR spectra of these dilute solutions were recorded over
a temperature range from 285 to 313 K.18 By use of van’t Hoff
plots ∆H values were evaluated over this temperature range.
These results are summarized in Table 1. ∆Hlinear values for
complexes with C8C become more negative, while ∆Hcyclic
values become less negative as the spacer length increases (Table
1, Figure 5b).
We interpret these results as being the result of an enthalpic
penalty for cyclization in addition to the expected entropic
penalty. Recent experimental19 and theoretical20 results for
systems which contain semirigid linkages provide precedence
for such observations. These workers found that for semirigid
systems the enthalpic penalty for formation of a ring grew with
the length of the species. In the present case rigidity is brought
about by the dibenzylammonium moieties themselves and the
fact that the charged sites Coulombically repel each other.
∆Hcyclic represents the difference in enthalpy between the linear
and cyclic dimers. If it is assumed that the enthalpic stabilization
from forming the cyclic species from the linear is equivalent to
that in forming the linear species from its components, as the
model systems indicate for forming [2]- and [3]pseudorotaxanes,
then ∆Hcyclic represents the difference between that value and
the enthalpic penalty. In other words the enthalpic penalty,
plexed end ) c, and uncomplexed end ) u). An expanded
version of the spectrum in Figure 4a in the region of 4.45-
4.75 ppm is shown in Figure 4a′. The 1H NMR spectra of three
sets of each pair (0.20 and 0.40 mM each) were recorded with
at least 50 min of acquisition time and integrated using a
deconvolution technique.
The ratios of each uncomplexed/complexed pair of signals
for lin-CxC•AyA was determined to be 1:1 ((5%), confirming
that they arise from this species and not from higher oligomers
which would yield ratios of <1:1. These spectroscopic observa-
tions allowed us to conclude that only lin-CxC•AyA and cyc-
CxC•AyA exist; larger cyclic15 and linear oligomers lin-
[CxC•AyA]n, n > 1 are not present in detectable amounts in
these dilute solutions. Since the concentrations of each species
(CxC, AyA, lin-CxC•AyA, and cyc-CxC•AyA) at equilibrium
are known, one can estimate the association constants Klinear
and Kcyclic (see Scheme 4).11 Klinear (Table 1) varies systemati-
cally, increasing as the length of the aliphatic spacer increases
(from A4A to A22A). Similarly, Kcyclic varies systematically,
decreasing as the length of the aliphatic spacer increases (from
A4A to A22A). With increasing spacer length y, ∆Glinear
becomes more negative and ∆Gcyclic becomes more positive
(Figure 5a). These observations are consistent with the impor-
(16) Early on in the macromolecular field, because of the importance of
cyclization as a competitive side reaction for step-growth polymerizations,
Stockmayer (Jacobson, H.; Stockmayer, W. H. J. Chem. Phys. 1950, 18,
1600-1606) devised a treatment relating the concentration at which intra-
and intermolecular reactions are equally probable, ceq, to the mean squared
end-to-end distance r2 : ceqR <r2 3/2. This relationship has been verified
in a number of experimental studies, e.g., Heath, R. E.; Wood, B. R.;
Semlyen, J. A. Polymer 2000, 1487-1495 and references therein. This
means that the tendency for cyclization decreases with increasing ring size.
(17) Kirby, A. J. AdV. Phys. Org. Chem. 1980, 17, 183-278. For examples of
applications of this concept see: Illuminati, G.; Mandolini, L. Acc. Chem.
Res. 1981, 14, 95-102. Mandolini, L. AdV. Phys. Org. Chem. 1986, 22,
1-111. Ercolani, G. J. Phys. Chem. B 1998, 102, 5699-5703.
(18) At lower temperatures linear oligomers (lin-[CxC•AyA]ng2) were formed,
complicating the analyses.
(15) Stockmayer (ref 16) has shown that under equilibrium conditions the
concentration of unstrained rings decreases exponentially (5/2 power) with
ring size. Hence the likelihood of the cyclic tetramer is 1/25/2 ) 1/6 and
the cyclic hexamer 1/16 that of the cyclic dimer. On this basis the
contribution of larger cyclic species is considered minimal.
(19) Bonnet, G.; Krichevsky, O.; Libchaber, A. Proc. Natl. Acad. Sci. U.S.A.
1998, 95, 8602-8606. Goddard, N. L.; Bonnet, G.; Krichevsky, O.;
Libchaber, A. Phys. ReV. Lett. 2000, 85, 2400-2403.
(20) Dua, A.; Cherayil, B. J. J. Chem. Phys. 2002, 117, 7765-7773.
9
J. AM. CHEM. SOC. VOL. 125, NO. 12, 2003 3527

A R T I C L E S
Gibson et al.
Table 1. Association Constants (Klinear, Kcyclic:Scheme 4), and Enthalpies for Linear and Cyclic Dimerization of Ditopic Hosts CxC with
Ditopic Guests AyA [(CD3)2CO:CDCl3 (1/1, v/v)]
a
b
c
c
d
K
∆G
(kcal/mol)
∆G
(kcal/mol)
EM
(mM)
∆H
∆H
∆H
cyclic penalty
(kcal/mol)
linear, 295K
linear, 295K
cyclic, 295K
295K
linear
cyclic
a
H/G x/y
(M-1
)
K
(kcal/mol)
(kcal/mol)
cyclic, 295K,
8/4
8/10
8/22
2/22
1.3 ( 0.3 × 103
3.7 ( 0.7 × 103
5.2 ( 0.6 × 103
8.9 ( 0.9 × 103
-4.2 ( 0.1
-4.9 ( 0.2
-5.0 ( 0.1
-5.3 ( 0.1
2.5 ( 0.4
1.7 ( 0.3
0.62 ( 0.03
0.86 ( 0.14
-0.54 ( 0.10
-0.31 ( 0.10
+0.28 ( 0.03
+0.088 ( 0.10
0.96 ( 0.50
0.23 ( 0.10
0.060 ( 0.010
0.050 ( 0.015
-8.0 ( 0.6
-11 ( 2
-10 ( 3
-
-8.6 ( 1.0
-3.7 ( 0.1
-2.1 ( 0.6
-
-0.6 ( 1.6
+7 ( 2
+8 ( 3
-
a Klinear ) [lin-CxC•AyA]/[CxC][AyA]. Kcyclic ) [cyc-CxC•AyA]/[lin-CxC•AyA]. (-Values represent standard deviations for determinations using three
independently prepared solutions at concentrations of 0.20 and 0.40 mM. b Calculated from EM ) Kcyclic/2Klinear. This assumes that the equilibrium constant
for addition of a unit of either CxC or AyA to linear dimer lin-CxC•AyA is identical to the measured Klinear. EM is the concentration of building blocks CxC
and AyA at which the total concentration of trimers lin-CxC•AyA•CxC and lin-AyA•CxC•AyA is equal to the concentration of cyclic dimer cyc-CxC•AyA.
Since Klinear ) [lin-CxC•AyA•CxC]/[lin-CxC•AyA][CxC] ) [lin-AyA•CxC•AyA]/[lin-CxC•AyA][AyA] and Kcyclic ) [cyc-CxC•AyA]/[lin-CxC•AyA],
when [lin-CxC•AyA•CxC] + [lin-AyA•CxC•AyA] ) [cyc-CxC•AyA], [CxC] ) [AyA] ) EM ) Kcyclic/2Klinear.
c (-Values represent errors estimated for
slopes of van’t Hoff plots of association constants determined using the three independently prepared solutions over the temperature range 285-313 K, r2
g 0.97. At lower temperatures higher linear species (lin-[CxC•AyA]n, n g 2) formed, preventing measurements of Klinear and Kcyclic
.
d ∆Hcyclic
)
penalty
∆Hcyclic - ∆Hlinear. (-Values represent errors derived by summing the errors in the individual terms on the right.
tional groups, in this case topological functional groups:
n ) 1/(1 - p) ) 2K[H]o/{(1 + 2K[H]o)1/2 - 1}
Now if 2K[H]o . 1, n ) 2K[H]o/{(2K[H]o)1/2 - 1}
and if (2K[H]o)1/2 . 1, n ) (2K[H]o)1/2
In this system [H]o ) 2[H-H]o.23
(1)
Therefore, since Klinear is on the order of 5.0 × 103 M-1 (Table
1), in the absence of cyclics, at 1.0 M concentration of CxC
and AyA we would expect n ) [(2)(1.0 M)(2)(5.0 × 103
M-1)]1/2 ) 1.4 × 102.
Figure 5. (Left) Dependence of ∆Glinear and ∆Gcyclic on y for complexation
of C8C and AyA. (Right) Dependence of ∆Hlinear, ∆Hcyclic and ∆Hlinear
∆Hcyclic on y for complexation of C8C and AyA. Data from Table 1.
-
IV. b. NMR Studies. By increasing the equimolar concentra-
tions as eq 1 requires to produce high supramolecular mass
assemblies, we then proceeded to construct linear arrays lin-
[Ax•Cy]n, using the sets of complementary homoditopic mol-
ecules, CxC and AyA, as represented in cartoon form in Scheme
4.
∆Hcyclic penalty, is ∆Hcyclic - ∆Hlinear. This parameter indeed
increases linearly as the length of AyA increases (Table 1, Figure
5b).
IV. Non-Covalent Polymers from CxC and AyA. IV. a.
Theoretical Expectations. A relationship among the association
constant between CxC and AyA, their concentrations and the
average degree of polymerization, n, of the resultant lin-
[CxC•AyA]n, is easily derived as follows, under the assumptions
that the same average association constant holds for each
successive step (isodesmic)21 and that cyclic species can either
be ignored or taken into account. Consider the generalized
equilibrium of Scheme 5, in which the concentrations of the
two homoditopic species are equal:
1
The H NMR spectra of equimolar solutions of C8C and
A10A in acetone-d6/chloroform-d (1/1, v/v) contained three sets
of signals for the NCH2 protons (e.g., see Figure 6d). As noted
above, two of these are attributed to complexed moieties, while
the other set corresponds to the signals for uncomplexed species.
The two sets of signals for complexed NCH2 units arise from
two different pseudorotaxane structures, the cyclic dimer cyc-
C8C•A10A, and linear dimer lin-C8C•A10A and linear species
lin-[C8C•A10A]n (Scheme 4), the latter two being indistin-
guishable.25 Cyclic dimer cyc-C8C•A10A was preferentially
formed in equimolar dilute solutions (<1.0 mM, see Figure 6c).
The 1H NMR spectra (Figure 6) recorded at different
equimolar concentrations of C8C and A10A in acetone-d6:
chloroform-d (1:1, v:v) revealed that both the ratio of cyc-
C8C•A10A to lin-[C8C•A10A]n and the extent of aggregation,
We can write K ) [H/G]/[H][G] ) [H/G]/[H]2 .
If we now define p ) extent of complexation,
2
K ) p[H]o/(1 - p)2[H]o .
Solving this quadratic equation leads to
1 - p ) {(1 + 2K[H]o)1/2 - 1}/2K[H]o
This relationship allows us to use the well-known Carothers
equation,22 which relates n to the extent of conversion of func-
Scheme 5. Generalized Representation of Self-Assembly of Linear Supramolecular Polymer from Complementary Homoditopic Building
Blocks. H ) host; G ) guest; H/G ) complex
9
3528 J. AM. CHEM. SOC. VOL. 125, NO. 12, 2003

Supramolecular Pseudorotaxane Polymers
A R T I C L E S
Figure 6. Partial 1H NMR spectra of solutions of C8C and A10A at (a) 10/0, (b) 0/10 mM/mM, and equimolar solutions at (c) 1.0 × 10-3, (d) 1.0 × 10-2
,
(e) 1.0 × 10-1, (f) 0.50, (g) 1.0, and (h) 2.0 M (400 MHz, acetone-d6:chloroform-d (1:1, v:v), 22 °C). The three sets of NCH2 signals are for (1) uncomplexed
moieties, i.e., end groups, (denoted by “u”) in A10A, lin-[A10A•C8C]n•A10A and lin-[C8C•A10A]n, (2) complexed moieties (denoted by “l”) in lin-
[C8C•A10A]n, lin-[A10A•C8C]n•A10A and lin-[C8C•A10A]n•C8C, and (3) complexed moieties in cyc-C8C•A10A (denoted by “c”). These assignments
were confirmed by the results presented in Figure 3.
n, of the latter are concentration-dependent as expected. At the
highest equimolar concentration we examined (2.0 M), the
signals for the cyclic dimer cyc-C8C•A10A are no longer
observed (Figure 6h), indicating its absence. It is noteworthy
that linear covalent polymerization is highly favored over
cyclization at high concentrations, e.g., 1.0 M.16,26 Therefore,
the assumption of the absence of the oligomeric cyclic com-
plexes is believed to be valid in the more concentrated
solutions.15 Since the concentrations of the end-groups in the
supramolecular polymer lin-[C8C•A10A]n can be determined
by integrating the signal for uncomplexed NCH2 moieties, one
can estimate its average molecular weight (Table 2). For
instance, in the 2.0 M equimolar solution Mn ) 18 kDa (n )
9.1). Substantially broadened signals in Figure 6, g and h,
support the formation of large supramolecular polymers lin-
[C8C•A10A]n in solution.
Equimolar solutions of C8C and the complementary homodi-
topic molecules A4A and A22A were also characterized by 1H
NMR spectroscopy in acetone-d6:chloroform-d (1:1, v:v). In the
case of C8C and A4A, the spectra10 confirmed that the
efficiency of formation of the cyclic dimer cyc-C8C•A4A was
greater than that of cyc-C8C•A10A from C8C and A10A (Table
2) at all concentrations. Equimolar solutions higher than 1.0 M
could not be achieved due to the poor solubility of the shorter
A4A. In the equimolar 0.5 M solution of C8C and A4A the
supramolecular polymer lin-[C8C•A4A]n has an average mo-
lecular weight (Mn) of 6.1 kDa, which is nearly a 2-fold decrease
with respect to that of lin-[C8C•A10A]n (Table 2). These
observations are consistent with the dilute solution studies above
(Table 1) in which it was demonstrated that the shorter the end-
to-end distance of the components, the higher the percentage
of the cyclic dimer complex (cyc-CxC•AyA) formation
(21) This term refers to linear self-assembly processes in which the successive
equilibrium constants are all the same. Other models include systematic
diminution of successive association constants, the attenuated K model;
see Martin, R. B. Chem. ReV. 1996, 96, 3043-3064.
(22) Carothers, C. H. Trans. Faraday Soc. 1936, 32, 39-53.
(23) Other authors have derived similar equations: Meijer et al., ref 24; Lehn,
ref 4, though not explicitly given.
(24) Sijbesma, R. P.; Beijer, F. H.; Brunsveld, L.; Folmer, B. J. B.; Hirschberg,
J. H. K. K.; Lange, R. F. M.; Lowe, J. K. L.; Meijer, E. W. Science 1997,
278, 1601-1604.
(25) Note that linear dimer lin-CxC•AyA in fact is identical to lin-[AyA•CxC)]n
n ) 1. lin-CxC•AyA is shown explicitly for clarity. Under isodesmic
conditions the concentrations of lin-[CxC•AyA]nCxC and lin-
[AyA•CxC)]n•AyA will always be equal when the concentrations of CxC
and AyA are equal. Note that species lin-[CxC•AyA]n•CxC and lin-
(AyA•CxC)n•AyA cannot directly cyclize, unlike the case of oligomers
derived from complementary heteroditopic monomers; this lack of direct
cyclizability of 50% of the species present is a previously unrecognized
advantage of the homoditopic pair approach.
In contrast, the spectra of equimolar solutions of C8C and
A22A (Figure 7) indicated quite convincingly that the formation
of the cyclic dimer complex cyc-C8C•A22A was reduced
compared to the systems C8C•A4A and C8C•A10A (Table 2).
(26) Flory, P. J. Principles of Polymer Chemistry; Cornell University Press:
Ithica, NY, 1953; pp 317-346.
9
J. AM. CHEM. SOC. VOL. 125, NO. 12, 2003 3529

A R T I C L E S
Gibson et al.
Table 2. Analyses of Percentages of Ammonium Ion Moieties in Cyclic Dimers ) CD (cyc-CxC•AyA) and Supramolecular Polymers ) SP
(lin-[CxC•AyA]n) and the Average Number of Repeat Units, n, in SP Derived from Ditopic Hosts CxC and Ditopic Guests AyA
c
equimolar
concentration (M)
%inCDat
+22/−40°C
% uncomp.
at +22/−40°C
nb of SP at
+22/−40 °C
M of SP at
predicted n of
SP @22°C
n
a
a
d
H/G x/y
8/4
+22/−40 °C (kDa)
0.50
0.10
17
25
78
13
14
7
3.2 ( 0.6
6.1 ( 1.2f
51 ( 7
2.7 ( 0.4
5.1 ( 0.8f
23 ( 3
1.0 × 10-2
1.0 × 10-3
2.0
1.6 ( 0.6
0.78 ( 0.28
3.0 ( 1.2f
7.2 ( 0.9
-
86
9
1.5 ( 0.6f
8/10
8/22
0/-e
3.4/-e
12/15
23/27
66/81
72/94
0
5.5/-e
6.4/-e
7.9/5.1
20/5.4
16/4.3
20/2.3
4.4
9.1 ( 3.8/-e
7.5 ( 2.7/- e
5.6 ( 1.6/8.3 ( 4.1
1.9 ( 0.2/6.8 ( 3.1
1.1 ( 0.2/2.2 ( 1.6
0.70 ( 0.12/1.3 ( 6.2
11 ( 6.6
18 ( 8/-e,g
1.7 ( 0.2 × 102
1.2 ( 0.1 × 102
86 ( 9
1.0
0.50
0.10
15 ( 5/-e,g
11 ( 3/16 ( 8g
3.8 ( 0.3/13 ( 6g
2.2 ( 0.4/4.4 ( 3.2g
1.4 ( 0.2/2.6 ( 12.4g
24 ( 14 h
38 ( 4
12 ( 1
1.0 × 10-2
1.0 × 10-3
1.0
3.8 ( 0.4
1.4 ( 0.1 × 102
1.0 ( 0.1 × 102
46 ( 3
0.50
0.10
0
8.3
28
38
4.7
20
34
43
11 ( 5.8
24 ( 12h
2.3 ( 0.3
4.9 ( 0.6h
1.0 × 10-2
1.0 × 10-3
0.50
1.1 ( 0.1
2.4 ( 0.2h
14 ( 1
0.72 ( 0.06
1.5 ( 0.2h
4.6 ( 0.3
1.3 ( 0.3 × 102
60 ( 3
2/22
0/-e
17/10
38/55
44/63
0
1.8/-e
9.2/4.3
24/4.7
39/13
45
28 ( ∼10/-e
4.5 ( 0.7/10 ( 6.5
1.3 ( 0.2/4.8 ( 2.8
0.72 ( 0.06/1.4 ( -0.3
1.1 ( 0.1
58 ( 20/-e,i
9.3 ( 2.4/21 ( 14i
2.7 ( 0.3/9.9 ( 5.7i
1.5 ( 0.1/2.9 ( 0.6i
3.7 ( 0.2k
0.10
1.0 × 10-2
1.0 × 10-3
0.10
19 ( 1
6.0 ( 0.3
-
8/EO35j
1.0 × 10-2
0
0
77
100
0.65 ( 0.02
2.2 ( 0.1k
-
-
-
1.0 × 10-3
0
a Calculated from the 1H NMR spectra of equimolar solutions (acetone-d6/chloroform-d, 1/1, v/v) of CxC and AyA. Typically [cyc-CxC•AyA] was
determined by direct integration of the NCH2 signals of the cyclic species vs the total of all of the NCH2 signals (checked vs the ArOCH2 signals) and/or
the Hy signals for the cyclic species vs the total of all of the NCH2 signals. Typically the uncomplexed NCH2 signals were determined either by direct
integration of the uncomplexed ends of lin-CxC•AyA and lin-[CxC•AyA]n and uncomplexed AyA vs the total of all of the NCH2 signals or by difference
between the complexed NCH2 signals and twice the integral for the ArOCH2 signals. Errors are estimated to be (2%. b n was determined by solving the
expression: n ) [100 - (% of ammonium ion units in lin-CxC•AyA)]/2(% uncomplexed ammonium ion units). Errors cited are the average deviations
using 2% errors in % lin-CxC•AyA and % uncomplexed. The maximum values were determined using the observed % lin-CxC•AyA less 2% and the
observed % uncomplexed less 2%. The minimum values were determined using the observed % lin-CxC•AyA plus 2% and the observed % uncomplexed
plus 2%. There is larger error toward the maximum values. c Errors cited are the average deviations of the maximum and minimum values calculated using
the maximum and minimum values of n based on 2% errors as noted in footnote b. There is larger error toward the maximum values. d Based on eq 1 and
Klinear value of Table 1; recall that eq 1 assumes no cyclics form. Therefore, this value is a reference point for the situation in which no cyclic forms and
Klinear is invariant with concentration. No calculations were done when the inequalities assumed in eq 1 were violated. e The spectra recorded at lower
temperatures exhibited severe signal broadening, preventing accurate signal integration. f The number average molecular weight (Mn) of the aggregate lin-
[C8C•A4A]n was calculated using Mn ) n(1896 Da). g The number average molecular weight (Mn) of the aggregate lin-[C8C•A10A]n was calculated using
Mn ) n(1980 Da). h The number average molecular weight (Mn) of the aggregate lin-[C8C•A22A]n was calculated using Mn ) n(2148 Da). i The number
average molecular weight (Mn) of the aggregate lin-[C2C•A22A]n was calculated using Mn ) n(2064 Da). j In CDCl3. k The number average molecular
weight (Mn) of the aggregate lin-[C8C•AEO35A]n was calculated using Mn ) n(3346 Da).
More importantly, in equimolar solutions of C8C and A22A at
greater than 0.5 M the cyclic dimer cyc-C8C•A22A was not
detected (Table 2, Figure 7e). From end group analysis, the
average molecular weight of supramolecular polymer lin-
[C8C•A22A]n is 24 kDa at 0.5 M; 2.0 M solutions could not
be made due to the limited solubility of A22A.
Finally solutions of the shortest bis(crown ether) C2C and
the longest diammonium salt A22A, as expected on the basis
of the EM results of Table 1, yielded the largest linear
aggregates. The highest degree of “noncovalent polymerization”,
n, was observed at 0.5 M concentrations, forming lin-
[C2C•A22A]n with a supramolecular molar mass of 58 kDa at
ambient temperature. Again, the limited solubility of A22A did
not allow higher concentrations to be studied.27
As expected on the basis of the low Ka observed in the model
system 1a/AEO35A noted above, homoditopic host C8C and
homoditopic guest AEO35A showed very small extents of
complexation as measured by the NCH2 signals of AEO35A.
At 22 °C in equimolar 100 mM CDCl3 solutions only 55%
complexation was observed versus 86, 80, and 80%, respec-
tively, for C8C with A4A, A10A, and A22A each at 100 mM
in the less polar solvent CDCl3:CD3COCD3 (1:1, v:v). Further-
more, at 1.0 mM in CDCl3 no complexation was observed for
C8C and AEO35A (Table 2).
The error bars for n and Mn in Table 2 are large. This is an
intrinsic problem in using NMR spectroscopy for end group
analysis of polymeric systems. Note that as n increases, the error
increases exponentially. Assuming that the error in end group
determination is 2% absolute, at n ) 25 the range would be
from 17 to 50. In many cases the error will be larger than this.
Therefore, NMR spectroscopy is at best a qualitative tool for
study of such supramolecular polymers. Other physical methods
are needed to establish the polymeric nature of the assemblies.
Nonetheless it is clear from Table 2 that degrees of nonco-
valent polymerization, n, are much less than predicted by eq 1
above, notwithstanding the intrinsic errors. For example, with
C8C/A22A at 1.0 M the experimental n value is only 20% of
that calculated by eq 1. Such comparisons for other results in
Table 2 indicate a diminution in the effective association
constant with increasing concentrations. In model monotopic
systems we have also noted such concentration dependences.28A
possible contributing factor is “exo complexation”, i.e., non-
pseudorotaxane hydrogen bonding of the crown ether and
ammonium salt moieties;13 this consumes “free” species and
reduces their true concentrations, increasingly as total concentra-
(27) Pseudorotaxane formation involving A22A was found to be time dependent
with both C8C and C2C. Approximately 18-24 h were required for
equilibration.
(28) Jones, J. W.; Huang, F.; Gibson, H. W. J. Am. Chem. Soc., submitted.
9
3530 J. AM. CHEM. SOC. VOL. 125, NO. 12, 2003

Supramolecular Pseudorotaxane Polymers
A R T I C L E S
Figure 8. Reduced viscosity as a function of equimolar host and guest
concentrations in acetone:chloroform (1:1, v:v) at 22 °C for (a) MC8MC
and A10A, (b) C8C and A10A, and (c) C2C and A22A. Error bars are
estimated to be 0.2% at 0.1 M, but perhaps as high as 5% at 0.5 M because
of capillary absorption effects.
Figure 7. The 1H NMR spectra of equimolar solutions of C8C and A22A
at (a) 1.0 × 10-3, (b) 1.0 × 10-2, (c) 0.10, (d) 0.50, and (e) 1.0 M each
(400 MHz, acetone-d6:chloroform-d (1:1, v:v), 22 °C). The three sets of
NCH2 signals are for 1) uncomplexed moieties, i.e., end groups, (denoted
by “u”) in A22A, lin-[A22A•C8C]n•A22A and lin-[C8C•A22A]n, (2)
complexed moieties (denoted by “l”) in lin-[C8C•A22A]n, lin-[A22A-
•C8C]n•A22A and lin-[C8C•A22A]n•C8C, and (3) complexed moieties in
cyc-C8C•A22A (denoted by “c”).
1
slope (Figure 8a) is consistent with the H NMR observation
that there is little or no complexation between MC8MC and
A10A in solution. It also indicates that the ionic strength per
se has a negligible effect on viscosity under these experimental
conditions. Therefore, the sharp increases in viscosity at higher
concentrations of C8C/A10A and C2C/A22A (Figure 8c) are
reflective of aggregation, producing structures of large hydro-
dynamic volume, i.e., lin-[C8C•A10A]n and lin-[C2C•A22A]n.
Cates and Candau have derived theoretical relationships for
the reduced viscosity of self-assembled noncovalent polymers
comprised of surfactant molecules in aqueous media as a
function of concentration.31 According to this mean field theory
if the association constants are high and the lifetimes of the
assemblies exceed the reptation times, a plot of log(reduced
viscosity) vs log(total concentration) should have a slope of 3.5-
3.7. Such log-log plots for MC8MC/A10A, C8C/A10A, and
C2C/A22A have slopes of 1.48, 2.83, and 2.64, respectively.
The fact that the slopes of the latter two systems are less than
the theoretical values may reflect several factors. First of all,
Cates and Candau note that for their systems in water there
was a strong inverse dependence of the exponent (slope) on
salt concentration.31 In our systems in organic solvents this effect
would be expected to be exaggerated, since one of the
components is in fact ionic and the concentrations are high. In
one reported experiment in an organic solvent an exponent
tions increase. On the basis of the Debeye-Hu¨ckel treatment
activity coefficients are sharply dependent on ionic concentration
in solvents of low dielectric constant.29 Clearly at concentrations
as high as 1 or 2 M the ionic strength effect may substantially
affect the activity coefficients and the apparent association
constants. As a result of these effects an “attenuated K” model
is probably more appropriate than the isodesmic treatment of
eq 1.21 These effects taken together present a serious challenge
in analyses of equilibria involving charged species for construc-
tion of supramolecular polymers in organic media.
IV. c. Solution Viscosity. For direct physical evidence of
the formation of large self-assembled noncovalent polymers,
we turned to viscometry. A plot of reduced viscosity versus
concentration of equimolar solutions of C8C and A10A in
acetone-d6:chloroform-d (1:1, v:v) is nonlinear (Figure 8b),
reflecting the increasing size of the supramolecular polymer lin-
1
[C8C•A10A]n with concentration as shown by the H NMR
data. In fact, the viscosities of 1.0 and 2.0 M equimolar solutions
were too high to measure by this method.
On the other hand, a completely different solution viscosity
profile was observed for equimolar solutions of MC8MC, a
constitutional isomer of C8C, and A10A (Figure 8a). It should
1
be noted that the H NMR spectrum of an equimolar solution
of C2C and dibenzylammonium hexafluorophosphate (6) at 10
mM in the same solvent system displayed no sign of complex-
ation, i.e., did not show extra sets of signals due to slow
exchange or changes in the chemical shifts due to fast exchange.
The m-phenylene linkage in the crown ether moieties of
MC8MC prevents or at least very strongly retards pseudoro-
taxane formation.30 In the viscosity plot a straight line with low
(30) Other workers have demonstrated that MC8MC does form a complex with
dibenzylammonium hexafluorophosphate, but with an immeasurably (by
NMR) low Ka: Ashton, P. R.; Bartsch, R. A.; Cantrill, S. J.; Hanes, R. E.,
Jr.; Hickingbottom, S. K.; Lowe, J. N.; Preece, J. A.; Stoddart, J. F.;
Talanov, V. S.; Wang, Z.-H. Tetrahedron Lett. 1999, 40, 3661-3664.
Cantrill, S. J.; Fulton, D. A.; Heiss, A. M.; Pease, A. R.; Stoddart, J. F.;
White, A. J. P.; Williams, D. J. Chem. Eur. J. 2000, 6, 2274-2287.
(31) Cates, M. E. Macromolecules 1987, 20, 2296-2300. Cates, M. E.; Candau,
S. J. J. Phys. Condens. Matter 1990, 2, 6869-6892. It should be noted
that although this theoretical approach is still in use (Narayanan, J.; Urbach,
W.; Langevin, D.; Manohar, C.; Zana, R. Phys. ReV. Lett. 1998, 81, 228-
231), it is not without its peculiarities and controversy (Aliotta, F.; Sacchi,
M. Colloid Polym. Sci. 1997, 275, 910-921. Ko¨per, G. J.; van der Ploeg,
J. P. O. M.; Cirkel, P. A. Prog.Colloid Polym. Sci. 1997, 1055, 294-297).
(29) Glasstone, S.; Lewis, D. Elements of Physical Chemistry; D. van Nostrand
Co., Inc.: New York, 1960; pp 507-516.
9
J. AM. CHEM. SOC. VOL. 125, NO. 12, 2003 3531

A R T I C L E S
Gibson et al.
Table 3. Supramolecular Formulas, Calculated Supramolecular Weights, and Observed (MS) Mass/Charge Ratios of Complexes from
Homoditopic Species C8C and A10A
structure
calcd MW
obs m/za
obs m/zb
[A10A-PF6]+
711.4
1686.9
1168.8
1226.7
1227.7
1595.8
1597.8
1687.9
1833.9
2054.2
2307.3
2398.3
2954.4
4958.2
711.3a1
[C8C•A10A-2HPF6]2+
844b1
1169b2
1227b3
[C8C•A10A-PF6-HPF6-DB24CH2OOCCH2]+
[C8C•A10A-PF6-HPF6-DB24CH2]+
[C8C•A10A-2PF6-DB24CH2]+
1228a2
[C8C•A10A-2HPF6-C6H5CH2]+
[C8C•A10A-2PF6-C6H5CH2]+
1596b4
1598.1a3
1688.2a4
1833.6a5
[C8C•A10A-PF6-HPF6]+
1688b5
1834b6
2054b7
2307b8
[C8C•A10A-PF6]+
[A10A•C8C•A10A-4HPF6-C6H4CH2NH2CH2C6H5]+
[A10A•C8C•A10A-2HPF6-PF6-C6H5CH2]+
[A10A•C8C•A10A-PF6-2HPF6]+
[C8C•A10A•C8C-H-HPF6]+
2398.5a6
2954.0a7
4959.0a8
[[C8C•A10A]2•C8C-C6H5CH2NH2CH2]+
a MALDI-TOF mass spectrum recorded in the positive ion mode using 2,5-dihydroxybenzoic acid (2,5-DHB) matrix (Washington U); sample prepared
in chloroform-acetone (1:1 v:v) at 2.0 M. 1) 9.6%, 2) 43%, 3) 7.8%, 4) 100%, 5) 5.8%, 6) 3.8%, 7) 2.5%, 8) 1.4%. “DB24” represents dibenzo-24-crown-8
moieties of A10A. b FAB mass spectrum recorded in the positive ion mode using 3-nitrobenzyl alcohol (3-NBA) as the matrix (Va Tech); sample prepared
in chloroform-acetone (1:1 v:v) at 2.0 M. Same sample as in (a). 1) 85%, 2) 7%, 3) 10%, 4) 6.7%, 5) 100%, 6) 12%, 7) 1%, 8) 45%. “DB24” represents
dibenzo-24-crown-8 moieties of A10A.
and alcohol 2a, as supported by 1H NMR spectral results. The
loss of HPF6 occurs at lower temperatures in the pseudorotaxane
complexes than in the ammonium salts by themselves; this is
probably because the anions are separated spatially from the
cations in the pseudorotaxanes and therefore not as tightly
bound.
Table 4. Thermogravimetric Analyses of Pseudorotaxanes and
Their Componentsa
b
T
other T
(°C)/%
5%
sample
(°C)
fragments lost (% th)
-
(C6H5CH2)2NH2+PF6
DB24C8
220
334
210
224
340
187
c
d
270f/32
343f/56
461f/80
300f/15
462f/80
1a•6 e
C6H5CH2NH2 (14) + HPF6 (18)
O(CH2)10 (20) + 2HPF6 (34)
2 DB24C8-CH2 (82)
2HPF6 (15)
A10A
IV. f. Differential Scanning Calorimetry (DSC). Solid
samples were prepared by freeze-drying 1.0 × 10-3, 0.10, and
0.50 M equimolar solutions of C8C and A10A in acetone:
C8C
C8C/A10Ag
-
1
chloroform (1:1, v:v) at -93 °C under high vacuum. The H
a Heating rate: 10 °C/min in N2 unless otherwise noted. b Temperature
of 5% weight loss. c Smooth weight loss to 3% residue at 289 °C. 32%
weight loss corresponds to 271°C. d 40 °C/min; smooth weight loss to 3%
residue at 409 °C. 32% weight loss corresponds to 392 °C. e Sample
consisted of crystals of the pseudorotaxane complex grown by vapor
diffusion of hexane into a chloroform solution 10 mM in both components.
f Plateau value. g Sample prepared by evaporation of a chloroform-acetone
(1:1 v:v) solution 1.0 M in both species.
NMR spectra of equimolar solutions of C8C and A10A at 1.0
× 10-3, 0.10, and 0.50 M in acetone-d6:chloroform-d (1:1, v:v)
were essentially unchanged from -40 to -60 °C (below this
temperature the 1H NMR spectra of the solutions could not be
recorded due to partial freezing of the solvent). Thus, Mn’s of
the supramolecular polymer lin-[A10A•C8C]n and the abun-
dance of the cyclic dimer cyc-C8C•A10A in the freeze-dried
samples were estimated by simple integration of relevant signals
of the spectra recorded at -40 °C (Table 2). The samples were
analyzed by DSC. To eliminate sample history, the freeze-dried
samples were heated to 100 °C and cooled to 30 °C at 10 °C
min-1. They were heated at 10 °C min-1 and the DSC
thermograms were recorded. The pure homoditopic molecules
C8C and A10A were crystalline and thus revealed sharp first-
order phase transitions. As a result of self-assembly, amorphous
polymer lin-[A10A•C8C]n with glass transition temperatures
(Tg) of 57 and 59 °C, respectively, were formed from the 0.10
and 0.50 M samples with Mn ) 13 and 16 kDa, respectively
(Table 2). In the case of the 10 mM sample (81% cyclic dimer
cyc-C8CA10A, Table 2), a high degree of crystallinity was
observed by optical microscopy
(slope) of 1.9 was observed, a value lower than those obtained
here.32 Meijer et al. reported a slope of 3.76 for a nonionic
23
H-bonded supramolecular polymer in chloroform. Another
possibility is that the lifetimes of the noncovalent polymers
described here are shorter than their reptation times.
IV. d. Mass Spectrometry. Due to the highly ionic nature
of the linear arrays and the normally used very polar matrixes,
which tend to dissociate the complexes, mass spectrometry on
samples prepared from ditopic host C8C and ditopic guest A10A
did not reveal assemblies of high mass. As shown in Table 3
MALDI-TOF and FAB techniques detected only species lin-
[A10A•C8C]n•A10A containing up to five total building blocks,
i.e., n ) 2.
IV. e. Thermogravimetric Analysis (TGA). As shown in
Table 4 TGA of samples prepared by evaporation of equimolar
solutions of homoditopic host C8C and homoditopic guest
A10A reflect the thermal stabilities of the two components.
Results for the model system consisting of DB24C8 (1a) and
dibenzylammonium hexafluorophosphate (6) are similar. Ther-
mal decomposition of the secondary ammonium salt moieties
of A10A gives HPF6, and the resultant secondary amine is
believed to attack the ester linkage of C8C to afford a polyamide
IV. g. Film and Fiber Formation. Flexible, creasible,
amorphous, and transparent films were cast from 1:1 stoichio-
metric solutions of C8C and A10A at concentrations >0.50 M.
Such film properties can only result from entanglement of
linearly connected macro-sized aggregates. Similarly, since a
polymeric structure of high molecular weight is necessary for
fiber formation, the scanning electron microscopic images
(Figure 9) of a fiber drawn from a concentrated equimolar
solution of C8C and A10A confirm the high degree of the linear
chain extension in lin-[C8C•A10A]n. Fibers of lin-[C8C•A22A]n
(32) Schurtenberger, P.; Scartazzini, R.; Magid, L. J.; Leser, M. E.; Luisi, P. L.
J. Phys. Chem. 1990, 94, 3695-3701.
9
3532 J. AM. CHEM. SOC. VOL. 125, NO. 12, 2003

Supramolecular Pseudorotaxane Polymers
A R T I C L E S
Figure 9. Scanning electron micrographs of (gold-coated) fibers pulled from concentrated solutions of (a) (left) C8C and A10A, (b) (center) C8C and
A22A, and (c) (right) C2C and A22A. The scale bar in each micrograph is 5 µm.
and lin-[C2C•A22A]n were also drawn from concentrated
solutions of C8C/A22A and C2C/A22A, respectively. As shown
in Figure 9, the morphology of these fibers was similar to that
of those from lin-[C8C•A10A]n. These fibers were formed in
a manner analogous to dry spinning33 used for covalently bonded
macromolecules. Unfortunately, insufficient quantities prevented
mechanical characterization of the films and fibers at this time.
partially overlapping signals for “lu” and “cc” of dilute solutions
of CxC/AyA of Figure 4, a and b, were carried out using
Lorentzian line shapes in the Varian NMR Software package,
version 6.1B.; the peak maxima were assigned using the
positions of the shoulders due to “lu”. The integration values
from the deconvolutions were checked by taking the total
integral in the region 4.35 to 4.45 ppm and subtracting from it
the integral for the other “cc” signal at 4.75 ppm. Reduced
viscosities were measured on a Cannon-Ubbelohde semi-micro
dilution viscometer with 200 centipoise inner diameter capillary.
Differential scanning calorimetry (DSC) was performed under
a nitrogen purge using indium as the calibration standard. For
scanning electron microscopy (SEM) fibers were drawn from
concentrated solutions using a boiling stick or tweezers and
placed on the mounting grid; the copper substrate was sputtered
with gold after sample deposition and before exposure to the
electron beam. Mass spectra were provided by the Washington
University Mass Spectrometry Resource, an NIH Research
Resource (Grant No. P41RR0954). Dialdehydes 3a and 3b were
prepared by the procedures reported in the literature.34 We
previously reported7a syntheses of 1a, 1b, 2a, 2b, C8C,
MC8MC, 4b, 5b, A10A, and 7a.
Conclusions
In summary, linear (lin-CxC•AyA) and cyclic (cyc-CxC•AyA)
dimeric complexes are preferentially formed in dilute equimolar
solutions (e1 mM) of bis(crown ether)s CxC and diammonium
salts A4A through A22A. Linear chain extension to lin-
[CxC•AyA]n is observed almost exclusively in concentrated
equimolar solutions (g0.5 M) of homoditopic hosts C8C or
C2C with homoditopic guests A4A through A22A. The solution
behaviors of the self-organized supramolecular assemblies lin-
[CxC•AyA]n were characteristic of large linear aggregates as
demonstrated by the viscosity experiments. Moreover, the
preparation of films and fibers corroborates the polymeric nature
of the self-organized supramolecules. Freeze-drying equimolar
solutions yielded linear assemblies lin-[CxC•AyA]n as amor-
phous solids. The efficiencies of the formation of the nonco-
valent polymers from complementary homoditopic molecules
increased with increasing end-to-end distance of the building
blocks. These studies provide proof of principle that self-
assembly via pseudorotaxane formation can produce noncovalent
polymers; however, there is a caveat in that exo complexation13
and ionic strength28 effects greatly reduce the effectiveness of
pseudorotaxane formation at high concentrations. Future im-
provements will involve host-guest pairs with higher associa-
tion constants to provide higher supramolecular weights.
Acknowledgment. We are grateful to the National Science
Foundation for support of this work through Grant CHE-
9521738. We thank Thomas Glass for expert assistance in
obtaining the NMR spectra and Professor James McGrath for
use of his thermal analysis equipment.
Supporting Information Available: Procedures for synthesis
of C2C, 3c, 3d, 4a, 4c, 5a, 5c, 5d, A4A, A22A, and AEO35A;
1H NMR and mass spectra (PDF). This material is available
free of charge via the Internet at http://pubs.acs.org.
Experimental Section
1
The 400 MHz H NMR spectra were recorded with tetram-
JA020900A
ethylsilane (TMS) as an internal standard. Deconvolution of the
(33) Allcock, H. R.; Lampe, F. W. Contemporary Polymer Chemistry, 2nd ed.;
Prentice-Hall: New York, 1990; pp 508-509.
(34) Guilani, B.; Rasco, M. L.; Hermann, C. F. K.; Gibson, H. W. J. Heterocycl.
Chem. 1990, 27, 1007-1009.
9
J. AM. CHEM. SOC. VOL. 125, NO. 12, 2003 3533