Synthesis of Conjugated Polyrotaxanes

Article abstract of DOI:10.1002/chem.200305245

A series of conjugated polyrotaxane insulated molecular wires are synthesised by aqueous Suzuki polymerisation, using hydrophobic binding to promote threading of the cyclodextrin units. These polyrotaxanes have conjugated polymer cores based on poly(para-

Full text of DOI:10.1002/chem.200305245

FULL PAPER
Synthesis of Conjugated Polyrotaxanes
Jasper J. Michels,[a] Michael J. O×Connell,[a] Peter N. Taylor,[a] Joanne S. Wilson,[b]
Franco Cacialli,[c] and Harry L. Anderson*[a]
Abstract: A series of conjugated poly-
rotaxane insulated molecular wires are
synthesised by aqueous Suzuki poly-
merisation, using hydrophobic binding
to promote threading of the cyclodex-
trin units. These polyrotaxanes have
conjugated polymer cores based on
spectra detect oligomers with up to ten
threaded cyclodextrins, and reveal the
presence of some defects that result for
oxidative homo-coupling of boronic
of these polyrotaxanes indicate that the
presence of the threaded cyclodextrin
macrocycles reduces the flexibility of
the conjugated polymer p-systems.
Both the solution and the solid-state
photoluminescence quantum yields are
enhanced upon threading of the conju-
gated polyaromatic cores through a- or
b-cyclodextrins, and the emission spec-
tra of the polyrotaxanes are blue-shift-
ed compared to the corresponding un-
threaded polymers. The greater weight
of the 0 0 transition in the emission
spectra, as well as the smaller Stokes
shift, indicate that the polyrotaxanes
are more rigid than the unthreaded
polymers.
acids.
Weight-average
molecular
weights were determined by analytical
ultracentrifugation, demonstrating that
step-growth polymerisation is efficient
enough to achieve degrees of polymeri-
sation up to ꢀ20 repeat units (84 para-
phenylenes). The fluorescence spectra
poly(para-phenylene),
polyfluorene,
and poly(diphenylene-vinylene),
threaded through 0.9 1.6 cyclodextrins
per repeat unit. Bulky naphthalene-3,6-
disulfonate endgroups prevent the mac-
rocycles from slipping off the conjugat-
ed polymer chains. Dialysis experi-
ments show that the cyclodextrins
become unthreaded only if smaller
stoppers are used. MALDI TOF mass
Keywords:
cyclodextrins
rotaxanes
¥
¥
nanotechnology
self-assembly
chemistry
¥
¥
supramolecular
Introduction
However, inter-chain charge-mobilities can also be fairly
high,[5] and p p stacking interactions between polymer
chains can dramatically affect the optical behaviour, for ex-
ample, by broadening the absorption and emission spectra
and quenching the fluorescence.[6] This makes it interesting
to study insulated molecular wires in which the conjugated
polymer backbone is encapsulated at the molecular level by
a protective sheath, limiting inter-chain interactions and en-
hancing the one-dimensional nature of the transport proper-
ties. Encapsulation can enhance the chemical stability[7] and
luminescent efficiency[8] of the conjugated core. When ex-
ploiting the electronic functionality of single molecules it is
important to prevent cross-talk, and the design of molecular-
ly isolated conducting structures may enable the size of elec-
tronic circuits to be reduced to molecular dimensions.[9] In-
sulated molecular wires have been created by threading con-
jugated polymers through cyclophanes,[10,11] cyclodextrins[12]
and zeolites,[9,13] by wrapping them in polymer helices,[14]
and by decorating them with dendritic side-chains.[15] Here,
we present some polyrotaxanes[16] consisting of a conjugated
polymer core threaded through a series of cyclodextrin mac-
rocycles.[17] Although many polyrotaxanes have been report-
ed,[18] and there are several examples of conjugated poly-
pseudorotaxanes,[11,12] to the best of our knowledge the ma-
Conjugated polymers have attracted widespread interest
due to their applications in light-emitting diodes,[1] field-
effect transistors[2] and photovoltaic devices.[3] They are
often loosely described as molecular wires because of the
high charge-mobility along individual polymer chains.[4]
[a] Dr. H. L. Anderson, Dr. J. J. Michels, M. J. O×Connell,
Dr. P. N. Taylor
Department of Chemistry, University of Oxford
Dyson Perrins Laboratory, South Parks Road, Oxford, OX1 3QY
(UK)
Fax : (+44)1865-275-674
E-mail: harry.anderson@chem.ox.ac.uk
[b] Dr. J. S. Wilson
Department of Physics, University of Cambridge
Cavendish Laboratory
Madingley Road, Cambridge, CB3 0HE (UK)
[c] Dr. F. Cacialli
Department of Physics and Astronomy, University College London
Gower Street, London, WC1E 6BT (UK)
Supporting information for this article is available on the WWW
under http://www.chemeurj.org/ or from the author. This includes syn-
thetic procedures and spectroscopic data for monomers 4a 6, and
crystallographic data for 4b and 4c.
6167
Chem. Eur. J. 2003, 9, 6167 6176
DOI: 10.1002/chem.200305245
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

FULL PAPER
terials reported here are the first genuine conjugated polyro-
taxanes with bulky endgroups preventing the macrocycles
from slipping off the chain.[17] This kinetic stability, with re-
spect to unthreading, is vital for purification and solution
processing.
We report the synthesis and chemical characterisation of
four water-soluble conjugated polyrotaxanes: 1ꢁb-CD, 2ꢁb-
CD, 3ꢁa-CD and 3ꢁb-CD (Figure 1), based on poly-para-
phenylene (PPP), polyfluorene (PF) and polydiphenylene-
vinylene (PDV) respectively, which illustrates the broad
scope of this synthetic approach. The molecular weights of
these polymers were determined by analytical ultracentrifu-
gation. In the case of 1ꢁb-CD, it also proved possible to
probe the degree of polymerisation by MALDI-TOF MS
and GPC. We prove that these insulated wires are true ro-
taxanes by testing for unthreading during dialysis as a func-
tion of the size of the endgroup. The absorption and emis-
sion spectra as well as luminescence efficiencies of the poly-
rotaxane are compared to those of the uninsulated wires.
Encapsulation was found to enhance the luminescence effi-
ciency and shifted the emission to shorter wavelengths.
Scheme 1. Synthesis of polyrotaxanes; polymerisation and capping are
aqueous Suzuki couplings.
esters 5a c were used instead of free boronic acids to facili-
tate monomer purification. Suzuki polymerisations were car-
ried out at 858C in aqueous lithium carbonate (0.17m)
under nitrogen. Similar polymers are obtained by running
the reaction at 458C, but a higher temperature was generally
used to ensure complete polymerisation and to increase the
solubility of the cyclodextrin. We employed palladium(ii)
acetate as the catalyst precursor, without adding phosphine
Results and Discussion
Synthesis and purification of polyrotaxanes: Our general
route to conjugated polyrotaxanes is an A+B step-growth
polymerisation, as outlined in Scheme 1. The hydrophobic
monomer (diboronic acid 5a c) is non-covalently bound
inside the macrocycle (a- or b-cyclodextrin) and polymer-
ised by Suzuki coupling[19] with a water-soluble monomer
(diiodide 4a c) to form a polypseudorotaxane. Finally, the
chains are terminated with bulky stoppers (iodonaphthalene
6) to prevent the macrocycles from slipping off the conjugat-
ed core. This whole sequence of reactions is carried out in
one pot, in an aqueous solution, and hydrophobic binding is
used to drive the threading process. The structures of the
components are summarised in Table 1.[20] Diboronic glycol
ligands.[21]
A polymerisation stoichiometry of 4a c:5a
c:6:Pd:CD = 1:(1+a/2+b):a:b:4.0, was used to give a 1:1
ratio of the two bifunctional monomers after some diboronic
acid has coupled to the stopper 6 and some has undergone
homo-coupling[22] reducing the PdII to the active Pd0 species
(evidence for this homo-coupling is provided by the
MALDI-TOF mass spectra discussed below). The predicted
number-average degree of polymerisation, n≈Nom (Figure 1)
can be calculated from the endgroup-to-monomer ratio (a)
with the simple equation n≈Nom = 2/a, assuming that the
Suzuki coupling reaction goes cleanly to completion, and ne-
glecting homo-coupling. Generally, we produced polyrotax-
Figure 1. Idealised structures of the four polyrotaxanes studied in this work. Note that these materials are polydisperse in two respects: each has a range
of chain lengths (n) and a range of cyclodextrin loadings. The relative orientation of cyclodextrins on each chain is probably random. When referring to
these polymers we generally include the nominal stoichiometry-predicted number-average degree of polymerisation, for example, n≈Nom = 10.
6168
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2003, 9, 6167 6176

Synthesis of Conjugated Polyrotaxanes
6167 6176
Table 1. Structures of component monomers and macrocycles for polyrotaxane synthesis.[20]
Diiodide
Diboronic
Endgroup
Macrocycle
Polyrotaxane
Polymer Backbone
b-CD
1ꢁb-CD
poly(para-phenylene) PPP
b-CD
2ꢁb-CD
poly(fluorene) PF
a-CD
b-CD
3ꢁa-CD
3ꢁb-CD
poly(diphenylene vinylene) PDV
anes with n≈Nom = 10 by the use of an endgroup-to-monomer
ratio of a = 0.2. In the next section we show that these pre-
dicted degrees of polymerisation compare well with experi-
mental values from analytical ultracentrifugation. Generally,
we used 10 mol% palladium catalyst (b = 0.1), although
similar results were also obtained with 3 mol% palladium.
With the PPP and PF polyrotaxanes, only b-cyclodextrin
was used to complex the aromatic core, whereas PDV-based
polyrotaxanes were produced with threaded a- and b-cyclo-
dextrin because stilbene units are slim enough to bind inside
the narrower a-cyclodextrin cavity.
Polyrotaxane characterisation: NMR and dialysis experi-
1
ments: A typical H NMR spectrum of 1ꢁb-CD[n≈Nom = 10]
is shown in Figure 2. Although the spectrum is poorly re-
solved, the ratio of the cyclodextrin and aromatic compo-
nents can readily be determined by integration. The small
Excess cyclodextrin was removed from the crude poly-
carboxylate polyrotaxanes 1ꢁb-CD and 2ꢁb-CD by precipi-
tation with acid and centrifugation. Redissolving the precipi-
tate in aqueous lithium carbonate and ultrafiltration through
a membrane with a nominal molecular weight cut-off
(NMWCO) of 5 kDa removed low molecular weight materi-
al, including any remaining unthreaded cyclodextrin. The
PDV-based rotaxanes 3ꢁa-CD and 3ꢁb-CD were not pre-
cipitated with acid on account of the low pKa of the sulfonic
acid residues. In these cases, free cyclodextrin was removed
by ultrafiltration. The crude products are generally contami-
nated with colloidal palladium catalyst residues, which are
too coarse to be removed by ultrafiltration and too fine to
be removed by a 0.1 mm filter. Several methods were tested
for the removal of these palladium particles. Dithiocarba-
mates have been found to be effective for dissolving palladi-
um,[23] so we tested N-dithiocarboxy glycine as a water-solu-
ble dithiocarbonate; however, it was ineffective at dissolving
these colloids. Fortunately, cyanide was found to be very ef-
fective. When aqueous alkaline solutions of the crude poly-
rotaxanes are treated with sodium cyanide, in the presence
of air, the solutions change from grey to clear in about five
minutes as the palladium dissolves to form Na2[Pd(CN)4],
which is removed by ultrafiltration. Analysis of the samples
before and after treatment with cyanide showed that the
wt% Pd decreased from 0.6% to <0.02%. Polymers 1[n≈Nom
= 10], 2[n≈Nom = 10] and 3[n≈Nom = 10] were synthesised and
purified as reference compounds using identical procedures.
Figure 2. 1H NMR (250 MHz, D2O) spectrum of 1ꢁb-CD[n≈ = 10], T =
208C (a residual HOD signal at d = 4.8 ppm has been deleted).
signal at dꢀ8.5 ppm is assigned to protons HA and HB of
the naphthalene disulfonate endgroups (see Table 1; this as-
signment is based on 2D NMR studies on the related [2]ro-
taxane[17a]). Integration of these endgroup signals provides
experimental values for the number-average degree of poly-
merisation n≈Nom, which compare well with values from the
polymerisation stoichiometry, n≈Nom, as shown in Table 2. In
general, endgroup integration is an unreliable method of
molecular weight determination because accidental termina-
tion or incomplete coupling[24] can result in chains with un-
expected termini. However, with these polyrotaxanes, the
fact that the cyclodextrins do not slip off the polymer back-
bones proves that the great majority of chains are terminat-
1
ed with bulky naphthalene stoppers, as shown below. The H
NMR spectra of the reference polymers 1 3 are broader
6169
Chem. Eur. J. 2003, 9, 6167 6176
www.chemeurj.org
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

H. L. Anderson et al.
FULL PAPER
Table 2. Molecular weight parameters for polyrotaxanes and reference polymers.[a]
er unthreading occurs during
dialysis. Two versions of poly-
n¯[mLgꢃ1
]
MW (AUC)
n≈AUC
≈
≈
Polymer
n≈Nom
y≈NMR
n≈NMR
MW(calcd)
para-phenylene
were synthesised, one with
large naphthalene disulfonate
polyrotaxane
1ꢁbCD
10
10
10
10
10
10
10
1.1
1.1
0.9
1.6
0
7.7
10.5
7.9
7.7
11
34700
36400
29500
49300
8540
0.61
0.66
0.61
0.62
0.62
0.65
0.55
36800
27100
25800
53500
13900
22500
13400
10.6
7.3
8.7
10.9
16.6
22.8
12.1
2ꢁbCD
3ꢁaCD
3ꢁbCD
endgroups, 1ꢁb-CD[n≈Nom
=
1
2
3
10], and one with small meta-
benzoic acid terminals 1aꢁb-
CD[n≈Nom = 10] (Figure 3). The
crude polyrotaxane reaction
mixture (which initially con-
tained excess cyclodextrin) was
0
0
10200
11100
[a] n≈Nom is the stoichiometry-predicted number-average degree of polymerisation; y≈NMR is the number-average
threading ratio determined by 1H NMR, ꢂ10%; n≈NMR is the number-average degree of polymerisation from
≈
endgroup integration, ꢂ10% for the polyrotaxanes and ꢂ20% for 1; MW (calcd) is the weight-average molec-
ular mass calculated from n≈Nom and y≈NMR ; n¯ is the partial specific volume from oscillating capillary densitome-
≈
try,ꢂ2%; MW (AUC) is the weight-average molecular mass from analytical ultracentrifugationꢂ10%; n≈AUC is
dialyzed
with
a
5 kDa
The
≈
the average degree of polymerisation calculated from MW (AUC) and y≈NMR, with an uncertainty ofꢂ10%.
NMWCO
membrane.
threading ratio, y≈NMR, was moni-
than those of the polyrotaxanes (probably as a result of ag-
gregation), making endgroup integration less reliable with 1
and impossible with 2 and 3.
tored as a function of the volume of water eluted through
the dialysis ultrafiltration cell (Figure 3). It is clear that in
the case of 1aꢁb-CD[n≈Nom = 10] the cyclodextrins com-
pletely unthread during dialysis, whereas the threading ratio
of 1ꢁb-CD[n≈Nom = 10] levels off at y≈NMR = 1.1. This proves
that polyrotaxanes with naphthalene disulfonate endgroups
are genuine rotaxanes, and that the great majority of chains
have naphthalene stoppers at both ends. This experiment
also demonstrates that b-cyclodextrin is able to slip over the
carboxylic acid substituents along the polymer backbone.
Similar experiments showed that polypseudorotaxane ana-
logues of 2ꢁb-CD and 3ꢁb-CD, which lack the naphthalene
endgroups, undergo slow unthreading, whereas those of
3ꢁa-CD do not unthread because a-cyclodextrin is too
narrow to slip over the sulfonated stilbene units. We also
tested the rate of unthreading in 1aꢁb-CD as a function of
the chain length, and showed that shorter chains unthread
more rapidly. For example, when the experiment shown in
Figure 3 was carried out with 1aꢁb-CD[n≈Nom = 4] the
threading ratio dropped to y≈NMR <0.1 after only 2.5 L of
water had been eluted.[25] Unthreading of higher molecular
weight polypseudorotaxanes is accompanied by precipitation
of the unthreaded polymer because the presence of the
threaded cyclodextrin increases the solubility.
The polyrotaxanes are shown in Figure 1 with one
threaded cyclodextrin per unsubstituted biphenyl, fluorene
or stilbene unit. We define the threadingratio as y = x/(n
+ 1), as shown in Figure 3, so this idealised representation
corresponds to a threading ratio of y = 1. The threading
ratio of each batch of polyrotaxane was determined by inte-
1
gration of the aromatic and cyclodextrin H NMR signals to
give the values listed in Table 2. All of the polyrotaxanes
have threading ratios of y≈NMR = 1.0ꢂ0.1, except for 3ꢁb-
CD, where the higher threading ratio (y≈NMR = 1.6) indicates
that two b-cyclodextrins are quite easily accommodated on
the unsubstituted stilbene unit.
To confirm that the naphthalene disulfonate endgroups
are bulky enough to retain b-cyclodextrin, we tested wheth-
MALDI-TOF mass spectra: Figure 4 shows a negative ion
MALDI-TOF MS of 1ꢁb-CD[n≈Nom = 10]. The spectrum
consists of families of peaks with different numbers of
threaded cyclodextrins (x) for each number of repeat units
(n). It is interesting that some peaks are observed with a
threading ratio greater than unity (x>n + 1) showing that
each unsubstituted biphenyl is able to accommodate up to
two cyclodextrin units. The space-filling model of 1ꢁb-
CD[n = 2; y = 1] in Figure 4 (top) shows that at a thread-
ing ratio of unity, there is still space between the cyclodex-
trins for a second macrocycle to be accommodated on at
least some repeat units. The number-average threading ratio
increases with increasing number of repeat units, n, and is
greater than unity when n>5. The lower threading ratios
for the shorter oligomers indicate that cyclodextrins are less
easily accommodated near the naphthalene endgroups. Only
one type of minor defect peak can be identified in the spec-
1
Figure 3. Plot of threading ratio from H NMR integration [y≈NMR = x/(n
+
1)] versus volume of water eluted during dialysis, with a 5 kDa
NMWCO membrane using polyrotaxane 1ꢁb-CD and pseudopolyrotax-
!
trum: peaks at M+152 marked ™ ∫ are assigned to chains
ane 1aꢁb-CD, both with n≈Nom = 10.
with one extra homo-coupled biphenyl unit, from oxidative
6170
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2003, 9, 6167 6176

Synthesis of Conjugated Polyrotaxanes
6167 6176
amide gel electrophoresis, rela-
tive to denatured protein mark-
ers, but this gave unrealistically
high
molecular
weights,[17a]
which is not surprising because
the polyrotaxanes do not re-
semble proteins. Next, we
turned to analytical GPC in a
variety of solvent systems and
calibrated with poly(ethylene
glycol) standards. When a high
ionic-strength eluant (400 mm
LiNO3, 1.5 mm LiOH, 1.0 mm
Li2CO3) was used to suppress
electrostatic interactions be-
tween polyelectrolyte chains,
we obtained plausible results
for 1ꢁb-CD[n≈Nom
= 10] and
≈
3ꢁb-CD[n≈Nom = 10] (Mw = 20
and 13 kDa, respectively), but
2ꢁb-CD[n≈Nom = 10] and 3ꢁa-
CD[n≈Nom = 10] gave very long
elution times corresponding to
unrealistically low molecular
≈
weights (M<1 kDa); evidently,
these materials adsorbonto the
column material. The unthread-
ed polymers 1 3 completely
adsorbonto the stationary
phase under these conditions
and gave no detectable peaks
(UV detector, 330 nm). GPC
with DMSO as the eluant also
failed to give satisfactory re-
Figure 4. MALDI-TOF MS spectrum of 1ꢁb-CD[n≈Nom = 10], obtained with a-cyano-4-hydroxycinnamic acid
as the matrix (negative mode). Triangles indicate molecular ions arising from chains with one defect as the
result of an extra biphenyl unit. Colours indicate peaks from oligomers with n = 2 (green), n = 3 (red), n =
4 (blue), n = 5 (purple), n = 6 (mustard) and n = 7 (turquoise). The space-filling model (top) shows an
energy-minimised structure of 1ꢁb-CD[n = 2; y = 1] with the cyclodextrin units shown in green (MM2 force
field, CACHe 4.1.1).
coupling of 4a by the PdII catalyst precursor. At first sight,
the pattern of these defect peaks seems surprising: for ex-
ample, there is a substantial M+152 defect peak at 5074 Da
corresponding to n = 2, x = 3, but no significant parent
peak for this n,x combination at 4922; similarly there is an
defect orphan peak at 7737 Da corresponding to n = 3, x =
5, and the defect peak at 10400 Da corresponding to n = 4,
x = 7 is larger than its parent. All these apparent anomalies
make sense when one realises that an extra hydrophobic bi-
phenyl unit is generally associated with an extra threaded b-
cyclodextrin (so that it is really an M+1287 defect). We
were unable to obtain satisfactory MALDI-TOF MS spectra
of the PDV and PF polyrotaxanes, or of any of the reference
polymers. The ability to transfer these polyrotaxanes into
the gas phase by MALDI is remarkably sensitive to the de-
tails of the molecular structure. Even when good MALDI
spectra are observed, they do not provide quantitative infor-
mation on the molecular weight distribution because shorter
chains are transferred into the gas phase more easily, as il-
lustrated by the data in Figure 4 (n≈NMR = 7.7; n≈AUC = 10.6;
n≈MALDI = 3.6).
sults. Molecular weight determination by small angle neu-
tron scattering (SANS) was also unsuccessful.[26]
Analytical ultracentrifugation (AUC) by equilibrium
sedimentation[27] proved to be the most practical method of
molecular weight determination. Although this technique is
routinely used to determine the molecular weights of biopol-
ymers, it is applied less often to synthetic polymers. When a
polymer solution is centrifuged for about 20 h at an angular
velocity of typically 1 3î104 rpm, equilibrium is reached be-
tween diffusion and sedimentation. A concentration gradi-
ent is established over the cell, which for a polydisperse
solute with ideal behaviour is given by Equation (1),[27a,28]
≈
where c(r) and Mw (r) are the concentration and the weight-
average molecular weight of the polymer as a function of r
the radial distance (normally given in cm from the centre of
rotation); n¯ is the partial specific volume of the polymer, 1
is the density of the solution, w is the rotor speed in radians
per second, R is the gas constant and T is the temperature.
2
ꢀ
ꢀ
dcðrÞ
MwðrÞð1ꢃn1Þw rcðrÞ
ð1Þ
¼
dr
RT
Molecular weight determination: It is often difficult to deter-
mine the molecular weights of polyelectrolytes, and this was
the case with our polyrotaxanes. Initially, we used polyacryl-
The weight-average molecular weight of the whole
≈
sample (Mw) is obtained from the average slope of a plot of
ln(c) versus r2. In practice, the concentration gradient is
6171
Chem. Eur. J. 2003, 9, 6167 6176
www.chemeurj.org
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

H. L. Anderson et al.
FULL PAPER
measured by UV/Vis absorption, assuming that the solute
ent mole ratios of the mono-functional stopper 6 (n≈Nom = 4,
follows Beer×s law, and c is substituted by absorbance (A),
so Mw is determined with Equation (2).
10, 15, 25 and 50) and comparing their molecular weights by
AUC. The variation in Mw(AUC) with n≈Nom is plotted in
[27d]
≈
≈
Figure 6. The straight line shows the ideal case, expected in
the absence of accidental termination. At n≈Nom
Mw(AUC) is higher than expected, because some short
oligomers are lost during precipitation and dialysis, whereas,
= 4,
2RT
dln A
dr2
ꢀ
MwðAUCÞ ¼
2 h
i
ð2Þ
≈
ꢀ
ð1ꢃn1Þw
All our polymers have high molar extinction coefficients,
so low concentrations of polymer can be used (0.5
2 mgmLꢃ1), minimising non-ideal effects such as aggregation
and electrostatic interactions between the polyelectrolyte
chains. The experiments were carried out with a background
salt (I = 6.0 mm) in order to further minimise non-ideal be-
haviour (see the Experimental Section). Figure 5 shows typi-
≈
Figure 6. Plot of Mw(AUC) versus n≈Nom for 1ꢁb-CD. The straight line
shows the ideal behaviour in the absence of accidental termination or
fractionation during purification (assuming a threading ratio of y≈ = 1.1).
≈
at high n≈Nom, the Mw(AUC) values become lower than pre-
dicted because accidental termination limits the molecular
≈
weight. Mw(AUC) reaches a limit of ꢀ70 kDa, which corre-
sponds to an average chain length of approximately
20 repeat units. This implies that side-reactions limit the effi-
ciency of the polymerisation reaction to ꢀ95%. The same
conclusion was drawn from GPC analysis of these 1ꢁb-CD
Figure 5. Sedimentation equilibrium profiles for the polyrotaxanes 3ꢁa-
CD and 3ꢁb-CD, both with a nominal degree of polymerisation of n≈Nom
= 10. A is the optical density of the solution at radial distance r; l =
398 nm; aqueous buffer: 1.25 mm LiOH, 1.58 mm Li2CO3; rotor speed =
15000 rpm; T = 208C; cell path length = 12 mm; stationary absorption
at 398 nm, A0 = 0.25.
≈
samples, although Mw values from GPC were systematically
lower by a factor of ꢀ2.5. Extensive dialysis of ™1ꢁb-
CD[n≈Nom = 50]∫ confirms that it is not a true polyrotaxane;
b-cyclodextrin slowly unthreads and results in precipitation.
cal AUC data for 3ꢁa-CD and 3ꢁb-CD; here the steeper
gradient for 3ꢁb-CD indicates a higher molecular weight,
mainly on account of the higher threading ratio and the
greater mass of b-CD.
Absorption and fluorescence: The electronic absorption and
emission spectra of these polyrotaxanes reveal how the con-
jugated p-system is affected by encapsulation. We have com-
pared the behaviour in dilute aqueous solution[31] and in the
solid state, as thin films spin-coated on silica substrates. The
absorption and emission bands of the polyrotaxanes are gen-
erally sharper than those of the unencapsulated reference
polymers. This can be attributed to reduced inter-chromo-
phore interaction, and increased conformational homogenei-
ty and rigidity in the polyrotaxanes. Examples of normalised
absorption and emission spectra for the PF derivatives 2ꢁb-
CD[n≈Nom = 10] and 2[n≈Nom = 10] are shown in Figures 7
and 8. Here the sharper emission of the polyrotaxane ena-
bles vibrational structure to be observed in solution at room
temperature. At low temperature (e.g. 10 K, Figure 8) the
vibrational structure is beautifully resolved in the solid-state
emission spectra of both the polyrotaxane and the reference
polymer. Cyclodextrin encapsulation increases the intensity
of the 0 0 band, relative to that of the 0 1 band; this is seen
most clearly in the low-temperature spectra, but is also evi-
The partial specific volumes of all seven polymers were
determined densitometrically with the oscillating capillary
technique.[29] This gave the values for n¯ in Table 2, which
≈
were used to determine Mw(AUC) with Equation (2). The
≈
expected molecular masses Mw(calcd) were calculated from
n≈Nom by numerical integration of the Flory distribution of
oligomers,[30] and experimental number-average degrees of
polymerisation n≈AUC were determined by calculating the
mole ratio of 6 that would have given a Flory distribution
≈
with the observed Mw(AUC). There is generally good agree-
≈
ment between the experimental and calculated values of Mw
and n≈. The slightly low values of n≈AUC for 2ꢁbCD and
3ꢁaCD may reflect some accidental termination or incom-
plete coupling, whereas the high apparent n≈AUC for polymers
1--3 are probably cased by the aggregation of these un-
threaded polymers during the AUC measurements.
We also tested the efficiency of the Suzuki polymerisa-
tion by preparing a range of samples of 1ꢁb-CD with differ-
6172
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2003, 9, 6167 6176

Synthesis of Conjugated Polyrotaxanes
6167 6176
The wavelengths of the absorption and emission maxima
increase in the order PPP<PF<PDV, reflecting the extent
of conjugation in their conjugated polymer backbones.[32]
For each polymer backbone, the emission spectra of the pol-
yrotaxane are shifted to shorter wavelength relative to the
reference polymer (Table 3). The smaller Stokes shift in the
polyrotaxanes is partly a result of the increased 0 0:0 1 in-
tensity ratio; however in those spectra where it is resolved,
the 0 0 peak is also blue-shifted in the case of the polyrotax-
anes, implying that encapsulation reduces the amount of re-
organisation in the excited state. The absorption spectra are
less affected by encapsulation.
The fluorescence quantum yields in Table 3 show that
the polyrotaxanes exhibit enhanced fluorescence efficien-
cies, both in solution and in the solid state. A similar effect
was observed in the analogous [2]rotaxanes[8] and can be at-
tributed to the reduced flexibility of the encapsulated p-
system, as well as to reduced solvent-quenching (in dilute
solution) and reduced aggregation (in the solid state). It is
interesting that 3ꢁb-CD has a significantly higher fluores-
cence quantum yield than 3ꢁa-CD, both in solution and in
the solid state, whereas in the [2]rotaxanes a-cyclodextrin
gives a greater fluorescence enhancement. The higher fluo-
rescence efficiency of the b-cyclodextrin polyrotaxane is
probably caused by its higher threading ratio (Table 2).
Light-emitting diodes have been fabricated from all seven of
these polymers and the electroluminescence quantum yields
of the polyrotaxanes are found to be enhanced more strong-
ly than their fluorescence quantum yields.[17b] The details of
these electroluminescence results will be reported shortly.
Figure 7. Normalised absorption spectra (bold) and emission spectra
(plain) of 2ꢁb-CD (c) and 2 (a) in aqueous solution (both n≈Nom
=
10; 1.66 mm LiOH, 2.11 mm Li2CO3 aqueous buffer, pH 9; emission spec-
tra with excitation at 361 nm).
Conclusion
An efficient preparation of water-soluble conjugated polyro-
taxanes, based on poly-para-phenylene, polyfluorene and
poly-para-diphenylenevinylene, has been developed. Polyro-
taxane synthesis is accomplished by polymerisation of cyclo-
dextrin-bound monomers by means of an aqueous Suzuki
1
coupling. H NMR shows that the polymer chains are effi-
ciently threaded with cyclodextrins, with approximately one
cyclodextrin per unsubstituted biphenyl, fluorene or stilbene
unit. This cyclodextrin-loading leaves some regions of the
conjugated backbone exposed, which probably facilitates
charge-transport through bulk polyrotaxanes.[17b] The naph-
thalene disulfonate endgroups are bulky enough to prevent
the cyclodextrins from unthreading, even after extensive di-
alysis. MALDI-TOF mass spectra of PPP-based polyrotax-
anes show the expected pattern of molecular ions for poly-
rotaxanes with different lengths and different numbers of
threaded cyclodextrins, and confirm the presence of some
homo-coupled biphenyl defects. Analytical ultracentrifuga-
tion gave weight-average molecular weights which compare
well with those predicted from the polymerisation stoichio-
metries, for polymers with up to 20 repeat units (which cor-
responds to 84 para-phenylene residues). The fluorescence
spectra of the polyrotaxanes show that the presence of the
cyclodextrins reduces the flexibility of the p system, as man-
ifested by the sharper emission spectra, the increased 0 0
Figure 8. Normalised emission spectra of 2ꢁb-CD (c) and 2 (a) in
the solid state at a) 290 K and b) at 10 K (both n≈Nom = 10).
dent at 290 K. According to the Franck Condon principle,
this change in band shape implies that the change in geome-
try between the excited state and the ground state is smaller
in the case of the polyrotaxane. In other words, encapsula-
tion increases the rigidity of the p system.
6173
Chem. Eur. J. 2003, 9, 6167 6176
www.chemeurj.org
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

H. L. Anderson et al.
FULL PAPER
Table 3. Wavelengths of the absorption and emission maxima and photoluminescence quantum yields.[a]
low molecular weight material, was re-
moved
CD[n≈Nom
obtained as
NMR (250 MHz, D2O): d
(4H), 8.6 6.7 ([16
by
=
ultrafiltration.
10] (230 mg, 66%) was
pale yellow film. 1H
8.6 8.4
18n]H;
3ꢁa-
Solution spectra
Solid-state thin-film spectra
Polymer
labs [nm]
lem [nm]
Fsolution
labs [nm]
lem [nm]
Ffilm
a
1ꢁb-CD
338
383
397
396
338
380
400
390
412
447
444
395
424
478
0.65
0.74
0.18
0.26
0.31
0.19
0.10
337
378
390
391
340
384
395
433
428
510
509
440
442
540
0.15
0.072
0.11
0.17
0.12
=
2ꢁb-CD
+
n =
3ꢁa-CD
7.9ꢂ1), 5.2 4.8 (6xH), 4.2 2.9 ppm
7.1ꢂ1); 13C NMR
3ꢁb-CD
(36xH;
x
=
1
2
3
(125 MHz, D2O): d = 138 142, 124
130, 102.0, 81.2, 73.6, 72.2, 71.8,
59.8 ppm; UV (H2O): lmax = 205, 234,
397 nm; elemental analysis calcd[33]
(%) for C586H693O323S21Li21¥56H2O: C
47.76, H 5.51, Li 1.00; found: C 47.79,
H 5.71, Li 0.64, B<0.1.
0.048
0.043
[a] Solution spectra were recorded in a 1.66 mm LiOH, 2.11 mm Li2CO3 aqueous buffer (pH 9) at concentra-
tions of ꢀ1 mm for absorption and 0.1 mm for emission. The estimated errors in the solution and solid-state
quantum yields (Fsolution and Ffilm) are less than 10%. All polymers are n≈Nom = 10.
Compound 3ꢁb-CD[n≈Nom = 10]: This
polyrotaxane was prepared by the
Franck Condon factors, and the higher fluorescence quan-
same procedure as 3ꢁa-CD[n≈Nom = 10], but with diiodide 4c (83 mg,
0.13 mmol), diboronic ester 5c (50 mg, 0.16 mmol), 1-iodonaphthalene-
3,6-disulfonate disodium salt (6, 12 mg, 0.026 mmol), b-cyclodextrin
(703 mg, 0.620 mmol), [Pd(OAc)2] (3.1 mg, 0.014 mmol), Li2CO3 (61 mg,
0.83 mmol) and NaCN (50 mg, 1.0 mmol) [Toxic!]. 3ꢁb-CD[n≈Nom = 10]
(160 mg, 48%) was obtained as a pale yellow film. 1H NMR (250 MHz,
D2O): d = 8.6 8.4 (4H), 8.6 6.3 ([16 + 18n]H; n = 7.7ꢂ1), 5.2 4.8
(7xH), 4.2 2.9 ppm (42xH, x = 12ꢂ1); 13C NMR (125 MHz, D2O): d =
tum yields, both in solution and in the solid state. The thor-
ough chemical characterisation of these insulated molecular
wires prepares the way for a full investigation of their solid-
state physics and semiconducting properties.
138 142, 124 130, 102.0, 80.7, 73.3, 72.2, 60.2 ppm; UV (H2O): lmax
=
205, 234, 397 nm; elemental analysis calcd[33] (%) for C1177H1601O769-
S27Li27¥86H2O: C 46.15, H 5.83, Li 0.60; found: C 46.17, H 5.29, Li 0.64,
B<0.1.
Experimental Section
Compound 1ꢁb-CD[n≈Nom = 10]: A two-necked flask was charged with
diiodide 4a (100 mg, 0.20 mmol), diboronic ester 5a (71 mg, 0.24 mmol),
1-iodonaphthalene-3,6-disulfonate disodium salt (6, 19 mg, 0.040 mmol),
b-cyclodextrin (929 mg, 0.818 mmol), [Pd(OAc)2] (4.5 mg, 0.020 mmol),
and Li2CO3 (90 mg, 1.2 mmol) and N2-saturated water (7 mL) was added
by syringe. The mixture was stirred over night at 858C. The colour
changed from reddish brown to black during the first hour. The solution
was diluted with water (10 mL) and passed through a 0.22 mm filter (cel-
lulose ester membrane). The solution was acidified to pH 1 with HCl (2m
aq.) to give a gelatinous precipitate, which was separated by centrifuga-
tion (5 min at 4000 rpm). The crude product was redissolved in aqueous
Li2CO3 (40 mL, 90 mm, pH 10) and stirred with NaCN (50 mg, 1.0 mmol)
[Toxic!] for 5 h. During this time, the colour changed from dark brown
to pale yellow. The product was further purified from low-molecular
weight material by ultrafiltration (polysulfone membrane; NMWCO
5 kDa), then passed through a 0.22 mm filter and evaporated to yield
Compound 1[n≈Nom
= 10]: This polymer was prepared analogously to
1ꢁb-CD[n≈Nom = 10], but with diiodide 4a (168 mg, 0.340 mmol), dibor-
onic ester 5a (120 mg, 0.408 mmol), 1-iodonaphthalene-3,6-disulfonate
disodium salt (6, 31 mg, 0.068 mmol), [Pd(OAc)2] (7.6 mg, 0.034 mmol),
Li2CO3 (151 mg, 2.04 mmol) and NaCN (80 mg, 1.6 mmol) [Toxic!].
1[n≈Nom = 10] (87 mg, 45%) was obtained as a pale beige film. 1H NMR
(250 MHz, D2O): d=8.6 8.4 (4H), 8.4 6.4 ([14 + 14n]H; n = 11ꢂ2);
13C NMR (125 MHz, D2O): d = 178, 142 136, 131, 130 124; UV (H2O):
lmax = 208, 337 nm; anal. calcd[33] (%) for C461H249O78S4Li37¥80H2O: C
62.37, H 4.66, Li 2.90; found: C 62.77, H 4.62, Li 2.91, B<0.1, Pd<0.02.
Compound 2[n≈Nom
= 10]: This polymer was prepared analogously to
1ꢁb-CD[n≈Nom = 10], but with diiodide 4b (184 mg, 0.327 mmol), dibor-
onic ester 5b (120 mg, 0.392 mmol), 1-iodonaphthalene-3,6-disulfonate
disodium salt (6, 30 mg, 0.065 mmol), [Pd(OAc)2] (7.3 mg, 0.033 mmol),
Li2CO3 (145 mg, 1.96 mmol) and NaCN (80 mg, 1.6 mmol) [Toxic!]. No
colour change was observed after stirring for 1 h with NaCN. After heat-
ing the solution for 1 min at 608C, the dark brown colour disappeared
readily. The solution was stirred for an additional 18 h at room tempera-
ture. 2[n≈Nom = 10] (177 mg, 97%) was obtained as a light brownish green
1ꢁb-CD[n≈Nom
=
10] (200 mg, 54%) as a pale beige film. 1H NMR
(250 MHz, D2O): d = 8.6 8.4 (4H), 8.4 7.2 ([14 + 14n]H; n = 7.7ꢂ1),
5.2 4.8 (7xH), 4.2 2.9 ppm (42xH; x = 8.5ꢂ1); 13C NMR (125 MHz,
D2O): d
= 178, 142 134, 131, 128 124, 102, 81, 74, 73, 60 ppm; UV
(H2O): lmax
=
211, 338 nm; elemental analysis calcd[33] (%) for
1
film. H NMR (250 MHz, D2O): d = 9.0 7.3 ([16 + 12n]H), 2.5 0.5 ppm
C774H953O449S4Li25¥53H2O: C 49.77, H 5.71, Li 0.91; found: C 49.77, H
5.71, Li 0.92, B<0.1, I<0.1, Pd<0.02.
([10 + 10n]H); 13C NMR (125 MHz, D2O): d = 185, 135 150, 120 130,
30 40 ppm; UV (H2O): lmax = 212, 379 nm; elemental analysis calcd[33]
(%) for C740H504O100S4Li48¥346H2O: C 50.20, H 6.81, Li 1.89; found: C
50.10, H 5.47, Li 1.59, B<0.1.
Compound 2ꢁb-CD[n≈Nom = 10]: This polyrotaxane was prepared by the
same procedure as 1ꢁb-CD[n≈Nom = 10] but with diiodide 4b (122 mg,
0.218 mmol), diboronic ester 5b (80 mg, 0.26 mmol), 1-iodonaphthalene-
3,6-disulfonate disodium salt (6, 20 mg, 0.044 mmol), b-cyclodextrin
(989 mg, 0.872 mmol), [Pd(OAc)2] (4.9 mg, 0.022 mmol), Li2CO3 (97 mg,
1.3 mmol) and NaCN (50 mg, 1.0 mmol) [Toxic!]. 2ꢁb-CD[n≈Nom = 10]
(340 mg, 86%) was obtained as a pale beige film. 1H NMR (250 MHz,
D2O): d = 8.6 8.4 (4H), 8.4 7.2 ([12 + 12n]H; n = 10.4ꢂ1), 5.2 4.8
(7xH), 4.3 2.9 ([2 + 2n + 42x]H, x = 11.5ꢂ1), 2.45 ([4 + 4n]H),
1.45 ppm ([4 + 4n]H); 13C NMR (125 MHz, D2O): d = 140 142, 120
Compound 3[n≈Nom
= 10]: This polymer was prepared analogously to
3ꢁa-CD[n≈Nom = 10], but with diiodide 4c (83 mg, 0.13 mmol), diboronic
ester 5c (50 mg, 0.16 mmol), 1-iodonaphthalene-3,6-disulfonate disodium
salt 6 (13 mg, 0.026 mmol), [Pd(OAc)2] (3.1 mg, 0.014 mmol), Li2CO3
(61 mg, 0.83 mmol) and NaCN (50 mg, 1.0 mmol) [Toxic!]. 3[n≈Nom = 10]
(45 mg, 57%) was obtained as a pale yellow film. 1H NMR (250 MHz,
D2O): d = 8.8 6.7 ppm ([18 + 18n]H); 13C NMR (125 MHz, D2O): d =
141, 137, 130 120 ppm; UV (H2O): lmax = 204, 400 nm; elemental analy-
sis calcd[33] (%) for C376H240O85S28Li28¥97H2O: C 50.26, H 4.88, Li 2.20;
found: C 50.36, H 5.44, Li 1.56, B<0.1.
128, 102.2, 81.2, 73.7, 72.5, 60.0 ppm; UV (H2O): lmax
= 215, 235,
383 nm; elemental analysis calcd[33] (%) for C682H842O367S4Li20¥61H2O: C
50.33, H 5.97, Li 0.84; found: C 50.36, H 6.14, Li 0.40, B<0.1.
Mass spectra: of 1ꢁb-CD were recorded on a Micromass Tof Spec2E
Compound 3ꢁa-CD[n≈Nom = 10]: This polyrotaxane was prepared by the
same procedure as 1ꢁb-CD[n≈Nom = 10], but with diiodide 4c (133 mg,
0.208 mmol), diboronic ester 5c (320 mg, 0.250 mmol), 1-iodonaphtha-
lene-3,6-disulfonate disodium salt (6, 19 mg, 0.040 mmol), a-cyclodextrin
(810 mg, 0.832 mmol), [Pd(OAc)2] (4.7 mg, 0.021 mmol), Li2CO3 (92 mg,
1.2 mmol) and NaCN (50 mg, 1.0 mmol) [Toxic!]. The PDV-based mate-
rials were not precipitated with acid. The excess cyclodextrin, as well as
MALDI-TOF mass spectrometer (with a nitrogen laser operating at
337 nm). a-Cyano-4-hydroxycinnamic acid was used as
a matrix. A
5 mgmLꢃ1 solution of polyrotaxane in water was mixed 1:1 with a
10 mgmLꢃ1 solution of the matrix in water/acetonitrile (70:30). 1.2 mL of
this mixture was applied to the target plate and allowed to dry. Negative
ion spectra were obtained by running the spectrometer in the linear
6174
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2003, 9, 6167 6176

Synthesis of Conjugated Polyrotaxanes
6167 6176
mode, with a laser power of typically 50% (course)/40 45% (fine), in
combination with a pulse voltage of 1200 V. Smoothing of the acquired
spectra revealed signals corresponding to high molecular weight material.
Acknowledgement
We thank Richard Friend (University of Cambridge, Physics) for gener-
ous support, Russell Wallis (University of Oxford, Biochemistry) for as-
sistance with AUC experiments, Alexander Krivokapic (University of
Oxford, Chemistry) for X-ray crystal structure determinations of 4b and
4c, Michael Hutchings (DyStar Ltd., Manchester) for generously provid-
ing disodium 1-aminonaphthalene-3,6-disulfonate as well as Stephen Har-
ding and Alan Rowe (University of Nottingham, National Centre for
Macromecular Hydrodynamics) for assistance with the determination of
partial specific volumes. We thank the EPSRC, the Marie Curie Fellow-
ship Association, the BBSRC and the Welcome Trust for financial sup-
port. F.C. is a Royal Society University Research Fellow.
Dialysis: was carried out in an Amicon 8200 ultrafiltration stirred cell
(200 mL) and polysulfone membranes (NMWCO 5 kDa). At least 3 L of
water was eluted through the ultrafiltration cell for each polymer sample
and under 4 bar nitrogen pressure.
Palladium analysis: A known mass of polymer (ꢀ10 mg) was dissolved in
hot, freshly prepared, piranha solution (4 mL, 7:3 conc. sulfuric acid/35%
aq. H2O2) in a 10 mL volumetric flask. Water (ꢀ6 mL) was added in
order to obtain the desired volume (10 mL) and the solution was homo-
genised by shaking the flask. The analysis was carried out on this solution
in a Thermo Jarrell Ash Atom Scan 16 spectrometer with inductive argon
plasma coupling.
Gel permeation chromatography: (GPC) was carried out on a Polymer
Laboratories instrument, equipped with refractive index and UV/Vis de-
tectors. Aqueous GPC was performed with
a basic buffer (400 mm
[1] J. H. Burroughes, D. D. C. Bradley, A. R. Brown, R. N. Marks, K.
Mackay, R. H. Friend, P. L. Burn, A. B. Holmes, Nature 1990, 347,
539 541; G. Gustafsson, Y. Cao, G. M. Treacy, F. Klavetter, N. Cola-
neri, A. J. Heeger, Nature 1992, 357, 477 479.
[2] H. Sirringhaus, N. Tessler, R. H. Friend, Science 1998, 280, 1741
1744.
[3] N. S. Sariciftci, D. Braun, C. Zhang, V. I. Srdanov, A. J. Heeger, G.
Stucky, F. Wudl, Appl. Phys. Lett. 1993, 62, 585 587; M. Granstrˆm,
K. Petritsch, A. C. Arias, A. Lux, M. R. Andersson, R. H. Friend,
Nature 1998, 395, 257 260.
[4] R. J. O. M. Hoofman, M. P. de Haas, L. D. A. Siebbeles, J. M.
Warman, Nature 1998, 392, 54 56.
[5] H. Sirringhaus, P. J. Brown, R. H. Friend, M. M. Nielsen, K. Bech-
gaard, B. M. W. Langeveld-Voss, A. J. H. Spiering, R. A. J. Janssen,
E. W. Meijer, P. Herwig, D. M. de Leeuw, Nature 1999, 401, 685
688.
[6] J. Cornil, D. A. dos Santos, X. Crispin, R. Silbey, J. L. Brÿdas, J. Am.
Chem. Soc. 1998, 120, 1289 1299.
[7] M. R. Craig, M. G. Hutchings, T. D. W. Claridge, H. L. Anderson,
Angew. Chem. 2001, 113, 1105 1108; Angew. Chem. Int. Ed. 2001,
40, 1071 1074.
[8] C. A. Stanier, M. J. O×Connell, W. Clegg, H. L. Anderson, Chem.
Commun. 2001, 493 494 and 787.
[9] T. Bein, P. Enzel, Angew. Chem. 1989, 101, 1737 1739; Angew.
Chem. Int. Ed. Engl. 1989, 28, 1692 1694.
[10] S. Anderson, R. T. Aplin, T. D. W. Claridge, T. Goodson, A. C.
Maciel, G. Rumbles, J. F. Ryan, H. L. Anderson, J. Chem. Soc.
Perkin Trans. 1 1998, 2383 2397.
[11] J. Buey, T. M. Swager, Angew. Chem. 2000, 112, 622 626; Angew.
Chem. Int. Ed. 2000, 39, 608 612; J.-P. Sauvage, J.-M. Kern, G.
Bidan, B. Divisia-Blohorn, P.-L. Vidal, New J. Chem. 2002, 26,
1287 1290.
[12] T. Shimomura, T. Akai, T. Abe, K. Ito, J. Chem. Phys. 2002, 116,
1753 1756; K. Yoshida, T. Shimomura, K. Ito, R. Hayakawa, Lang-
muir 1999, 15, 910 913; H. Okumura, Y. Kawaguchi, A. Harada,
Macromol. Rapid Commun. 2002, 23, 781 785.
[13] V. S.-Y. Lin, D. R. Radu, M.-K. Han, W. Deng, S. Kuroki, B. H.
Shanks, M. Pruski, J. Am. Chem. Soc. 2002, 124, 9040 9041; D. J.
Cardin, S. P. Constantine, A. Gilbert, A. K. Lay, M. Alvaro, M. S.
Galletero, H. Garcia, F. Marquez, J. Am. Chem. Soc. 2001, 123,
3141 3142; D. J. Cardin, Adv. Mater. 2002, 14, 553 563; C.-G. Wu,
T. Bein, Science 1994, 264, 1757 1759; T.-Q. Nguyen, J. Wu, S. H.
Tolbert, B. J. Schwartz, Adv. Mater. 2001, 13, 609 611.
[14] J. Stahl, J. C. Bohling, E. B. Bauer, T. B. Peters, W. Mohr, J. M.
MartÌn-Alvarez, F. Hamplel, J. A. Gladysz, Angew. Chem. 2002, 114,
1951 1957; Angew. Chem. Int. Ed. 2002, 41, 1871 1876.
[15] A. P. H. J. Schenning, J.-D. Arndt, M. Ito, A. Stoddart, M. Schreiber,
P. Siemsen, R. E. Martin, C. Boudon, J.-P. Gisselbrecht, M. Gross, V.
Gramlich, F. Diederich, Helv. Chim. Acta 2001, 84, 296 334; A. D.
Schl¸ter, J. P. Rabe, Angew. Chem. 2000, 112, 860 880; Angew.
Chem. Int. Ed. 2000, 39, 864 883; P. R. L. Malenfant, J. M. J. Frÿ-
chet, Macromolecules 2002, 35, 3634 3640.
LiNO3, 1.5 mm LiOH, 1.0 mm Li2CO3) as the eluant on PLaquagel-OH-
30 8 mm and PLaquagel-OH-40 8 mm columns (300î7.5 mm), calibrated
with poly(ethylene oxide) standards. 4-Acetylamino-5-hydroxy-naphtha-
lene-2,7-disulfonic acid was used as a flow-rate marker. Organic GPC
was performed with DMSO as the eluant on PLgel 5 mm MIXED-D col-
umns (300î7.5 mm), calibrated with polystyrene standards.
Analytical ultracentrifugation: Weight-average molecular weights were
determined by sedimentation equilibrium at 208C in a Beckman Coul-
terXLA Analytical Ultracentrifuge, equipped with an AN60 rotor and a
UV/Vis detection system. The samples were prepared by dissolving poly-
mer in aqueous buffer (1.25 mm LiOH/1.58 mm Li2CO3, pH 8.5). Each
sample was spun at three different initial loading concentrations, typically
in the range 0.5 2 mgmLꢃ1. Rotor speeds were chosen in the range
15000 20000 rpm for the polyrotaxanes, and 25000 30000 rpm for the
reference polymers; at higher rotor speeds, high molecular weight frac-
tions are deposited on the base of the cell resulting in a decrease in the
apparent molecular weight. Concentration profiles were monitored by
measuring the absorbance across the cell at lmax of the high-wavelength
absorption bands of the polymers (338 nm for 1ꢁb-CD and 1, 384 nm for
2ꢁb-CD and 2), and 398 nm for 3ꢁa-CD, 3ꢁb-CD, and 3. No residual
(baseline) absorption was observed after equilibrating at 48000 rpm. Mo-
lecular weight averages were obtained from the average slopes of lnA
versus r2 by means of Equation (2).[27] No dependence of the molecular
weight on either rotor speed or loading concentration was observed. Par-
tial specific volumes (n¯) were determined densitometrically[29] at the Na-
tional Centre for Macromolecular Hydrodynamics (Nottingham, UK), in
an Anton Paar DMA02C Precision Density Meter. All measurements
were carried out at 208C.
Photophysical measurements: UV/Vis and luminescence spectra were re-
corded on Perkin Elmer Lambda20 and Fluoromax-2 spectrometers. The
luminescence quantum yields of the polymers (FA) in basic aqueous solu-
tion (1.66 mm LiOH/2.11 mm Li2CO3, pH 9) were determined relative to
the fluorescence quantum yield (FB) of a solution of quinine hemisulfate
salt in 0.5m H2SO4 as a standard (FB
= 0.546), according to FA =
[(ABFAnA2)/(AAFBnB2)]FB, where A, F and n are absorbance, integrated
fluorescence intensity and refractive index, respectively. The refractive
indices of solutions of polymer and standard were assumed to be equal.
In a typical experiment, the fluorescence intensity of the sample and the
standard was measured at four different concentrations at low optical
density (A<0.1). FA was subsequently obtained from the slopes of plots
of F versus A. Solid-state fluorescence spectra were measured by means
of excitation with UV lines (355 365 nm) of an argon ion laser. Variable-
temperature spectra were recorded by placing the sample in a continu-
ous-flow helium cryostat capable of cooling to around 9 K, under a pres-
sure of ꢀ8 mbar of helium. Emission spectra were recorded on an Oriel
InstaspecIV spectrometer at a fixed geometry such that the relative pho-
toluminescence intensity at different temperatures can be compared.
Solid-state fluorescence quantum yields were measured with an integrat-
ing sphere
[16] A polyrotaxane is defined as a supramolecular structure with a poly-
mer backbone threaded through several macrocyclic units. There is
no covalent link between the macrocycles and the polymer chain,
6175
Chem. Eur. J. 2003, 9, 6167 6176
www.chemeurj.org
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

H. L. Anderson et al.
FULL PAPER
but the macrocycles are unable to slip off the ends of the polymer
on account of the presence of bulky terminal units.
on 1ꢁb-CD[n≈Nom = 10] and 1[n≈Nom = 10] in D2O by Dr D. G. Buck-
nall (Department of Materials, University of Oxford), with momen-
tum transfer range 0.009 0.25 äꢃ1. Scattering data from 1.25, 2.5
and 5% wt./wt. solutions all indicated concentration dependent ag-
gregation behaviour, as observed by excess scattering at momentum
transfer values of approximately 0.03, 0.05 and 0.07 äꢃ1, respective-
ly. This non-ideal behaviour prevents determination of single-chain
molecular dimensions; however, further analysis may allow extrac-
tion of characteristic length scales and apparent average molecular
weights of aggregates.
[17] For preliminary communications on the synthesis and electrolumi-
nescence of these polyrotaxanes, see: a) P. N. Taylor, M. J. O×Con-
nell, L. A. McNeill, M. J. Hall, R. T. Aplin, H. L. Anderson, Angew.
Chem. 2000, 112, 3598 3602; Angew. Chem. Int. Ed. 2000, 39,
3456 3460; b) F. Cacialli, J. S. Wilson, J. J. Michels, C. Daniel, C.
Silva, R. H. Friend, N. Severin, P. SamorÏ, J. P. Rabe, M. J. O×Con-
nell, P. N. Taylor, H. L. Anderson, Nat. Mater. 2002, 1, 160 164. The
synthesis of a b-cyclodextrin-threaded poly(azomethine) polyrotax-
ane has been reported recently: D. Nepal, S. Samal, K. E. Geckeler,
Macromolecules 2003, 36, 3800 3802; A. Farcus, M. Grigoras, J. Op-
toelectron. Adv. Mater. 2000, 2, 525 530; A. Farcus, M. Grigoras,
Polymer Int. 2003, 52, 1312 1320.
[27] a) M. Wales, J. Phys. Colloid Chem. 1948, 52, 235 246; b) C. Tan-
ford, Physical Chemistry of Macromolecules, Wiley, New York, 1961;
c) J. M. Creeth, R. H. Pain, Prog. Biophys. Mol. Biol. 1967, 17, 217
287; d) Analytical Ultracentrifugation in Biochemistry and Polymer
Science (Eds: S. E. Harding, A. J. Rowe, J. C. Horton), Royal Soci-
ety of Chemistry, Cambridge, 1992; e) D. Schubert, C. Tziatzios, P.
Schuck, U. S. Schubert, Chem. Eur. J. 1999, 5, 1377 1383.
[18] F. M. Raymo, J. F. Stoddart, Chem. Rev. 1999, 99, 1643 1663; K.
Kim, Chem. Soc. Rev. 2002, 31, 96 107; A. Harada, Acc. Chem. Res.
2001, 34, 456 464.
[19] A. Suzuki, J. Organomet. Chem. 1999, 576, 147 168; M. B. Goldfin-
ger, K. B. Crawford, T. M. Swager, J. Am. Chem. Soc. 1997, 119,
4578 4593; C. Kowitz, G. Wegner, Tetrahedron 1997, 53, 15553
15574; N. Miyaura, A. Suzuki, Chem. Rev. 1995, 95, 2457 2483;
J. C. Ostrowski, M. R. Robinson, A. J. Heeger, G. C. Bazan, Chem.
Commun. 2002, 784 785.
[28] For systems containing a polyelectrolyte with a background salt, it is
usually more accurate to utilise the density increment at constant
chemical potential of diffusible components, (d1/dc)m, instead of the
buoyancy factor, (1ꢃv≈1), due to the Donnan effect (see: Budd,
P. M. in Analytical Ultracentrifugation in Biochemistry and Polymer
Science (Eds.: S. E. Harding, A. J. Rowe, J. C. Horton), Royal Soci-
ety of Chemistry: Cambridge, 1992, pp.593 608). However, we
assume variations in n¯ across the cell at sedimentation equilibrium
to be small and not to be contributing significantly to the experi-
mental error.
[20] The synthesis and characterisation of compounds 4a c, 5a c and 6
are detailed in the Supporting Information to this article.
[21] N. A. Bumagin, V. V. Bykov, Tetrahedron 1997, 53, 14437 14450;
T. I. Wallow, B. M. Novak, J. Org. Chem. 1994, 59, 5034 5037; A. D.
Child, J. R. Reynolds, Macromolecules 1994, 27, 1975 1977; S. Kim,
J. Jackiw, E. Robinson, K. S. Schanze, J. R. Reynolds, J. Baur, M. F.
Rubner, D. Boils, Macromolecules 1998, 31, 964 974.
[29] O. Kratky, H. Leopold, H. Stabinger, Methods in Enzymology 1973,
27, 98 110.
≈
[30] Predicted Mw(calcd) values were obtained from the mole fraction of
[22] M. Moreno-MaÊas, M. Pÿrez, R. Pleixats, J. Org. Chem. 1996, 61,
2346 2351; K. A. Smith, E. M. Campi, W. R. Jackson, S. Marcuccio,
C. G. M. Naeslund, G. B. Deacon, Synlett 1997, 131 132.
[23] J. C. Nelson, J. K. Young, J. S. Moore, J. Org. Chem. 1996, 61, 8160
8168.
[24] We were unable to detect iodine or boron in these polymers by ele-
mental analysis (<0.1%) demonstrating the absence of unreacted
boronic acid or aryl iodide endgroups.
[25] The rate of unthreading in a-cyclodextrin/poly(imine) pseudorotax-
anes also decreases with decreasing chain length: G. Wenz, B.
Keller, Angew. Chem. 1992, 104, 201 203; Angew. Chem. Int. Ed.
Engl. 1992, 31, 197 199; W. Herrmann, B. Keller, G. Wenz, Macro-
molecules 1997, 30, 4966 4972.
mono-iodide 6 used in the polymerisation reactions by numerical in-
tegration of the Flory distribution of oligomers, assuming that all
aryl iodide units react at equal rate, that Suzuki coupling goes to
completion and that there are no accidental termination reactions.
[31] The fluorescence quantum yields were determined in aqueous solu-
tion at pH 9 (1.66 mm LiOH, 2.11 mm Li2CO3), which guaranteed
complete deprotonation of the polymer chains. Under more acidic
conditions, the quantum yields proved to be sensitive to pH, but no
changes were observed above pH 8. Fluorescence quantum yields
are independent of ionic strength up to 2m NaCl.
[32] M. Fukuda, K. Sawada, K. Yoshino, J. Polym. Sci. Polym. Chem.
Ed. J. Polym. Sci. Polym. Chem. 1993, 31, 2465 2471.
[33] Elemental analyses are compared with formulae calculated from
y≈NMR and n≈AUC (Table 2).
[26] Small-angle neutron scattering measurements (using the LOQ in-
strument, Rutherford Appleton Laboratory, UK) were performed
Received: June 18, 2003 [F5245]
6176
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2003, 9, 6167 6176