Controlling factors in the synthesis of cucurbituril and its homologues

Article abstract of DOI:10.1021/jo015897c

The acid-catalyzed synthesis of cucurbit[n]urils from formaldehyde and glycoluril is poorly understood. In this paper, we examine a wide range of reaction conditions that include the effects of acid type, acid concentration, reactant concentrations, and temperature to both probe the mechanism and optimize the yields of isolated cucurbit[n]urils, where n = 5-10. A mechanism for the formation of these cucurbit[n]urils is presented. Individual cucurbit[n]urils were unambiguously identified in reaction mixtures using ESMS and ¹³C NMR.

Full text of DOI:10.1021/jo015897c

8094
J . Org. Chem. 2001, 66, 8094-8100
Con tr ollin g F a ctor s in th e Syn th esis of Cu cu r bitu r il a n d Its
Hom ologu es
Anthony Day,* Alan P. Arnold,* Rodney J . Blanch,* and Barry Snushall
School of Chemistry, University College, University of New South Wales,
Australian Defence Force Academy, Canberra, ACT 2600, Australia
a-day@adfa.edu.au
Received J uly 6, 2001
The acid-catalyzed synthesis of cucurbit[n]urils from formaldehyde and glycoluril is poorly
understood. In this paper, we examine a wide range of reaction conditions that include the effects
of acid type, acid concentration, reactant concentrations, and temperature to both probe the
mechanism and optimize the yields of isolated cucurbit[n]urils, where n ) 5-10. A mechanism for
the formation of these cucurbit[n]urils is presented. Individual cucurbit[n]urils were unambiguously
identified in reaction mixtures using ESMS and 13C NMR.
Cucurbituril 1 is a remarkably robust macrocyclic host
Me10Q5. Until recently, all published work that refers to
molecule that has a rigid cavity capable of binding small
guests under a variety of conditions.1-12 It is synthesized
from glycoluril 2 (R ) H) and formaldehyde under acidic
conditions. Cucurbituril was first synthesized in 1905 by
Behrend et al.13 as an acid-soluble, crystalline product
that they were unable to fully characterize. It was not
until 1981, when Freeman, Mock, and Shih14 reexamined
this product that the highly ordered and symmetrical
structure of cucurbituril was found to be six glycoluril
units linked by pairs of methylene bridges derived from
formaldehyde. The structure of 1 resembles a flattened
sphere, truncated top and bottom, with a hollow core. In
1992, Flinn et al.15 synthesized a methyl-substituted
cucurbituril 3 that was determined to have five, rather
than six, units of dimethylglycoluril 2 (R ) Me). Accord-
ing to the nomenclature suggested by Flinn, 3 is
decamethylcucurbit[5]uril and Berhend’s original com-
pound 1 is cucurbit[6]uril. For brevity hereafter, we will
abbreviate cucurbit[n]uril as Qn, so 1 is Q6 and 3 is
cucurbituril really only describes work with 1, Q6.
With the discovery of Qn,16,17 we were puzzled by the
observation that Q6 is produced in preference to any
other Qn. The reasons for this have been unclear.
Semiempirical AM1 calculations suggest that there should
be no preference for a particular ring size, and the gas-
phase heats of formation, per glycoluril unit, appear to
be almost identical for Q4-10. The only substituted Qn
known at present is 3 and, semiempirical AM1 calcula-
tions for Me2nQn are similar to the results for Qn. The
report of 3, rather than the hypothetical Me12Q6, suggests
that methyl-methyl clashes may dominate the synthetic
process. However, molecular modeling does not support
this explanation for Me2nQn smaller than Me16Q8 (Sup-
porting Information, Table S1). The effects of hydrogen-
bonding interactions, selective solvation, and counterions
nevertheless could be controlling factors favoring the
preferential synthesis of Q6. These factors are not easily
included in AM1 calculations.
(1) Buschmann, H. J .; J ansen, K.; Schollmeyer, E. Thermochim. Acta
1998, 317, 95-98.
(2) Buschmann, H. J .; Schollmeyer, E. J . Inclusion Phenom. Mol.
Recognit. Chem. 1997, 29, 167-174.
(3) Cintas, P. J . Inclusion Phenom. Mol. Recognit. Chem. 1994, 17,
205-20.
(4) J eon, Y.-M.; Kim, J .; Whang, D.; Kim, K. J . Am. Chem. Soc. 1996,
118, 9790-9791.
(5) J eon, Y.-M.; Whang, D.; Kim, J .; Kim, K. Chem. Lett. 1996, 503-
504.
(6) Karcher, S.; Kornmueller, A.; J ekel, M. Acta Hydrochim. Hy-
drobiol. 1999, 27, 38-42.
(7) Meschke, C.; Buschmann, H.-J .; Schollmeyer, E. Polymer 1998,
40, 945-949.
(8) Meschke, C.; Buschmann, H. J .; Schollmeyer, E. Thermochim.
Acta 1997, 297, 43-48.
(9) Mock, W. L. Top. Curr. Chem. 1995, 175, 1-24.
(10) Neugebauer, R.; Knoche, W. J . Chem. Soc., Perkin Trans. 2
1998, 529-534.
The reaction conditions described for the original
synthesis of 1,13,18 and the minor variations to synthesize
(11) Taketsuji, K.; Tomioka, H. Nippon Kagaku Kaishi 1998, 670-
678.
(12) Whang, D.; Heo, J .; Park, J . H.; Kim, K. Angew. Chem., Int.
Ed. 1998, 37, 78-80.
(16) Day, A. I.; Blanch, R. J .; Arnold, A. P. PCT Int. Appl. WO 2000-
2000AU412 20000505. Priority: AU 99-232 19990507, Unisearch
Limited, Australia, 2000, p 112.
(17) Kim, J .; J ung, I.-S.; Kim, S.-Y.; Lee, E.; Kang, J .-K.; Sakamoto,
S.; Yamaguchi, K.; Kim, K. J . Am. Chem. Soc. 2000, 122, 540-541.
(18) Buschmann, H.-J.; Fink, H. Ger. Offen. DE 90-4001139 19900117.
Priority: DE 89-3912784 19890419, Deutsches Textilforschungszen-
trum Nord-West e. V., Fed. Rep. Ger., 1990, p 10.
(13) Behrend, R.; Meyer, E.; Rusche, F. J ustus Liebigs Ann. Chem.
1905, 339, 1-37.
(14) Freeman, W. A.; Mock, W. L.; Shih, N. Y. J . Am. Chem. Soc.
1981, 103, 7367-8.
(15) Flinn, A.; Hough, G. C.; Stoddart, J . F.; Williams, D. J . Angew.
Chem., Int. Ed. Engl. 1992, 31, 1475-7.
10.1021/jo015897c CCC: $20.00 © 2001 American Chemical Society
Published on Web 11/07/2001

Synthesis of Cucurbituril and Its Homologues
J . Org. Chem., Vol. 66, No. 24, 2001 8095
Ta ble 1. Obser ved P ositive-Ion ESMS P ea k s (m /z) of a
CsCl Solu tion of Q5-8.a ,b
Ta ble 2. Effect of Acid Typ e a n d Rea cta n t
Con cen tr a tion s on P r od u ct Distr ibu tion
[P + Cs]+ [P + 2Cs + Cl]+ [P + 2Cs]2+
(P + 132.9) (P + 300.8) (P + 265.8)/2
weight %a
[glycoluril]
mg/mL acid
n
parent ion, P
acidb
Q5
Q6
Q7
Q8c
5 C30H30N20O10 (830)
6 C36H36N24O12 (996)
7 C42H42N28O14 (1162) 1295.0
963.0
1129.1
1131.0
1297.0
1463.0
1628.9
548.0
631.0
714.0
797.2
155-190
155-190
155-190
155-190
155-190
155-190
155-190
0.125
concd H2SO4
9 M H2SO4
concd HCl
9 M HCl
12
23
17
17
9
28
5
58
3
88
44
48
50
52
43
61
42
44
<1
25
28
25
35
24
27
<1
8
7
8
4
8 C48H48N32O14 (1328)
1461.0
a
5 M HCl
Q8 is almost insoluble in aqueous CsCl solutions, and these
b
50% HBF4
melt TsOH
concd HCl
concd HCl
5
7
ions were observed only for pure material. Lower cone voltages
predominantly gave the double-charged ions.
1700d
33
12
Q5-8,17 were scrutinized in order to develop an under-
standing of the factors acting to produce Qn. In reexam-
ining the original, particularly harsh, reaction conditions,
we observed substantial quantities of ammonium salts
of Q6, suggesting decomposition of starting materials or
products. Milder reaction conditions might allow the
formation of other Qn or prevent their possible destruc-
tion. We examined a wide range of reaction conditions,
including the effects of acid type, acid concentration,
reactant concentrations, and temperature, to both probe
the mechanism and optimize the yields of individual Qn.
Early in this study, we recognized previously unde-
scribed trends in electrospray mass spectroscopy (ESMS)
and 13C NMR data that allowed us to identify known Qn
and hitherto unknown Q9 and Q10 in reaction mixtures.
Alkali metal cations have been shown to solubilize 1 by
complexation with the carbonyl portals,4,12,19,20 and we
find that they also solubilize Q5-10 to varying degrees.
ESMS analyses of CsCl/Qn solutions gave positive ions
(at a cone voltage of 100 V), with a major set of peaks for
each Qn as the associated Cs+, (2Cs)2+, and (2Cs + Cl)+
ions (Table 1).
a
Weight percentages were determined by 13C NMR and include
b
a correction factor (Supporting Information). Reaction temper-
atures were at 100 °C over a 2-3 h period, and in all cases, the
reaction mixtures were homogeneous throughout the process.
c Less than 1% of products that could be attributed to Qn where
n > 8 (except for 1700 mg/mL glycoluril in concentrated HCl).
Solid paraformaldehyde was used rather than the equally
effective 40% aqueous solution of formaldehyde. Solid paraform-
aldehyde enabled a high concentration of glycoluril.
d
in H2SO4 are unnecessary. We find that the reaction can
be carried out in one pot with any individual strong acid
and can be driven to completion to form a mixture of Qn
(Table 2). If heated to 100 °C, the reaction proceeds
through a number of physical changes and is virtually
complete within 2 h to give a nearly quantitative yield
of Qn. Following the progress of the reaction by either
1H or 13C NMR helped to define the intermediate oligo-
mers and the conditions under which they form.
The 13C NMR spectrum of the initial product(s) gener-
ated at room temperature showed four major broad, cone-
shaped peaks at 52-55.2, 67.0-70.0, 70.05-73.0, and
157.5-158.5 ppm. After heating to 50 °C over 1-2 h was
carried out, the spectrum resolves into three sharper
cones. The resonances between 67 and 70.0 almost
These sets of ions, and the propensity to form double-
charged ions, aid identification of new Qn as pure
material as well as Qn in mixtures. We also found that
the identification of unknown Qn was supported by the
use of an observed trend in the 13C NMR resonances. The
13C NMR resonances for each of the three chemically
different carbons of Qn show a progressive downfield shift
as the value of n increases from 5 to 8. The downfield
shift is most pronounced for the methylene and methine
carbons (Supporting Information, Table S3, Figure S1).
We observed a strong correlation (R2 > 0.99) between
the chemical shifts of the methylene and methine carbons
and the number of units, n, in Qn. We predicted the
chemical shifts of previously unknown Qn by extrapolat-
ing the curves. This alerted us to weak 13C resonances,
indicating potentially new Qn. The intensities of 13C
1
disappear. The H NMR at the same early stages shows
three broad, overlapping cones with no fine structure
between 3.9 and 6.1 ppm. After further heating was
carried out, the center broad peak almost disappears and
is replaced by two relatively sharp peaks with some
developing fine structure. These two peaks are similar
in shape and chemical shift positions to those of a poorly
resolved mixture of Qn, suggesting that the oligomer at
this stage is simpler and perhaps linear. No Qn formed
after 2 h at 50 °C. After prolonged heating (several days)
was carried out, the broad 13C NMR peaks decrease in
intensity and sharp signals attributable to each of the
Qn appear. The relative proportions of the Qn varied
according to the acid type, concentration of reagents, and
temperature of reaction.
Effect of Acid Typ e. Table 2 summarizes the effects
of varying the acid type and reactant concentration on
the final distribution of product Qn. More extensive data
are given in Supporting Information, Table S4.
Concentrated H2SO4 gave only 12% of Q5 using our
milder conditions, with the remainder of Qn being Q6
(Q7 and Q8 were undetectable). In HCl, the product
distribution was shifted as the acid concentration was
1
NMR resonances, rather than H NMR resonances, were
used to determine the product distributions (Supporting
1
Information, Table S4). While the H NMR spectra were
characteristic of each of the pure Qn, it was not possible
to use these data to determine product distributions due
to peak overlap in the spectrum of mixtures.
F a ctor s In flu en cin g P r od u ct Distr ibu tion . A ma-
jor aim of our study was to determine which factors
govern the proportional distribution of each cucurbituril
homologue. The previously published two-step proce-
dures13,17,21 of first reacting formaldehyde and glycoluril
in HCl to form an oligomer22 and then synthesizing Qn
(21) Buschmann, H.-J .; Fink, H.; Schollmeyer, E. Ger. Offen. DE
19603377 A1 19970807. Application: DE 96-19603377 19960131,
Deutsches Textilforschungszentrum Nord-West e. V., Germany, 1997,
p 4.
(22) Taketsuji, K. J pn. Kokai Tokkyo Koho J P 11217557 A2
19990810 Heisei. Application: J P 98-20949 19980202, Hakuto K. K.,
J apan, 1999, p 6.
(19) Buschmann, H. J .; Cleve, E.; Schollmeyer, E. Inorg. Chim. Acta
1992, 193, 93-7.
(20) Hoffmann, R.; Knoche, W.; Fenn, C.; Buschmann, H.-J . J . Chem.
Soc., Faraday Trans. 1994, 90, 1507-11.

8096 J . Org. Chem., Vol. 66, No. 24, 2001
Day et al.
Sch em e 1
decreased, favoring the production of Q6 and Q7. The
product distribution for any given set of conditions was
reproducible ((3%) provided that the glycoluril was
completely dissolved in HCl before the addition of form-
aldehyde. If the glycoluril was not dissolved, the propor-
tion of Q5 decreased by 10% or more (related to reactant
concentration effects). Below 5 M HCl, the final steps in
the reaction to form Qn became inefficient and incom-
plete, and it appears that the acid concentration was
insufficient to catalyze conversion of the oligomer into
the Qn. Concentrated HCl, 9 M HCl, 9 M H2SO4, and ∼6
M HBF4 show similar product distributions (Table 2).
The formation of Qn was also achieved in a melt of
para-toluenesulfonic acid by combining the reactants and
heating the mixture to 110 °C. The product distribution
markedly favored Q6 and Q7. By contrast, TFA is unable
to catalyze the reaction beyond the oligomer stage and
no Qn was formed.
Effect of Rea cta n t Con cen tr a tion s. If the length
of the oligomer is the main determinant of the size of
the Qn, then varying the reactant concentrations should
change the ratio and even the range of Qn formed. The
distribution of reaction products previously discussed was
obtained with glycoluril concentrations ranging from 155
to 190 mg/mL of acid. When the glycoluril concentration
was reduced to 125 µg/mL of acid, considerable decom-
position of the glycoluril occurred, and the only Qn that
were detected were Q5 and Q6 (Table 2). Increasing the
concentration of glycoluril to 1.7 g/mL of acid significantly
changed the Q5-8 distribution, discriminating against
Q5 and producing higher Qn. Two new sets of charac-
teristic Qn 13C NMR resonances were observedsone at
a higher field than Q8, the other below Q5. The chemical
shifts of the high-field set agree with those predicted for
Q10, but the other set did not agree with a hypothetical
Q4 (Supporting Information, Figure S1). Repeated re-
crystallization of the reaction mixture in order to isolate
Q10 gave a crystalline material that contained two sets
of Qn resonances. ESMS established a double-charged
molecular ion ([P + 2Cs]2+, m/z ) 1378) for Q15, sug-
gesting that the structure was Q10 tightly associated
with Q5. This was verified by displacing the bound Q5
with 13C-labeled Q5. A feature of guests inside Qn is an
upfield shift of the resonances in NMR spectra23 consis-
tent with our proposal for Q5 inside Q10, Q5@Q10. A
detailed characterization and discussion of this unprec-
edented new product is described in a subsequent paper.24
Interestingly, a set of 13C NMR resonances for Q9 could
be observed only when the high-concentration reaction
was carried out with 13C-enriched paraformaldehyde or
in concentrated filtrates from successive recrystalliza-
tions. 13C-Enrichment allows the observation of methyl-
ene resonances that fit the chemical shift trend for Qn
up to a possible Q16, (<1%).
to the Qn was 100 °C, although conversion was possible
at 50 °C with extended reaction times (greater than 4
weeks to reach completion). Under these cooler reaction
conditions, the product distribution was essentially the
same as that for 100 °C, except for a small set of
methylene and methine resonances, consistent with those
of Q9 (∼1%). In the first few days, Q5 was considerably
less abundant compared to the final proportion. Heating
Q5, Q6, or Q7 individually in concentrated HCl for 24 h
at 100 °C showed no interconversion and no detectable
decomposition. By contrast, Q8 in concentrated HCl at
100 °C contracted to Q5, Q6, and Q7 (4, 13, and 38%,
respectively) after 24 h, with 45% of Q8 remaining
unchanged. This conversion of Q8 to smaller homologues
is too slow to significantly contribute to the synthetic
outcome, although it is possible that larger homologues
are less thermally stable. The very low level of Q9 (<1%)
suggests that under these reaction conditions, the larger
Qn, or at least the open oligomers of a unit length of 9,
are thermally unstable.
At room temperature, no Qn were formed in concen-
trated HCl after 1 month; but in concentrated H2SO4, Q6
was almost quantitatively produced (>95%) after 2
months. We expect the dehydrating ability of concen-
trated H2SO4 to drive the equilibrium toward Qn prod-
ucts, whereas the high concentration of water in other
acid systems favors the reverse reaction.
F a ctor s In flu en cin g th e Rea ction Mech a n ism .
The primary stage of the reaction to Qn is an acid-
catalyzed condensation of glycoluril 2 and formaldehyde,
which is rapid and facile (Scheme 1).
Where R ) H, type A intermediates are formed under
basic conditions.25 No intermediates of type A or B have
been observed under acidic conditions as they react too
rapidly, even at room temperature, beyond the monomer
stage to form oligomers (Scheme 2). Where R ) Me, the
initial isolated product is of type B under acidic condi-
tions.26 In our experience, for R ) Me, the tetracyclic
diether B can be converted to Me10Q5 but at much higher
yields (85%) than previously reported (16%).15 This
demonstrates the reversible nature of the hydroxymeth-
ylation and subsequent ether formation, as in this case,
there are 2 equiv more of the methylene linkers than are
necessary to form Me10Q5.
The average length of the oligomers formed from the
acid-catalyzed glycoluril 2 (R ) H) and formaldehyde
condensation has been difficult to determine, as they are
only soluble in aqueous acid solutions and formic acid.
Both TOF-MALDI and ESMS have been unsuccessful.
Determination of the oligomer length by end-group
Nonhomogeneous reactions, where the glycoluril is not
completely dissolved before the addition of formaldehyde,
resemble concentrated reactions with regard to product
ratios.
Effect of Rea ction Tem p er a tu r e. In concentrated
HCl, the ideal conversion temperature of the oligomer
1
functionalization was also unsuccessful. H NMR of the
oligomer mixture showed <2% Qn, using dioxane as a
probe (see Experimental Section). Despite washing the
oligomer with aqueous CsCl solutions in which the Qn
(23) Mock, W. L.; Shih, N.-Y. J . Org. Chem. 1986, 51, 4440-46.
(24) Day, A.; Blanch, R. J .; Arnold, A. P.; Lorenzo, S.; Lewis, G. R.;
Dance, I. Ang. Chem., Int. Ed. 2001, in press. Crystallographic data
(excluding structure factors) for the structure reported in this paper
have been deposited with the Cambridge Crystallographic Data Centre
as supplementary publication no. CCDC-169807.
(25) Poskrobko, M.; Dejnega, M. J . Liq. Chromatogr. Relat. Technol.
1998, 21, 2725-2731.
(26) Niele, F. G. M.; Nolte, R. J . M. J . Am. Chem. Soc. 1988, 110,
172-7.

Synthesis of Cucurbituril and Its Homologues
J . Org. Chem., Vol. 66, No. 24, 2001 8097
Sch em e 2
F igu r e 1. Formation of cyclic Qn from endo-exo oligomer
ribbons.
orientation of the glycoluril moieties. The glycolurils are
composed of two cis-fused five-membered rings, and they
can orient themselves in two ways when linked by
methylene bridges: with both concave faces on the same
side (endo) or on opposite sides (exo). AM1 calculations
suggest that the all-endo intermediates are more stable
than endo-exo combinations, particularly if solvation is
taken into account (Supporting Information, Table S2).
As there are many combinations possible for endo-exo
oligomers, the 13C NMR spectrum is complex. The 1H
NMR spectrum shows that the methylene protons are
anisotropic, as they are in all Qn. Kinetically, any endo-
exo arrangement is initially possible but only the all-endo
conformation can close to form a Qn (Figure 1). Any
oligomer with concave faces on opposite sides (exo-endo
combinations) must disconnect and reconnect for it to flip
into the all-endo conformer (Scheme 2). Since Q8 can
contract to smaller Qn, acid-catalyzed disconnection and
reconnection of the methylene linkers is demonstrated
as a reversible step. The oligomeric, ribbonlike structures
do not have to become all-endo over their entire lengths
but only over a length sufficient to allow them to close
to a Qn. Upon closure of the oligomer, the remaining
fragment is clipped off, either enabling it to condense
with other oligomers or, if it is of a suitable length,
rearrange and self-condense to form another Qn.
Qn might be formed by head-to-tail condensation of the
oligomer, condensation at an intermediate point of an
extended oligomer, or both (Figure 1). In principle,
oligomers could be formed from a large number of
glycoluril units and, through a process of stepwise
disconnection and self-condensation, cyclize to form Qn.
Our high-dilution experimental conditions limited the
rate at which the oligomer achieved a length beyond 5
or 6 units before it condensed. As expected, smaller Qn
were favored and Q5 was the major product. When
Me10Q5 is formed from tetracyclic diether B, no oligomers
or MexQn homologues were detected under the reported
are soluble and the oligomers are not, we observed only
the ions for the Qn (as cesium complexes), albeit weakly.
A repetitive series of reactions leads to the precursor
oligomers required for the formation of Qn (Scheme 2).
We show intermediates containing ethers in our repre-
sentation of the mechanism for the sake of simplicity;
however, it is clear that these can be hydrated under
acidic conditions to diols and tetraols of type A or even
be absent with only terminal NH functionality. Reaction
equilibrium arrows demonstrate our view that all the
reactions are reversible.
A weighting (as indicated by the equilibrium arrows)
has been given for these reactions, on the basis of
predicted thermodynamic outcomes combined with a
perceived chemical outcome.
The same types of nucleophiles and electrophiles shown
in Scheme 2 are likely to occur repeatedly until stable
Qn are formed. Not all possible combinations or inter-
mediates leading to oligomerization are represented, but
it should be inferred that many variations exist.
In ter p r eta tion s. We have sought to identify why Q6
is the favored product in the synthesis of Qn. The answer
might lie in the range of thermochemical stabilities of
Qn, oligomer length, templating effects, or a combination
of these. Although none of the individual precursor
oligomers have been isolated, they are observed in the
13C NMR spectra as complex mixtures. There is no reason
to suspect the formation of short oligomers as they
remain in solution during reaction and are not termi-
nated by precipitation. One reason for the apparent
complexity is likely to be associated with the relative

8098 J . Org. Chem., Vol. 66, No. 24, 2001
Day et al.
reaction conditions. This suggests that the oligomer is
built up to a unit length of 5, where it immediately
condenses to form Me10Q5. We are continuing to inves-
tigate the effect of reactant concentration on this system.
Q5-7 were thermally stable in concentrated HCl for
at least 24 h at 100 °C. Under the same conditions, Q8
began to contract to smaller Qn. The rate of ring
contraction of Q8 was slowed when the reaction mixture
was heated in concentrated HCl in the presence of 10
mol equiv of formaldehyde. The formation of smaller Qn
requires reformation of the oligomer by cleavage of two
adjacent methylene linkers. This process is suppressed
by excess formaldehyde. The rate of ring contraction of
Q8 is slow and is therefore an unlikely source of smaller
Qn. However, Qn larger than Q8 are potential sources
of smaller Qn. The trace amount of Q9 (<1%) suggests
that higher homologues are unstable and break down to
the more thermally stable lower homologues. However,
significant amounts of Q10 (∼8%) were detected and
isolated. We suggest that Q10 is most likely stabilized
by the encapsulation of Q5.
Generally, high reactant concentrations increase the
probability of extended oligomers before ring closure
occurs. High concentrations most likely have two effects
on the equilibrium. They either allow sufficient time for
the oligomer to reach lengths greater than the lengths
required for the most commonly found Qn or reduce the
rate at which longer oligomers are broken, thus enabling
the formation of larger Qn. An alternative explanation
for the formation of Qn larger than Q6 is through the
insertion of additional glycoluril units (or short oligomers)
into small Qn to form larger Qn. However, this mecha-
nism seems improbable given the stability of Q5-7.
There are many equilibrium steps that could be
influenced by factors such as concentration, temperature,
acid type, or any combination of these. The reaction
temperature and the acid concentration have a profound
effect on Qn formation but are less important in oligomer
formation. Nevertheless, these factors may be important
for the stability of the oligomer. Both the reaction
temperature and the concentration of H+ ions are likely
to limit the rate and the potential for the glycoluril units
to invert through methylene linker disconnection. This
was evident in the formation of the oligomer at 50 °C
over 12 h where <2% of Qn was formed. In addition, no
Qn are produced in <4 M HCl.
spectra of Q5 in DCl/D2O. Pure Q5 in 22% DCl/D2O gave
two similar intensity sets of resonances for both H and
1
13C NMR spectra. The intensities of these resonances
were found to be markedly dependent on chloride ion
concentration. One set progressively diminished as the
concentration of DCl was increased to 37%. Conversely,
the other set of resonances progressively diminished as
the concentration was decreased to <15% DCl/D2O. The
distinct sets of resonances show the exchange rate of the
DCl to be slow on the NMR time scale. Similar effects
2-
were observed for SO4
and BF4-. Samples of Q5
containing these larger ions showed a relatively slow
exchange with Cl- when dissolved in DCl. We expect the
binding abilities of the Qn cavity and the emerging cavity
of an oligomer precursor to be similar. It is likely that
the influences of anions and hydrogen-bonded clusters
affect the balance of the competing forces of kinetics vs
thermodynamics.
The catalytic solvent TsOH favors the synthesis of Q6
and Q7 (Table 2). The minimal formation of Q5 might
reflect the templating effect of a hydrophobic aromatic
group sitting in a relatively hydrophobic emerging cavity.
Further work is being conducted in this area to fully
understand the influences of anions and organic tem-
plates.
Con clu sion
We expect that the number of repeat glycoluril units
in Qn significantly impacts the properties of these
macrocyclic molecules and that the newly emerging area
of host-guest cucurbituril chemistry will require the
ready synthesis of specific Qn species. This study provides
a foundation for understanding the synthetic processes
to form Qn, enabling the synthesis to be tailored to
efficiently generate particular Qn. We have identified
reaction conditions (acid concentration, acid type, tem-
perature, and reactant concentration) that determine the
relative proportions of the Qn produced. Manipulation
of these influences allows a degree of selectivity to be
achieved. A better understanding of the formation of the
oligomer and the influences of anions and templates
would further advance the ability to selectively synthe-
size any one of the Qn. The limit to the size of the Qn
(isolated Q5-8 and 10, detected Q9-16) is unknown, and
the factors effecting their stability and methods of
synthesis are the subject of further work.
The basicity of the oligomer carbonyls and the accom-
modating nature of the concave face of the glycoluril units
to a molecular or ionic guest are factors that could lower
the energy difference between those oligomers that are
oriented to cyclize to Qn, and those oriented in other
ways. The array of carbonyls on the top and bottom edges
of an oligomer ribbon, to which hydrogen bonding and
protonation can occur, combined with the association of
the cation formed with a close or a loosely hydrated
counteranion could affect the thermodynamics of the
oligomer closure. AM1 calculations show a significant
positive electrostatic potential for the inner, concave cis-
fused junction of the glycoluril moieties (Supporting
Information, Table S2). This positive region will be
attractive to small anions that might then stabilize a
looped oligomeric structure that can close to form a Qn
(Figure 1).
Exp er im en ta l Section
Gen er a l P r oced u r es. Unless otherwise indicated, com-
mercial materials were used as received. For ESMS, samples
were dissolved in CsCl/H2O except where specified. 1H and 13C
NMR spectra were recorded in 35% DCl/D2O at 20 °C at 300
and 80 MHz, respectively. Chemical shifts are reported in
parts per million, relative to external dioxane (3.56 ppm for
protons and 67.4 ppm for carbon) in D2O. Chemical shifts were
very dependent on pH and temperature. All new compounds
decomposed above 250 °C. Microanalysis was performed on
weight-stabilized air-dried samples except where specified.
H2O of crystallization for Q5 and Q7 was consistent with the
findings for Q6.27 The crystalline compounds Q8 and Q5@Q10
have a high capacity for incorporating H2O into their struc-
tures. This was verified by gravimetric analysis in conjunction
with desiccation.
According to this model, the size of the newly formed
Qn should be influenced by the nature of the anion guest.
Evidence for anion binding in Qn is found in NMR
(27) Germain, P.; Letoffe, J . M.; Merlin, M. P.; Buschmann, H. J .
Thermochim. Acta 1998, 315, 87-92.

Synthesis of Cucurbituril and Its Homologues
J . Org. Chem., Vol. 66, No. 24, 2001 8099
St a n d a r d R ea ct ion Con d it ion s for t h e Syn t h esis of
Cu cu r bit[n ]u r ils. The reactions have been carried out on a
scale from 10 mg up to 1 kg. An example of isolated yields can
be found in method 2 below; all others where determined by
13C NMR [see Supporting Information]. For purposes of
analysis, the Qn were recrystallized from HCl, except in the
case of Q7, which was purified on Dowex 50WX2 ion-exchange
resin, eluting with 1:1 1 M HCl/formic acid. Fractions were
evaporated and dried at 100 °C in vacuo (0.3 mmHg) for 12 h.
Cucurbit[5]uril: 1H NMR δ 4.18 (d, 2 H, J ) 15.5 Hz, CH2),
5.15 (d, 2 H, J ) 15.5 Hz, CH2), 5.34 (s, 2 H, CH); 13C NMR
51.0 (CH2), 69.9 (CH), 157.0 (CO). Anal. Calcd for C30H30N20O10‚
11H2O‚HCl: C, 32.92; H, 4.95; N, 25.60; Cl, 5.83. Found: C,
32.26; H, 4.66; N, 25.11; Cl, 5.85. Cucurbit[7]uril: 1H NMR δ
4.00 (d, 2 H, J ) 15.5 Hz, CH2), 5.18 (d, 2 H, J ) 15.5 Hz,
CH2), 5.28 (s, 2 H, CH); 13C NMR 53.5 (CH2), 72.0 (CH), 157.5
(CO). Anal. Calcd for C42H42N28O14‚8H2O‚0.5HCl: C, 38.06; H,
4.45; N, 29.59; Cl, 1.34. Found: C, 38.17; H, 4.59; N, 29.48;
Cl, 1.30. Cucurbit[8]uril: 1H NMR δ 3.97 (d, 2 H, J ) 15.5
Hz, CH2), 5.21 (d, 2 H, J ) 15.5 Hz, CH2), 5.27 (s, 2 H, CH);
13C NMR 54.5 (CH2), 72.5 (CH), 158.0 (CO). Anal. Calcd for
C48H48N32O16‚30H2O‚2.5HCl: C, 29.41; H, 5.68; N, 22.86; Cl,
4.52. Found: C, 29.87; H, 5.15; N, 23.25; Cl, 4.64. Cucurbit-
[9]uril: NMR assignments only, the compound has not been
isolated; 13C NMR 55.1 (CH2), 73.3(CH). Cucurbit[5]uril@
cucurbit[10]uril: 1H NMR δ 3.70 (d, 2 H, J ) 15.5 Hz, CH2,
Q5), 4.00 (d, 2 H, J ) 15.5 Hz, CH2, Q10), 4.91 (s, 2 H, CH,
Q5), 5.01 (d, 2 H, J ) 15.5 Hz, CH2, Q5), 5.30 (d, 2 H, J ) 15.5
Hz, CH2, Q10), 5.32 (s, 2 H, CH, Q10); 13C NMR 50.7 (CH2,
Q5), 56.4 (CH2, Q10), 69.3 (CH, Q5), 74.0 (CH, Q10), 156.5
tallization from HCl solutions, each component was isolated
in a pure form from the solid fractions Z1-3. Fractionations
were achieved in 60% aqueous formic acid, which, through
repetition, enabled the separation of Q8 and the Q5@Q10
complex from the other Qn as insoluble solids. Hot 20%
aqueous glycerol was used to separate Q7 from all of the other
components to achieve > 95% purity. Q7 was isolated from
the aqueous glycerol by the addition of MeOH and the
precipitate collected by filtration. Washing three times with
MeOH × removed the glycerol, and the solid was dried at 60
°C for 12 h to give 276 g (24%) of an off-white solid. Q6 and
Q5 accumulated in the formic acid soluble fractions and were
separated by fractional crystallization from HCl solutions, Q6,
541 g (46%), and Q5, 94 g (8%). The formic acid insoluble Q8
and the Q5@Q10 complex were also separated by fractional
crystallization from HCl solutions, Q8, 146 g (12%), and Q5@
Q10, 65 g (5%). Method 3. Dry, powdered glycoluril (1.02 g,
7.2 mmol) and paraformaldehyde (442 mg, 14.7 mmol) were
mixed together in a paste with the addition of concentrated
HCl (0.6 mL). The paste set as a solid and was then heated in
a closed reaction vessel to 100 °C for 15 h with occasional
manual mixing. The total reaction mixture was analyzed by
13C NMR and showed weight percentages of 3, 44, 33, 12, and
8% for Q5, 6, 7, 8, and 10, respectively.
Meth od for Differ en t Acid s a n d Dilu tion s. Method 1
was used in all cases except where specified. In all cases, the
reaction mixtures were homogeneous throughout the entire
process. Q5-8 were the major products, with less than 1% of
other Qn as determined by 13C NMR. Method 1, was also used
for para-toluenesulfonic acid except that the volume of acid
was substituted for a weight of acid (10 g) and the mixture
heated to 110 °C. The products were isolated by MeOH
precipitation as described. The weight percentages of Q5-8
products are reported in Supporting Information, Table S4.
Ch lor id e Bin d in g. Pure samples of Q5 were dissolved in
different concentrations of DCl/D2O and examined by NMR.
At 22% DCl/D2O, 1H NMR spectra showed the following: δ
4.13 (d, 2 H, J ) 15.5 Hz, CH2), 4.18 (d, 2 H, J ) 15.5 Hz,
CH2), 5.28 (d, 2 H, J ) 15.5 Hz, CH2), 5.30 (d, 2 H, J ) 15.5
Hz, CH2), 5.32 (s, 2 H, CH), 5.34 (s, 2 H, CH); 13C NMR δ 50.8
(CH2), 51.1 (CH2), 69.8 (CH), 69.9 (CH) 157.2 (CO) 157.9 (CO).
From 22 to 37% DCl/D2O, the upfield CH2 resonance and the
downfield CO resonance increase to become the only CH2 and
CO resonances observed; from 22 to 15%, the converse applies.
In the 1H NMR spectrum, the upfield resonances only are
found for 37% DCl/D2O; the converse applies for <15% DCl/
D2O.
(CO, Q5), 158.2 (CO, Q10). Anal. Calcd for C90H90N60O30
‚
28H2O‚3HCl: C, 34.81; H, 4.83; N, 27.06; Cl, 3.42. Found: C,
34.22; H, 4.49; N, 27.19; Cl, 3.25. ESMS m/z: 1378 ([P + 2Cs]+/
2). ESMS data for cucurbit[n]urils where n ) 5-8 can be found
in Table 1. Method 1. To glycoluril (77 mg, 0.54 mmol)
dissolved in the acid (0.4 mL) at 25-30 °C, with the aid of a
sonic bath, was added formaldehyde either as a 40% aqueous
solution (85 µL) or as solid paraformaldehyde (34 mg), and
the solution was immediately mixed. In some cases, a gel forms
after 20 min. In all cases, an exothermic reaction ensues. (On
a larger scale, the reaction mixture was cooled in an ice bath
during the addition of either form of formaldehyde). The
mixture was heated to 100 °C for 2.5-3 h, or until the reaction
was shown to be complete by NMR. Reactions on this scale
were conducted in NMR tubes and analyzed as total reaction
mixtures by 13C NMR using D2O as an external lock. Reactions
carried out on a scale 19.5 times greater than that above were
cooled and poured into MeOH (30 mL), and the precipitate
was collected by vacuum filtration. After the precipitate was
air-dried, the components were separated by fractional crystal-
lization from hydrochloric acid solutions. When HCl was the
solvent, it was either evaporated, and the residue was exam-
ined and purified as above, or the crystallization procedure
was commenced upon cooling of the reaction mixture. Method
2. Finely powdered glycoluril (1000 g, 7.04 mol) and powdered
paraformaldehyde (422 g, 14.07 mol) were added together and
the dry powders thoroughly mixed. To this was added ice-cold
concentrated HCl (1420 mL) while the mixture was stirred
rapidly. Stirring was continued until the mixture set as a gel.
After 30 min, the mixture was heated to 100 °C for 18 h. The
composition of the total reaction mixture at this point was 9,
48, 23, 14, and 6% w/w Q5, 6, 7, 8, and 10, respectively. The
mixture was cooled and filtered to collect the first crop (Z1) of
colorless crystals as a composite of 10, 40, 4, 32, and 14% w/w
Q5, 6, 7, 8, and 10, respectively, totaling 278 g. The filtrate
was then evaporated until the volume had been reduced to
one-fourth of the initial volume. To this was added water (500
mL), and the colorless crystalline suspension was collected by
filtration to give 324 g (Z2) as a composite of 5, 82, 6, and 7%
w/w Q5, 6, 7, and 8, respectively. The mother liquors contained
Q7 as the major homologue at 59%. The volume was reduced
to ∼1.2 L, poured into MeOH (3.5 L), and allowed to stand at
room temperature for 18 h. The off-white precipitate was
collected by filtration to give 654 g (Z3). Through a repetitive
process of fractionation by differential solubility and recrys-
13C-En r ich ed P a r a for m a ld eh yd e Rea ction . Method 1
was followed except for the substitution for 99% 13C-enriched
paraformaldehyde. In the methylene region of the 13C NMR
spectrum, resonances were found at 50.9, 52.4, 53.5, 54.5, 55.4,
56.8, 57.8, 58.9, and 59.9 (CH2) ppm of approximate ratios of
7, 60, 25, 7, and <1% for each of the last 5 peaks. The
remaining Q resonances were at 70.0, 71.1, 72.0, and 72.8 (CH)
ppm, ∼1% of the intensity of the methylene carbon.
Deca m eth ylcu cu r bit[5]u r il, 3. The dimethyltetracyclic
ether B (R ) Me, 5.2 g, 23.4 mmol) was placed in HCl 36%
(21 mL) and H2O (6.3 mL) and heated for 1.5 h after which all
volatiles were evaporated in vacuo and the residue was
1
examined by H and 13C NMR.15 Using an external sample of
dimethylglycoluril of known quantity, we determined the yield
of Me10Q5 to be 85%. ESMS of the residue gave m/z 618
[(Me10Q5 + 2Cs+)/2, 100] and 1103 [Me10Q5 + Cs+, 1%].
Oligom er . Powdered glycoluril (12.2 g, 85.9 mmol) and
paraformaldehyde (5.15 g, 171.8 mmol) were mixed together,
and ice-cold 36% HCl (75 mL) was added with stirring. The
solid material gradually dissolved, and the solution set as a
clear gel that was allowed to sit at room temperature for 10
min before it was heated to 50 °C in an oil bath for 19 h. The
gel produced a homogeneous solution that was poured into
methanol and the resultant precipitate collected by filtration,
washed with methanol, water, methanol, and ether, and air-
dried for several days to give 15.4 g of colorless powder that
was dried further at 80 °C in vacuo at 0.3 mmHg. Anal. Calcd
for C6H6N4O2‚4H2O: C, 30.26; H, 5.92; N, 23.52. Found: C,

8100 J . Org. Chem., Vol. 66, No. 24, 2001
Day et al.
29.87; H, 5.15; N, 23.25; Cl, 0.00. 1H NMR: δ 4.2-4.51 (m,
methylene), 5.31-5.53 (m, methylene), 5.56 (br s, methine).
The three sets of peaks had identical integrals. 13C NMR: δ
52.0-55.2, 70.05-73.0, 157.5-158.5. The addition of dioxane
(10 µL) resulted in a bound dioxane singlet at 2.52 ppm. In
comparison with the integral of the multiplet centered at 4.36
ppm, this corresponded to ∼2% w/w of Qn. (When 2 mol equiv
of dioxane are added to a pure Q6 sample in 35% DCl, a free
and a bound dioxane resonance are observed; the bound signal
integrates for 1:0.77 Q6:bound dioxane. Q7 is less sensitive to
dioxane as a probe, with bound dioxane resonances occurring
between 2.6 and 2.75 ppm, in a ratio of 1:0.4 Q7:bound dioxane.
Q5 does not bind dioxane at all.) Washing the solid oligomer
(1.03 g) with boiling aqueous solutions of CsCl and filtering
yielded a filtrate containing only Q5-7 (calculated from the
concentration to be in total 28 mg with weight percentages of
4, 85, and 11%, respectively). Addition of dioxane to the
remaining solid gave an 1H NMR spectrum that was es-
sentially unchanged and had no detectable bound dioxane
signal. This oligomer sample was dissolved in formic acid and
1462.3 [(Q7 + Cs2Cl), 10%], 1463.9 [9%], 1793.5 [(Q10 + Cs+),
3%].
Ack n ow led gm en t . We thank Dr. J ohn MacLeod,
Research School of Chemistry, ANU, for the mass-
spectral measurements and J odi Smith for technical
assistance. This research was supported in part by an
Australian Research Council Small Grant.
Su p p or tin g In for m a tion Ava ila ble: HyperChem 6.0.1
gas-phase AM1 heats of formation (∆Hf, kcal/mol) for cucurbit-
[n]urils, Qn, and methylated cucurbit[n]urils, Me2nQn, n )
3-10 after geometry optimization; MacSpartan Pro AM1
geometry optimizations of glycoluril-formaldehyde trimers
with terminal diether bridges; 13C NMR chemical shifts (ppm)
of the three carbons of the cucurbit[n]urils, n ) 5-8, in 1:1
v/v D2O/concentrated DCl; dependence of the methylene and
methine 13C NMR chemical shifts on the number of repeat
units, n, in cucurbit[n]urils; and the effects of varying acid type
and acid concentration on the weight percentages of Q5-8
products. This material is available free of charge via the
Internet at http://pubs.acs.org.
analyzed by ESMS (cone 100 V): m/z 1116.8 [(C35H36N24O12
+
Cs+), 100%], 1129 [(Q6 + Cs+), 32%], 1282.4 [8%], 1283.6
[(C41H42N28O14 + Cs+), 16%], 1284.3 [21%], 1294.9 [(Q7+ Cs+),
41%], 1296.4 [(Q6 + Cs2Cl), 45%], 1297.8 [21%], 1299.1 [12%],
J O015897C