Kinetic solvent effects on 1,3-dipolar cycloadditions of benzonitrile oxide

Article abstract of DOI:10.1002/poc.917

The kinetics of 1,3-dipolar cycloadditions of benzonitrile oxide with a series of N-substituted maleimides and with cyclopentene are reported for water, a wide range of organic solvents and binary solvent mixtures. The results indicate the importance of both solvent polarity and specific hydrogen-bond interactions in governing the rates of the reactions. The aforementioned reactions are examples for which these factors often counteract, leading to a complex dependence of rate constants on the nature of the solvent. For the reactions of N-ethylmaleimide and N-n-butylmaleimide with benzonitrile oxide, isobaric activation parameters have been determined in several organic solvents, water, and water-1-propanol mixtures. Interestingly, the activation parameters reveal significant differences in solvation in different solvents that are not clearly reflected in the rate constants. In highly aqueous mixtures, enforced hydrophobic interactions lead to an increase in rate constant, relative to organic solvents. However, the overall rate enhancement in water is modest, if present at all, because the solvent polarity diminishes the rate constant. This pattern contrasts with common Diels-Alder reactions, where polarity, hydrogen-bond donor capacity and enforced hydrophobic interactions work together, which can result in impressive rate accelerations in water. Copyright

Full text of DOI:10.1002/poc.917

JOURNAL OF PHYSICAL ORGANIC CHEMISTRY
J. Phys. Org. Chem. 2005; 18: 908–917
Published online 13 April 2005 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/poc.917
Kinetic solvent effects on 1,3-dipolar cycloadditions
of benzonitrile oxide
Theo Rispens and Jan B. F. N. Engberts*
Physical Organic Chemistry Unit, Stratingh Institute, University of Groningen, Nijenborgh 4, 9747 AG Groningen, The Netherlands
Received 21 December 2004; revised 7 February 2005; accepted 7 February 2005
ABSTRACT: The kinetics of 1,3-dipolar cycloadditions of benzonitrile oxide with a series of N-substituted
maleimides and with cyclopentene are reported for water, a wide range of organic solvents and binary solvent
mixtures. The results indicate the importance of both solvent polarity and specific hydrogen-bond interactions in
governing the rates of the reactions. The aforementioned reactions are examples for which these factors often
counteract, leading to a complex dependence of rate constants on the nature of the solvent. For the reactions of N-
ethylmaleimide and N-n-butylmaleimide with benzonitrile oxide, isobaric activation parameters have been deter-
mined in several organic solvents, water, and water–1-propanol mixtures. Interestingly, the activation parameters
reveal significant differences in solvation in different solvents that are not clearly reflected in the rate constants. In
highly aqueous mixtures, enforced hydrophobic interactions lead to an increase in rate constant, relative to organic
solvents. However, the overall rate enhancement in water is modest, if present at all, because the solvent polarity
diminishes the rate constant. This pattern contrasts with common Diels–Alder reactions, where polarity, hydrogen-
bond donor capacity and enforced hydrophobic interactions work together, which can result in impressive rate
accelerations in water. Copyright # 2005 John Wiley & Sons, Ltd.
KEYWORDS: Diels–Alder reaction; cycloaddition; water; hydrophobic interactions; binary solvent mixtures; isobaric
activation parameters
INTRODUCTION
moments, which are (partially) lost during the activation
process. The latter accounts for the inverse dependence
on polarity. Rate constants plotted against ET(30) (a mea-
sure of the solvent polarity, see below) usually show a fair
correlation. However, rate constants in protic solvents
sometimes deviate from this trend.2 Lewis acids are also
known to sometimes induce either accelerations or in-
hibitions of DC reactions.11–13
Diels–Alder (DA) reactions are often described as rela-
tively insensitive towards the nature of the solvent,
although rate enhancements of the order of 102–103 on
going from n-hexane to water are nevertheless commonly
found.1 Rate constants of 1,3-dipolar cycloadditions
(DC) are even less dependent on the the solvent.2–5
Even rate constants in water hardly differ from those in
other solvents and, when accelerations are observed,6–8
they are modest, compared with those for DA reactions.
Notably, DC reactions sometimes show a reverse depen-
dence of rate constant on the polarity of the medium, the
reactions being slowed in polar media.2,7,9,10 Intermedi-
ate cases are also known, leading to almost negligible
changes in rate constants. For example, for the reaction
between phenyldiazomethane and norbornene, rate con-
stants vary only by a factor of 1.8 over a wide range of
solvents (water not included).2 The inverse dependence
of rate constant on the polarity of the solvent is most
pronounced for nitrones and nitrile oxides as dipolaro-
philes. These compounds possess relatively high dipole
1,3-Dipolar cycloadditions of benzonitrile oxide
Benzonitrile oxide (1, Scheme 1), a very reactive 1,3-
dipole, was first prepared in 1886 by Gabriel and
Koppe.14 Benzonitrile oxide is often generated in situ,
because it dimerizes quickly. In fact, it dimerizes in
solution so easily that its reactivity was at first not
recognized,15 but cycloadditions proceed smoothly even
with completely unactivated dipolarophiles such as
ethene under ordinary laboratory conditions.16 Cycload-
ditions with 1 were explored in the 1950s (for an over-
view, see in Ref. 17); early mechanistic studies include
Refs 18 and 19.
*Correspondence to: J. B. F. N. Engberts, Physical Organic Chemistry
Unit, Stratingh Institute, University of Groningen, Nijenborgh 4, 9747
AG Groningen, The Netherlands.
Solvent effects on DC reactions with 1 and derivatives
are remarkably small.2 Studies of solvent effects have
often been brief and inconclusive about the different
E-mail: j.b.f.n.engberts@rug.nl
Copyright # 2005 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2005; 18: 908–917

SOLVENT EFFECTS ON 1,3-DIPOLAR CYCLOADDITIONS
909
philes, rate constants in n-hexane, ethanol and water are
nearly equal contradicts the explanation of the lowering
of the rate constant due to the hydrogen-bond interactions
with 1. A larger destabilization of the hydrophobic initial
state, relative to the less hydrophobic transition state
(enforced hydrophobic interactions), may explain why
in water the rate is not much lower than in organic
solvents, but for ethanol such a counteracting effect on
the rate is not possible. In summary, hydrogen bonding
and hydrophobic interactions are important factors that
influence rate constants in water, but in general, solvent
effects on DC reactions of 1 are still only partially
understood.
As mentioned, rate constants for the reaction of 1 with
N-methylmaleimide in several solvents have been pre-
viously determined.9 The complicated results prompted a
more detailed study. Here, an extensive study is presented
of the influence of the medium (pure solvents and
mixtures of solvents) on rate constants of 1,3-dipolar
cycloadditions of benzonitrile oxide (1) with N-alkyl-
substituted maleimides (2a–c) and with cyclopentene (4)
(Scheme 1). Emphasis is placed on the complex interplay
of different factors that control the rate, in particular
hydrogen bonding and polarity. In this regard, the reac-
tions of 1 with 2a–c are of particular interest, because
both substrates are susceptible to hydrogen-bond forma-
tion. The reaction of 1 with 4 is a convenient reference,
because 4 does not form significant hydrogen bonds.
Scheme 1
factors that control rates.8,20,21 In a detailed kinetic study,
the DC reactions between 1 and several electron-rich and
electron-poor dipolarophiles have been studied for a
number of solvents, including water, and also for mix-
tures of ethanol and water.9 The dipolarophiles include
cyclopentene, methyl vinyl ketone and N-methylmalei-
mide. Whereas reactions involving an electron-rich di-
polarophile are still 3–10 times faster in water than in
most organic solvents, reactions involving electron-poor
dipolarophiles are slightly decelerated. This difference
was rationalized on the basis of FMO theory;22 1 is a
good hydrogen-bond acceptor, and its FMOs are lowered
in energy when dissolved in a protic solvent. On reacting
with the electron-rich dipolarophile cyclopentene, the
RESULTS AND DISCUSSION
Solvent dependence of the rate constant
In Fig. 1, logk2 is plotted against ET(30)23 for a wide
range of solvents for the reaction of 1 with 2a and with
4 (Table 1). The solvents roughly form two groups: protic
and aprotic solvents.
dominating interaction is LUMO1–HOMOcyclopentene
.
Consequently, the energy gap, and hence the Gibbs
energy of activation, are smaller in a protic solvent. In
the case of an electron-poor dipolarophile, the dominat-
ing interaction is LUMOdipolarophile–HOMO1. FMO en-
ergies of both reactants are lowered in a protic solvent
(the electron-poor dipolarophiles studied are also suscep-
tible to hydrogen-bond formation), but it was proposed
that this occurs more effciently for 1, leading to a rate
retardation. (The relative energies of the FMOs of 1 and
electron-poor dipolarophiles are such that both LUMO1–
HOMOdipolarophile and HOMO1–LUMOdipolarophile inter-
actions may contribute significantly. The focus on only
one of these HOMO–LUMO interactions may therefore
not be fully justified, although this simplification was
sufficient to interpret the data presented in this paper.)
However, this explanation does not account for the
complicated dependence of the rate constants on the
solvent; for instance, k(n-hexane) ꢀ k(ethanol) ꢀ k(water) >
k(1,4-dioxane) > k(dichloromethane) ꢀ k(2,2,2-trifluoroeth-
anol) for the reaction of 1 with N-ethylmaleimide (2a,
Scheme 1). The fact that, for electron-poor dipolaro-
First, when considering the group of apolar solvents
[with values of ET(30) below ꢁ40], logk2 decreases
roughly linearly with ET(30). This pattern is indicative
of a polar initial state (1) and a less polar activated
complex, in which the charge separation of the 1,3-dipole
has partly disappeared. Note that DA reactions are almost
invariably faster in a more polar solvent. In the activated
complex, some charge separation developes, that is
stablilized by polar interactions. One reason why rates
of DC reactions (proceeding via an analogous mechan-
ism) are so weakly dependent on the solvent is that this
charge separation is mediated by the partial disappear-
ance of the 1,3-dipole, leading to only small accelera-
tions, or even, as for the reactions of 1 with 2a and 4, to a
decrease in rate on going to a more polar solvent.
For solvents where ET(30) > 40, the medium effects
are more complicated. A comparison of the reactions
between 1 and 2a and between 1 and 4 sheds some light
on this phenomenon. The latter reaction is classified as
Copyright # 2005 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2005; 18: 908–917

910
T. RISPENS AND J. B. F. N. ENGBERTS
Figure 1. Values of logk2 (Mꢂ1 sꢂ1) for the reaction of 2a (left) and of 4 (right) with 1 vs the ET(30) values23 of various solvents at
25 ꢃC. DMSO is left out, as side-reactions interfered with the cycloaddition; the choice of solvents is further limited by the
requirement of being able to monitor the reaction at 273 nm (closed circles are values for N-methyl- rather than N-
ethylmaleimide9)
inverse electron demand (IED). Therefore, hydrogen
bonding to 1 is favorable. Furthermore, 4 is not capable
of forming hydrogen bonds. Compared with the aprotic
solvents, alcohols show larger rate constants for the
reaction of 1 with 4. The more polar the alcohol, the
smaller is the rate constant (except for TFE). This trend
may be the continuation of the effect of the polarity (see
below; Fig. 3). Two solvents stand out. 2,2,2-Trifluoro-
ethanol (TFE), a very potent hydrogen-bond donor,
causes a larger rate constant than the other alcohols.
This is also true for water. In addition, a further accel-
eration may be ascribed to enforced hydrophobic inter-
actions.9 Hence both solvent polarity and the hydrogen
bond-donating capacity of the solvent affect the reaction
rates and, interestingly, in this particular case in opposite
directions (the hydrogen bond-donating capacity, of
course, also contributes to the solvent polarity). For
most DA reactions, both factors increase the rate of the
reaction. For DC reactions, it appears that usually these
two factors either enhance or diminish the rate constants.
The DC reaction of 4 with 1 is an example where these
factors are opposed, and therefore generate a much more
complex dependence of rate constant on solvent.
For the reaction of 1 with 2a, hydrogen bonding is also
possible for the dipolarophile, which introduces further
complexity. In the absence of hydrogen-bond donors, the
same dependence of rate constant on solvent polarity is
found as for the reaction of 1 with 4. Again, hydrogen
bond-donating solvents produce an additional rate in-
crease, but now the rate constant in TFE is much lower
than that in other alcohols.
The two simplest explanations for this pattern are: (i)
the reaction is still mainly IED; hydrogen bonding occurs
both to 1 and 2a, affecting 1 more than 2a; only in the
case of TFE is hydrogen bonding more efficient to 2a; (ii)
the reaction is mainly NED, hydrogen bonding also
occurs both to 1 and 2a, but affects 2a more than 1.
Only in the case of TFE does hydrogen bonding to 1
supersede hydrogen bonding to 2a. The kinetic data can
be understood using either explanation, but UV–visible
spectra of 1 in different solvents support the latter
explanation (Fig. 2). The maximum in the absorption
Table 1. Rate constants for the reaction of 1 with 2a and 4 in various solvents at 25 ꢃC
b
c
b
c
Solvent
ET(30)a
k2,2a
k2,4
Solvent
ET(30)a
k2,2a
k2,4
n-Hexane
CCl4
1,4-Dioxane
THF
Chloroform
CH2Cl2
t-BuOH
CH3CN
30.9
32.5
36.0
37.4
39.1
41.1
43.3
45.6
0.330
0.210
0.12d
0.100
0.059
0.07d
0.254
0.10
0.333
0.255
0.170
n.d.
0.127
0.120
n.d.
2-BuOH
2-PrOH
1-BuOH
1-PrOH
EtOH
MeOH
TFE
Water
47.1
48.4
49.7
50.7
51.9
55.5
59.8
61.3
0.308
0.289
0.310
0.320
0.22d
0.196
0.061
0.350
0.334
0.274
n.d.
0.259
0.265
0.229
0.380
0.978
0.124
aValues from Ref. 23.
bUnits Mꢂ1 sꢂ1
.
cUnits 10ꢂ2
M .
ꢂ1 sꢂ1
dValue for N-methylmaleimide.9
Copyright # 2005 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2005; 18: 908–917

SOLVENT EFFECTS ON 1,3-DIPOLAR CYCLOADDITIONS
911
Figure 2. UV spectra of 1 in various solvents at 25 ꢃC. The inset shows the corresponding transition energies [(ET (1) plotted
against ET(30)]
band of 1 shifts from 252 nm in n-hexane to 248 nm in
acetonitrile, 250 nm in 1-propanol, 243 nm in water and
240.5 nm in TFE, indicating both a relatively weak
interaction with 1-propanol and an efficient interaction
with TFE (even more efficient than with water). The
energy of the transition [ET(1)] plotted against ET(30)
clarifies this pattern. For 1 þ 4, the rate constant for TFE
deviates positively from the trend found among the
alcohols, whereas for 1 þ 2a, a negative deviation is
found (Fig. 3). This pattern is also in line with a relatively
efficient binding of TFE to 1, and supports the NED
mechanism for the reaction of 1 with 2a.
and polarizability–dipolar interactions. The contribution
of the hydrogen bond-donating capacity of the solvent in
just providing a more polar reaction environment could
be smaller than indicated by ET(30), as betaine-30 is
rather sensitive to these interactions. In fact, the fair
correlation of logk with ET(30), including alcohols, found
for many other DA and DC reactions,2 may be the result
A sharp deviation from a general trend in a plot of logk
against solvent polarity (Fig. 3) may indicate a change in
mechanism. In this case, one may perhaps regard the
introduction of hydrogen bonds as a change in mechanism.
A similar (but reversed) pattern was found for the reaction
of phenylazide with norbornene, both experimentally6 and
theoretically.24 Nevertheless, the deviation remains an unu-
sual observation. An alternative explanation, based on a
change from a concerted mechanism to a mechanism
involving a zwitterionic intermediate, may be rejected on
several grounds: (i) the dependence on the solvent polarity
should be much larger for such a mechanism;5 (ii) Hammett
ꢀ values for the reaction of 1 with electron-poor styrenes
(in CCl4)19 and with acrylonitrile (in water)21 are small
and similar, contradicting a (change to a) zwitterionic
mechanism; (iii) activation entropies are large and negative
over the full range of solvents (see below) and are
characteristic of a concerted reaction mechanism.
The effect of solvent polarity is extrapolated from the
range of solvents with ET(30) values between 30 and 40
(Fig. 3), to illustrate the divergence from this trend for the
more polar solvents [ET(30) values > 40]. The ET(30)
scale is based on one parameter, that includes hydrogen-
bond donor capacity, hydrogen-bond acceptor capacity
Figure 3. Data from Fig. 1, with various trends indicated.
Main plot, 1 þ 2a; inset, 1 þ 4
J. Phys. Org. Chem. 2005; 18: 908–917
Copyright # 2005 John Wiley & Sons, Ltd.

912
T. RISPENS AND J. B. F. N. ENGBERTS
of solvent polarity (including hydrogen bonds) playing a
smaller role in determining the rate in alcohols than
estimated on the basis of ET(30), together with hydrogen
bonds inducing (catalyze/inhibit) additional effects,
which work in the same direction as the solvent polarity.
This is supported by many other cases, in which a
difference in slope is found among the alcohols, and is
consistent with the idea that in general both non-specific
(polarity) and specific (‘catalytic’ hydrogen bonds) sol-
vation is important.
Acetonitrile is the odd one out, in particular for the
reaction of 1 with 2a. A similar pattern was found for a
related reaction.8 A specific accelerating effect of aprotic
dipolar solvents has been suggested,25 but the effect is far
from general (in the case of 1 þ 4, the rate constant for
acetonitrile deviates only slightly. Examples where the
effect is absent are given in Refs 2, 5 and 7).
Several multiparameter analyses have also been under-
taken (results not shown), using the Abraham–Kamlett–
Taft model, extended with the solvent parameter Sp. These
models discern different aspects of polarity, e.g. the
hydrogen-bond donor capacity (ꢁ). Usually, decent corre-
lations are found for DA reactions, but for the DC reactions
descibed in this paper no satisfactory fit was obtained.
play an important role, although the overall rate constant
need not be affected to a large extent, because of this
compensating behavior.
In TFE, the rate constants of these reactions are low,
which has been explained in terms of FMO theory.9 The
activation parameters reveal that the decrease in rate
constant is entirely due to a more unfavorable entropy
of activation, which seems hard to reconcile with a larger
difference in energies between the HOMO and LUMO of
the reactants. [Note that for ordinary Diels–Alder reac-
tions, a larger hydrogen bond-donating capacity of a
solvent leads to an increase in rate because of TS
stabilization and that this is reflected in a decrease in
the enthalpy of activation. In terms of FMO theory, the
LUMO of the dienophile is lowered in energy because of
(stronger) hydrogen bonding and the energy gap with the
HOMO of the diene is decreased.] Instead, the high
solvent polarity may be responsible for the low rate
constant, but this is expected to lead to an increase in
ÁzHꢂo also. Moreover, for the reaction of 1 with 4 no
corresponding decrease in rate constant is found. Yet
another explanation is that the activated complex is more
strongly solvated by TFE than are the reactants, but that
this enthalpic advantage is overcompensated by an un-
favorable entropic effect, leading to an increase in the
Gibbs energy of activation. This explanation is highly
speculative, but in line with the large negative entropy of
activation in TFE. Of course, this effect ultimately may
be present together with the lowering of the HOMO of 1.
As mentioned, it is possible that the changes in desolva-
tion cause (large) differences in ÁzHꢂo and TÁzSꢂo that
nearly compensate each other.
Isobaric activation parameters
Isobaric activation parameters (ÁzGꢂo, ÁzHꢂo and ÁzSꢂo)
for the reactions of 1 with 2a and 2b have been deter-
mined in different solvents (Table 2). On going from n-
hexane to chloroform, the rate decreases, and the accom-
panying increase in activation enthalpy is in line with a
stabilization of the polar initial state (with respect to the
activated complex) by the more polar chloroform. Curꢂo-
iously, for the reactions in 1-propanol, ÁzHꢂo and ÁzS
are the same as those in n-hexane. When compared with
chloroform, the enthalpy of activation is decreased, as a
result of hydrogen bonding (to the dipolarophiles), in
such a way that the FMO interaction energyꢂois lowered.
Note that in both cases, the changes in ÁꢂozH and ÁzSꢂo
are much larger than the changes in ÁzG , but strongly
compensating. This indicates that differences in solvation
The activation parameters for the reactions in water are
more in line with expectation, with relatively small
negative entropies of activation and relatively large en-
thalpies of activation. Water and aqueous mixtures will be
discussed later.
Isobaric activation parameters for the reactions of 1
with 2a and 2b follow the same pattern. Differences in
solvation (hydration) due to the presence of a larger alkyl
substituent apparently do not affect the activation pro-
cess. The n-butyl tail is still too small to fold back or have
any interaction with 1.
Table 2. Isobaric activation parameters for 1 þ 2a and 2b at 25 ꢃC
Compounds
Solvent
ÁzGꢂo (kJ molꢂ1 a
)
ÁzHꢂo (kJ molꢂ1 b
)
ꢂTÁzSꢂo (kJ molꢂ1 b
)
1 þ 2a
n-Hexane
75.8
80.0
75.9
79.9
75.7
75.5
75.2
79.5
74.6
42.5
51.8
42.7
34.4
50.2
40.8
40.5
33.0
48.0
33.3
28.2
33.1
45.5
25.5
34.7
34.8
46.4
26.6
Chloroform
1-Propanol
Trifluoroethanol
Water
1 þ 2b
n-Hexane
1-Propanol
Trifluoroethanol
Water
aStandard error < 0.1 kJ molꢂ1
.
bStandard error 1.5–2 kJ molꢂ1 for 2a, 1 kJ molꢂ1 for 1b.
Copyright # 2005 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2005; 18: 908–917

SOLVENT EFFECTS ON 1,3-DIPOLAR CYCLOADDITIONS
913
Binary solvent mixtures
To investigate further the complex kinetic behavior of the
present reactions, rate constants for reaction of 1 with 2a
were determined in the solvent mixtures chloroform–
CCl4, chloroform–TFE, chloroform–1-propanol and 1-
propanol–TFE (Figs 4 and 5). A comparison with ET(30)
values in these mixtures provides information concerning
the influence of the solvent polarity on the reaction. The
ET(30) values already offer clues concerning the nature of
the mixtures. In all cases, ET(30) shows a linear depen-
dence on the composition, except for a small range of
compositions (Figs 4 and 5). In case of chloroform–TFE,
the formation of strong hydrogen bonds between TFE and
betaine-30 is definitely responsible for the sharp increase
in ET(30) at low mole fractions of TFE. In mixtures of
acetonitrile with 1-propanol26 or chloroform with etha-
nol27 the same pattern was observed, with a dependence
of ET(30) on the mole fraction resembling a binding
curve at low mole fractions of 1-propanol or ethanol, and
a more gradual, linear dependence once ‘saturation’ has
been reached. Strong hydrogen bonds between the alco-
hol and the negatively charged phenolic oxygen of
betaine-30 are responsible for this pattern. In the other
mixtures, a strong preference for one of the solvents is not
observed because (i) neither of the two solvents is a
(strong) hydrogen bond donor or (ii) both solvents are
(strong) hydrogen bond donors. The chloroform–CCl4
mixture is an interesting case, because chloroform is a
relatively weak hydrogen bond donor. Nevertheless,
some kind of binding is observed.28
Analogously to betaine-30, 1 can be expected to form
strong hydrogen bonds. Also 2a is a good hydrogen bond
acceptor, but probably to a lesser degree.
In chloroform–CCl4 mixtures, two linear relationships
are observed in a plot of logk2 versus solvent composition,
with slightly different slopes. No strong interactions
Figure 4. Logarithms of the bimolecular rate constants
(Mꢂ1 sꢂ1) of the reaction of 2a with 1 in mixtures of chloro-
form with (a) CCl4, (b) 1-propanol and (c) TFE at 25 ꢃC. (In all
cases, the plots hardly differ when converted to a molar
scale, as in all cases the molar volumes of the solvents are
similar and the extent of non-ideal mixing limited.) The insets
show the corresponding ET(30) values and logk2 vs ET(30)
Figure 5. Logarithms of the bimolecular rate constants
(Mꢂ1 sꢂ1) of the reaction of 2a with 1 in mixtures of 1-
propanol with TFE at 25 ꢃC. The insets show the correspond-
ing ET(30) values and logk2 vs ET(30)
Copyright # 2005 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2005; 18: 908–917

914
T. RISPENS AND J. B. F. N. ENGBERTS
(hydrogen bonds) between solvents (the hydrogen bond
donor capacity ꢁ is 0.78 for 1-propanol, 0.93 for metha-
nol, 1.17 for water and 0.44 for chloroform; this indicates
that chloroform has an ability to form hydrogen bonds,
but to a much lesser extent than the alcohols) and
reactants are anticipated, and the rate constant is gov-
erned primarily by the polarity of the medium. As
reactants and betaine-30 are preferentially solvated to
different degrees, a linear correlation between logk2
and ET(30) over the full range of composition is not
observed.
In chloroform–1-propanol mixtures, 1-propanol will be
the better hydrogen-bond donor. The FMOs of both
reactants will be affected by the formation of hydrogen
bonds. In this case, a net accelerating effect results. A plot
of logk2 versus composition reveals that no strong bind-
ing of 1-propanol to either reactant occurs, because logk2
gradually increases towards the value of logk2 in pure 1-
propanol. The increased slope for x1-PrOH > 0.3 may
result from additional hydrogen bonding with 2a starting
to be significantly involved in the activation process.
Solvent polarity also plays a role, but the ‘catalytic’
effect of 1-propanol is more prominent.
In mixtures of chloroform and TFE rate constants for
reaction of 2a with 1 [Fig. 4(c)] are lower than those in
either pure solvent, passing through a minimum at
xTFE ¼ 0.5. ET(30) values are indicative of preferential
solvation of betaine-30 by TFE, and the same may well
be the case for 1. However, the changes in ET(1) indicate
that although a significant degree of preferential solva-
tion of 1 occurs, the preferential solvation is not as
dramatic as for betaine-30 (Fig. 6). For the sake of
comparison, the reaction of 1 with 4 was also studied
using chloroform–TFE mixtures (Fig. 6). A nearly linear
dependence of logk2 on the mole fraction of TFE was
found, with only a small deviation in the chloroform-rich
region. This pattern rules out any irregular effects of TFE
on 1. Perhaps the catalytic effect of TFE on 2a is small
initially, but gains importance at higher mole fractions of
TFE, bearing in mind that a similar pattern is found for
mixtures of chloroform with 1-propanol, albeit without a
minimum.
These observations lead to the following explanation:
the rate-accelerating effect induced by hydrogen bonding
depends on the solvent and is different for the different
reactants. TFE interacts efficiently with 1 and induces
catalytic effects that enhance the reacion with 4 but
reduce the rate constants for 2a. The interactions are
not so strong that extensive preferential solvation of 1 by
TFE occurs in mixtures of chloroform and TFE. 1-
Propanol interacts with 1 less efficiently, inducing smal-
ler effects. By contrast, 1-propanol interacts efficiently
with 2a, and an overall accelerating effect is found for 1-
propanol. In mixtures of chloroform with TFE, the rate
initially drops (1 þ 2a) because of hydrogen bonding of
TFE to 1. However, at higher mole fractions of TFE, the
rate constant again increases, most likely because then
(additional) hydrogen bonds between TFE and 2a are
involved.
Compared with 1-propanol, the rate inhibition in TFE
due to hydrogen bonding to 1 seems to dominate. How-
ever, note that in the hypothetical case that only the
polarity would affect the rate constant (as illustrated in
Fig. 3) the rate constant would be (much) lower. The
‘catalytic’ effect (hydrogen bonding with 2a) therefore
still contributes more than inhibition due to hydrogen
bonding to 1. The inhibition is just more efficient with
TFE than with other alcohols.
In mixtures of 1-propanol with TFE, the many possi-
bilities for TFE to form hydrogen bonds to 1-propanol
will reduce peculiarities in solvating 1 or 2a, and a linear
dependence of logk2 on xTFE is found for the complete
solvent composition range (Fig. 5).
Water and aqueous solvent mixtures
Rate constants for the reactions of 1 with 2a–c in water
and 1-propanol are given in Table 3. The rate constants
show a slight increase with increasing tail length R (Et, n-
Bu, Bz) in 1-propanol, and a more pronounced increase in
water. In other words, the rate constants in water com-
pared with 1-propanol (kw/kalcohol) increase with increas-
ing hydrophobicity of the dipolarophile. This pattern has
Figure 6. Left: values of logk2 (Mꢂ1 sꢂ1) for the reaction of 1 with 4 at 25 ꢃC in mixtures of chloroform and TFE. Right:
corresponding values of ET(1) (&) and ET(30) (*)
Copyright # 2005 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2005; 18: 908–917

SOLVENT EFFECTS ON 1,3-DIPOLAR CYCLOADDITIONS
915
Table 3. Rate constants for the reaction of 1 with 2a–c in water and in 1-propanol at 25 ꢃC
2a
2b
2c
Medium
k2 (Mꢂ1 sꢂ1
)
k/kw
k2 (Mꢂ1 sꢂ1
)
k/kw
k2(Mꢂ1 sꢂ1
)
k/kw
Water
1-Propanol
0.35
0.30
1
0.86
0.55
0.37
1
0.67
0.73
0.45
1
0.59
also been observed for DA reactions of 2,3-dimethylbu-
tadiene with N-alkylmaleimides,29 and are possibly due
to an additional hydrophobic interaction between the
reactants, lowering the Gibbs energy of activation.
Another, extreme example of this phenomenon is ob-
served for the DC reactions of C,N-diphenylnitrone with
dimethyl fumarate and dibutyl fumarate, for which kw/
kalcohol is 12 and 108, respectively.7 Inspection of Fig. 3
reveals that, compared with the trend found among the
alcohols, the rate constant in water shows a positive
deviation. This minor effect may be attributed to enforced
hydrophobic interactions.9 Such a small contribution is to
be expected for two polar reactants. From elaborate
calculations on DA reactions involving the hydrophobic
cyclopentadiene, the contribution of enforced hydropho-
bic interactions was estimated to be a factor of 5–6 in
rate.30 In this case, the contribution is estimated to be a
factor of 2–3. For the reaction of 1 with 4, the effect
appears larger, which can be attributed to the fully apolar
character of 4.
effect is absent.31 Plotted versus the mole fraction of
water, the positions of the maxima seem to depend on the
hydrophobicity of the alcohol. However, Fig. 7 reveals
that the maximum appears at roughly equal concentra-
tions of water. In other words, the positions of the
maxima correlate with the size of the cosolvent mole-
cules, if the mole fraction scale is used. The mole fraction
scale may not be the best choice for discussing medium
effects in aqueous solvent mixtures.
There is also a shallow minimum in rate constant
around 15 M of water. In this concentration range, the
hydrogen-bond network of water becomes completely
disrupted and hydrophobic effects no longer play a role.
Compared with 1-propanol, addition of a small amount of
water increases the hydrogen bond-donating capacity of
the solvent mixture, which for these reactions is both
favorable (activation of the dipolarophile) and unfavor-
able (deactivation/stabilization of the 1,3-dipole), and a
net unfavorable effect results.9 In methanol–water mix-
tures, the minimum is absent, and the rate constant
increases monotonically on adding water, up to 40 M.
Once more, an unusually complex dependence of rate
constant on the nature of the reaction medium is found.
Rate constants in aqueous alcohol mixtures
In mixtures of water and 1-propanol or 2-methyl-2-
propanol, with [alcohol] 2 M, the reactions of 1 with 2a
and 2b are slightly accelerated, compared with water
(Fig. 7), leading to a maximum in rate constant, at around
45 M water. Similar maxima have been observed for many
cycloadditions, although there are examples where the
Isobaric activation parameters in mixtures
of water and 1-propanol
Isobaric activation parameters have also been determined
for the reactions of 1 with 2a and 2b in water–1-propanol
mixtures (Fig. 8). Up to 40 M watꢂeor, the variation in
activation parameters is small. ÁzH slightly decreases
and ꢂTÁzSꢂo incꢂoreases accordingly. In the water-rich
mixtures, ꢂTÁzS suddenly drops significantly, accom-
panied by a largely compensating increase in ÁzHꢂo. In
the water-rich reꢂgoime, hydrophobic effects come into
play, and ꢂTÁzS drops (initial state destabilization by
hydrophobic hydration). Whereas at around 40 M of water
the solvent mixture behaves as a ‘normal’ polar solvent,
at higher water concentrations the characteristic features
of water at room temperature (i.e. hydration of apolar
compounds accompanied by a large unfavorable entropy
term) become apparent. The patterns in the activation
parameters are in part similar for 2a and 2b, as was found
among the pure solvents, but significant differences are
observed for ÁzHꢂo and ÁzSꢂo in the concentration range
where hydrophobic effects are important. The alkyl
substituent in the dipolarophile clearly influences the
activation process in these aqueous mixtures, most likely
Figure 7. Rates in water–1-propanol mixtures (squares) and
water–2-methyl-2-propanol mixtures (circles), for the reac-
tion of 1a with 2a (closed symbols) and 2b (open symbols)
Copyright # 2005 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2005; 18: 908–917

916
T. RISPENS AND J. B. F. N. ENGBERTS
similarly to the impact of a Lewis acid catalyst, and this
specific effect can either accelerate or inhibit the reaction.
For the present reactions these two factors oppose each
other, which partially explains the modest solvent effects
on these reactions, and leads to the complex dependence
of rate constants on solvent.
In the case of the reactions of the N-alkylmaleimides
with benzonitrile oxide, hydrogen bonding (with corre-
sponding changes in the FMOs) to both reactants occurs,
with opposite effects: hydrogen bonding to the dipolar-
ophiles is favorable, but hydrogen bonding to the benzo-
nitrile oxide unfavorable. This pattern can be rationalized
with FMO theory, assuming the reaction has normal
electron demand (NED).
In mixtures of solvents, of which only one is a hydrogen-
bond donor, logk varies gradually with the composition
(often linearly), indicating the absence of significant
preferential solvation of the reactants due to hydrogen-
bond interactions. The specific, ‘catalytic’ effects of
hydrogen bonds are not accompanied by strong binding
or complexation of the hydrogen bond-donating solvent
to the reactants.
Interestingly, isobaric activation parameters reveal sig-
nificant differences in solvation in different solvents that
are not reflected in the rate constants.
In (highly) aqueous mixtures, hydrophobic effects are
important, but this does not lead to large increases in rate,
because the contribution of these effects is modest, and
these effects are counteracted by other factors. This
pattern contrasts with common DA reactions, where
polarity, hydrogen-bond donor capacity and enforced
hydrophobic interactions work together, which can result
in impressive rate accelerations in water.
Figure 8. Activation parameters for the reaction of 1 with
(a) 2ꢂoa and (b) 2b inꢂowater–1-propanol mixtures at 25 ꢃC:
ÁzG ꢂ30 (^); ÁzH (&); ꢂTÁzSꢂo
by inducing differences in solvation/hydration. The butyl
group could, for example, induce a larger preference of
the dipolarophile for being solvated by propanol. These
differences in solvation are again ‘innocent’, i.e. not
reflected in the rate constants. Nevertheless, these data
show that despite the absence of large effects of water on
the rate constants of these reactions, typical ‘aqueous’
behavior is occurring in water-rich mixtures.
EXPERIMENTAL
Materials
N-n-Butylmaleimide (2b)29 has been synthesized pre-
viously. The ET(30) probe was kindly provided by Prof.
Dr Chr. Reichardt (University of Marburg, Germany). All
other materials were obtained from commercial suppliers
and were of the highest purity available. Solvents were
either of analytical grade or distilled. Acetonitrile was run
over basic aluminium oxide prior to use. Cyclopentene
(4) was distilled before use.
CONCLUSIONS
A systematic study of solvent effects on the reactions of
benzonitrile oxide (1) with N-alkylmaleimides (2) and
cyclopentene (4) has provided additional insights into the
factors that determine the rates of cycloadditions in
different media. This study emphasizes the importance
of both polarity and hydrogen-bond donating capacity: (i)
differences in charge distributions for reactants and
activated complex cause polar interactions with the
solvent (including hydrogen bonds) to enlarge or reduce
the energy gap between both states; (ii) a hydrogen bond
formed between reactant and solvent (hydrogen-bond
donor) affects the HOMO and LUMO of the reactant,
Kinetic experiments
The procedure, described in the literature,9 where 1 is
generated in situ in a CH2Cl2–bleach two-phase system,
and small aliquots of the organic layer are transferred to
the reaction mixture, was found to lead to poor kinetics
for aqueous solutions because of solubility problems.
Instead, the preparation of 1 was performed by dissolving
Copyright # 2005 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2005; 18: 908–917

SOLVENT EFFECTS ON 1,3-DIPOLAR CYCLOADDITIONS
917
benzaldoxime in a bleach–1-propanol mixture in a test-
tube and shaking this tube for a few seconds. After the
addition of sodium chloride a two-phase system quickly
emerged, and 0.5–1 ml of the organic layer was trans-
ferred to a quartz cuvet, which contained the reaction
mixture with the dipolarophile. This method led to
excellent kinetics and was used for most kinetic measure-
ments. No differences in rate constants were found in
aprotic solvents when using 1-propanol rather than di-
chloromethane for this method.
3. Huisgen R, Fisera L, Giera H, Sustmann R. J. Am. Chem. Soc.
1995; 117: 9671–9678.
4. Rispens T, Engberts JBFN. J. Phys. Org. Chem. in press. DOI:
10.1002/POC.917
5. Huisgen R. Pure Appl. Chem. 1980; 52: 2283–2302.
6. Wijnen JW. PhD Thesis, University of Groningen, 1997.
7. Gholami MR, Yangheh AH. Int. J. Chem. Kinet. 2001; 33: 118–
123.
8. Araki YIK, Shiraishi S. Bull. Chem. Soc. Jpn. 1991; 64: 3079–
3083.
9. van Mersbergen D, Wijnen JW, Engberts JBFN. J. Org. Chem.
1998; 63: 8801–8805.
10. Elender K, Riebel P, Weber A, Sauer J. Tetrahedron 2000; 56:
4261–4265.
Kinetic measurements were performed using UV–
visible spectroscopy (Perkin-Elmer ꢂ5 or ꢂ12 spectro-
photometer). The dipolarophile was used in excess, and
reactions were monitored at 273 nm. The reactions were
followed for at least four half-lives and pseudo-first-order
rate constants were obtained using a fitting program.
Typical conditions were [dipolarophile] ¼ 1–10 mM and
[1,3-dipole] ꢀ 0.025–0.05 mM. Activation parameters
were calculated from 4–5 rate constants in the tempera-
ture range 20–40 ꢃC.
UV–visible spectra of 1 and the ET(30) probe were
recorded on a Perkin-Elmer ꢂ5 spectrophotometer at
25 ꢃC. ET(30) values were calculated from the longest
wavelength charge-transfer absorption band of the dye,
ET(30) (kcal molꢂ1) ¼ 28591/ꢂmax (nm);23 ET(1) values
were calculated accordingly from its longest wavelength
absorption band. A few microliters of solvatochromic dye
was injected into a known volume of solution or solvent
mixture. In the case of betaine-30, stock solutions were
prepared in ethanol. For 1, the procedure described above
was used.
11. Seerden J-PG. PhD Thesis, Catholic University of Nijmegen,
1995.
12. Curran DP, Kim BH, Piyasena HP, Loncharich RJ, Houk KN.
J. Org. Chem. 1987; 52: 2137–2147.
13. Kanemasa S, Nishiuchi M, Wada E. Tetrahedron Lett. 1992; 33:
1357–1360.
14. Gabriel S, Koppe M. Ber. Dtsch. Chem. Ges. 1886; 19: 1145–
1148.
15. Wieland H. Ber. Dtsch. Chem. Ges. 1907; 40: 1667.
16. Stagno d’Alcontres G. Gazz. Chim. Ital. 1982; 82: 627–634.
17. Quilico A. Five- and Six-Membered Compounds with Nitrogen
and Oxygen. Interscience: New York, 1963.
18. Christl M, Huisgen R. Tetrahedron Lett. 1968; 9: 5209–5213.
19. Dondoni A, Barbaro G. J. Chem. Soc., Perkin Trans. 2 1973;
1769–1773.
20. Barbaro B, Battaglia A, Dondoni A. J. Chem. Soc. B 1970;
588–592.
21. Dignam KJ, Hegarty AF, Quain PL. J. Org. Chem. 1978; 43:
388–393.
22. Houk KN. Acc. Chem. Res. 1975; 8: 361–369.
23. Reichardt C. Chem. Rev. 1994; 94: 2319–2358.
24. Pekasky MP, Jorgensen WL. Faraday Discuss. 1998; 110:
379–389.
25. Kadaba PK. Synthesis 1973; 71–84.
26. Elias H, Gumbel G, Neitzel S, Volz H. Fresenivs’ Z. Anal. Chem.
1981; 306: 240–244.
27. Balakrishnan S, Easteal AJ. Aust. J. Chem. 1981; 34:
933–941.
´
´
28. Maksimovic ZB, Reichardt C, Spiric A. Z. Anal. Chem. 1974;
270: 100–104.
REFERENCES
29. Meijer A, Otto S, Engberts JBFN. J. Org. Chem. 1998; 63: 8989–
8944.
30. Chandrasekhar J, Shariffskul S, Jorgensen WL. J. Phys. Chem. B
2002; 106: 8078–8085.
31. Rispens T. PhD Thesis, University of Groningen, 2004.
1. Otto S, Engberts JBFN. Pure Appl. Chem. 2000; 72: 1365–1372.
2. Padwa A. 1,3-Dipolar Cycloaddition Chemistry. Wiley: New
York, 1984.
Copyright # 2005 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2005; 18: 908–917