Enhanced bulk catalyst dissolution for self-healing materials

Article abstract of DOI:10.1039/c0jm00521e

A model was developed to aid in the selection of healing monomers that can rapidly dissolve catalysts in self-healing materials. Predictions are made regarding dissolution rates of Grubbs' catalyst in a small library of ring-opening metathesis polymerizat

Full text of DOI:10.1039/c0jm00521e

View Online / Journal Homepage / Table of Contents for this issue
PAPER
www.rsc.org/materials | Journal of Materials Chemistry
Enhanced bulk catalyst dissolution for self-healing materials†
a
Timothy C. Mauldin and Michael R. Kessler
b
*
Received 25th February 2010, Accepted 11th March 2010
First published as an Advance Article on the web 12th April 2010
DOI: 10.1039/c0jm00521e
A model was developed to aid in the selection of healing monomers that can rapidly dissolve catalysts in
self-healing materials. Predictions are made regarding dissolution rates of Grubbs’ catalyst in a small
library of ring-opening metathesis polymerization (ROMP)-active norbornenyl-based healing
monomers. The Grubbs’ catalyst and the healing monomers were experimentally assigned sets of
Hansen parameters, and it was observed that healing monomers and blends of monomers with
parameters similar to the catalyst were able to rapidly dissolve the catalyst. The model is limited in its
ability to predict dissolution trends of healing monomers with substantially different viscosities.
catalyst combination used thus far in self-healing systems is
Introduction
dicyclopentadiene (DCPD) and Grubbs’ catalyst, the former of
which undergoes ring-opening metathesis polymerization
(ROMP)8–10 in the presence of the latter.11–17 This system worked
well to first demonstrate self-healing as both the catalyst and
DCPD satisfy many of the unique and complex requirements of
this healing mechanism, but owing to the price of Grubbs’
catalyst, the research community has been slowly moving away
from ROMP-based self-healing in favor of healing based on
other chemistries. But while healing with these other chemistries
(notably siloxane polycondensation,18 epoxy ring-opening poly-
merization,19 solvent activation of residual matrix reaction,20 and
click chemistry,21 among others) has indeed proven fruitful, the
superior healing precedent set by the seminal Grubbs’ catalyst
should disallow it from being deemed impractical for self-healing
applications simply due to economic considerations.
In the past decade, polymers and composites that can repair
themselves with complete, or nearly complete, autonomy have
been extensively studied in academia and received significant
commercial interest.1–3 Perhaps the most ubiquitous self-healing
mechanism developed to date incorporates liquid monomer-filled
vessels and catalyst particles into a polymer matrix. Upon
material fracture, vessels rupture, followed by flow of the liquid
monomer into the crack volume. When the monomer contacts
the catalyst particles it polymerizes and adheres the crack faces
together (Fig. 1).4–7 The first and most well-studied monomer/
One approach to minimize the economic obstacle of using
Grubbs’ catalyst is to reduce the amount of catalyst required to
achieve maximum healing potential. Unfortunately, efficient use
of the catalyst in a self-healing system is often difficult due to the
fact that the monomer needs to dissolve the catalyst in a very
non-ideal mixing scenario; that is, elevated temperatures or
external agitation cannot realistically be applied to aid catalyst
dissolution during the healing event. In addition, sluggish cata-
lyst dissolution can also likely be attributed to the fact that the
catalyst particle is embedded in the crack surface, limiting the
surface area of catalyst exposed to the liquid monomer during
healing to one cross-sectional area of the particle. Nevertheless,
others have developed rather clever techniques to promote effi-
cient use of Grubbs’ catalyst during healing. For example, Rule
et al. showed that by first encapsulating Grubbs’ catalyst parti-
cles in wax microspheres, a 10-fold decrease in catalyst loading
can achieve similar or better healing relative to the original, wax-
free self-healing systems.22 Catalyst was used more efficiently due
to the catalyst particles being smaller in size in the wax micro-
spheres, which increases the surface area of catalyst exposed to
liquid monomer. Furthermore, the Grubbs’ catalyst was pro-
tected by the wax casing from surface layer decomposition
caused by contact with the polymer matrix resin during
composite fabrication. Another approach to more efficiently use
catalyst is to incorporate different healing monomers and
Fig. 1 Schematic representing self-healing of a polymer.
aDepartment of Chemistry, Iowa State University, Ames, IA, 50011, USA
bDepartment of Materials Science and Engineering, Iowa State University,
2220 Hoover Hall, Ames, IA, 50011, USA. E-mail: mkessler@iastate.edu
† Electronic supplementary information (ESI) available: Density and
molar volume values for the Healing Agent Library and the calculation
of Beerbower group contributions and Hansen Parameters for
norbornene. See DOI: 10.1039/c0jm00521e
4198 | J. Mater. Chem., 2010, 20, 4198–4206
This journal is ª The Royal Society of Chemistry 2010

View Online
monomer blends that inherently require less catalyst to achieve
a high degree of polymerization, and therefore require lower
loadings of catalyst in the self-healing polymer. For example, Liu
et al. have identified a low viscosity, highly reactive ROMP
monomer called ethylidene norbornene that can be used either
neat or in blends with DCPD to, among other things, signifi-
cantly reduce the required catalyst loading in a self-healing
polymer.23–27
In this paper, we create a small library of healing monomers
and develop a model to make predictions regarding their ability
to dissolve a catalyst in a self-healing polymer. While we antici-
pate our model can be versatile enough to be applicable to any
type of healing chemistry, due to the robust and well-studied
healing performance of ROMP-based self-healing, we focus our
discussion here only on the dissolution of Grubbs’ catalyst in
ROMP-active healing monomers. In order to create this model,
we borrow several ideas from the concept of solubility para-
meters, otherwise widely known in the realm of polymer physics
as a means to make predictions regarding solute–solvent inter-
actions and the thermodynamics of polymer mixing.30 Details of
the calculations and the strengths and limitations of the model
are discussed in detail. And finally, the model is experimentally
validated by directly measuring the dissolution rate of a struc-
turally modified Grubbs’ catalyst in various monomers and
monomer blends. The structurally modified catalyst was used to
inhibit polymerization of the ROMP-active monomers during
the course of the dissolution.
An alternate approach to efficiently using the catalyst in a self-
healing polymer is to increase the rate at which catalyst is dis-
solved into the liquid monomer. Presumably, the reason why
such a large amount of Grubbs’ catalyst, relative to DCPD, is
necessary in self-healing polymers is because the catalyst particles
exposed on the crack surface are not entirely dissolved before
DCPD undergoes appreciable amounts of polymerization, and
therefore cannot easily dissolve more catalyst. Hence, the effec-
tive catalyst concentration is considerably lower than the feed
concentration that is initially added to the self-healing polymer.
So to reach maximum healing capabilities, comparatively large
amounts of catalyst are necessary. Increasing the dissolution rate
of the catalyst would then cause the effective concentration to
approach the feed catalyst concentration, thus requiring less
overall catalyst to achieve similar levels of healing. Jones et al.
attempted to increase dissolution kinetics by grinding, freeze-
drying, and recrystallizing catalyst particles to smaller sizes,
thereby increasing their surface area.28 Smaller catalyst particles
were shown to dissolve faster in the healing monomer than the
large, as-received Grubbs’ catalyst particles. In some cases, this
faster dissolution also improved healing by ensuring a relatively
even distribution of dissolved catalyst throughout the monomer,
which led to a continuous polymer film in the crack volume. This
was unlike early self-healing systems with large, slow-dissolving
catalyst particles, whose reaction with monomer was largely
heterogeneous, leading to intermittent patches of polymer
surrounding catalyst embedded in the crack plane. However,
reducing catalyst particle size and increasing catalyst surface area
are not without drawbacks. For example, as mentioned above,
the extent of catalyst surface layer decomposition is increased
with a greater surface area, and the smaller size potentially leads
to catalyst dissolution in the polymer matrix resin in which it is
embedded.
Experimental
General considerations
Bis(tricyclohexylphosphine)benzylidene ruthenium dichloride
(1st generation Grubbs’ catalyst), bicyclo[2.2.1]hept-2-ene
(norbornene), dicyclopentadiene, 5-vinyl-2-norbornene, 5-ethyl-
idene-2-norbornene, 5-norbornene-2-carboxylic acid, 5-norbor-
nene-2-methanol, tetramethylsilane, ethyl vinyl ether, and all
solvents/starting materials/reagents used for synthesis were
purchased from Aldrich and, unless stated otherwise, used as-
received. While determining Grubbs’ catalyst’s parameters, all
solvents known to not readily absorb atmospheric moisture were
used as-received from freshly opened bottles. All other solvents
were either purchased anhydrous or dried accordingly and were
handled under a nitrogen gas purge. 5-Norbornene-2-carbox-
aldehyde,31 2-(chloromethyl)-5-norbornene,32 2-(bromomethyl)-
5-norbornene,33 ethyl 5-norbornene-2-carboxylate,34 5-(meth-
oxymethyl)-2-norbornene,35 and 2-acetyl-5-norbornene36 were
synthesized according to literature procedures. All 5-substituted
norbornenyl-derivatives, either purchased or synthesized, were
obtained as a mixture of exo- and endo-isomers. All other
syntheses are described below. 1H NMR spectra were recorded at
either 300 or 400 MHz with a Varian Spectometer (Palo Alta,
CA) using CDCl3 as a solvent and residual chloroform as an
internal reference. HR-MS was done on a Finnigan TSQ700
mass spectrometer (San Jose, CA). Unless otherwise stated,
In addition to decreasing the amount of catalyst required for
self-healing polymers, improving the bulk catalyst dissolution
kinetics can also increase the speed of healing, which greatly
expands self-healing polymer’s potential application base.
Ideally, self-repair should be as rapid as possible, especially for
applications where the polymer may be subject to rapid or
periodic stress. But it was recently observed that healing kinetics
is not a simple phenomenon and dependent on many factors,
notably the sensitive interplay between catalyst dissolution
kinetics and bulk polymerization kinetics of the healing mono-
mer.29 In particular, it was noted that the bulk polymerization
kinetics of the healing monomer should not greatly exceed the
catalyst dissolution kinetics. Otherwise the system would result
in a polymer with poor properties and ‘‘spotty’’ healing, similar
to the aforementioned case of healing with large, slow-dissolving
catalyst particles. Thus, catalyst dissolution kinetics is a signifi-
cant (but often ignored) consideration when determining the
speed and quality of self-healing.
ꢀ
room temperature is defined as 23–24 C.
Synthesis of phenyl vinyl ether
A modified version of the original preparation37 was used.
b-Bromophenetole (15 g, 75 mmol) was added to KOH (20 g,
0.36 mol), which had been ground into a fine powder. The
ꢀ
suspension was heated to a reflux (ꢁ150 C) for 12 hours. The
crude product was isolated by vacuum filtration and purified by
silica gel column chromatography (hexanes) to yield a clear,
colorless liquid (3.1 g, 35%). 1H NMR (300 MHz, CDCl3): d ppm
7.35 (apparent t, J ¼ 9 Hz, 2H), 7.10 (t, J ¼ 9 Hz, 1H), 7.04 (d,
J ¼ 9 Hz, 2H), 6.68 (dd, J ¼ 6, 15 Hz, 1H), 4.79 (dd, J ¼ 3, 15 Hz,
This journal is ª The Royal Society of Chemistry 2010
J. Mater. Chem., 2010, 20, 4198–4206 | 4199

View Online
Table 1 Ranking system for dissolution of Grubbs’ catalyst in various
solvents
1H), 4.45 (dd, J ¼ 3, 6 Hz, 1H). High-resolution mass spec-
trometry (HRMS) expected 120.0575, found 120.0571.
Category
Rapid
Description
Synthesis of (PCy3)2Cl2Ru]C(H)OPh
Fully dissolved during, or within
a few seconds after, solvent
addition
30–90 seconds
2–5 minutes
>5 minutes
Phenyl vinyl ether (2.5 g, 21 mmol) was slowly added to a CH2Cl2
(10 ml) solution of Grubbs’ 1st generation catalyst, (PCy3)2Cl2-
Ru]C(H)Ph (2 g, 2.4 mmol). This solution was allowed to stir at
room temperature for 30 minutes, during which time the purple
solution changed to a dark red color. The solvent was evaporated
in vacuo, and the resulting solid was washed with 4 ꢂ 10 ml of
Moderate
Slow
Deficient
1
cold (ꢃ25 ꢀC) pentane to reveal a red solid (1.90 g, 93%). H
categorized as either rapid, moderate, slow or deficient, based on
Table 1:
NMR (300 MHz, CDCl3): d ppm 14.84 (s, 1H), 7.36 (apparent t,
J ¼ 7.8 Hz, 2H), 7.21 (t, J ¼ 7.5 Hz, 1H), 7.09 (d, J ¼ 7.8 Hz, 2H),
2.72–2.64 (m, 6H), 2.01–1.97 (m, 12H), 1.81–1.71 (m, 19H), 1.61–
1.51 (m, 13H), 1.35–1.20 (m, 16H). High resolution mass spec-
trometry (HRMS) expected 838.3479, found 838.3484.
Additionally, an extra dissolution experiment was performed
with double the amount of each solvent in order to determine the
dependence of the dissolution rate on the degree of under-
saturation.
Synthesis of N,N-dimethyl-5-norbornene-2-carboxamide
Dissolving Grubbs’ catalyst in monomer
N,N-Dimethylacrylamide (15 g, 0.15 mol) was added to 250 ml
ethyl acetate. Over the course of 30 minutes, freshly distilled
cyclopentadiene (11 g, 0.17 mol) was added dropwise. After
complete addition, the solution was brought to a gentle reflux
and allowed to stir for 5 days in the dark, with an additional 1 ml
of cyclopentadiene being added daily. Complete consumption of
N,N-dimethylacrylamide was observed by silica gel thin-layer
chromatography (ethyl acetate, Rf ¼ 0.6), and volatiles were
removed in vacuo. The resulting crude product was cooled to
ꢁ5 ꢀC and triturated three times with pentane (1 ꢂ 250 ml, 2 ꢂ
100 ml). Upon warming to room temperature, a dark brown,
viscous liquid was isolated (19.5 g, 78%) with sufficient purity
250 ꢄ 1 mg of modified Grubbs’ 1st generation catalyst,
(PCy3)2Cl2Ru]C(H)OPh, were weighed into a flame-dried vial,
and the catalyst particles were evenly distributed throughout the
area of the bottom of the vial. Exactly 5 ml of monomer or
monomer blend containing 0.2 ml tetramethylsilane internal
standard were quickly added to the catalyst in a manner that did
not disturb the even distribution of catalyst on the bottom of the
vial. At 5 minute intervals up to 30 minutes, 0.1 ml of liquid was
withdrawn from the center of the solution. To these aliquots were
added 10 ml of ethyl vinyl ether to ensure minimal or no poly-
merization during further characterization. 1H NMR spectra
(400 MHz, CDCl3, 64 scans) were taken of the aliquots, and the
amount of dissolved catalyst is defined as the molar ratio of
catalyst to TMS internal standard, denoted herein as ‘‘relative
intensity.’’ Relative intensity was calculated by
1
(>95%). Product was a mixture of endo/exo-isomers (ꢁ4/1). H
NMR (300 MHz, CDCl3) for the endo-isomer: d ppm 6.15 (dd, J
¼ 3.0, 5.7 Hz, 1H), 6.02 (dd, J ¼ 3.0, 5.7 Hz, 1H), 3.10–3.01 (m,
2H), 3.07 (s, 3H), 2.90–2.83 (m, 1H), 2.88 (s, 3H), 1.97–1.89 (m,
1H), 1.41–1.32 (m, 2H), 1.28 (apparent d, J ¼ 8.1 Hz, 1H). High-
resolution mass spectrometry (HRMS) expected 165.1154, found
165.1160.
P
IRu]CðHÞR
ITMS=12
Relative intensity ¼
(1)
where SIRu]C(H)R is the sum of all ruthenium carbene integra-
tions, and ITMS is the integration value of tetramethylsilane.
Reported relative intensity values are the average of three
dissolution experiments.
Dissolving Grubbs’ catalyst in solvent
5 mg of Grubbs’ 1st generation catalyst were weighed into a small,
flame-dried vial, and the catalyst particles were evenly distributed
throughout the bottom area of the vial. A given solvent was
added to the vial via syringe (0.5 ml). The solvent addition was
rapid enough to quickly cover the catalyst layer, but not so rapid
that the even distribution of the catalyst along the bottom of the
vial was disturbed. For those solvents whose saturation point
was not reached after dissolving the 5 mg of catalyst (i.e. solu-
bility of catalyst in solvent >5 mg per 0.5 ml), the approximate
time for complete dissolution was recorded. For solvents that
became saturated prior to completely dissolving the 5 mg cata-
lyst, the saturation level was determined, and a similar dissolu-
tion experiment was rerun with half the amount of catalyst that
can saturate the solvent (e.g. if the saturation level of Grubbs’
catalyst in a solvent was determined to be 4 mg catalyst per 0.5 ml
solvent, the subsequent dissolution test was done with 2 mg
catalyst in 0.5 ml solvent). Three dissolution experiments were
performed with each solvent. Each solvent’s dissolution rate was
Results and discussion
Healing monomer library dissolution parameters
For each monomer in the healing monomer library (shown in
Fig. 2) we calculated a set of Hansen parameters, which were
Fig. 2 ROMP-active, healing monomer library.
4200 | J. Mater. Chem., 2010, 20, 4198–4206
This journal is ª The Royal Society of Chemistry 2010

View Online
Table 2 Complete Hansen parameters for the healing monomer library
originally developed to provide a semi-quantitative model for
predicting cohesive compatibility between solvents, plasticizers,
polymers, dyes, etc.38 Hansen parameters, denoted as dx, are
defined as the square root of the cohesive energy density related
to an x-type intermolecular attraction, where chemicals with
similar parameters have both similar cohesive energies and, more
importantly for the present work, greater chemical compatibility.
A conceptual explanation of Hansen parameters and their
calculation is discussed briefly here, and in much greater detail in
ref. 39.
Monomera
dT
dD
dP
dH
–H
16.0
17.8
16.2
16.7
16.9
22.9
21.4
19.3
20.4
18.2
18.5
16.9
22.7
15.5
17.1
15.4
15.9
15.3
18.0
18.5
17.5
16.1
16.7
16.7
15.7
18.5
1.4
3.7
–CH]CH–CH2–
–CH]CH2
]CH–CH3
–COOEt
–CH2–OH
–COOH
–COCH3
–CHO
–CH2–Cl
–CH2–Br
–CH2–O–CH3
–CON(CH3)2
1.8
1.7
2.0
3.2
5.1
3.5
5.7
10.4
6.3
6.2
3.9
9.8
4.7
4.6
4.7
6.4
13.2
10.3
6.0
6.9
3.7
5.0
4.8
8.8
A Hansen parameter of a chemical’s x-type cohesive force can
be directly calculated from the x-type force’s contribution to the
total cohesive energy of a chemical, Ex, normalized by its molar
volume, V:
a
The functional group listed refers to a derivative of 2-norbornene
functionalized with this group at the 5-position.
rffiffiffiffiffi
Ex
dx ¼
(2)
V
There are three parameters historically defined as Hansen
parameters: dD, the parameter related to intermolecular disper-
sion forces; dP, the parameter related to fixed-dipole forces; and
dH, a parameter classically known as the hydrogen-bonding
parameter, but often regarded as an electron-transfer parameter.
As shown in eqn (3), the sum of the squares of the Hansen
parameters is equivalent to the total parameter, dT, which is
known as the one-dimensional, Hildebrand parameter. Addi-
tionally, using eqn (2) and (3), the relationship can be rewritten
directly as a function of cohesive energy densities (eqn (4)),
although dealing with the smaller Hansen parameters is often
more convenient.
presumably a result of the norbornenyl functionality present on
all of our healing monomers being a highly strained, bicyclic
structure, while group contributions are generally intended for
use with small, linear, and strain-free molecules. Hence, we
experimentally calculated the ‘‘atomic group contribution’’ for
the 2-norbornenyl-group with covalent connectivity at the
5-position (the functionalized position in the majority of our
2-norbornenyl-based library). Then, our 2-norbornenyl ‘‘group
contribution’’ was used concurrently with the Beerbower group
contributions to build Hansen parameters for our healing
monomer library. Further details regarding the calculation of the
2-norbornenyl ‘‘group contribution’’ are available in the ESI† for
this paper, and the full set of Hansen parameters for all healing
monomers is shown in Table 2.
2
2
2
dT ¼ dD2 + dP + dH
ET ED EP EH
(3)
(4)
The parameters for the healing monomer library are shown in
Fig. 3, expressed as a 2-dimensional Hansen Parameter Map,
which we refer to as a Healing Monomer Map. Since the dD
Hansen parameters for all of the healing monomers are very
¼
þ
þ
V
V
V V
Then, chemicals with similar parameters can be considered
chemically compatible and have some favorable intermolecular
interactions. While Hansen parameters are most often used to
make predictions regarding thermodynamic-based interactions
of two or more chemicals (e.g. solubility, polymer swelling in
a solvent, polymer mixing, etc.), it has been shown that they can
also potentially be useful in linking together chemicals’ kinetic
processes,40 but to the best of our knowledge, have never been
used to make predictions on dissolution. However, it should be
noted that these parameters are only meant to be a useful guide
for making predictions, and their specific values should be
considered approximations that cannot predict minor differences
in behavior.
½
similar (16.7 ꢄ 1.1 MPa ), they are intentionally ignored both in
While Hansen parameters can be calculated experimentally,
reasonably accurate values can be obtained using the Beerbower
group contribution method, where atomic groups (e.g. –CH2–,
]CH–, –OH, –C6H5, –NH2, etc.) have fixed cohesive energy
contributions related to a specific Hansen intermolecular force
(i.e. dispersion, fixed-dipole or hydrogen-bonding). Lists of the
Beerbower group contributions can be found in numerous
sources.39 In order to determine Ex for a given chemical, the sum
of the appropriate atomic group contributions for that chemical
is determined, and the corresponding Hansen parameters are
calculated using eqn (2). However, our initial efforts to directly
use this group contribution approach were unsuccessful,
Fig. 3 Healing Monomer Map for the ROMP-active library used in this
study.
This journal is ª The Royal Society of Chemistry 2010
J. Mater. Chem., 2010, 20, 4198–4206 | 4201

View Online
Fig. 3 and when comparing the parameters of different chemicals
(vide infra), which is commonly done with Hansen parameters. It
is noteworthy that, for all healing monomers, dH > 3. This is true
even for the hydrocarbon monomers that are not expected to
show significant hydrogen bonding. But as mentioned above, the
‘‘hydrogen-bonding’’ parameter nomenclature is something of
a misnomer, used largely due to historical precedent, and instead
should be considered as a parameter related to electron-transfer
intermolecular forces. Therefore, the unexpectedly large dH
values can be explained by the norbornenyl-group’s double bond
acting as a Lewis base, which is essentially defined as a species
that can undergo electron-transfer. Furthermore, the Lewis
basicity of the norbornenyl double bond is especially high,
relative to other double bonds, because donation of its
p-bonding electrons would result in a moderate relief of ring-
strain in the highly strained norbornenyl functionality. All other
Hansen parameters for the healing monomers fall within
expected ranges for the functional groups on each monomer (i.e.
hydrocarbons have low dP values, the amide has a high dP value,
monomers with a removable hydrogen have high dH values, etc.).
seemed to have minimal effect on the dissolution. At the very
least, the different degrees of undersaturation did not create
ambiguity as to which of the four qualitative dissolution cate-
gories the solvents should be placed.
It is clear from Fig. 4b that there are localized regions of
rapid, moderate, slow and deficient dissolution on the Hansen
Parameter Map; the regions of rapid dissolution are marked by
arrows. Additionally, the regions of rapid dissolution seem to be
roughly circular in shape, and proceeding outside the ‘‘rapid’’
dissolving circle are sequential regions of moderate, slow, and
deficient dissolution. This is consistent with what is observed for
polymers, which commonly have ‘‘spheres’’ on their Hansen
Parameter Maps that correspond to regions of good solubility,
swelling, etc. in solvents whose parameters are within the
‘‘sphere,’’ and the center coordinates of the ‘‘sphere’’ correlates
to the polymer’s Hansen parameters.39 While the existence of
multiple spheres, as in this work, is very uncommon with
polymers, there is some precedent for multiple spheres in
a
Hansen Parameter Map for other organometallic
compounds.41 Therefore, we confidently assign Grubbs’ catalyst
two sets of Hansen parameters, which are roughly calculated to
be (dP ¼ 2.9, dH ¼ 3.3) and (dP ¼ 7.0, dH ¼ 7.1).
Catalyst dissolution parameters
The Hansen parameters for Grubbs’ Catalyst were determined
by qualitatively measuring dissolution of the catalyst in various
common organic solvents with well known parameters. A total of
34 solvents with a wide array of Hansen parameters were chosen
(Hansen parameters for common solvents can be found in
numerous sources, for example ref. 39), which are plotted in
Fig. 4a. While it was difficult to quantify dissolution rate of the
catalyst in the solvents for a variety of reasons (some extremely
rapid dissolution, some extremely slow dissolution, and an
overall great deal of scatter in quantitative measurements), it was
surprisingly easy and reliable to qualitatively determine disso-
lution rates. Dissolution rates of the catalyst in the solvents
naturally fell into four categories, which we denote herein as
rapid, moderate, slow and deficient (Fig. 4b), which are described
in more detail in the experimental section of this paper. Assign-
ment of the solvents into these four categories was straightfor-
ward, and very few solvents were at the ‘‘borderline’’ of two
categories, but any type of ranking of the solvents within each
category is futile due to the amount of scatter during attempts to
quantify dissolution. On a side note, several of the solvents used
in this study are known to react with Grubbs’ catalyst (e.g.
acetonitrile, acrylonitrile, aniline, pyridine, piperidine and
pyrrole), and can potentially skew their inherent dissolution
kinetics. However, we found that our interpretation of the
Hansen Parameter Map both including and excluding the data
points accrued from these solvents was essentially identical.
During dissolution measurements, which are described in
more detail in the experimental section of this paper, temperature
(room temperature), agitation rate (no agitation), and interfacial
area of the catalyst available to the monomer were kept constant,
forcing the dissolution rate to be dependent solely on chemical
compatibility between the solvents and the catalyst. However,
one potential concern while taking dissolving measurements is
that the degree of undersaturation may affect the dissolution
rate. But increasing the amount of solvent available to dissolve
the catalyst, which would increase the degree of undersaturation,
Fig. 4 (a) Solvent parameter map and (b) results of dissolution
measurements of Grubbs’ catalyst in each solvent. The arrows denote
regions of superior dissolution.
4202 | J. Mater. Chem., 2010, 20, 4198–4206
This journal is ª The Royal Society of Chemistry 2010

View Online
Table 3 Hansen parameters for three healing monomers and their
volume fractions required to make a blend that approximately matches
one of the Hansen parameters of the Grubbs’ catalyst (dP ¼ 2.9, dH ¼ 3.3).
The ‘‘R ¼’’ corresponds to the functional group at the 5-position of the
norbornenyl-based monomer
chosen whose overall Hansen parameters are reasonably close to
that of the Grubbs’ catalyst; possible errors resulting from this
modification are discussed below.
An obvious problem with measuring the dissolution rate of
Grubbs’ catalyst with a ROMP-reactive monomer is the ensuing
polymerization reaction affecting dissolution. While the dynamic
structural and rheological environment resulting from bulk
polymerization is indeed indicative of what is actually occurring
during the complex self-healing process, for the purposes of this
study where we focus solely on dissolution rates (and need
a suitable method to measure dissolution), the polymerization
reaction must be temporarily stopped. To achieve this, a modi-
fied version of Grubbs’ catalyst with extremely low ROMP-
reactivity was synthesized (shown in Fig. 5 along with the
structure of the Grubbs’ catalyst). While this catalyst is not
exactly inactive, the placement of an oxygen atom between the
carbene and the phenyl group places the catalyst into a thermo-
dynamic well, which was sufficient to inhibit an appreciable
amount of polymerization in the time-scale of our experiments.
In fact, for other alkoxy carbene-derivatives of Grubbs’ catalyst,
this ROMP-inhibition was observed for higher catalyst concen-
trations, longer times, and higher temperatures than those in this
study.42 Of course, this structural modification should not come
at the expense of significantly changing the Hansen parameters of
the catalyst. Thus, dissolution tests of the modified catalyst in
solvent similar to those described in the previous section were
done, which yielded similar results as the unmodified catalyst.
Volume fraction
½
½
Entry R ¼ H R ¼ COOEt R ¼ COCH3 dP/MPa
dH/MPa
1a
2
3
4
1
0
0
0.62
0
1
0
0.06
0
0
1
0.32
1.4
3.2
5.7
2.9
3.7
6.5
6.0
4.6
a
Test conducted at ꢁ50 ꢀC.
Model validation
By comparing Fig. 3 and 4b, it is apparent that no healing
monomer’s Hansen parameters coincide very well with Grubbs’
catalyst’s parameters, implying that the maximum dissolution
kinetics cannot be achieved with a one-component healing
monomer. However, Hansen parameters are known to obey
a simple rule of mixing; in other words, any miscible blend of
liquids takes on the Hansen parameters intermediate between the
separate components of the blend,39 as shown in eqn (5).
P
dx,blend
¼
qidx,i
(5)
In this equation, dx,blend is the Hansen parameter of the x-type
cohesive force (i.e. Dispersion, Dipole or Hydrogen bonding
forces) in the blend of liquids, qi is the volume fraction of the i-th
liquid in the blend, and dx,i is the Hansen parameter of the x-type
cohesive force for the i-th liquid in the blend. Therefore, to
experimentally validate our model, various healing monomers
were blended in the appropriate volume fractions that would
cause the blend to have Hansen parameters similar to the cata-
lyst’s parameters. The two healing monomer blends chosen for
this study (one blend to match each of the Grubbs’ catalyst’s two
Hansen parameters) are shown in Tables 3 and 4. It should be
noted that the blend represented in Table 3 (Hansen parameters:
(dP ¼ 2.9, dH ¼ 4.6)) does not exactly match the estimated
Hansen parameters of Grubbs’ catalyst (dP ¼ 2.9, dH ¼ 3.3). This
is because this set of Grubbs’ catalyst parameters falls out of the
range of the parameters of our healing monomer library, making
it mathematically impossible to find a monomer blend that
precisely matches the catalyst’s parameters. Hence, a blend was
Fig. 5 (a) Grubbs 1st generation catalyst and (b) a ROMP-inactive,
modified version of Grubbs’ catalyst.
Table 4 Hansen parameters for three healing monomers and their
volume fractions required to make a blend that matches one of the
Hansen parameters of the Grubbs’ catalyst (dP ¼ 7.0, dH ¼ 7.1). The
‘‘R ¼’’ corresponds to the functional group at the 5-position of the
norbornenyl-based monomer
Volume fraction
½
½
Entry R ¼ COOH R ¼ CHO R ¼ CH2Br dP/MPa
dH/MPa
Fig. 6 Dissolution of a modified Grubbs’ catalyst in various monomers
and monomer blends. The ‘‘R ¼’’ group in the legend denotes the func-
tionality at the 5-position of the norbornenyl group. The monomer blend
has Hansen parameters of (dP ¼ 2.9, dH ¼ 4.6), which is similar to that of
the catalyst.
1
2
3
4
1
0
0
0.27
0
1
0
0.36
0
0
1
0.37
3.5
10.4
6.2
10.3
6.9
5.0
7.1
7.0
This journal is ª The Royal Society of Chemistry 2010
J. Mater. Chem., 2010, 20, 4198–4206 | 4203

View Online
Therefore the modified Grubbs’ catalyst shown in Fig. 5 was
used in all dissolution tests of catalyst in healing monomer.
Ideally, monomers and monomer blends with parameters
closer to the catalyst’s parameters would dissolve the catalyst
fastest. NMR-based dissolution tests (see Experimental section)
of the first monomer blend (dP ¼ 2.9, dH ¼ 4.6; shown in Table 3)
and the individual components of this blend are shown in Fig. 6.
As mentioned earlier, it was not possible to construct a monomer
blend with Hansen parameters that equaled the catalyst’s first set
of parameters, so the blend in Table 3 was chosen as having
parameters close to the catalyst; at the very least, the blend’s
parameters were closer to the catalyst than any of the compo-
nents of the blend. As seen in Fig. 6, the dissolution rate of the
catalyst by the blend and by norbornene (R ¼ H) were
approximately equal, followed by the acetyl-functionalized nor-
bornenyl derivative (R ¼ COCH3) and the ester-functionalized
derivative (R ¼ COOEt). However, it is important to note that
because norbornene is a solid at room temperature, its dissolu-
tion experiments had to be carried out at elevated temperatures
(ꢁ50 ꢀC). Although it is hard to distinguish differences in the
dissolution rates of the blend and of norbornene from experi-
mental error, it is definitely reasonable to presume that norbor-
nene’s theoretical room temperature dissolution rate would be
slower than that of the blend, which is in agreement with our
model predictions. Furthermore, the proximity of norbornene’s
Hansen parameters to the catalyst is not too dissimilar from the
proximity of the blend’s parameters to the catalyst. So even if the
reduction in norbornene’s dissolution rate from ꢁ50 ꢀC to room
temperature was small enough to still be indistinguishable from
that of the blend, our model would still predict dissolution
behavior reasonably well. The slower dissolution rates of the
acetyl and ester-functionalized norbornenyl groups are also in
agreement with predictions—both monomers have Hansen
parameters that are, comparably, far from that of the catalyst. In
fact, the catalyst’s parameters (dP ¼ 2.9, dH ¼ 3.3) are slightly
closer to that of the acetyl-functionalized derivative than the
ester-derivative. This predicts that the dissolution rate of the
catalyst by the acetyl-derivative would be slightly faster than the
ester, which is also in agreement with what is experimentally
observed.
A blend whose Hansen parameters matched the second set of
Grubbs’ catalyst’s parameters (dP ¼ 7.0, dH ¼ 7.1) was con-
structed from three of the healing monomers, which are shown in
Table 4. Dissolution experiments of the modified Grubbs’ cata-
lyst with this blend and the individual components of the blend
were conducted, and this data is shown in Fig. 7. The acid-
functionalized norbornenyl-derivative (R ¼ COOH) was obvi-
ously the slowest to dissolve the catalyst, which is expected from
the large distance of the acid’s Hansen parameters from that of
the catalyst. The aldehyde-functionalized monomer (R ¼ CHO)
was much faster to dissolve the catalyst than the acid, but slower
than both the blend and the bromomethyl-functionalized nor-
bornene (R ¼ CH2Br). This is also predicted with our Hansen
parameter-based model, as the aldehyde-derivative’s Hansen
parameters are closer to the catalyst’s parameters than the acid,
but further than both the blend and the bromomethyl-derivative.
However, the dissolution rates of the catalyst by the blend and
the bromomethyl-norbornene monomer are nearly identical. The
Hansen parameters of the bromomethyl-derivative are closer to
the catalyst’s parameters than almost all of the other components
in the healing monomer library, so a relatively rapid dissolution
rate is expected, but the fact that it has a dissolution rate as rapid
as the blend (which exactly matches the catalyst’s parameters)
falls outside of what is expected with our model. There are
several possible explanations for this behavior. First, the possi-
bility that the second set of Grubbs’ catalyst Hansen parameters
(dP ¼ 7.0, dH ¼ 7.1) is not entirely accurate cannot be ignored. As
mentioned earlier, Hansen parameters are a semi-quantitative
technique in that the large degree of error in their calculation
makes it difficult to predict minor differences in behavior.
Perhaps the difference in the proximities of the parameters
between the catalyst/blend and the catalyst/bromomethyl-deriv-
ative is too small, thereby causing the inaccuracies in parameter
calculations to become more evident. But the most likely expla-
nation for the blend’s dissolution rate being closer than expected
to the bromomethyl-derivative’s dissolution rate is viscosity.
Dissolution rates are known to be viscosity-dependant,43 with
trends of generally slower dissolution with increasing viscosity. It
is easily observed that the blend, which contains 27% of the
highly viscous acid-functionalized norbornene, is of a higher
viscosity than the bromomethyl norbornenyl-derivative. Hence,
it is reasonable to assume that the blend’s rate of dissolving
catalyst is being effectively reduced as a result of its higher
viscosity, and the model predictions were not necessarily incor-
rect. Nonetheless, it should be noted in future utilizations of this
model that dissolution rates will probably fall short of Hansen
parameter-based predictions when using healing monomers of
higher viscosities. While this is indeed a limitation of using the
model presented here, most of the healing monomers in the
library of Fig. 1 were liquids of relatively low viscosity. In fact,
this viscosity effect probably has not manifested itself in the other
experiments presented in this paper because all of the other
healing monomers tested have similar viscosities, at least by
visual inspection. Furthermore, high-viscosity healing monomers
and monomer blends are already considered unfavorable for self-
healing for a number of reasons, and they would therefore likely
Fig. 7 Dissolution of a modified Grubbs’ catalyst in various monomers
and monomer blends. The ‘‘R ¼’’ group in the legend denotes the func-
tionality at the 5-position of the norbornenyl group. The monomer blend
has Hansen parameters of (dP ¼ 7.0, dH ¼ 7.1), which matches that of the
catalyst.
4204 | J. Mater. Chem., 2010, 20, 4198–4206
This journal is ª The Royal Society of Chemistry 2010

View Online
14 M. R. Kessler and S. R. White, Cure kinetics of ring-opening
metathesis polymerization of dicyclopentadiene, J. Polym. Sci., Part
A: Polym. Chem., 2002, 40, 2373–2383.
be prematurely excluded as a possibility when considering
a feasible healing monomer/catalyst combination.
15 G. E. Larin, M. R. Kessler, N. Bernklau and J. C. DiCesare,
Rheokinetics of ring-opening metathesis polymerization of
norbornene based monomers intended for self-healing applications,
Polym. Eng. Sci., 2006, 46, 1804–1811.
16 T. C. Mauldin and M. R. Kessler, Latent catalytic systems for ring-
opening metathesis-based thermosets, J. Therm. Anal. Calorim.,
2009, 96, 705–713.
17 M. R. Kessler and S. R. White, Self-activated healing of delamination
damage in woven composites, Composites, Part A, 2001, 32, 683–699.
18 S. H. Cho, H. M. Andersson, S. R. White, N. R. Sottos and
P. V. Braun, Polydimethylsiloxane-based self-healing materials,
Adv. Mater., 2006, 18(8), 997–1000.
19 Y. C. Yuan, M. Z. Rong, M. Q. Zhang, J. Chen, G. C. Yang and
X. M. Li, Self-healing polymeric materials using epoxy/mercaptan
as the healant, Macromolecules, 2008, 41(14), 5197–5202.
20 M. M. Caruso, B. J. Blaiszik, S. R. White, N. R. Sottos and
J. S. Moore, Full recovery of fracture toughness using a nontoxic
solvent-based self-healing system, Adv. Funct. Mater., 2008, 18,
1898–1904.
21 X. Sheng, T. C. Mauldin and M. R. Kessler, J. Polym. Sci., Part A:
Polym. Chem., 2010, submitted.
22 J. D. Rule, E. N. Brown, N. R. Sottos, S. R. White and J. S. Moore,
Wax-protected catalyst microspheres for efficient self-healing
materials, Adv. Mater., 2005, 17, 205–208.
23 X. Liu, J. K. Lee, S. H. Yoon and M. R. Kessler, Characterization of
diene monomers as healing agents for autonomic damage repair,
J. Appl. Polym. Sci., 2006, 101, 1266–1272.
Conclusion
In this paper, a model was developed using the concept of
Hansen parameters in order to make predictions regarding the
dissolution of a solid catalyst in liquid monomers for self-healing
applications. More specifically, we developed our model in
reference to the dissolution of Grubbs’ catalyst in a library of
ROMP-active monomers, which are often used as a catalyst/
healing monomer combination in other self-healing works.
Hansen parameters were calculated for a small library of nor-
bornenyl-based monomers, which are generally reactive via the
ROMP reaction. The Grubbs’ catalyst was found to have two
sets of Hansen parameters, and blends of healing monomers
created to match these parameters were found to dissolve the
catalyst rapidly. However, the model is limited in its ability to
predict and compare dissolution rates of liquids with significantly
different viscosities. While our model was developed for
a ROMP-based self-healing system, we believe this Hansen
parameter technique is fundamental and versatile enough to be
applied to other types of self-healing chemistries.
24 J. K. Lee, X. Liu, S. H. Yoon and M. R. Kessler, Thermal analysis of
ring-opening metathesis polymerized healing agents, J. Polym. Sci.,
Part B: Polym. Phys., 2007, 45, 1771–1780.
25 X. Liu, X. Sheng, J. K. Lee, M. R. Kessler and J. S. Kim, Rheokinetic
evaluation of self-healing agents polymerized by Grubbs’ catalyst
embedded in various thermosetting resins, Compos. Sci. Technol.,
2009, 69, 2102–2107.
26 X. Liu, X. Sheng, M. R. Kessler and J. K. Lee, Isothermal cure
characterization of ROMP healing agents for autonomic damage
repair: the glass transition temperature and conversion, J. Therm.
Anal. Calorim., 2007, 89, 453–457.
Acknowledgements
This work was supported by The American Chemical Society
Petroleum Research Fund (ACS PRF# 47700-AC7). The
authors gratefully acknowledge Dr Xia Sheng, Dr Wonje Jeong,
and Dr Malika Jeffries-EL for helpful discussions.
References
27 X. Liu, X. Sheng, J. K. Lee and M. R. Kessler, Synthesis and
characterization of melamine–urea–formaldehyde microcapsules
containing ENB-based self-healing agents, Macromol. Mater. Eng.,
2009, 294, 389–395.
1 M. R. Kessler, Self-healing: a new paradigm in materials design, Proc.
Inst. Mech. Eng., Part G, 2007, 221(4), 479–495.
2 R. P. Wool, Self-healing materials: a review, Soft Matter, 2008, 4,
400–418.
28 A. S. Jones, J. D. Rule, J. S. Moore, S. R. White and N. R. Sottos,
Catalyst morphology and dissolution kinetics of self-healing
polymers, Chem. Mater., 2006, 18, 1312–1317.
29 T. C. Mauldin, J. D. Rule, N. R. Sottos, S. R. White and J. S. Moore,
Self-healing kinetics and the stereoisomers of dicyclopentadiene,
J.R. Soc., Interface, 2007, 4, 389–393.
30 J. Hildebrand and R. L. Scott, The Solubility of Nonelectrolytes,
Reinhold, New York, 3rd edn, 1950.
31 P. V. Bonnesen, C. L. Puckett, R. V. Honeychuck and W. H. Hersh,
Catalysis of Diels–Alder reactions by low oxidation state transition-
metal Lewis acids: fact and fiction, J. Am. Chem. Soc., 1989, 111,
6070–6081.
3 S. D. Bergman and F. Wudl, Mendable polymers, J. Mater. Chem.,
2008, 18, 41–62.
4 S. R. White, N. R. Sottos, P. H. Geubelle, J. S. Moore, M. R. Kessler,
S. R. Sriram, E. N. Brown and S. Viswanathan, Autonomic healing of
polymer composites, Nature, 2001, 409, 794–797.
5 M. R. Kessler, N. R. Sottos and S. R. White, Self-healing structural
composite materials, Composites, Part A, 2003, 34, 743–753.
6 K. S. Toohey, N. R. Sottos, J. A. Lewis, J. S. Moore and S. R. White,
Self-healing materials with microvascular networks, Nat. Mater.,
2007, 6, 581–585.
7 K. S. Toohey, C. Hansen, N. R. Sottos, J. A. Lewis and S. R. White,
Delivery of two-part self-healing chemistry via microvascular
networks, Adv. Funct. Mater., 2009, 19, 1399–1405.
8 E. L. Dias, S. T. Nguyen and R. H. Grubbs, Well-defined ruthenium
olefin metathesis catalysts: mechanism and activity, J. Am. Chem.
Soc., 1997, 119, 3887–3897.
32 P. K. Freeman, V. N. Mallikarjuna Rao, D. E. George and
G. L. Fenwick, The reactions of exo- and endo-5-
chloromethylnorbornene with sodium, J. Org. Chem., 1967, 32,
3958–3963.
33 S. J. Dolman, K. C. Hultzsch, F. Pezet, X. Teng, A. H. Hoveyda and
R. R. Schrock, Supported chiral Mo-based complexes as efficient
catalysts for enantioselective olefin metathesis, J. Am. Chem. Soc.,
2004, 126, 10945–10953.
34 I. S. Ponticello, The preparation of a-substituted acrylic esters,
J. Polym. Sci., Polym. Chem. Ed., 1979, 17, 3509–3518.
35 S. Oh, J.-K. Lee, P. Theato and K. Char, Nanoporous thin films based
on polylactide-grafted norbornene copolymers, Chem. Mater., 2008,
20, 6974–6984.
9 T. M. Trnka and R. H. Grubbs, The development of L2X2Ru]CHR
olefin metathesis catalysts: an organometallic success story, Acc.
Chem. Res., 2001, 34, 18–29.
10 C. W. Bielawski and R. H. Grubbs, Living ring-opening metathesis
polymerization, Prog. Polym. Sci., 2007, 32, 1–29.
11 E. Kirkby, V. Michaud, J. A. Manson, J. D. Rule, N. R. Sottos and
S. R. White, Embedded shape-memory allow wires for improved
performance of self-healing polymers, Adv. Funct. Mater., 2008, 18,
2253–2260.
36 P. Beslin, D. Lagain, J. Vialle and C. Minot, Preparation of
a-unsaturated acyclic thioketones—regioselective dimerization of
4H-1,3-dithiin, Tetrahedron Lett., 1981, 37, 3839–3845.
37 W. M. Lauer and M. A. Spielman, Synthesis of a,b-unsaturated
ethers, J. Am. Chem. Soc., 1931, 53, 1533–1536.
12 X. Sheng, J. K. Lee and M. R. Kessler, The influence of cross-link
density on the properties of ROMP thermosets, Polymer, 2009, 50,
1264–1269.
13 G. O. Wilson, J. S. Moore, S. R. White, N. R. Sottos and
H. M. Andersson, Adv. Funct. Mater., 2008, 18, 44–52.
This journal is ª The Royal Society of Chemistry 2010
J. Mater. Chem., 2010, 20, 4198–4206 | 4205

View Online
38 (a) C. M. Hansen, The three dimensional solubility parameter—key to
paint component affinities I. Solvents, plasticizers, polymers, and
resins, J. Paint Technol., 1967, 39, 104–117; (b) C. M. Hansen, The
three dimensional solubility parameter—key to paint component
affinities II. Dyes, emulsifiers, mutual solubility and compatability,
and pigments, J. Paint Technol., 1967, 39, 505–510; (c) C. M. Hansen
and K. Skaarup, The three dimensional solubility parameter—key to
paint component affinities III. Independent calculation of the
parameter components, J. Paint Technol., 1967, 39, 11–514.
40 A. F. M. Barton, Applications of solubility parameters and other
cohesion parameters in polymer science and technology, Pure Appl.
Chem., 1985, 57, 905–912.
41 Personal communication between Alan Beerbower and Charles
Hansen, which revealed two distinct solubility spheres for lithium
stearate (further details in ref. 39).
42 J. Louie and R. H. Grubbs, Metathesis of electron-rich olefins:
structure and reactivity of electron-rich carbene complexes,
Organometallics, 2002, 21, 2153–2164.
39 C. M. Hansen, Hansen Solubility Parameters: A User’s Handbook,
CRC Press, Boca Raton, FL, 2nd edn, 2007.
43 U. V. Banakar, Pharmaceutical Dissolution Testing, Marcel Dekker,
Inc., New York, 1992, vol. 49.
4206 | J. Mater. Chem., 2010, 20, 4198–4206
This journal is ª The Royal Society of Chemistry 2010