Article

pubs.acs.org/Macromolecules

A Simple Relationship Relating Linear Viscoelastic Properties and

Chemical Structure in a Model Diels−Alder Polymer Network
Richard J. Sheridan and Christopher N. Bowman*
Department of Chemical and Biological Engineering, University of Colorado, 3415 Colorado Ave., UCB 596, Boulder, Colorado
80303-0596, United States
*
S Supporting Information

ABSTRACT: Although the gel point conversion of a

See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

thermoreversible polymer network is certainly a key parameter

Downloaded via CHALMERS UNIV OF TECHNOLOGY on October 5, 2020 at 08:56:39 (UTC).

in determining the material properties, it is not a conventional

liquid−solid transition as in common, irreversible networks.
Rather, the material’s viscosity is time-dependent and finite at
the gel point and beyond, as bond breakage works in concert
with diffusion to relax stresses imposed on the forming
transient network of the material. For example, in a model
Diels−Alder network with functionality 3.8 and a stoichio-
metric ratio of 10:6 furan:maleimide used here, a crossover
frequency (0.52, 0.32, and 0.12 rad/s) was measured below the
temperature corresponding to gelation (3, 5, and 10 °C below, respectively). In this work, we describe this complex process
occurring in model thermoreversible networks with a simple relationship from the work of Semenov and Rubinstein on
associative transient networks. This relationship provides a toolkit for the prediction of the important engineering and rheological
properties of the material in the postgel regime, such as viscosity, plateau modulus, and relaxation time, based upon the
straightforward estimation of two material-dependent parameters: the gel point conversion pgel and a proportionality constant C.
We show key agreement between theory and experiment as the gel point conversion estimated from network dynamics matches
the classical prediction of the gel point within 4% conversion. We discuss the applicability criteria of the Semenov−Rubinstein
scaling relationship and compare it to time−temperature superposition methods of describing transient network relaxation.

■ INTRODUCTION
Polymer networks composed of thermally reversible cross-
size)14 clusters and network chains and “backbone”15 transient
bonds within those clusters (Figure 1), a scaling relationship
between bond lifetime and relaxation in the sol and gel states
linksincluding noncovalent interactions such as hydrogen
near the gel point can be derived:
bonds1 or π stacking,2 or reversible covalent reactions like the
Diels−Alder,3 boronate−diol,4 and most recently transester- τl ∝ τb|ε| (1)
ification5 reactionsare becoming popular as mendable,6
recyclable,7 and smart8 materials. The ability these various In eq 1, the bond lifetime τb is the decay constant or the inverse
materials have to respond to an external stimulus such as of the kinetic rate constant for the thermoreversible links in the
exposure to light, changes in pH, or an increase in temperature
with a change in material properties8,9 makes them potentially
valuable in many fields, from biomaterials to photolithography.
As these materials are unified, at least in part, by common
attributes of their chemical mechanism, a common structure−
property relationship should be applicable to each of them.
The theoretical foundations for the behavior of “transient”,10
“thermoreversible”,11 “dynamic”,12 or “adaptable”8 networks
have been derived, most recently and completely by Semenov
and Rubinstein.13 Their theory of linear dynamics for
associating polymers is suitable for many aspects and details Figure 1. Visualization of transient cluster relaxation via backbone
specific to their associative network of interest; however, the breakage. Bolded bonds represent critical connections which, when
derivation begins with classical polymer scaling theory that split, transform one transient cluster into two roughly equal parts.
should also be widely applicable to most thermoreversible
networks, as long as they can be approximated by mean-field Received: June 28, 2012
assumptions. The crucial insight they provide is that by Revised: August 12, 2012
considering only “typical” (characteristic percolation cutoff Published: September 14, 2012

© 2012 American Chemical Society 7634 dx.doi.org/10.1021/ma301329u | Macromolecules 2012, 45, 7634−7641
Macromolecules Article

cluster, and τl is the rheological lifetime of the cluster as a Scheme 1. Diels−Alder Reaction between Maleimide and
whole. In their paper, Semenov and Rubinstein make the Furana
argument that the two time scales can be related through ε = p
− pgel, the distance from the Flory16 gel point in terms of
conversion, in the postgel regime. If a thermoreversible network
is considered as a generalized Maxwell material,17 the longest
relaxation time, τmax, represents the transition time scale
between viscoelastic and purely viscous behavior and should
correspond to τl. If we rewrite the scaling relationship as an
explicit proportionality, with a material-dependent constant, C, a
The [4 + 2] tricyclic adduct structure on the right generally forms at
we can algebraically rearrange it to obtain a simple equation in high yield at ambient temperature. At increased temperature, the
terms of the most important material parameters: conversion adduct reversibly decouples to re-form the reactant species at a rate
(p), bond lifetime (τb), gel point conversion (pgel), and the and quantity which are determined by standard chemical equilibrium
longest relaxation time in the network (τmax). and kinetics.
τ
p = C max + pgel and maleimide moieties. Multifunctional furan and maleimide
τb (2) monomers have been used in a number21−23 of applications to
In circumstances where a material is far above its gel point, and form thermoreversibly cross-linked polymers. This reaction
the lifetime of a bond is orders of magnitude longer than the generally follows standard chemical equilibrium behavior as in
time frame of interest, the viscoelastic spectrum is well eq 3, which has previously been reported.3,24
predicted by classical cross-link calculations without consid- x o o
K C(T ) = −1
e(−ΔH − T ΔS )/ RT
eration of the reversible bonds. To calculate elasticity based on [L]0 (1 − x)(r − x) (3)
conversion and the distribution of multifunctional monomers,
we can employ the Miller and Macosko18 recursive method, Here, x is the conversion of a limiting reagent, [L]0 is the initial
solving for the cross-link density, υ, numerically. This value may concentration of that limiting reagent, r is the ratio of limiting
be converted into a “plateau” modulus via the relationship Gp = to excess reagent, ΔH° is the heat of reaction, and ΔS° is the
υRT, disregarding other potential contributions to the modulus entropy of the reaction. This approach describes the
as a first-order approximation.19 At lower temperatures, equilibrium constant KC as a function of absolute temperature
deviations would be seen as the glass transition begins to T with the ideal gas constant R.
interfere with measurement of the purely rubbery behavior.3 For this work we use a commercially available bismaleimide
Similarly, the main property of interest in the regime far away and a low molecular weight, but flexible multifuran cross-linker
from the gel point, i.e., the viscosity, is not substantially (Schemes 2 and 3). We synthesize the multifuran via the
determined by the bond breakage rate. Rather, the viscosity is
largely determined by the properties of the unreacted Scheme 2. Multifuran Monomer SRF: Pentaerythritol Tri-/
monomers. Nearer to the gel point, a simple scaling exponent tetraacrylate (Sartomer SR295) Substituted via a Michael
adequately3 describes viscosity until very close to the gel point Addition with Furfuryl Mercaptan and Octanethiola
(|p − pgel| < εmin),13b where the largest molecular clusters are
big enough that bond breakage begins to have an effect.
In this report, we aim to demonstrate the versatility of the
Semenov−Rubinstein (SR) theory, particularly as expressed in
eq 2, as an accurate description for the viscoelastic behavior of
thermoreversibly cross-linked materials, which represent an
ever-increasingly important class of materials. This approach
will use a chemically well-defined model Diels−Alder network
to couple experiments with theory and demonstrate the validity
of the theory for these materials.

■ BACKGROUND
The Diels−Alder20 cycloaddition reaction, often occurring
a
Functionality is adjusted by varying the ratio of furan to octane
moieties in the reaction step. These compositions are distinguished by
between maleimide and furan functional groups (Scheme 1), displaying the resultant effective functionality in parentheses, e.g.,
has become a common feature of the thermoreversible SRF(3.0).
networks literature, thanks to its relatively ideal and orthogonal
character as an addition reaction. Polymers featuring the Diels− Scheme 3. Bismaleimide Monomer DPBM: (1,1′-
Alder reaction are fairly old, with a patent by Craven21 (Methylenedi-4,1-phenylene)bismaleimide)a
appearing in 1969. However, thanks to work done by Wudl’s
group,6 the field has seen a revitalization, with many different
groups demonstrating different polymer architectures and
Diels−Alder pairs.3,12,22
Arising from an exothermic thermoreversible reaction, the
equilibrium conversion of Diels−Alder functional groups is a
strong function of temperature (eq 3). At low temperatures, the a
Whenever a composition’s stoichiometric ratio is indicated, as in
equilibrium is shifted toward the adduct state, while at elevated SRF(3.8) r = 0.6, the ratio indicates the amount of maleimide used, as
temperatures adducts tend to break apart into the original furan a fraction of the furan moieties used.

7635 dx.doi.org/10.1021/ma301329u | Macromolecules 2012, 45, 7634−7641

Macromolecules Article

Michael addition of furfuryl mercaptan to a simple commercial Table 1. Compositions and Their Propertiesa (All Values in
tetra-acrylate, presenting several helpful features. The synthesis %)
is straightforward and scalable and requires only limited
SRF(3.8) SRF(3.0) SRF(3.0) SRF(3.5) SRF(3.5)
equipment and knowledge to perform. By limiting the reaction composition name r = 0.6 r = 1.0 r = 0.78 r = 0.91 r = 0.6
with a less-than-stoichiometric amount of furfuryl mercaptan,
furan excess 167 100 128 110 167
reduced functionality analogues can be produced simply by
gel point as 77 71 80 67 82
eliminating residual acrylate groups in a second Michael predicted by
addition step, using a thiol inert to the Diels−Alder reaction, eq 5
such as octanethiol. gel point as 81 72 81 67 86
The cross-link density is high enough in the fully predicted by
Figure 6
functionalized case to form a glassy polymer at ambient a
Furan excess is the amount of furan moieties available to react, as a
temperature;24 however, its functionality is low enough that a
percentage of maleimide groups present. It can be calculated via 100/r,
modest inequality of stoichiometry (in our case r = 0.36 to 0.5) where r is the ratio of the concentration of limiting moieties to excess
is sufficient to prevent gelation, even at the equilibrium moieties, because furan is always in excess in these studies.
conversion of the limiting reactant at all temperatures studied.
By combining the effects of these two capabilities for
controlling the system, a gel point conversion and its
corresponding gel temperature can be targeted during the functionality (fe) of 3.8 is achieved in the fully functionalized
formulation step, without any additional synthesis. We took multifuran SRF(3.8) (which is a mixture of tri- and tetrafunc-
advantage of this effect to test the Semenov and Rubinstein tional molecules in a ratio of 0.2:0.8), whereas the lowest
theory with analogous materials, but at several different effective functionality monomer SRF(3.0) is achieved when
conditions. 70% of the acrylates are reacted with furfuryl mercaptan.

■ EXPERIMENTAL AND THEORETICAL METHODS

Synthesis of Monomers. DPBM (1,1′-(methylenedi-4,1-
SRF(3.5) results when the SRF(3.8) and SRF(3.0) are mixed in
equal parts by weight and the functionality of the mixture is
back-calculated. (See Supporting Information for details of the
phenylene)bismaleimide), triethylamine, furfuryl mercaptan, and calculation.) Stoichiometric ratios of e.g. 6:10 maleimide:furan
octanethiol were purchased from Sigma-Aldrich. SR295 was received (r = 0.6) were chosen as intermediate values that would cause a
from Sartomer. Dichloromethane supplied by Fisher was used as significant change in the gel point conversion and cross-link
solvent. density while ensuring the ability to reach gelation with all SRF
SRF(3.8) was synthesized as reported previously.24 Briefly, SR295 monomer mixtures in the temperature range of interest.
and 5% molar excess of freshly distilled furfuryl mercaptan were mixed Furthermore, the use of maleimide as the limiting reagent in
with an equal volume of dichloromethane. This solution was stirred
chilled in an ice bath, and 1 wt % triethylamine was added to catalyze
the Diels−Alder reaction served to limit the amount of
the Michael addition. After 24 h, the reaction was complete as verified potential high-temperature side reactions that involve the
by proton NMR. Excess furfuryl mercaptan and triethylamine were maleimide’s highly electron deficient double bond.25 Notably
removed under high vacuum. absent are the “obvious” choices of the on-stoichiometry, fully
SRF(3.0) was synthesized as described above, but using 70% of the functionalized and the far-off-stoichiometry, partially function-
stoichiometric amount of furfuryl mercaptan, substituting ∼35% alized systems. These systems had important measurables, such
octanethiol. Furfuryl mercaptan reacts much faster, as it has a lower as gel point conversion or low-temperature modulus crossover
pKA, and no excess furfuryl mercaptan peaks were seen in proton frequency, outside the operating limits needed for accurate
NMR upon reaction completion, so the monomer is therefore a 30% assessment of the material behavior.
octanethiol-substituted version of SRF(3.8).
Fourier Transform Infrared Spectroscopy. Infrared spectra
Thermochemistry. To make use of the relationship in eq 2,
were collected on a Thermo-Nicolet Magna-IR 750 Series II it is clear that equilibrium, kinetic, and structural parameters of
instrument with a custom sample stage and temperature controller the monomers must be combined with mechanical measure-
combination. Sample preparation is described in a previous ments of the polymer mixture. Estimates for the heat and
publication.3 entropy of reaction, as well as Arrhenius activation energy and
Rheometry. A TA ARES rheometer measured the complex kinetic prefactors for forward and reverse rate constants, of the
modulus of the materials in this study, using a poly(p-phenylene Diels−Alder reaction are available in the literature. However,
sulfide) (PPS) 18.7 mm plate geometry with a Peltier temperature for the sake of internal consistency and because of variability
controller to minimize thermal gradients across the sample. Sample associated with the specific functionalization and molecular
thickness was between 0.35 and 0.45 mm. Monomers were mixed at
ambient temperature, heated for 4 min at 170 °C, and poured onto the
environment, we re-estimated the parameters for these specific
Peltier plate, which had been preheated to 130 °C. The samples were chemical systems by evaluating temperature-dependent FTIR
then brought to experimental temperature and allowed to remain in data for the equilibrium and time-dependent concentrations of
this state for 15 min to achieve a steady thermal and chemical state. the reactants. The maleimide ring-stretching peak at 690 cm−1
Strain sweeps were performed on representative samples to ensure served as the most reliable source of conversion data. A
that frequency sweeps were linear. nonlinear least-squares fit of the concentration data ([M],
Cross-Link Density. The calculation of cross-link density for use [M]0, [F]0 = [M]0/r, [A] = [M]0 − [M]) to a kinetic model
in determining the theoretical plateau modulus Gp,theor is included as
Supporting Information. d[A] d[F] d[M]

■
− = = = k r[A] − kf [M][F]
dt dt dt (4)
RESULTS AND DISCUSSION (see Figure 2) was used to estimate the forward and reverse
In our experimental design, we chose compositions at rate constants kf and kr corresponding to each measurement,
representative points within the functionality/stoichiometry which also allowed for the calculation of the equilibrium
parameter space (see Table 1). The maximum effective constant KC.
7636 dx.doi.org/10.1021/ma301329u | Macromolecules 2012, 45, 7634−7641
Macromolecules Article

leaving only the measurement of the longest relaxation time as

a function of temperature before the gel point, pgel, and the
material-dependent proportionality constant, C, to be esti-
mated.
Dynamic Mechanical Analysis. The viscoelastic spectra of
our model networks have a number of important features
predicted by SR theory that arise within this experimental
window. As an example, the complex shear modulus of
SRF(3.8) r = 0.6 measured at 115 °C is shown in Figure 4,

Figure 2. Example kinetic parameters for SRF(3.5) r = 0.91 at 100 °C.

The parameters were obtained via nonlinear least-squares fit (solid
line) to the FTIR concentration data (× markers) obtained from
measurements of the maleimide peak found at 690 cm−1.

Figure 4. Shear modulus of SRF(3.8) r = 0.6 at 115 °C. Gp,theor is the

modulus corresponding to a cross-link density calculated from classical
gelation theory. The crossover, G″ local minimum, and cluster
relaxation frequencies (vertical solid and dashed lines) are marked as
2π/τmax, 2π/τ2, and 2π/τrelax, respectively. These features are in
agreement with SR theory predictions.

with those features marked. Gp,theor = υRT and is the theoretical

plateau modulus expected from an equivalent irreversible
phantom network with cross-link density υ or, equivalently,
the instantaneous number of network chains in the material,
calculated according to Miller and Macosko.18 (A description of
the approach to this calculation can be found in the Supporting
Information.) We chose the crossover frequency, ωx, as our
estimate of the longest relaxation time, τmax, when converted to
a characteristic time scale via the relationship τmax = 2π/ωx.
This assumption is most accurate when the crossover occurs at
the maximum of a distinct final G″ peak, which is valid for our
systems in the regime of interest. SR theory expects G′ to scale
as ω0 over the region 2π/τmax to 2π/τrelax (the characteristic
time scale of network chain and sol cluster relaxations) at the
level of Gp,theor. As G′ crosses over Gp,theor in this range, the level
of the complex modulus is in agreement with theoretical
Figure 3. Van’t Hoff and Arrhenius plots for a variety of SRF systems. expectations. Likewise, G″ should scale as ω−1 and ω1 over the
ΔH° = −60 ± 20 kJ/mol, ΔS° = −116 ± 60 J/(mol K), Ea = 88 ± 20 same frequency range, with a minimum occurring at 2π/τ2.
kJ/mol, and k0 = 1.3 × 1011±2 min−1. Wide parameter confidence Because τ2 is the geometric mean of τmax and τrelax, and since
intervals were due to the relatively narrow experimental temperature τrelax does not have a distinctive transition feature, we estimated
range, resulting in high parameter covariance. the chain relaxation frequency as ω22/ωx and indicated it in
Figure 4 with a dashed line. Qualitatively, then, this frequency
sweep paints a picture of (from right to left) a material that
The main outcome of this FTIR analysis was an equilibrium settles into its network-constrained configuration in response to
temperature−conversion relationship p(T) and a temperature− stress on the order of 0.3 s. The material exhibits a relatively
lifetime relationship τb(T) = 1/kr for the Diels−Alder adducts. elastic response to deformations persisting as long as 3−6 s;
These parameters were consistent among the various however, it relaxes any longer-lasting stresses as network chains
compositions and agreed with literature values. This disconnect and rearrange due to the retro-Diels−Alder reaction.
information allows us to rewrite eq 2 Figure 5 shows additional complex modulus curves of the
τ SRF(3.8) r = 0.6 composition, measured at two additional
p(T ) = C max + pgel temperatures. The crossover frequency shifts left as temper-
τb(T ) (5) ature decreases or, equivalently, as equilibrium conversion
7637 dx.doi.org/10.1021/ma301329u | Macromolecules 2012, 45, 7634−7641
Macromolecules Article

is highly sensitive to small errors in the gel point. A

temperature−conversion relationship is required to estimate
the gel point unless conversion can be measured in situ in the
application in question. Alternatively, the analysis can be
framed in terms of “temperature distance” from the “gel
temperature” at equilibrium, with the assumption that change
in conversion is linear with temperature over the narrow
temperature range of interest. This assumption is typically valid
for the materials and conditions of this study and many others,
as shown in Figure 7.

Figure 5. Shear modulus of SRF(3.8) r = 0.6 at 110, 115, and 117 °C.
The crossover point (vertical solid lines) shifts up and to the left as a
temperature decreases or, equivalently, as the equilibrium conversion
increases.

increases. As we discussed in an earlier24 report, the apparent

activation energy of this process (∼260 kJ/mol according to the
data of Figure 5) is much higher than can be explained simply
by the Diels−Alder kinetics alone.
According to the SR theory (eq 5), a plot of the ratio of the Figure 7. Theoretical conversion versus temperature for SRF(3.5) r =
longest relaxation time to the adduct lifetime as a function of 0.6 compositions in the temperature range of interest. The line is a
conversion should result in a straight line, if the material is linear fit. Although curvature in the points is apparent, the deviations
beyond its Flory gel point conversion for an An + An−1 + ... + A1 from a local linear approximation are negligible.
+ B2 step-growth polymerization.
1
pgel = Equation 2 can also be used to estimate the gel point for
r(fe − 1) (6) thermoreversible materials where the exact value is unknown,
and without having to target the precise conditions of gelation.
Figure 6 demonstrates that our experiments with several By the same token, the true effective functionality of a mixture
compositions are consistent with this expectation. An additional can be calculated via the estimation of the gel point, without
implication of these plots is that the rate of cross-link requiring exact knowledge of the molecular structure. Once eq
dissociation will be about 20× the retro-Diels−Alder rate at a 2 has been used, the estimate of pgel can be converted into an
distance from the gel point ε ≈ 3%, but drops to 10× at ε ≈ 5% effective functionality via the inverse of eq 6:
in these compositions.
Applicability. Although the relationship described by eq 2 1
fe = +1
is simple, a number of material parameters must be known rpgel 2 (7)
before its predictive capacity becomes useful. The activation
energy and (to an extent) the kinetic prefactor for the reversible which, in the case of an An + An−1 + ... + A1 + B2 step-growth
bond are vital information that must be known. Attempts to polymerization, is the effective functionality of the A
estimate the retro-Diels−Alder activation energy based solely components in aggregate. This approach allows structural
on G* and distance from the gel point showed that the estimate information to be determined from empirical dynamic

Figure 6. Time scale ratio τmax/τb versus conversion, showing linear dependence on conversion beyond the gel point for various compositions. Data
for τmax were collected via dynamic mechanical analysis, while maleimide conversion and τb were calculated from the experimental temperature using
the van’t Hoff and Arrhenius relationships with the thermochemical parameters calculated in Figure 3. In fact, at any given temperature or
conditions, τb is simply the reciprocal of the retro-Diels−Alder rate constant kr. Comparisons of fitted intercepts to expected gel points are presented
in Table 1.

7638 dx.doi.org/10.1021/ma301329u | Macromolecules 2012, 45, 7634−7641

Macromolecules Article

mechanical tests and published thermodynamic and kinetic standing the temperature dependence of properties in polymer
parameters as demonstrated in Figure 8. science. Some groups26 have used this empirical process with
success on materials where bond dynamics and conversion are
not strong functions of temperature or when the crossover
frequency is out of the range of measurement. Clearly, if the
reversible bond is thermally responsive within the temperature
range considered, the result of conventional time−temperature
superpositions will be skewed. Another superposition principle
commonly used in the literature is to normalize the frequency
by the cross-link reverse rate constant, as developed by Yount
et al.26a This principle has been shown1,4,27 to be effective in the
treatment of many systems from hydrogen-bonded networks to
boronate−diol systems. However, these systems are typically
highly functionalized or very far from their gel point in the
range of conditions studied. Several of our systems that are
closer to the gel point show good overlap at the crossover
Figure 8. Determination of effective functionality from the intercept frequency; however, superposition is unable to account for the
(implied gel point) of a time scale ratio versus conversion plot of the G* features measured at higher frequencies. Figure 9 shows
SRF(3.5) r = 0.91 composition. By inverting the Flory−Stockmayer
equation (solid line), the intercept of the linear fit (dotted line) of the
data (triangles), which represents the gel point conversion, can be used
to calculate (dashed lines) the overall effective functionality fe of an
unknown An + An−1 + ... + A1 + B2 mixture.

Even given good estimates of p(T) and kr(T), the longest

relaxation time is not a typically accessible parameter in many
situations where temperature-dependent properties are crucial
information. To convert the relationship into a more widely
applicable form, it is possible to use the relationship
η′(ω) = G″(ω)/ω (8)
and the assumption that the material behaves according to a
generalized Maxwell model to rewrite it in terms of the steady
shear viscosity and cross-link density. If the plateau modulus
grows according to an appropriate scaling law G′ ∼ |ε|t, and the
viscosity at the temperature of interest can be measured, then in
the first-order approximation the longest relaxation time can be
written as
Tη0
τmax ∝
|ε|t (9)
where the absolute temperature factor normalizes the temper-
ature dependence of rubbery elasticity. The scaling exponent t
has a value of 3 according to classical theory but may be Figure 9. Results of attempting time−temperature superposition
measured for an individual system. Equation 2 can therefore be SRF(3.5)/DPBM model thermoreversible materials. (top) The raw
rewritten without any mention of characteristic times as data for the material at 113 and 116 °C. (bottom) The data are
normalized to the crossover point. Low-frequency behavior shows
η0 ⎛ E ⎞
p=C k 0 exp⎜ − a ⎟ + pgel good overlap, but the high-frequency behavior (of G″ especially)
|ε|t T ⎝ RT ⎠ (10) overlaps much better without the shift. Curves are representative of
similar comparisons for multiple compositions.
Equation 10 can be rearranged to produce an expression for the
viscosity of the material as a function of temperature, or an even
simpler expression proportional to the reverse rate constant and
the distance from the gel point, as shown in eq 11. Note that how G′ and G″ curves shifted to match at the crossover fail to
this equation does not imply zero viscosity at the gel point, as
overlap at the upper range of the frequency sweeps in our
this development only applies after a minimum distance εmin
beyond the gel point where breakage of network chains is the materials. Indeed, the G″ curves from 2 to 100 rad/s had better
dominant relaxation mechanism.13b overlap before the shift. This result can be interpreted as
evidence that the mechanisms governing τmax and τrelax, namely
|ε||ε|t εt + 1 cutoff cluster backbone breakage and small cluster diffusion,
η0 = ∝
CTk 0 exp( −Ea /RT ) Tk r (11) have distinct temperature dependencies. Thus, it is clear that
Relationship to Time−Temperature Superposition. the use of the superposition concept is inappropriate for
Time−temperature superposition is a staple tool of under- networks with temperature-dependent chemical structure.
7639 dx.doi.org/10.1021/ma301329u | Macromolecules 2012, 45, 7634−7641
Macromolecules

■
Article

CONCLUSION (5) Montarnal, D.; Capelot, M.; Tournilhac, F. o.; Leibler, L. Silica-
Like Malleable Materials from Permanent Organic Networks. Science
Semenov and Rubinstein’s theory of linear dynamics for 2011, 334 (6058), 965−968.
reversible networks provides an appropriate description of the (6) Chen, X.; Dam, M. A.; Ono, K.; Mal, A.; Shen, H.; Nutt, S. R.;
behavior of this model thermoreversible Diels−Alder network Sheran, K.; Wudl, F.; Thermally Re-mendable, A. Cross-Linked
in a variety of situations near the gel point. Because of the Polymeric Material. Science 2002, 295 (5560), 1698−1702.
simplicity of the model system, this confirmation leads us to (7) Zhang, Y.; Broekhuis, A. A.; Picchioni, F. Thermally Self-Healing
expect that the simple scaling relationship used here has wide Polymeric Materials: The Next Step to Recycling Thermoset
applicability to reversibly bonded polymer networks, whether Polymers? Macromolecules 2009, 42 (6), 1906−1912.
covalent or noncovalent. By simple measurement of the (8) Kloxin, C. J.; Scott, T. F.; Adzima, B. J.; Bowman, C. N. Covalent
crossover time at one or two points, parameters can be Adaptable Networks (CANs): A Unique Paradigm in Cross-Linked
obtained that enable calculation of the important material Polymers. Macromolecules 2010, 43 (6), 2643−2653.
(9) Bowman, C. N.; Kloxin, C. J. Covalent Adaptable Networks:
properties over a wide range of temperature, conversion, and Reversible Bond Structures Incorporated in Polymer Networks. Angew.
bond breakage rate values. Furthermore, the ratio of time scales Chem., Int. Ed. 2012, 51 (18), 4272−4274.
can be rewritten in terms of more accessible measurables, (10) Stadler, R. Transient networks by hydrogen bond interactions in
providing a useful qualitative description of the behavior of polybutadiene-melts. In Permanent and Transient Networks; Springer:
gelling thermoreversible materials for applications lacking Berlin, 1987; Vol. 75, pp 140−145.
sophisticated rheological equipment. (11) te Nijenhuis, K. Calculation of network parameters in
Although other groups have had success with less thermoreversible gels. Polym. Gels Networks 1996, 4 (5−6), 415−433.
fundamental approaches to the description of thermoreversible (12) Reutenauer, P.; Buhler, E.; Boul, P. J.; Candau, S. J.; Lehn, J. M.
network viscoelasticity, it appears that these approaches are Room temperature dynamic polymers based on Diels-Alder chemistry.
functionally subsets of the predictions provided by eq 2. We Chemistry 2009, 15 (8), 1893−900.
(13) (a) Semenov, A. N.; Rubinstein, M. Thermoreversible Gelation
believe that this further demonstrates its value as a more flexible
in Solutions of Associative Polymers. 1. Statics. Macromolecules 1998,
tool for predicting the properties of transient covalent (or 31 (4), 1373−1385. (b) Rubinstein, M.; Semenov, A. N.
physical) networks in their most characteristic regime. It Thermoreversible Gelation in Solutions of Associating Polymers. 2.
represents a valuable tool in the structure−property toolkit for Linear Dynamics. Macromolecules 1998, 31 (4), 1386−1397.
this class of materials, filling in the gap between the limiting (14) Stauffer, D. Scaling theory of percolation clusters. Phys. Rep.
cases where the classic toolscross-link density calculations, 1979, 54 (1), 1−74.
small molecule viscosity measurements, and scaling expo- (15) Herrmann, H. J. Geometrical cluster growth models and kinetic
nentsapply. gelation. Phys. Rep. 1986, 136 (3), 153−224.

■
(16) Flory, P. J. Molecular Size Distribution in Three Dimensional
ASSOCIATED CONTENT Polymers. I. Gelation1. J. Am. Chem. Soc. 1941, 63 (11), 3083−3090.
(17) Ferry, J. D. Viscoelastic Properties of Polymers, 3rd ed.; Wiley:
*
S Supporting Information New York, 1980; p 641.
Methods of calculation of cross-link density for finding Gp,theor (18) Miller, D. R.; Macosko, C. W.; New, A. Derivation of Post Gel
and of effective functionality fe. This material is available free of Properties of Network Polymers. Macromolecules 1976, 9 (2), 206−
charge via the Internet at http://pubs.acs.org. 211.

■
(19) Dotson, N. A.; Galvan, R.; Laurence, R. L.; Tirrell, M.
Polymerization Process Modeling; Wiley-VCH: New York, 1995.
AUTHOR INFORMATION (20) Diels, O.; Alder, K. Synthesen in der hydroaromatischen Reihe.
Corresponding Author Justus Liebigs Ann. Chem. 1928, 460 (1), 98−122.
*E-mail: [email protected]. (21) Craven, J. M. Cross-Linked Thermally Reversible Polymers
Produced from Condensation Polymers with Pendant Furan Groups
Notes
Cross-Linked with Maleimides. 3435003, 1969.
The authors declare no competing financial interest.

■
(22) (a) Goiti, E.; Heatley, F.; Huglin, M. B.; Rego, J. M. Kinetic
aspects of the Diels-Alder reaction between poly(styrene-co-furfuryl
ACKNOWLEDGMENTS methacrylate) and bismaleimide. Eur. Polym. J. 2004, 40 (7), 1451−
The authors of this paper thank NSF Grant CBET-0933828 as 1460. (b) Goiti, E.; Huglin, M. B.; Rego, J. M. Some properties of
networks produced by the Diels-Alder reaction between poly(styrene-
their funding source.

■
co-furfuryl methacrylate) and bismaleimide. Eur. Polym. J. 2004, 40
(2), 219−226. (c) Costanzo, P. J.; Beyer, F. L. Thermoresponsive,
REFERENCES Optically Active Films Based On Diels-Alder Chemistry. Chem. Mater.
(1) van Beek, D. J. M.; Spiering, A. J. H.; Peters, G. W. M.; te 2007, 19 (25), 6168−6173. (d) Peterson, A. M.; Jensen, R. E.;
Nijenhuis, K.; Sijbesma, R. P. Unidirectional Dimerization and Palmese, G. R. Reversibly cross-linked polymer gels as healing agents
Stacking of Ureidopyrimidinone End Groups in Polycaprolactone for epoxy-amine thermosets. ACS Appl. Mater. Interfaces 2009, 1 (5),
Supramolecular Polymers. Macromolecules 2007, 40 (23), 8464−8475. 992−5. (e) Inglis, A. J.; Nebhani, L.; Altintas, O.; Schmidt, F. G.;
(2) Burattini, S.; Colquhoun, H. M.; Greenland, B. W.; Hayes, W. A Barner-Kowollik, C. Rapid Bonding/Debonding on Demand:
novel self-healing supramolecular polymer system. Faraday Discuss. Reversibly Cross-Linked Functional Polymers via Diels,àiAlder ́
2009, 143, 251−264. Chemistry. Macromolecules 2010, 43 (13), 5515−5520. (f) Peterson,
(3) Adzima, B. J.; Aguirre, H. A.; Kloxin, C. J.; Scott, T. F.; Bowman, A. M.; Jensen, R. E.; Palmese, G. R. Room-Temperature Healing of a
C. N. Rheological and chemical analysis of reverse gelation in a Thermosetting Polymer Using the Diels-Alder Reaction. ACS Appl.
covalently crosslinked Diels-Alder polymer network. Macromolecules Mater. Interfaces 2010, 2 (4), 1141−1149.
2008, 41 (23), 9112−9117. (23) (a) Jones, J. R.; Liotta, C. L.; Collard, D. M.; Schiraldi, D. A.
(4) Jay, J. I.; Langheinrich, K.; Hanson, M. C.; Mahalingam, A.; Kiser, Cross-Linking and Modification of Poly(ethylene terephthalate-co-2,6-
P. F. Unequal stoichiometry between crosslinking moieties affects the anthracenedicarboxylate) by Diels-Alder Reactions with Maleimides.
properties of transient networks formed by dynamic covalent Macromolecules 1999, 32 (18), 5786−5792. (b) Swanson, J. P.;
crosslinks. Soft Matter 2011, 7 (12), 5826−5835. Rozvadovsky, S.; Seppala, J. E.; Mackay, M. E.; Jensen, R. E.; Costanzo,

7640 dx.doi.org/10.1021/ma301329u | Macromolecules 2012, 45, 7634−7641

Macromolecules Article

P. J. Development of Polymeric Phase Change Materials On the basis

of Diels-Alder Chemistry. Macromolecules 2010, 43 (14), 6135−6141.
(c) Goussé, C.; Gandini, A. Diels−Alder polymerization of difurans
with bismaleimides. Polym. Int. 1999, 48 (8), 723−731. (d) McElha-
non, J. R.; Russick, E. M.; Wheeler, D. R.; Loy, D. A.; Aubert, J. H.
Removable foams based on an epoxy resin incorporating reversible
Diels-Alder adducts. J. Appl. Polym. Sci. 2002, 85 (7), 1496−1502.
(24) Sheridan, R. J.; Adzima, B. J.; Bowman, C. N. Temperature
Dependent Stress Relaxation in a Model Diels-Alder Network. Aust. J.
Chem. 2011, 64 (8), 1094−1099.
(25) Hopewell, J. L.; Hill, D. J. T.; Pomery, P. J. Electron spin
resonance study of the homopolymerization of aromatic bismalei-
mides. Polymer 1998, 39 (23), 5601−5607.
(26) (a) Yount, W. C.; Loveless, D. M.; Craig, S. L. Small-Molecule
Dynamics and Mechanisms Underlying the Macroscopic Mechanical
Properties of Coordinatively Cross-Linked Polymer Networks. J. Am.
Chem. Soc. 2005, 127 (41), 14488−14496. (b) Roberts, M. C.;
Hanson, M. C.; Massey, A. P.; Karren, E. A.; Kiser, P. F. Dynamically
Restructuring Hydrogel Networks Formed with Reversible Covalent
Crosslinks. Adv. Mater. 2007, 19 (18), 2503−2507. (c) Roberts, M. C.;
Mahalingam, A.; Hanson, M. C.; Kiser, P. F. Chemorheology of
Phenylboronate-Salicylhydroxamate Cross-Linked Hydrogel Networks
with a Sulfonated Polymer Backbone. Macromolecules 2008, 41 (22),
8832−8840. (d) Cordier, P.; Tournilhac, F.; Soulie-Ziakovic, C.;
Leibler, L. Self-healing and thermoreversible rubber from supra-
molecular assembly. Nature 2008, 451 (7181), 977−80.
(27) Sijbesma, R. P.; Beijer, F. H.; Brunsveld, L.; Folmer, B. J. B.;
Hirschberg, J. H. K. K.; Lange, R. F. M.; Lowe, J. K. L.; Meijer, E. W.
Reversible Polymers Formed from Self-Complementary Monomers
Using Quadruple Hydrogen Bonding. Science 1997, 278 (5343),
1601−1604.

7641 dx.doi.org/10.1021/ma301329u | Macromolecules 2012, 45, 7634−7641