Article https://doi.org/10.

1038/s41467-023-43117-1

Unraveling the rheology of inverse

vulcanized polymers

Received: 4 July 2023 Derek J. Bischoff 1, Taeheon Lee2, Kyung-Seok Kang 2, Jake Molineux2,
Wallace O’Neil Parker Jr.3, Jeffrey Pyun 2 & Michael E. Mackay 1,4
Accepted: 1 November 2023

Multiple relaxation times are used to capture the numerous stress relaxation
Check for updates modes found in bulk polymer melts. Herein, inverse vulcanization is used to
synthesize high sulfur content (≥50 wt%) polymers that only need a single
1234567890():,;
1234567890():,;

relaxation time to describe their stress relaxation. The S-S bonds in these
organopolysulfides undergo dissociative bond exchange when exposed to
elevated temperatures, making the bond exchange dominate the stress
relaxation. Through the introduction of a dimeric norbornadiene crosslinker
that improves thermomechanical properties, we show that it is possible for the
Maxwell model of viscoelasticity to describe both dissociative covalent
adaptable networks and living polymers, which is one of the few experimental
realizations of a Maxwellian material. Rheological master curves utilizing time-
temperature superposition were constructed using relaxation times as non-
arbitrary horizontal shift factors. Despite advances in inverse vulcanization,
this is the first complete characterization of the rheological properties of this
class of unique polymeric material.

The development of stimuli-responsive macromolecules composed of The Maxwell model of viscoelasticity predicts a single relaxa-
reversible covalent bonds in the polymer backbone has prompted a tion time, τ m , that can be determined from the crossover frequency
paradigm shift in the understanding of rheological and thermo- as τ m = 1=ωm , where ωm is the frequency defined at the crossover of
mechanical properties of these materials. The vast majority of work in the storage modulus (G0 ) and loss modulus (G00 ). The model is often
this field has investigated covalent adaptable networks (CANs), where depicted as a spring and dashpot in series representing an ideal
bond reorganization within a crosslinked polymer network has a pro- solid that follows Hooke’s law of elasticity and an ideal liquid that
found impact on the rheological properties by introducing additional follows Newton’s law of viscosity, respectively (Fig. 1a). Conven-
stress relaxation modes1–16. To describe the frequency dependent, tional polymer melts have multiple relaxation times, or a spectrum,
viscoelastic properties of polymer melts, multiple relaxation times are where no one mode is sufficiently dominant to fully describe its
necessary to capture the complex interactions over a variety of length viscoelastic properties20. Generalized Maxwell models combine
and time scales. Reptation theory espoused by de Gennes17 predicts an several Maxwell elements in parallel to better capture the physics in
infinite number of relaxation times even for a monodisperse polymer real materials, typically, one element is added per decade of fre-
melt18. However, the stress relaxation of certain polymers with rever- quency. The moduli of a single relaxation time Maxwell model are
sible bonds is dominated by the dynamic bond reorganization and may given by Eqs. 1 and 2
follow the simple Maxwell model of viscoelasticity to afford a single,
dominant relaxation time. One of the very few examples of this
behavior is the polymeric allotrope of elemental sulfur that is solely
2 
2 
composed of dynamic S-S bonds19. G0 = G0N ωτ m = 1 + ωτ m ð1Þ

1
Department of Materials Science and Engineering, University of Delaware, Newark, DE 19716, USA. 2Department of Chemistry and Biochemistry & Wyant
College of Optical Sciences, University of Arizona, Tucson, AZ 85721, USA. 3Physical Chemistry Department, Eni S.p.A., San Donato Milanese 20097, Italy.
4
Department of Chemical and Biomolecular Engineering, University of Delaware, Newark, DE 19716, USA. e-mail: [email protected]; [email protected]

Nature Communications | (2023)14:7553 1

Article https://doi.org/10.1038/s41467-023-43117-1

a chosen since the maximum of G00 is easily identifiable in the rheological

data. Upon plotting Eq. 3, a Cole-Cole representation21 is achieved
whereby the normalized imaginary component, G00 , is plotted against
the normalized real part, G0 , of the complex modulus on a linear scale
Viscosity Modulus (Fig. 1c). The Maxwell model predicts a semicircular shape, affording a
= Pa · s = Pa
convenient graphical method to determine if a material follows the
b Maxwell model and has a single, Debye-like22 relaxation mode. Poly-
100 Storage Modulus meric materials that exhibit a single relaxation time and yield a
Loss Modulus relaxation spectrum as shown in Fig. 1b, c are highly unusual in poly-
mer science and are herein referred to as Maxwellian polymers.
Independent of the Maxwell model and the field of CANs, Cates23
G'/G0N or G''/G0N

Eq. 2 defined the term living polymers that describe linear polymer chains
with reversible main chain scission on experimental time scales with
= presumably fast bond exchange to suppress depolymerization
10-1
(Fig. 2a). These rheologically living polymers are definitionally dis-
tinct from polymers possessing active propagating chain ends for-
med via living anionic polymerizations or related controlled
polymerizations24,25. Cates’ model was developed to specifically
Eq. 1 address surfactant organization in microemulsions and micelles as well
-2
as homogenous equilibrium polymers such as liquid sulfur that has
10
polymerized above its λ-transition23,26. Cates’ theory has been suc-
10-2 10-1 100 101 102 cessfully used to experimentally describe the viscoelastic behavior of
ωWm
worm-like micelles21,27–34 that break and diffuse during deformation,
c 1.0 however, the realization of bulk living polymer melts remains elusive,
with the exception of pure sulfur22. Application of Cates’ model yields
Normalized Loss Modulus G''/G''max

the curious prediction that the geometric mean of the breakage and
0.8
reptation times, under certain conditions, dominate the rheology of
living polymers and thus exhibit a single relaxation time. As such, the
0.6 Eq. 3 experimental rheological properties would follow the Maxwell model
when the bond scission/recombination kinetics are fast compared to
the viscoelastic relaxation time. Notably certain polysulfide rubbers
0.4 show Maxwellian stress relaxation35,36. It should be noted that even a
living polymer system has other relaxation times associated with
0.2
submolecular modes that can be absorbed into Cates’ model37,38,
however, the dominant mode is predicted to be associated with the
living relaxation time. Hence, living polymers can be described as
0.0 Maxwellian polymers that possess dynamic reversible bonds within a
0.0 0.5 1.0 1.5 2.0 linear, or non-crosslinked polymer architecture. Therefore, living
Normalized Storage Modulus G'/G'' max polymers represent a new class of condensed matter with dynamic
bonds that can be simply thought of as non-crosslinked analogues of
Fig. 1 | Maxwellian material. a a spring and dashpot combine in series to form a CAN materials (Fig. 2b).
single Maxwell element to model a viscoelastic material’s response to an applied
With both the Maxwell and living polymer models in mind, we set
stress (b) rheological master curve for a Maxwellian material highlighting the
forth to engineer a bulk, robust polymeric material that could be
relaxation time as determined from the crossover in moduli (c) Cole-Cole repre-
described by a single relaxation time, in this case with inexpensive,
sentation for a Maxwellian material giving a semicircular shape.
dynamic covalent S-S bonds made from elemental sulfur (S8). The
inverse vulcanization process, developed by Pyun et al.39 is ideally

2  suited for this purpose where S8 is used as the monomeric reaction
G00 = G0N ðωτ m Þ= 1 + ωτ m ð2Þ medium for bulk homolytic ring-opening polymerizations with organic
unsaturated monomers to afford inverse vulcanized polymers or
where G0N is the rubbery plateau modulus and ω is frequency (Fig. 1b). In organopolysulfides40–44. Inverse vulcanized polysulfides inherently
the terminal regime (at longer timescales or lower frequencies), G0 and contain dynamic covalent S-S bonds in the copolymer backbones
G00 are predicted to have a slope of two and one, respectively, on a log-log offering a route to dissociative bond exchange without the need to
plot of moduli versus frequency. Notably, this master curve shows a explicitly engineer reversibly covalent moieties. However, the intract-
unique maximum in G00 followed by a linear decay at higher frequencies, able nature of these organopolysulfides stemming from their poor
which is a distinctive feature of the Maxwell model, as most experimental solubility in common solvents and dynamic S-S bond exchange has
polymer melts exhibit an inflection in G00 causing its value to rise at complicated detailed experimental and mechanistic characterization
higher frequency. The Maxwell relationship can be written to eliminate methods to understand their true microstructure. These inherent
frequency dependence as shown in Eq. 3. This equation is dimensionless challenges have prevented unambiguous classification of the inverse
with G00max = G0N =2 since the maximum occurs at a frequency of 1=τ m . vulcanized polysulfides as thermoplastics or thermosets (or neither).
To date, the only rheological study of inverse vulcanized polysulfides
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
G00 = G0 × ð2  G0 Þ ð3Þ was conducted by Mackay and Pyun et al.45 in which polymers con-
taining a higher sulfur rank were found to be more dynamic when
exposed to heat and shear deformation. Ultimately, no frequency
This equation was produced by normalizing the moduli with G00max sweeps or master curves were included leaving an experimental need
and accounts for the * notation. The normalization technique was to probe the terminal region and definitive flow behavior.

Nature Communications | (2023)14:7553 2

Article https://doi.org/10.1038/s41467-023-43117-1

a incorporation of both cyclic aliphatic and aromatic moieties (Supple-

mentary Figs. 3–4). NMR spectroscopic, mechanistic, and computa-
Linear Polymers Dynamic S-S Bonds Exchanged Polymers tional evidence for this inverse vulcanization mechanism has been
discussed elsewhere47, but indicates that copolymerization of S8 with
either Sty or DIB affords linear type units, while NBD2 increases
Δ Δ
crosslinking density.
Figure 4 shows the first bulk polymeric organopolysulfide mate-
rial that can be classified as both a living and a Maxwellian polymer:
Living Polymer: Thermoplastic with reversible bonds poly(S70-r-DIB30). Frequency sweep data collected using SAOS rheol-
ogy provides the frequency dependent G0 and G00 values at various
b
temperatures. Replotting the data in a Cole-Cole representation,
shown in Fig. 4a, exhibits the characteristic semicircular shape as dis-
Network Polymer Dynamic S-S Bonds Altered Topology
cussed in Fig. 1c for a Maxwellian material. Following the same pro-
cedure for each temperature that experimentally shows a maximum in
Δ Δ G00 enables the clear realization that the material has a single dominant
relaxation mode. At higher values of normalized storage modulus,
small deviations from the semicircle are attributed to Rouse-like
motions and other (sub)molecular processes that contribute to
Dissociative Covalent Adaptable Network: Thermoset with reversible bonds relaxation at shorter time scales23,38. This behavior is also observed in
elemental sulfur as described by Stashick and Marriott22. By dividing
Fig. 2 | Comparison of polymer types. a living polymers are linear polymers with Eq. 1 by Eq. 2, the linear relationship in Eq. 4 is obtained that can be
exchangeable bonds that undergo reversible chain scission reactions and recom- plotted to obtain τ m as the slope.
bine with other dynamic linear chains (b) dissociative covalent adaptable networks
(CANs) are covalently crosslinked polymer networks with exchangeable bonds that
G0 =G00 = τ m × ω ð4Þ
when exposed to a stimulus (i.e., heat) alter their topology and enable stress
relaxation/flow.
Figure 4b shows the linear fit for each temperature at which a
maximum in G00 is observed. To keep in the spirit of the Maxwell model,
Herein, we report the first detailed rheological analysis of inverse a linear line of best fit with its intercept forced through zero is used.
vulcanized polysulfides of varying sulfur composition, organic Using a forced zero intercept (0.149 s) showed little change in the
comonomer ratio, and polymer architecture. Using small amplitude relaxation time compared to not forcing the intercept (0.145 s) for the
oscillatory shear (SAOS) rheology, we demonstrate that through data at 160 °C. (Supplementary Figs. 30 and 31 for poly(S50-r-DIB50)).
control of co- and terpolymer composition, inverse vulcanized poly- To capture G0 and G00 data over a wider frequency range than is
sulfides can be engineered to behave as classical thermoplastics or experimentally possible, it is customary to employ time-temperature
dissociative CANs by altering crosslinking density. This study also superposition (TTS). The Maxwell relaxation times were used as hor-
indicates that inverse vulcanized polysulfides of certain compositions izontal shift factors to generate the master curve (Fig. 4c) and were not
are also Maxwellian and living polymers, which is one of first experi- arbitrarily determined. The vertical shift factors are then determined
mental realizations of these rheological models. The key molecular by empirically shifting the data to obtain good overlay amongst the
causality of these bulk rheological properties arises from the sig- different temperatures used to construct the dimensionless master
nificant number of dynamic S-S bonds present in these materials curve. TTS often fails to describe the rheology of dissociative CANs
which is an important finding in the fields of polymer chemistry and made entirely from organic moieties. The resulting master curve
materials science. shows good agreement after shifting except after passing the max-
imum in G” with dimensionless frequency. It was hypothesized that
Results and Discussion this could be caused by solubilized sulfur in the polymer leading to
A series of co- and terpolymers for these rheological studies were bloom, however, deviations are still observed in poly(S50-r-DIB50) even
prepared via inverse vulcanization of S8 (50 wt% or 70 wt%) with when new material is used for each oscillatory frequency sweep that
combinations of styrene (Sty), 1,3-diisopropenylbenzene (DIB), or drastically reduced the exposure time of the material to elevated
dimeric norbornadiene (NBD2) to determine design criteria for Max- temperatures (Supplementary Fig. 32). The empirical vertical shift
wellian behavior. The terpolymer poly(S50-r-DIB25-r-NBD225) contains factors are significant in the material, which span from 0.36 to 3.2 over
50 wt% sulfur, 25 wt% DIB, and 25 wt% NBD2, for example. Poly(S-r-DIB) only a 30 °C temperature range as shown in Fig. 4d. To our knowledge,
copolymers prepared by inverse vulcanization were previously this is a rare example of a soft material requiring these large shift
thought to be branched or network polymers as determined by curing factors, the validity of which is demonstrated by the good fit of the
conditions and therefore were assumed to be dissociative CANs46. dimensionless master curve.
However, a recent mechanistic report by Pyun et al.47. has proven that The dominant stress relaxation mechanism is through rapid S-S
poly(S-r-DIB) is primarily composed of linear bis-cumyl units, which bond exchange compared to reptation in agreement with Cates26 as
disqualifies these copolymers as CANs. Alternatively, poly(S-r-NBD2) well as Stashick and Marriott22. The single relaxation time predicted by
was more recently prepared where NBD2 was presumed to serve as a the adherence to the Maxwell model indicates that the chemorheo-
crosslinking agent to improve thermomechanical properties48. The logical relaxation mechanism (bond scission/recombination) is inde-
glass transition temperature (Tg) of poly(S-r-NBD2) is ~100 °C, which is pendent of the polymer chain length and therefore of the diffusional
vastly improved over that of ~30 °C for poly(S-r-DIB), preventing reflow reptation time. As the rank of sulfur (i.e., the number of consecutive S-S
issues when molded or processed in the solid state48. The reaction bonds in a chain) increases, the S-S bonds become more dynamic in
scheme for all synthesized materials is shown in Fig. 3 along with nature49, where the S-S bond farthest from the C-S bonds connecting
revised structures for DIB units incorporated into poly(S-r-DIB-r- to the organic comonomer possesses the lowest bond dissociation
NBD2), where the idealized microstructure of DIB is replaced with a energy. The S-S exchange reaction is dissociative, meaning that a long
bis-cumyl type unit47. Solid-state 13C nuclear magnetic resonance linear polymer chain can transiently break into two shorter, distinct
(NMR) spectroscopy was conducted on the terpolymers to confirm the polymer chains without needing to interact with another chemical

Nature Communications | (2023)14:7553 3

Article https://doi.org/10.1038/s41467-023-43117-1

DIB

Tg = 20-35 °C
Tdeg = 213-215 °C
' Living Polymer

Sty Poly(S-r-DIB) Copolymers

Tg = -8 °C
Tdeg = 161 °C
' Thermoplastic

NBD2 Poly(S-r-Sty) Copolymer

Elemental Sulfur Tg = 105-112 °C

(S8) Tdeg = 294-320 °C
' Dissociative CAN

Poly(S-r-NBD2) Copolymers
DIB
+
NBD2 Tg = 58-105 °C
Tdeg = 217-292 °C
' Dissociative CAN

Poly(S-r-DIB-r-NBD2) Terpolymers

Fig. 3 | Synthesis of inverse vulcanized polysulfides. Reaction scheme to syn- thermoplastics, Maxwellian/living polymers, or dissociative covalent adaptable
thesize the inverse vulcanized copolymers and terpolymers using S8, Sty, DIB, and networks.
NBD2. Through the control of composition, the polysulfides can behave as

a b
1.2 3
130 °C Linear Fit
y = 4.987x
Normalized Loss Modulus G''/G''max

1.0 140 °C Linear Fit

150 °C Linear Fit
160 °C Linear Fit
0.8 2
G'/G''

0.6 y = 0.353x

y = 1.109x y = 0.112x
0.4 1
130 °C
140 °C
0.2 150 °C
160 °C
Maxwell Model
0.0 0
0.0 0.5 1.0 1.5 2.0 2.5 0 2 4 6 8 10
Normalized Storage Modulus G'/G''max Angular Frequency (rad/s)
c d
10
0
10 T ref=140°C
Shift Factor or Relaxation Time (s)

-1
10
G'/G0N or G''/G0N

-2
1
10

G' 130 °C G'' 130 °C

10-3 G' 140 °C G'' 140 °C
G' 150 °C G'' 150 °C
G' 160 °C G'' 160 °C Maxwell Relaxation Time
G' Model G'' Model Empirical Vertical Shift Factor
-4
10 0.1
-2 -1 0 1 2 3
10 10 10 10 10 10 130 140 150 160
Wm Temperature (°C)

Fig. 4 | Adherence of the poly(S70-r-DIB30) copolymer to the Maxwell model of modulus as predicted by the Maxwell model. c Dimensionless master curve utiliz-
viscoelasticity. a Cole-Cole plot showing the semicircular shape indicating that the ing Maxwell relaxation times for horizontal shifting and empirical shifting along the
material is Maxwellian and dominated by a single relaxation mode. b Determination vertical axis. Solid black and blue lines show the Maxwell model. d Resulting shift
of the Maxwell relaxation time at each temperature displaying a maximum in loss factors showing the importance of the vertical shift factor.

moiety belonging to another chain (Fig. 2a). As such, when the modulus with temperature, discussed next. The combination of the
dynamic covalent chemistry is activated by thermal stimulus, the single relaxation time, the S-S bond exchange, and the linear polymer
transient average molecular weight of the polymer decreases before architecture indicate that poly(S70-r-DIB30) is indeed a living polymer.
subsequent thiyl radical coupling upon cooling. As temperature Likewise, poly(S50-r-DIB50) exhibits similar rheological properties as
increases, the equilibrium constant begins to shift more and more the poly(S70-r-DIB30) composition and is therefore both Maxwellian
towards the dissociated, oligomeric form which is not captured by and living despite the difference in sulfur rank (Supplemen-
Cates’ theory, and it does not account for the rapid decrease in plateau tary Fig. 30).

Nature Communications | (2023)14:7553 4

Article https://doi.org/10.1038/s41467-023-43117-1

Figure 5 was constructed to show the original, unshifted G0 data 110 °C45,46,51,52. S-S bond scission in poly(S-r-Sty) is exploited in dynamic
for poly(S70-r-DIB30) as a representative example of these organopo- covalent polymerizations at temperatures between 110-130 °C52.
lysulfides which prominently illustrates the variance of rubbery pla- Moreover, the mechanism for stress relaxation is likely through dif-
teau modulus as a function of temperature providing insight into the fusion/reptation and not S-S bond exchange. The horizontal and ver-
large vertical shift factors observed experimentally (Supplementary tical shift factors (Fig. 6b) were determined empirically and the
Fig. 43 for poly(S50-r-DIB50) which exhibits identical behavior). Not horizontal shift factors follow the standard Williams–Landel–Ferry
seen in this plot is the increase in moduli associated with shorter length (WLF) model53. The vertical shift factors do not vary significantly over a
scales and glassy dynamics at higher frequency that superimpose in 30 °C temperature change, in contrast to the living polymers discussed
the overall master curve (Supplementary Fig. 18). Since Cates’ model previously since the dissociative bond exchange is not active, making
did not account for the reversible equilibrium between S-S bonds and the average molecular weight of the polymer a constant. The non-
thiyl radicals, the strong temperature dependence of the rubbery Maxwellian behavior is further emphasized in Fig. 6c where the data
plateau modulus was not captured, explaining why the vertical shift shown using a Cole-Cole representation does not form a semicircular
factor is so large relative to traditional polymer melts where the shift shape. Therefore, the lack of thermally activated S-S bond exchange
factor is usually within the range of 0.9 to 1.1 (Supplementary Fig. 27 combined with the fact that poly(S-r-Sty) is linear makes the material a
for traditional polystyrene). The large vertical shift factors, essentially traditional thermoplastic under ambient conditions. To further high-
the ratio of the rubbery plateau modulus at a given temperature light the tunable rheological properties between poly(S-r-Sty) and
compared to that at the reference temperature, are shown in Fig. 4d. poly(S-r-DIB), a series of linear poly(S-r-DIB-r-Sty) terpolymers was
Figure 5 also shows the terminal behavior at different temperatures. At synthesized via inverse vulcanization. The inclusion of more rigid DIB
higher temperatures, terminal behavior is achieved on a shorter units into the poly(S-r-Sty) backbone predictably raised the Tg of the
timescale which is a feature of CANs as the kinetics of the bond terpolymers in turn suppressing terminal flow before the onset of S-S
exchange reaction increases with temperature. The trend of terminal bond exchange. Hence, we demonstrate the recovery of Maxwellian
behavior occurring at shorter timescales also occurs in traditional behavior at the appropriate terpolymer composition of poly(S50-r-
thermoplastic melts due to the increased reptation rate of polymer DIB40-r-Sty10) (Supplementary Fig. 36) and Tg (17 °C; Supplementary
chains, but the effect is enhanced by the increasing rate of bond Fig. 11). Furthermore, the synthesis of terpolymers with higher Sty
exchange reactions. compositions (poly(S50-r-DIB25-r-Sty25) and poly(S50-r-DIB10-r-Sty40))
The importance of the thermal stimulus in activating the S-S bond lowered the Tg of these materials (Tg = 0 °C and –18 °C, respectively) so
exchange is demonstrated using the soluble, linear thermoplastic that terminal flow was observed before S-S bond exchange was acti-
poly(S50-r-Sty50), which possesses a subambient Tg of −8 °C (Fig. 3)50. vated and Maxwellian relaxation was no longer observed (Supple-
The material was studied between −20 °C and 10 °C utilizing liquid mentary Figs. 22–25) which is also true in poly(S50-r-Sty50).
nitrogen to realize the subambient temperatures. The master curve Having established that Maxwellian behavior is obtainable in lin-
constructed using standard TTS where horizontal shift factors are ear living polymers of poly(S-r-DIB) due to the thermal activation of the
arbitrarily determined, at a reference temperature of −10 °C, in Fig. 6a S-S bond exchange, rheological characterization of the crosslinked
shows a crossover between G0 and G00 giving a relaxation time of organopolysulfide poly(S-r-NBD2) was conducted. In contrast to styr-
1
τ = ½0:872rad=s = 1:15 sec, a terminal viscosity of ~1.54×108 Pa-s, and a ene or DIB that possess a monomer functionality of two or less when
rubbery plateau modulus of 171 MPa. The rheological behavior of used for inverse vulcanization, we presume that the two olefinic units
poly(S50-r-Sty50) is not Maxwellian due to the low Tg of this sulfur of NBD2 can form four C-S bonds thereby affording NBD2 an effective
copolymer, enabling terminal flow before significant S-S bond functionality of four, which is qualitatively supported by solid-state
exchange occurs. The S-S bonds have not been observed to be NMR spectroscopy of poly(S-r-NBD2) (Supplementary Figs. 3 and 4).
dynamic below ~90 °C in poly(S-r-DIB)46, but have been reported at Compared to poly(S70-r-DIB30) or poly(S50-r-DIB50), poly(S50-r-NBD250)
has a lower sulfur rank due to the higher monomer functionality
resulting in stronger S-S bonds that are less likely to dissociate. Higher
106 bond dissociation energies are observed for shorter S-S chains, greatly
reducing the number of S-S bond exchange events in the crosslinked
5 organopolysulfide54. The master curve shown in Fig. 7a, constructed
10
using standard TTS, does not show Maxwellian behavior as it does not
Storage Modulus (Pa)

share the master curve shape shown in Fig. 1b. The Cole-Cole plot of
104 the master curve (Fig. 7c) shows an incomplete semicircle and there-
fore is not a Maxwellian polymer, although very large vertical shift
factors are required to shift the poly(S50-r-NBD250) data to form the
10
3
master curve (Fig. 7b).
120 °C Interestingly, the master curve for the crosslinked poly(S50-r-
130 °C NBD250) does have a terminal region, indicating that complete stress
2
10 140 °C relaxation is possible over a sufficient time scale. The terminal
150 °C relaxation behavior observed here indicates that this copolymer is a
160 °C dissociative covalent adaptable network whereby the terminal
101
relaxation is achieved via reversible S-S bond scission. A typical
10-1 100 101 102 crosslinked network where the crosslinking density is constant
Angular Frequency (rad/s) regardless of temperature will show an equilibrium modulus at low
frequency that grows in value with the degree of crosslinking20. Only
Fig. 5 | Terminal region of the Maxwellian copolymer poly(S70-r-DIB30). The 1
drop off in storage modulus, here due to the organopolysulfide exchange reaction,
a single crossover is observed giving τ = ½1:339rad=s = 0:747 sec
is a feature of polymers containing dynamic covalent bonds in general that show a (Tref = 230 °C), along with a terminal viscosity of ~1.14 × 108 Pa-s. The
more rapid stress relaxation as the rate of reaction increases with temperature, thus rubbery plateau modulus for poly(S50-r-NBD250) is 99.3 MPa com-
shortening the rubbery plateau region. This unshifted data also emphasizes the pared to only 4.57 MPa for poly(S70-r-DIB30). The results for poly(S70-
importance of the vertical shift factor as its value increases as the equilibrium r-NBD230) are shown in Supplementary Fig. 35 which reveal Max-
constant shifts towards the dissociated thiyl radicals. wellian behavior due to the higher sulfur content that imparts more

Nature Communications | (2023)14:7553 5

Article https://doi.org/10.1038/s41467-023-43117-1

a 103 109 b 103

Tref = -10 °C
102 102

Complex Viscosity (Pa-s)

108
1
10 101

Moduli (MPa)

Shift Factor
100 107 100

10-1 10-1

Storage Modulus 106

10-2 10-2 Horizontal
Loss Modulus
Complex Viscosity Vertical
10-3 105 10-3
10-4 10-3 10-2 10-1 100 101 102 103 104 -20 -15 -10 -5 0 5 10
Angular Frequency (rad/s) Temperature (°C)

c 1.2

Normalized Loss Modulus G''/G''max

0.9

0.6

0.3

poly(S50-r-Sty50)
Maxwell Model
0.0
0 1 2 3 4 5
Normalized Storage Modulus G'/G''max

Fig. 6 | Copolymer poly(S50-r-Sty50) rheological results. a master curve for linear poly(S50-r-Sty50) at Tref = −10 °C (b) the horizontal and vertical shift factors were
determined empirically using standard rheological techniques. c Cole-Cole representation showing that the material is not a Maxwellian polymer.

a 103 10
9 b 10 2

Tref = 230 °C
1
10
Complex Viscosity (Pa-s)

102 10
8
Moduli (MPa)

100
Shift Factor

101 10
7

10-1

100 Storage Modulus 106

Loss Modulus 10-2
Complex Viscosity Horizontal
10-1 10
5
-3
Vertical
10-2 10-1 100 101 102 103 10
210 220 230 240 250 260
Angular Frequency (rad/s)
Temperature (°C)
c
1.0
Normalized Loss Modulus G''/G''max

0.5

poly(S50-r-NBD250)
Maxwell Model
0.0
0.0 0.5 1.0 1.5 2.0 2.5
Normalized Storage Modulus G'/G''max

Fig. 7 | Crosslinked copolymer poly(S50-r-NBD250) rheological results. a master curve of at Tref = 230 °C (b) shift factors used to make the master curve following
standard TTS (c) Cole-Cole plot of the shifted data revealing an incomplete semicircular shape and therefore non-Maxwellian behavior.

Nature Communications | (2023)14:7553 6

Article https://doi.org/10.1038/s41467-023-43117-1

dynamic S-S bonds that outweigh the restriction from the NBD2 shifted using the same procedure outlined above and behaved as a
crosslinks. dissociative CAN material as discussed for poly(S50-r-NBD250). The
Thus far, we have shown that the linear copolymers poly(S70-r- master curves of all materials tested, generated using standard,
DIB30) and poly(S50-r-DIB50) are living, Maxwellian polymers and that empirical TTS, are shown in Supplementary Figs. 13–22 and 24. The
the crosslinked network poly(S50-r-NBD250) is not Maxwellian, but presence of a terminal region in the master curves of all organopoly-
rheologically behaves as a dissociative CAN. Terpolymers of poly(S-r- sulfides tested indicates the materials undergo complete relaxation
DIB-r-NBD2) with varying sulfur (50 wt%, 70 wt%), DIB (10–40 wt%), and imply that polymer processing techniques like extrusion are
and NBD2 (10–40 wt%) composition were prepared to ascertain the amenable. There is expected to be a maximum amount of NBD2 that
rheological effects of NBD2 comonomer units and prepare a thermo- can be added before a departure from Maxwellian behavior is
mechanically enhanced, melt-processable material. While the poly(S70- observed, likely due to the shortening of linear S-S chains and asso-
r-DIB30) copolymer possesses favorable optical properties, the low Tg ciated higher bond dissociation energies. The limit appears to be
of this material precludes its use for plastic optics55. Fig. 8 shows the between 25 and 40 wt% NBD2 since poly(S50-r-DIB25-r-NBD225) is
rheological results of the terpolymer poly(S70-r-DIB15-r-NBD215). Unlike Maxwellian, but poly(S50-r-DIB10-r-NBD240) is not. We have alluded to
the highly crosslinked poly(S50-r-NBD250) just discussed, the terpoly- the point that the thermomechanical properties of the terpolymers are
mer shows the recovery of Maxwellian stress relaxation behavior. The tunable, and those trends will now be discussed.
adherence to the Maxwell model was confirmed by the semicircular fit Aside from the tuning of the Maxwellian behavior, the thermo-
of Cole-Cole plot in Fig. 8a. Determination of the Maxwell relaxation mechanical properties of these terpolymers were likewise controllable
time at each temperature (Fig. 8b) enabled the construction of the by adjusting composition. The reflow issues stemming from the rela-
master curve in Fig. 8c which afforded good shifting of the data tively poor thermomechanical properties of living poly(S-r-DIB) were
obtained between 150-180 °C using a reference temperature of 160 °C. improved by using NBD2, however, the CAN copolymers poly(S-r-
Again, the Maxwell relaxation times are used to shift horizontally. NBD2) are phenomenologically thermosets48. Because of this, the
Large vertical shift factors were also required for this terpolymer as addition of NBD2 comonomer to poly(S-r-DIB) was employed to raise
observed for DIB- and NBD2-containing inverse vulcanized copoly- the Tg and degradation temperature (Tdeg) of these materials while
mers (Fig. 8d). The introduction of the DIB units into the crosslinked retaining melt processability. Thermogravimetric analysis (TGA) for
polymer network increases the length of the linear segments between sulfur co- and terpolymers was conducted to determine the Tdeg of
crosslinks, along with the recovery of Maxwellian rheological behavior. each material, as determined by a 5% drop in mass. These results also
Due to the presence of NBD2 crosslinker in the terpolymer, the provide the upper limit for the differential scanning calorimetry (DSC)
material is crosslinked and is therefore a Maxwellian CAN. heating ramps and rheological testing (Supplementary Figs. 5–7). The
The other terpolymer CAN compositions (both poly(S50-r-DIB40-r- soluble oligomeric poly(S50-r-Sty50) exhibited the lowest Tdeg (161 °C)
NBD210) and poly(S50-r-DIB25-r-NBD225)) that show Maxwellian beha- and the least amount of char yield, only 0.03%. For polymers com-
vior are shown in Supplementary Figs. 33 and 34. These compositions prised of 50 wt% and 70 wt% sulfur, Tdeg values were observed to
behave similarly in that they are Maxwellian, shift to form a dimen- increase with higher NBD2 comonomer content due to the higher
sionless master curve using relaxation times as horizontal shift factors, crosslinking density and reduced sulfur rank in the polymer network.
and they have large vertical shift factors. The most highly crosslinked All inverse vulcanized organopolysulfides containing DIB and/or
terpolymer composition tested, poly(S50-r-DIB10-r-NBD240), does not NBD2 show a higher Tdeg than elemental sulfur (194 °C, Supplementary
reveal Maxwellian behavior (Supplementary Fig. 16) and cannot be Fig. 8), along with an increase in the char yield, which is also indicative

a 1.2 b 1.8
150 °C Linear Fit
Normalized Loss Modulus G''/G''max

1.0 160 °C Linear Fit

y = 1.477x 170 °C Linear Fit
0.8 1.2 y = 0.479x 180 °C Linear Fit

y = 0.151x
G'/G''

0.6
y = 0.058x

0.4 150 °C 0.6

160 °C
0.2 170 °C
180 °C
Maxwell Model
0.0 0.0
0.0 0.5 1.0 1.5 2.0 2.5 0 5 10 15 20
Normalized Storage Modulus G'/G''max Angular Frequency (rad/s)
1
c d 10
100 Tref = 160 °C
Shift Factor or Relaxation Time (s)

10-1
100
G'/G0N or G''/G0N

10-2

G' 150 °C G'' 150 °C 10-1

10-3 G' 160 °C G'' 160 °C
G' 170 °C G'' 170 °C
G' 180 °C G'' 180 °C Maxwell Relaxation Time
10-4 G' Model G'' Model Empirical Vertical Shift Factor
10-2
10-2 10-1 100 101 102 103 150 160 170 180
Wm Temperature (°C)

Fig. 8 | Adherence of the terpolymer poly(S70-r-DIB15-r-NBD215) to the Maxwell Maxwell model. c Dimensionless master curve utilizing Maxwell relaxation times
model of viscoelasticity. a Cole-Cole plot showing the semicircular shape indi- for horizontal shifting and empirical shifting along the vertical axis; solid black and
cating that the material is Maxwellian. b Determination of Maxwell relaxation time blue lines show the Maxwell model. d Resulting shift factors showing the impor-
at each temperature displaying a maximum in loss modulus as predicted by the tance of the vertical shift factor.

Nature Communications | (2023)14:7553 7

Article https://doi.org/10.1038/s41467-023-43117-1

a 350 350 b 105 105

70wt% S 50wt% S 70wt% S 50wt% S
300 Tg 300
104 104

Temperature (°C)
250 Tdeg 250

Mx (g/mol)
3 3
200 200 10 10

150 150 102 102

100 100
1 1
10 10
50 50
0 0
0 0 10 10
A B C D E F G H A B C D E F G H

Increasing NBD2 Increasing NBD2 Increasing NBD2 Increasing NBD2

Fig. 9 | Trends in thermal properties and molecular weight between topolo- Figs. 13–21) at their respective reference temperatures when G00 goes through a
gical constraints. a Degradation and glass transition temperatures for each minimum. A = poly(S70-r-DIB30); B = poly(S70-r-DIB15-r-NBD215); C= poly(S70-r-
polymer. A 5% mass loss and second heating scans were used to determine the NBD230); D = poly(S50-r-DIB50); E = poly(S50-r-DIB40-r-NBD210); F = poly(S50-r-
degradation and glass transition temperatures, respectively. b Values of G0N used DIB25-r-NBD225); G = poly(S50-r-DIB10-r-NBD240); H = poly(S50-r-NBD250).
in the calculations are obtained from the master curves (Supplementary

of a crosslinked system56. Poly(S50-r-DIB10-r-NBD240) and poly(S50-r- The resulting values of M x are displayed in Fig. 9b assuming that
NBD250) had similar char yields from TGA measurements most likely the density of each polymer is the same at 1.3 g/cc, as determined by
due to the similar NBD2 content in these materials. These structure- compression molding a cylinder of poly(S50-r-DIB50) material and
thermal property trends were further corroborated with the calcula- weighing it. Within the 50 wt% sulfur compositions, the value of M x
tion of molecular weight between topological constraints discussed decreases accordingly with an increasing amount of NBD2 crosslinker
later for these two systems. with the exception of poly(S50-r-DIB10-r-NBD240). Similarly, within the
DSC was utilized to determine the Tg of sulfur co- and terpoly- 70 wt% sulfur compositions, M x decreases with an increasing amount
mers, where materials possessing 50 wt% sulfur (Supplementary of NBD2. An increase in crosslinking density is synonymous with
Fig. 9) show a trend of increasing Tg with higher NBD2 comonomer decreasing M x between topological constraints. Equation 5 gives very
compositions. A similar trend was observed with the 70 wt% series low values for M x in the case of poly(S50-r-DIB10-r-NBD240) and
where increasing Tg correlates with higher NBD2 content (Supple- poly(S50-r-NBD250) that are on the order of a few atoms between
mentary Fig. 10). This general trend is consistent with other cross- crosslinks. Physically these results are nonsensical and are most likely
linkers used in polysulfides39,57,58 and more broadly with increasing due to excluding the effects of dynamic bonds in rubber network
crosslinking density in traditional thermosets59. Interestingly, the Tg theory. The incorporation of such effects and insights from a micro-
of materials with a fixed 50 wt% sulfur exhibited a broad, tunable scopic perspective61 in the development of a new dynamic framework
range from 35 to 112 °C by varying the ratio of NBD2 crosslinker with may be a worthwhile theoretical endeavor to undertake. It is also likely
DIB comonomers. Likewise, Tg was tunable between 20 and 105 °C that the network is saturated with crosslinks in these two highly
with the 70 wt% sulfur materials while varying feed ratios of DIB and crosslinked systems which makes it difficult to differentiate the two.
NBD2 (Fig. 9a). Additionally, the combination of DIB and NBD2 The stress relaxation in the polysulfides is due to S-S bond rear-
comonomers can successfully consume S8 to (nearly) complete rangement, however, due to differences in sulfur rank and chemical
conversion over the compositional range tested which effectively structure, the temperature dependence of the horizontal and vertical
eliminates residual S8 blooming in the final material. Crystalline S8 shift factors may be different and require investigation. The horizontal
following bloom would be observable by a melting peak in DSC near shift factors (τ m ), determined from fitting the Maxwell model to the
~113 °C corresponding to the melting of sulfur (Supplementary data, rely on the fact that one relaxation mode is dominant and is
Fig. 12). The linear polymer poly(S50-r-Sty50) exhibited a low Tg value attributed to the S-S bond exchange reactions and therefore have a
(-8 °C) due to the lack of crosslinks, increasing the overall mobility of single temperature dependency. As shown above, this time scale was
the material (Supplementary Fig. 11). The low Tg value makes it a determined by fitting the low frequency data and is not an arbitrary
viscous liquid at room temperature and thus the thermally activated constant. Its temperature dependence was found from the Arrhenius
S-S bond exchange is not needed for stress relaxation to occur in the equation:
material. All of the polysulfide polymers display a higher Tg than that
of polymeric sulfur (Tg = –30 °C), referring to the quick-quenched τ m or bT = A × expðE a =RTÞ ð6Þ
morphology from 200 °C60.
Since adherence to the Maxwell model is a function of cross- where A is a pre-exponential factor, E a is the activation energy, R is the
linking density, it is useful to understand how the crosslinking density gas constant, and bT is the vertical shift factor. Remarkably, the
changes with NBD2 composition. Using the plateau modulus from the activation energy for the Maxwell relaxation times is typical of linear
master curves (marked by a star in Supplementary Figs. 13–22 and 24) covalent polymers62 as shown in Supplementary Table 1. Stern and
of sulfur co- and terpolymers, the molecular weight between topolo- Tobolsky have shown that the activation energy of the stress relaxation
gical constraints (M x ) (i.e., crosslinks/entanglements) was calculated in polysulfide rubbers is approximately the same despite different
for each material. These effective Mx can be determined using Eq. 5 chemical structures indicating that the same type of bond is
where ρ is the polymer density, R is the ideal gas constant, and T is responsible for the decay35,36. Their tabulated energies range between
temperature20. The value of G0 in the rubbery plateau where G00 is at a 85 and ~108 kJ/mol. Tobolsky et al. have shown that the bond
minimum is used as the plateau modulus, G0N . dissociation energy of S-S bonds in methyl tetrasulfide (a small
molecule model system) is approximately 150 kJ/mol which is similar
to previous reports of typical activation energies of ~146 kJ/mol63.
ρRT While it has been shown that the flow activation energy of a polymer
G0N = ð5Þ
Mx can differ from small molecule studies10,64, there is an order of

Nature Communications | (2023)14:7553 8

Article https://doi.org/10.1038/s41467-023-43117-1

magnitude difference between our energies. This could potentially be temperatures. Polymer, in powdered form following synthesis, was
due to the higher rank of sulfur in our materials or the amount of added to the bottom plate utilizing a polymer melt ring. After allowing
oligomeric cyclized sulfur in the polymer. the material to soften and completely relax at an elevated temperature,
The vertical shift factors (bT ) are found to vary significantly in a the ring was removed, the material was trimmed, and the test gap was
very small temperature range. Since the vertical shift factors typically set. At each temperature, the linear viscoelastic region (LVR) of the
do not vary this much in carbon-based polymers (Supplementary polymer was determined by performing strain sweeps from 0.01 to
Fig. 27 for polystyrene), it was hypothesized that these too, in addition 100% at a constant frequency of 100 rad/s. Frequency sweeps were
to the horizontal shift factors, are due to the S-S bond exchange then performed from 100 to 0.1 rad/s within the LVR at varying tem-
reactions. Plots of the vertical shift factors versus temperature (Sup- peratures from low to high. Data collected under the minimum torque
plementary Figs. 37–42) were also modeled using an Arrhenius equa- of the instrument are excluded.
tion to determine the activation energy and pre-exponential factor in
Eq. 6 with results shown in Supplementary Table 1. It is clear this Solid-state Nuclear Magnetic Resonance (NMR) Spectroscopy
activation energy is for some systems a factor of two or more higher The synthesized material was ground to a fine powder by pressing it
compared to that for the relaxation times and although hypothesized lightly with a pestle in a mortar. Approximately 30 mg of fine powder
to be affected by S-S bond exchange, its contribution to each factor was loaded into a 4 mm zirconia rotor with Kel-F cap. 13C observed at
must be different. The vertical shift factors for linear polymers are 125 MHz, in 12 T magnetic field using a Varian spectrometer. Irradiation
typically near one, so, this demonstrates another difference our sys- powers were 4.0 (13C) and 3.0 (1H) microseconds for 90° pulses. 800-
tems have to traditional polymer melt rheology. Indeed, Cates and co- 3200 scans were accumulated with spinal 1H decoupling during
workers21,23,38 do not predict such an effect for dynamic covalent bonds acquisition (20 ms) and 1 s recycle delay. Observations were made at 7
in a linear polymer system. At this point we can only make an obser- and/or 12 kHz spin rate. Magic angle was set using the outer spinning
vation of this effect and cannot hypothesize why this behavior occurs. sidebands of NaNO3. Chemical shifts referenced externally to TMS (0.0
ppm) using the amide carbon signal of glycine (176.3 ppm). RF pulse
Methods powers and cross-polarization (CP) were verified using a glycine
Materials standard. 13C chemical shifts were predicted using ACD Labs suite 2018
Elemental sulfur was acquired from Sigma-Aldrich, 1,3-diisoprope- with a maximum expected error of +/-3 ppm.
nylbenzene was acquired from TCI America, 2,5-Norbornadiene was
acquired from ThermoScientific, and polystyrene (Styron 685D) was Data availability
acquired from AmSty. All data necessary to support the conclusions of this paper are avail-
able within the paper and in the Supplementary Information, including
General inverse vulcanization synthetic procedure detailed material synthesis and characterization. All other data that
Elemental sulfur (in the desired wt%) was added to a 20 mL glass vial support the findings of this study are available from the corresponding
equipped with a magnetic stir bar and was heated to 165 °C in an oil authors upon request.
bath until a clear yellow molten phase was formed. Norbornadiene
(NBD2) was then added to the molten phase (in the desired wt%). 1,3- References
diisopropenylbenzene (DIB) (in the desired wt%) was then injected into 1. Kloxin, C. J., Scott, T. F., Adzima, B. J. & Bowman, C. N. Covalent
the molten phase via a syringe. The resulting mixture was stirred at adaptable networks (CANs): A unique paradigm in cross-linked
165 °C for 10-15 minutes (depending on exact composition) until stir- polymers. Macromolecules 43, 2643–2653 (2010).
ring stopped due to the increased viscosity of the reaction mixture. 2. Montarnal, D., Capelot, M., Tournilhac, F. & Leibler, L. Silica-like
After cooling to room temperature, the polysulfide was extracted from malleable materials from permanent organic networks. Science
the vial to yield the polysulfide. Yields and exact procedures for each 334, 965–968 (2011).
composition are in the Supplementary Information Methods. 3. Post, W., Susa, A., Blaauw, R., Molenveld, K. & Knoop, R. J. I. A review
on the potential and limitations of recyclable thermosets for
Thermogravimetric Analysis (TGA) structural applications. Polym. Rev. 60, 359–388 (2020).
A TA Instruments Discovery TGA 5500 (New Castle, Delaware) was 4. Wu, S. & Chen, Q. Advances and new opportunities in the rheology
used for thermal analysis at a constant ramp rate of 10 °C/min from 30 of physically and chemically reversible polymers. Macromolecules
to 800 °C. Platinum sample pans were used with approximately 10 mg 55, 697–714 (2022).
of material. 5. Zhang, Z. P., Rong, M. Z. & Zhang, M. Q. Polymer engineering based
on reversible covalent chemistry: A promising innovative pathway
Differential Scanning Calorimetry (DSC) towards new materials and new functionalities. Prog. Polym. Sci.
A TA Instruments Discovery DSC 2500 (New Castle, Delaware) equip- 80, 39–93 (2018).
ped with a RCS 40 cooler was used for thermal analysis. The samples of 6. Khan, A., Ahmed, N. & Rabnawaz, M. Covalent adaptable network
approximately 5 mg were sealed hermetically in aluminum Tzero pans. and self-healing materials: current trends and future prospects in
A heat-cool-heat cycle was run on each material at a ramp rate of 10 °C/ sustainability. Polymers 12, 2027 (2020).
min from −10 °C to 150 °C and the second heat was used for analysis 7. Podgórski, M. et al. Toward stimuli‐responsive dynamic thermosets
of Tg. through continuous development and improvements in covalent
adaptable networks (CANs). Adv. Mater. 32, (2020).
Rheological experiments 8. Kloxin, C. J. & Bowman, C. N. Covalent adaptable networks: smart,
Small amplitude oscillatory rheology was performed using a TA reconfigurable and responsive network systems. Chem. Soc. Rev.
Instruments (New Castle, Delaware) ARES-G2 rheometer. Tests were 42, 7161–7173 (2013).
conducted using 8 mm stainless steel parallel plates with a forced 9. McBride, M. K. et al. Enabling applications of covalent adaptable
convection oven accessory for temperature control above ambient. networks. Annu. Rev. Chem. Biomol. Eng. 10, 175–198 (2019).
Cool nitrogen gas from liquid nitrogen was blown through the con- 10. Scheutz, G. M., Lessard, J. J., Sims, M. B. & Sumerlin, B. S. Adaptable
vection oven to access subambient temperatures. When necessary, crosslinks in polymeric materials: resolving the intersection of
320 grit 3M sandpaper was attached to the 8 mm parallel plates to thermoplastics and thermosets. J. Am. Chem. Soc. 141,
acquire measurements free of slip at higher frequencies at low 16181–16196 (2019).

Nature Communications | (2023)14:7553 9

Article https://doi.org/10.1038/s41467-023-43117-1

11. Wojtecki, R. J., Meador, M. A. & Rowan, S. J. Using the dynamic bond 37. Turner, M. S. & Cates, M. E. Linear viscoelasticity of living polymers:
to access macroscopically responsive structurally dynamic poly- a quantitative probe of chemical relaxation times. Langmuir 7,
mers. Nat. Mater. 10, 14–27 (2011). 1590–1594 (1991).
12. Chakma, P. & Konkolewicz, D. Dynamic covalent bonds in polymeric 38. Granek, R. & Cates, M. E. Stress relaxation in living polymers:
materials. Angew. Chem. - Int. Ed. 58, 9682–9695 (2019). Results from a Poisson renewal model. J. Chem. Phys. 96,
13. Wanasinghe, S. V., Dodo, O. J. & Konkolewicz, D. Dynamic bonds: 4758–4767 (1992).
adaptable timescales for responsive materials. Angew. Chemie - Int. 39. Chung, W. J. et al. The use of elemental sulfur as an alternative
Ed. 61, (2022). feedstock for polymeric materials. Nat. Chem. 5, 518–524 (2013).
14. Denissen, W. et al. Vinylogous Urethane Vitrimers. Adv. Funct. 40. Lee, T., Dirlam, P. T., Njardarson, J. T., Glass, R. S. & Pyun, J. Poly-
Mater. 25, 2451–2457 (2015). merizations with elemental sulfur: from petroleum refining to
15. Röttger, M. et al. High-performance vitrimers from commodity polymeric materials. J. Am. Chem. Soc. 144, 5–22 (2022).
thermoplastics through dioxaborolane metathesis. Science 356, 41. Griebel, J. J., Glass, R. S., Char, K. & Pyun, J. Polymerizations with
62–65 (2017). elemental sulfur: A novel route to high sulfur content polymers for
16. Denissen, W. et al. Chemical control of the viscoelastic properties of sustainability, energy and defense. Prog. Polym. Sci. 58,
vinylogous urethane vitrimers. Nat. Commun. 8, 14857 (2017). 90–125 (2016).
17. de Gennes, P. G. Reptation of a polymer chain in the presence of 42. Worthington, M. J. H., Kucera, R. L. & Chalker, J. M. Green chemistry
fixed obstacles. J. Chem. Phys. 55, 572–579 (1971). and polymers made from sulfur. Green. Chem. 19, 2748–2761 (2017).
18. Doi, M. & Edwards, S. F. The Theory of Polymer Dynamics. (Clar- 43. Chalker, J. M., Worthington, M. J. H., Lundquist, N. A. & Esdaile, L. J.
endon Press, 1988). Synthesis and applications of polymers made by inverse vulcani-
19. Meyer, B. Elemental sulfur. Chem. Rev. 76, 367–388 (1976). zation. Top. Curr. Chem. 377, 1–27 (2019).
20. Ferry, J. D. Viscoelastic Properties of Polymers. (John Wiley & Sons, 44. Zhang, Y., Glass, R. S., Char, K. & Pyun, J. Recent advances in the
Inc., 1980). polymerization of elemental sulphur, inverse vulcanization and
21. Cates, M. E. & Candau, S. J. Statics and dynamics of worm-like methods to obtain functional Chalcogenide Hybrid Inorganic/
surfactant micelles. J. Phys. Condens. Matter 2, 6869–6892 (1990). Organic Polymers (CHIPs). Polym. Chem. 10, 4078–4105 (2019).
22. Stashick, M. J. & Marriott, R. A. Viscoelastic behavior corresponding 45. Griebel, J. J. et al. Preparation of dynamic covalent polymers via
to reptative relaxation times across the λ -transition for liquid ele- inverse vulcanization of elemental sulfur. ACS Macro Lett. 3,
mental sulfur. J. Chem. Phys. 152, 044503 (2020). 1258–1261 (2014).
23. Cates, M. E. Reptation of living polymers: dynamics of entangled 46. Griebel, J. J. et al. Dynamic covalent polymers via inverse vulcani-
polymers in the presence of reversible chain-scission reactions. zation of elemental sulfur for healable infrared optical materials.
Macromolecules 20, 2289–2296 (1987). ACS Macro Lett. 4, 862–866 (2015).
24. Grubbs, R. B. & Grubbs, R. H. 50th anniversary perspective: living 47. Bao, J. et al. On the Mechanism of the Inverse Vulcanization of
polymerization—emphasizing the molecule in macromplecules. Elemental Sulfur: Structural Characterization of Poly(sulfur- ran-
Macromolecules 50, 6979–6997 (2017). dom -(1,3-diisopropenylbenzene). J. Am. Chem. Soc. 145,
25. Szwarc, M. ‘Living’ polymers. Nature 178, 1168–1169 (1956). 12386–12397 (2023).
26. Cates, M. E. Theory of the viscosity of polymeric liquid sulfur. 48. Kleine, T. S. et al. Infrared fingerprint engineering: a molecular‐
Europhys. Lett. 4, 497–502 (1987). design approach to long‐wave infrared transparency with poly-
27. Berret, J. F., Appell, J. & Porte, G. Linear rheology of entangled meric materials. Angew. Chem. Int. Ed. 58, 17656–17660 (2019).
wormlike micelles. Langmuir 9, 2851–2854 (1993). 49. Lundquist, N. A. et al. Reactive compression molding post‐inverse
28. Kern, F., Lemarechal, P., Candau, S. J. & Cates, M. E. Rheological vulcanization: a method to assemble, recycle, and repurpose sulfur
properties of Semidilute and concentrated aqueous solutions of polymers and composites. Chem. – A Eur. J. 26,
Cetyltrimethylammonium Bromide in the presence of Potassium 10035–10044 (2020).
Bromide. Langmuir 8, 437–440 (1992). 50. Zhang, Y. et al. Inverse vulcanization of elemental sulfur and styrene
29. Schurtenberger, P., Scartazzini, R., Magid, L. J., Leser, M. E. & Luisi, for polymeric cathodes in Li‐S batteries. J. Polym. Sci. Part A Polym.
P. L. Structural and dynamic properties of polymer-like reverse Chem. 55, 107–116 (2017).
micelles. J. Phys. Chem. 94, 3695–3701 (1990). 51. Kang, K. et al. Segmented polyurethanes and thermoplastic elas-
30. Schurtenberger, P., Scartazzini, R. & Luisi, P. L. Viscoelastic prop- tomers from elemental sulfur with enhanced thermomechanical
erties of polymerlike reverse micelles. Rheol. Acta 28, properties and flame retardancy. Angew. Chem. Int. Ed. 60,
372–381 (1989). 22900–22907 (2021).
31. Shikata, T., Hirata, H. & Kotaka, T. Micelle formation of detergent 52. Zhang, Y., Konopka, K. M., Glass, R. S., Char, K. & Pyun, J. Chalco-
molecules in aqueous media. 3. Viscoelastic properties of aqueous genide hybrid inorganic/organic polymers (CHIPs) via inverse vul-
Cetyltrimethylammonium Bromide-salicylic acid solutions. Lang- canization and dynamic covalent polymerizations. Polym. Chem. 8,
muir 5, 398–405 (1989). 5167–5173 (2017).
32. Shikata, T., Hirata, H. & Kotaka, T. Micelle formation of detergent 53. Williams, M. L., Landel, R. F. & Ferry, J. D. The temperature depen-
molecules in aqueous media: viscoelastic properties of aqueous dence of relaxation mechanisms in amorphous polymers and other
cetyltrimethylammonium bromide solutions. Langmuir 3, glass-forming liquids. J. Am. Chem. Soc. 77, 3701–3707 (1955).
1081–1086 (1987). 54. Yamada, S. et al. Characterization of sulfur exchange reaction
33. Rehage, H. & Hoffmann, H. Rheological properties of viscoelastic between polysulfides and elemental sulfur using a 35S radio-
surfactant systems. J. Phys. Chem. 92, 4712–4719 (1988). isotope tracer method. Chem. Commun. 3, 842–843 (2003).
34. Fischer, P. & Rehage, H. Rheological master curves of viscoelastic 55. Kleine, T. S. et al. 100th anniversary of macromolecular science
surfactant solutions by varying the solvent viscosity and tempera- viewpoint: high refractive index polymers from elemental sulfur for
ture. Langmuir 13, 7012–7020 (1997). infrared thermal imaging and optics. ACS Macro Lett. 9,
35. Stern, M. D. & Tobolsky, A. V. Stress-time-temperature relations in 245–259 (2020).
polysulfide rubbers. J. Chem. Phys. 14, 93–100 (1946). 56. Smith, J. A., Wu, X., Berry, N. G. & Hasell, T. High sulfur content
36. Mochulsky, M. & Tobolsky, A. V. Chemorheology of polysulfide polymers: The effect of crosslinker structure on inverse vulcaniza-
rubbers. Rubber Chem. Technol. 22, 712–730 (1949). tion. J. Polym. Sci. Part A Polym. Chem. 56, 1777–1781 (2018).

Nature Communications | (2023)14:7553 10

Article https://doi.org/10.1038/s41467-023-43117-1

57. Parker, D. J. et al. Low cost and renewable sulfur-polymers by Competing interests
inverse vulcanisation, and their potential for mercury capture. J. The authors declare no competing interests.
Mater. Chem. A 5, 11682–11692 (2017).
58. Crockett, M. P. et al. Sulfur-Limonene polysulfide: a material syn- Additional information
thesized entirely from industrial by-products and its use in remov- Supplementary information The online version contains
ing toxic metals from water and soil. Angew. Chem. - Int. Ed. 55, supplementary material available at
1714–1718 (2016). https://doi.org/10.1038/s41467-023-43117-1.
59. Nielsen, L. E. Cross-linking–effect on physical properties of poly-
mers. J. Macromol. Sci. Part C. 3, 69–103 (1969). Correspondence and requests for materials should be addressed to
60. Tobolsky, A. V., MacKnight, W., Beevers, R. B. & Gupta, V. D. The Jeffrey Pyun or Michael E. Mackay.
glass transition temperature of polymeric sulphur. Polymers 4,
423–427 (1963). Peer review information Nature Communications thanks Dhriti Nepal
61. Zaccone, A. & Terentjev, E. M. Disorder-assisted melting and the and the other, anonymous, reviewer(s) for their contribution to the peer
glass transition in amorphous solids. Phys. Rev. Lett. 110, review of this work. A peer review file is available.
178002 (2013).
62. Wang, J. & Porter, R. S. On the viscosity-temperature behavior of Reprints and permissions information is available at
polymer melts. Rheol. Acta 34, 496–503 (1995). http://www.nature.com/reprints
63. Kende, I., Pickering, T. L. & Tobolsky, A. V. The dissociation energy of
the tetrasulfide linkage. J. Am. Chem. Soc. 87, 5582–5586 (1965). Publisher’s note Springer Nature remains neutral with regard to jur-
64. Jourdain, A. et al. Rheological properties of covalent adaptable isdictional claims in published maps and institutional affiliations.
networks with 1,2,3-Triazolium cross-links: the missing link between
Vitrimers and dissociative networks. Macromolecules 53, Open Access This article is licensed under a Creative Commons
1884–1900 (2020). Attribution 4.0 International License, which permits use, sharing,
adaptation, distribution and reproduction in any medium or format, as
Acknowledgements long as you give appropriate credit to the original author(s) and the
J.P. and M.M. disclose support for the research of this work from the source, provide a link to the Creative Commons licence, and indicate if
National Science Foundation (NSF) [PFI-RP 1940942] and the Air Force changes were made. The images or other third party material in this
Research Laboratory through DMREF-2118578, the Air Force Research article are included in the article’s Creative Commons licence, unless
Laboratories [FA8650-16-D-5404], and ENI S.p.A. indicated otherwise in a credit line to the material. If material is not
included in the article’s Creative Commons licence and your intended
Author contributions use is not permitted by statutory regulation or exceeds the permitted
D.B. performed rheological characterization, thermal analysis, data use, you will need to obtain permission directly from the copyright
analysis, and wrote the manuscript. T.L. performed thermal analysis and holder. To view a copy of this licence, visit http://creativecommons.org/
material synthesis. K.K. and J.M. performed material synthesis and 1H & licenses/by/4.0/.
13C NMR analysis. W.P. performed ssNMR characterization and analysis.
J.P. and M.M. contributed to data analysis and wrote the manuscript. © The Author(s) 2023

Nature Communications | (2023)14:7553 11