Ultralight Metallic Microlattices

Abstract

Ultralight (<10 milligrams per cubic centimeter) cellular materials are desirable for thermal insulation; battery electrodes; catalyst supports; and acoustic, vibration, or shock energy damping. We present ultralight materials based on periodic hollow-tube microlattices. These materials are fabricated by starting with a template formed by self-propagating photopolymer waveguide prototyping, coating the template by electroless nickel plating, and subsequently etching away the template. The resulting metallic microlattices exhibit densities ρ ≥ 0.9 milligram per cubic centimeter, complete recovery after compression exceeding 50% strain, and energy absorption similar to elastomers. Young's modulus E scales with density as E ~ ρ(2), in contrast to the E ~ ρ(3) scaling observed for ultralight aerogels and carbon nanotube foams with stochastic architecture. We attribute these properties to structural hierarchy at the nanometer, micrometer, and millimeter scales.

Full text

DOI: 10.1126/science.1211649
, 962 (2011);334 Science , et al.T. A. Schaedler
Ultralight Metallic Microlattices
This copy is for your personal, non-commercial use only.
clicking here.colleagues, clients, or customers by , you can order high-quality copies for yourIf you wish to distribute this article to others
here.following the guidelines can be obtained byPermission to republish or repurpose articles or portions of articles
): November 18, 2011 www.sciencemag.org (this infomation is current as of
The following resources related to this article are available online at
http://www.sciencemag.org/content/334/6058/962.full.html
version of this article at: including high-resolution figures, can be found in the onlineUpdated information and services,
http://www.sciencemag.org/content/suppl/2011/11/16/334.6058.962.DC1.html
can be found at: Supporting Online Material
http://www.sciencemag.org/content/334/6058/962.full.html#ref-list-1
, 2 of which can be accessed free:cites 23 articlesThis article
http://www.sciencemag.org/cgi/collection/mat_sci
Materials Science subject collections:This article appears in the following
registered trademark of AAAS. is aScience2011 by the American Association for the Advancement of Science; all rights reserved. The title CopyrightAmerican Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005.
(print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by theScience
on November 18, 2011www.sciencemag.orgDownloaded from

Ultralight Metallic Microlattices
T. A. Schaedler,
1
*A. J. Jacobsen,
1
A. Torrents,
2
A. E. Sorensen,
1
J. Lian,
3
J. R. Greer,
3
L. Valdevit,
2
W. B. Carter
1
Ultralight (<10 milligrams per cubic centimeter) cellular materials are desirable for thermal
insulation; battery electrodes; catalyst supports; and acoustic, vibration, or shock energy damping.
We present ultralight materials based on periodic hollow-tube microlattices. These materials are
fabricated by starting with a template formed by self-propagating photopolymer waveguide
prototyping, coating the template by electroless nickel plating, and subsequently etching away
the template. The resulting metallic microlattices exhibit densities r≥0.9 milligram per cubic
centimeter, complete recovery after compression exceeding 50% strain, and energy absorption similar
to elastomers. Young’s modulus Escales with density as E~r
2
,incontrasttotheE~r
3
scaling
observed for ultralight aerogels and carbon nanotube foams with stochastic architecture. We attribute
these properties to structural hierarchy at the nanometer, micrometer, and millimeter scales.
The effective properties of low-density ma-
terials are defined both by their cellular
architecture (i.e., the spatial configuration
of voids and solid) and the properties of the solid
constituent (e.g., stiffness, strength, etc.). In the
ultralight regime below 10 mg/cm
3
,veryfewma-
terials currently exist: silica aerogels [density r≥
1mg/cm
3
(1,2)], carbon nanotube aerogels [r≥4
mg/cm
3
(3)], metallic foams [r≥10 mg/cm
3
(4,5)],
and polymer foams [r≥8mg/cm
3
(6,7)]. These
materials have a wide range of applications, such
as thermal insulation, shock or vibration damp-
ing, acoustic absorption, and current collectors in
battery electrodes and catalyst supports (8). All of
the ultralow-density materials mentioned above
have random cellular architectures. This random
cell structure results in some beneficial properties
(e.g., high specific surface area and restriction of
gas flow), but generally the inefficient distri-
bution of the constituent results in specific prop-
erties (e.g., stiffness, strength, energy absorption,
and conductivity) far below those of the bulk
material (8). As an example, Young’s modulus,
E, of ultralight stochastic materials scales poorly
with density, generally following E~r
3
(9), in
contrast to the well-known E~r
2
relationship for
random open-cell foams with higher relative
densities (8). In large-scale structures, it has been
shown that introducing order and hierarchy can
substantially improve material utilization and re-
sultant properties. For instance, the Eiffel Tower
possesses a relative density similar to that of low-
density aerogels (10) but is clearly structurally
robust. The size difference between the smallest
and largest structural features will determine the
degree of hierarchy that can be achieved. In this
paper, we present a method for creating ordered
hollow-tube lattice materials with a minimum
scale of ~100 nm. Coupled with control over mm-
to cm-scale structural features, this enables us to
bring the benefits of order and hierarchy down
to the materials level. The result is an ultralight-
weight cellular material with efficient material
utilization, a Young’s modulus that follows E~
r
2
, and the ability to recover from >50% com-
pression while demonstrating large energy absorp-
tion upon cyclic loading.
The base architecture of our metallic micro-
lattices consists of a periodic array of hollow
tubes that connect at nodes, forming an octahe-
dral unit cell without any lattice members in the
basal plane. Figure 1 illustrates how the micro-
lattice architecture can be distilled into three levels
of hierarchy at three distinct length scales: unit
cell (~mm to cm), hollow tube lattice member
(~mm to mm), and hollow tube wall (~nm to mm).
Each architectural element can be controlled in-
dependently, providing exceptional control over
the design and properties of the resulting micro-
lattice. The architecture determines the relative den-
sity of the lattice, with the absolute density dictated
by the film material.
The fabrication process begins with solid mi-
crolattice templates fabricated by using a self-
propagating photopolymer waveguide technique.
In this method, a thiol-ene liquid photomonomer
is exposed to collimated ultraviolet (UV) light
through a patterned mask, producing an inter-
connected three-dimensional photopolymer lat-
tice (11). A wide array of different architectures
with unit cell dimensions ranging from 0.1 to >10
mm can be made by altering the mask pattern and
the angle of the incident light (12,13). Here, we
focus on architectures with 1- to 4-mm lattice
member length L, 100- to 500-mm lattice member
diameter D, 100- to 500-nm wall thickness t,
and 60° inclination angle q, similar to the mi-
crolattices depicted in Fig. 1. Conformal nickel-
phosphorous thin films were deposited on the
polymer lattices by electroless plating, and the
polymer was subsequently etched out (table S1).
The autocatalytic electroless nickel-plating reac-
tion enables deposition of thin films with con-
trolled thickness on complex shapes and inside
pores without noticeable mass transport limita-
tions. The ultralight microlattice essentially trans-
lates the deposited nanoscale thin film in three
dimensions to form a macroscopic material where
the base structural elements are hollow tubes. By
controlling the reaction time, a 100-nm-thick uni-
form conformal coating can be achieved, resulting
in a cellular material with a density of 0.9 mg/cm
3
(Fig. 1). The density is calculated by using the
weight of the solid structure but not including the
Fig. 1. Design, processing, and cellular architecture of ultralight microlattices. (A) Polymer microlattice
templates are fabricated from a three-dimensional array of self-propagating photopolymer waveguides. (B)
The open-cellular templates are electroless plated with a conformal Ni-P thin film followed by etch removal
of the template. (C) Image of the lightest Ni-P microlattice fabricated with this approach: 0.9 mg/cm
3
.
(Dand E) Images of two as-fabricated microlattices along with a breakdown of the relevant architec-
tural elements.
1
HRL Laboratories Limited Liability Company, Malibu, CA 90265,
USA.
2
Department of Mechanical and Aerospace Engineering,
University of California, Irvine, CA 92697, USA.
3
Division of En-
gineering and Applied Sciences, California Institute of Technol-
ogy, Pasadena, CA 91125, USA.
*To whom correspondence should be addressed. E-mail:
taschaedler@hrl.com
18 NOVEMBER 2011 VOL 334 SCIENCE www.sciencemag.org962
REPORTS
on November 18, 2011www.sciencemag.orgDownloaded from

weight of air in the pores, adhering to the standard
practice for cellular materials. The density of air at
ambient conditions, 1.2 mg/cm
3
, multiplied by its
volume fraction would need to be added to ex-
press the density of the solid-air composite. This
method to form microlattices allows significantly
more control than typical methods for forming
other ultralightweight materials, such as foams
and aerogels, where nominally random processes
govern porosity formation.
Characterization of the base constituent by
transmission electron microscopy (TEM) revealed
that as-deposited electroless nickel thin films have
an average grain size of ~7 nm, which is consistent
with literature reports (14,15). Energy-dispersive
x-ray spectroscopy confirmed that the film com-
position is 7% phosphorous and 93% nickel by
weight. Because the films were not annealed after
deposition, they remained as a supersaturated
solid solution of phosphorous in a crystalline
face-centered cubic nickel lattice with no Ni
3
Ppre-
cipitates present (14). The 7-nm grain size ren-
ders electroless nickel thin films harder and more
brittle than typical nano- and microcrystalline
nickel. A hardness of 6 GPa and a modulus of
210 GPa were measured by nanoindentation
and hollow tube compression experiments, re-
spectively (16).
Compression experiments on the as-formed
microlattices showed a nearly complete recovery
Fig. 2. Nickel microlattices exhibit recoverable deformation. (A) Before deformation. (B)15%
compression. (C) 50% compression. (D) Full recovery after removal of load. (E) Optical image of
unit cell unloaded. (F) Example of node buckling under compression. (G) SEM image of node before testing. (H) SEM image of node after six compression
cycles at 50% strain. (The compression test is shown in movie S1.)
Fig. 3. Multicycle compression test results of nickel microlattices.
(A) Stress-strain curves of a microlattice with 14 mg/cm
3
ex-
hibiting recoverable deformation (compare with Fig. 2). (B)
History of compressive modulus, yield stress, maximum stress,
and energy loss coefficient during the first six compression
cycles shown in (A). (C) Stress-strain curve of a microlattice
with a density of 1.0 mg/cm
3
.(D) Stress-strain curves of a mi-
crolattice with 43 mg/cm
3
showing deformation more typical
for metallic cellular materials. (E)SEMmicrographofpost-
nanoindentation mark in 500-nm-thick electroless nickel film
demonstrating brittle behavior.
www.sciencemag.org SCIENCE VOL 334 18 NOVEMBER 2011 963
REPORTS
on November 18, 2011www.sciencemag.orgDownloaded from

from strains exceeding 50% (movie S1). Figure
2, A to D, provides images of a 14 mg/cm
3
mi-
crolattice sample (L=1050mm, D= 150 mm, t=
500 nm) during compression testing, and Fig. 3A
conveys the corresponding stress-strain curve
measured at a prescribed displacement rate of
10 mm/s. In this experiment, the sample was not
constrained by face sheets or attached to any
compression platens. Because of a small taper
in lattice strut diameter, the deformation typ-
ically initiates at a particular surface. Upon first
compression, the lattice exhibits a compressive
modulus of 529 kPa, with deviations from lin-
ear elastic behavior starting at ~10 kPa. The stress
decreases slightly after the peak, which is as-
sociated with buckling and node fracture events,
and a broad plateau is subsequently established in
the stress-strain curve as buckling and localized
node fracture events spread through the lattice.
Figure 2C shows the microlattice at 50% com-
pression. Upon unloading, the stress drops rapid-
ly but does not approach zero until the platen is
close to its original position. After removing the
load, the microlattice recovers to 98% of its orig-
inal height and resumes its original shape (Figs.
2D and 3A).
The stress-strain behavior corresponding to
the first cycle is never repeated during subsequent
testing. Rather, during a second compression, the
peak stress is absent and the “pseudohardening”
behavior changes, but the stress level achieved at
50% strain is only 10% lower than that after the
first cycle. Consecutive compression cycles ex-
hibit stress-strain curves nearly identical to those
of the second compression. Stiffness and strength
diminish with cycle number but remain almost
constant after the third cycle (Fig. 3B). The mi-
crolattice also shows significant hysteresis during
compression experiments. For the first cycle, we
estimate the work done in compression to be
4.6 mJ/cm
3
and the energy dissipation to be
3.5 mJ/cm
3
,yieldinganenergylosscoefficient
(Du/u) of 0.77. This large energy dissipation re-
sults from extensive node microcracking and
thus is limited to the first cycle. After three cy-
cles, a nearly constant energy loss coefficient
of ~ 0.4 is calculated (Fig. 3B). From this esti-
mate, we extract a loss coefficient (tan d)of
~0.16 (17), an order of magnitude higher than
for typical nickel foams (18). Figure 3C shows
the stress-strain response of a 1.0 mg/cm
3
sample
with larger unit cells (L=4mm,D= 500 mm, t=
120 nm), illustrating that different microlattice
architectures in the ultralow-density regime result
in similar behavior [although this sample under-
went an additional freeze-drying process step that
caused some damage (table S1)]. Increasing the
density and wall thickness eventually led to a com-
pression behavior more typical for metallic cellular
materials. Figure 3D shows the stress-strain curve
for a 43-mg/cm
3
sample (L= 1050 mm, D= 150 mm,
t= 1400 nm), for which recovery upon unloading
from 50% strain is essentially absent.
Optical examination of the ultralight micro-
lattices during deformation suggests that defor-
mation initiates by local buckling at the nodes
(Fig. 2, E and F). A closer inspection of the
microlattices by scanning electron microscopy
(SEM) shows that cracks and wrinkles are intro-
duced primarily at the nodes during compres-
sion (Fig. 2, G and H). This damage is responsible
for the 1 to 2% residual strain observed after the
first compression cycle, as well as for the drop in
yield strength and modulus during subsequent
compression cycles. Once stable “relief cracks”
form at the nodes, the bulk microlattice material
can undergo large effective compressive strains
through extensive rotations about remnant node
ligaments, but with negligible strain in the solid
nickel-phosphorous material, thus requiring no
further fracture or plastic deformation. This prop-
erty results in the reversible compressive behavior
showninFigs.2and3.Theextremelysmallwall
thickness–to–diameter ratio is essential for this de-
formation mechanism. Increasing this aspect ratio
leads to excessive fracture and loss of recover-
ability (Fig. 3D).
The rise in stress at strains of ~ 40% (Fig. 3A)
is a result of increased interaction between lattice
members after localized compression at the nodes
and should not be confused with densification,
which in these samples does not occur until strains
exceed 90% (fig. S2).
Similar stress-strain curves as presented in
Fig. 3A are typical for viscoelastic polymer foams
(19) and carbon nanotube forests (20) but not
for metal-based materials, implying a nonconven-
tional loss mechanism is present. Two energy-loss
mechanisms could possibly explain the energy
dissipation during compression cycles: (i) struc-
tural damping because of snapping events (e.g.,
kinking or local buckling of the trusses) and (ii)
mechanical or Coulomb friction between con-
tacting members (or a combination of both). This
mechanical behavior is especially unexpected con-
sidering the relatively brittle nature of the constit-
uent electroless nickel thin film, as evidenced by
the formation of cracks near a residual indenta-
tion mark (Fig. 3E) and the rapid collapse upon
single hollow truss member compressions (16).
The brittle nature of the film arises from its nano-
grained microstructure (~7 nm) that hinders plastic
deformation by dislocation motion (21). How-
ever, at the bulk scale microlattices exhibit com-
pletely different properties, because the cellular
architecture effectively transforms this brittle thin
film into a ductile, superelastic lattice by enabling
sufficient freedom for deformation and tolerance
to local strains through formation of stable relief
cracks.
Plotting the relative compressive modulus,
E/E
s
, of various fabricated microlattices versus
their relative density, r/r
s
, shows that the modulus
scales with (r/r
s
)
2
(Fig. 4). This scaling law in-
dicates bending-dominated mechanical behavior
similar to open-cell stochastic foams (8). In con-
trast, other materials with densities < 10 mg/cm
3
,
such as aerogels and carbon nanotube (CNT)
foams, exhibit a steeper scaling of E/E
s
~(r/r
s
)
3
(Fig. 4) (22,23), because of inefficient load transfer
between ligaments. (Incidentally, this also affects
the structural stability of a self-supporting cellular
material, imposing a lower bound on density.) We
notice that, although the relative modulus of
topologically designed periodic lattice materials
[such as octet-truss lattices (24)] typically scales
linearly with relative density (25,26), we do not
expect the same scaling for the microlattices
described here because of the absence of struts in
the basal plane (26) and the ultrathin-walled
hollow nodes, both of which facilitate localized
bending deformation (Fig. 2F). Nonetheless, the
marked improvement in mechanical efficiency
over any other existing ultralightweight material,
achieved by controlling both dimensions and
periodicity of the architecture, enabled a self-
supporting cellular material with a relative density
an order of magnitude lower than previously re-
alized. Further, by transforming a brittle Ni-P thin
film into a three-dimensional cellular material and
designing a hierarchical cellular architecture at three
Fig. 4. Relative compres-
sive modulus (defined as
the measured Young’smod-
ulus, E,dividedbythe
Young’s modulus of the
constituent solid, E
s
)ofse-
lected cellular materials
at low relative density.
18 NOVEMBER 2011 VOL 334 SCIENCE www.sciencemag.org964
REPORTS
on November 18, 2011www.sciencemag.orgDownloaded from

distinct length scales, we demonstrated the emer-
gence of entirely different mechanical properties.
In addition to possible applications for an ultra-
light material with high energy absorption and
recoverability, we anticipate that these results will
help reshape our understanding of the interac-
tion between material properties and structural
architecture.
References and Notes
1. Guinness Book of World Records, Least Dense Solid
(2003), www.guinnessworldrecords.com/records-1/least-
dense-solid/.
2. T. M. Tillotson, L. W. Hrubesh, J. Non-Cryst. Solids 145,
44 (1992).
3. J. Zou et al., ACS Nano 4, 7293 (2010).
4. A. Verdooren, H. M. Chan, J. L. Grenestedt, M. P. Harmer,
H. S. Caram, J. Am. Ceram. Soc. 89, 3101 (2006).
5. B. C. Tappan et al., J. Am. Chem. Soc. 128, 6589 (2006).
6. M. Chanda, S. K. Roy, Plastics Technology Handbook
(CRC Press, Boca Raton, FL, 2007).
7. B. A. S. F. Corporation, Materials Safety Data Sheet for
Basotect V3012 (2007), www.basf.co.kr/02_products/
01_thermoplastics/spe/document/MSDS-Basotect%
20V3012.pdf.
8. L. J. Gibson, M. F. Ashby, Cellular Solids: Structure and
Properties (Cambridge Univ. Press, Cambridge, 1997).
9. H.-S. Ma, J.-H. Prévost, R. Jullien, G. W. Scherer,
J. Non-Cryst. Solids 285, 216 (2001).
10. R. S. Lakes, Nature 361, 511 (1993).
11. A. J. Jacobsen, W. Barvosa-Carter, S. Nutt, Acta Mater.
55, 6724 (2007).
12. A. J. Jacobsen, W. Barvosa-Carter, S. Nutt, Adv. Mater.
19, 3892 (2007).
13. A. J. Jacobsen, W. Barvosa-Carter, S. Nutt, Acta Mater.
56, 2540 (2008).
14. S. H. Park, D. N. Lee, J. Mater. Sci. 23, 1643 (1988).
15. S. Y. Chang, Y. S. Lee, H. L. Hsiao, T. K. Chang,
Metall. Mater. Trans. A 37A, 2939 (2006).
16. J. Lian et al., Nano Lett. 11, 4118 (2011).
17. G. F. Lee, B. Hartmann , J. Sound Vibrat. 211, 265
(1998).
18. M. F. Ashby et al., Metal Foams: A Design Guide
(Butterworth-Heinemann, Burlington, MA, 2000), p. 43.
19. N. J. Mills, Cell. Polym. 5, 293 (2006).
20. A. Cao, P. L. Dickrell, W. G. Sawyer, M. N. Ghasemi-Nejhad,
P. M. Ajayan, Science 310, 1307 (2005).
21. J. R. Trelewicz, C. A. Schuh, Acta Mater. 55, 5948
(2007).
22. J. Gross, T. Schlief, J. Fricke, Mater. Sci. Eng. A 168,
235 (1993).
23. M. A. Worsley, S. O. Kucheyev, J. H. Satcher Jr.,
A. V. Hamza, T. F. Baumann, Appl. Phys. Lett. 94,
073115 (2009).
24. V. S. Deshpande, N. A. Fleck, M. F. Ashby, J. Mech.
Phys. Solids 49, 1747 (2001).
25. V. S. Deshpande, M. F. Ashby, N. A. Fleck, Acta Mater.
49, 1035 (2001).
26. V. S. Deshpande, N. A. Fleck, Int. J. Solids Struct. 38,
6275 (2001).
27. L. J. Gibson, M. F. Ashby, Proc. R. Soc. London Ser. A
382, 43 (1982).
28. Hexcel Corporation, HexWeb Honeycomb Attributes and
Properties (1999); datasheet available at www.hexcel.
com/Resources/DataSheets/Brochure-Data-Sheets/
Honeycomb_Attributes_and_Properties.pdf.
29. M. Moner-Girona, A. Roig, E. Molins, E. Martinez,
J. Esteve, Appl. Phys. Lett. 75, 653 (1999).
Acknowledgments: The authors gratefully acknowledge
the financial support by Defense Advanced Research
Projects Agency under the Materials with Controlled
Microstructural Architecture program managed by
J. Goldwasser (contract no. W91CRB-10-0305) and thank
J. W. Hutchinson and C. S. Roper for useful discussions.
A patent application regarding the structure and
formation process of the ultralight microlattices has been
submitted to the U.S. Patent and Trademark Office.
The polymer waveguide process has been patented (U.S.
Patent 7,382,959, U.S. Patent 7,653,279, and U.S.
Patent 8,017,193), but the template can be fabricated
in other ways.
Supporting Online Material
www.sciencemag.org/cgi/content/full/334/6058/962/DC1
Materials and Methods
Fig. S1
Table S1
References
Movie S1
25 July 2011; accepted 12 October 2011
10.1126/science.1211649
Silica-Like Malleable Materials from
Permanent Organic Networks
Damien Montarnal, Mathieu Capelot, François Tournilhac, Ludwik Leibler*
Permanently cross-linked materials have outstanding mechanical properties and solvent resistance,
but they cannot be processed and reshaped once synthesized. Non–cross-linked polymers and
those with reversible cross-links are processable, but they are soluble. We designed epoxy networks
that can rearrange their topology by exchange reactions without depolymerization and showed
that they are insoluble and processable. Unlike organic compounds and polymers whose viscosity
varies abruptly near the glass transition, these networks show Arrhenius-like gradual viscosity
variations like those of vitreous silica. Like silica, the materials can be wrought and welded to make
complex objects by local heating without the use of molds. The concept of a glass made by
reversible topology freezing in epoxy networks can be readily scaled up for applications and
generalized to other chemistries.
Thermoset polymers such as Bakelite must
be polymerized in a mold having the
shape of the desired object because once
the reaction is completed, the polymer cannot be
reshaped or reprocessed by heat or with solvent.
In contrast, thermoplastics, when heated, can flow,
which permits extrusion, injection, and molding
of objects. Depending on the chemical nature of
the plastic, during cooling, solidification occurs
by crystallization or by glass transition. During
vitrification, as the temperature is lowered below
the glass transition, the viscosity abruptly in-
creases in a narrow temperature range, and the
material becomes so viscous that it behaves
essentially like a solid with an elastic modulus
of about 10
9
to 10
10
Pa (1). Nevertheless, com-
pared to processable plastics, cross-linked poly-
mers have superior dimensional stability; have
high-temperature mechanical, thermal, and envi-
ronmental resistance; and are irreplaceable in many
demanding applications, such as in the aircraft
industry. High-performance coatings, adhesives,
rubbers, light-emitting diode lenses, and solar cell
encapsulants are made of permanently cross-linked
polymer networks as well.
Making covalent links reversible could pro-
vide a way to combine processability, reparability,
and high performance (2–6). Networks with bonds
able to break and reform (7–9) or to exchange
pairs of atoms (10) can relax stresses and flow.
The challenge is to allow rapid reversible reac-
tions at high temperatures or by a convenient
stimulus and to fix the network at service condi-
tions. In this context, cleavage or exchange reac-
tions by addition-fragmentation in the presence of
radicals offer interesting possibilities (5,11–14).
Scott et al. demonstrated photoinduced plas-
ticity in cross-linked polymers (11). Similarly,
reparability and self-healing can be induced either
thermally (13) or photochemically (15,16)in
radical systems. However, these systems undergo
unavoidable termination reactions that limit re-
versibility of the networks.
In parallel, a completely different concept
based on chemical equilibrium between bond
breaking and reforming without irreversible side
reactions has been developed (17–20). In these
systems, heating has two effects: It displaces
the equilibrium toward depolymerization and it
accelerates the bond breaking and reforming rate
(8,9). The advantage of such reversible links is
that both above-mentioned effects act together to
bring fluidity and thus processability (5,17–19).
They are, however, detrimental to the network
integrity and performance. Chen et al.haveshown
that to avoid flow and creep at service temper-
atures, one can rely, as in thermoplastics, on glass
transition to quench the system (17). Unfortu-
nately, the systems based on chemical equilibri-
um between bond breaking and reforming are
sensitive to solvents because in the presence of a
solvent, the chemical equilibrium is displaced
toward network depolymerization and dissolu-
tion (19).
We sought to show that reversible networks
can flow while maintaining their integrity and
insolubility at high temperature. The idea is to
Matière Molle et Chimie, UMR 7167 CNRS-ESPCI, Ecole
Supérieure de Physique et Chimie Industrielles, 10 rue
Vauquelin, 75005 Paris, France.
*To whom correspondence should be addressed. E-mail:
ludwik.leibler@espci.fr
www.sciencemag.org SCIENCE VOL 334 18 NOVEMBER 2011 965
REPORTS
on November 18, 2011www.sciencemag.orgDownloaded from