Shrinky Dink Molecula

Progress in Polymer Science 49–50 (2015) 79–120
Contents lists available at ScienceDirect
Progress in Polymer Science

journal homepage: www.elsevier.com/locate/ppolysci
Review
Recent progress in shape memory polymer: New behavior,

enabling materials, and mechanistic understanding
Qian Zhao a , H. Jerry Qi b , Tao Xie a,∗
a
State Key Laboratory of Chemical Engineering, Department of Chemical and Biological Engineering, Zhejiang University, 38 Zheda
Road, Hangzhou, 310027 PR China
b
The George W. Woodruff School of Mechanical Engineering, Georgia Institute of Technology, 801 Ferst Drive, Atlanta, GA 30332, USA
a r t i c l e i n f o a b s t r a c t
Article history: Shape memory polymers (SMPs), as a class of programmable stimuli-responsive shape
Available online 11 April 2015 changing polymers, are attracting increasing attention from the standpoint of both fun-
damental research and technological innovations. Following a brief introduction of the
Keywords: conventional shape memory effect (SME), progress in new shape memory enabling mech-
Shape memory polymer anisms and triggering methods, variations of in shape memory forms (shape memory
Smart materials surfaces, hydrogels, and microparticles), new shape memory behavior (multi-SME and two-
Stimuli-responsive polymer
way-SME), and novel fabrication methods are reviewed. Progress in thermomechanical
Polymer modeling
modeling of SMPs is also presented.
© 2015 Elsevier Ltd. All rights reserved.
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
2. General aspects of SMPs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
2.1. Programming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
2.2. Classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
2.3. Basic molecular requirements and mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
2.4. SMP characterization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
3. Shape memory switches and triggers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
3.1. Phase versus molecular switches and thermal versus non-thermal triggers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
3.2. Direct/indirect trigger methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
4. Variants in shape memory forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
4.1. Shape memory surfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
Abbreviations: SCPs, shape changing polymers; LCEs, liquid crystalline elastomers; SMP, shape memory polymer; SME, shape memory effect; 2W,
two-way; 1W, one-way; Ttrans , transition temperature; Tg , glass transition temperature; Tm , melting temperature; Tcl , liquid crystal cleaning temperature;
Td , deformation temperature; Tf , shape fixing temperature; Tc , crystallization temperature; Rf , shape stability ratio; Rr , shape recovery ratio; Tsw , switching
temperature; εmax , maximum recoverable strain; max , maximum recovery stress; SMC, shape memory cycle; εload , strain under load; ε, fixed strain; Tr ,
recovery temperature; εrec , recovered strain; Vr , strain recovery rate; Tmax , temperature corresponding to ␴max ; Tsw,app , apparent switching temperature;
PCL, poly(␧-caprolactone); SMPU, shape memory polyurethane; EMU, elemental memory unit; TME, temperature memory effect; PU, polyurethane; Ti ,
liquid-crystal isotropic transition temperature; Tv , vitrification temperature; CIE, crystallization induced elongation; MIC, melting induced contraction.
∗ Corresponding author.
E-mail address: [email protected] (T. Xie).
http://dx.doi.org/10.1016/j.progpolymsci.2015.04.001
0079-6700/© 2015 Elsevier Ltd. All rights reserved.
80 Q. Zhao et al. / Progress in Polymer Science 49–50 (2015) 79–120
4.2. Shape memory hydrogels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92

4.3. Shape memory microparticles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
5. Multi-SMPs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
5.1. Intrinsic multi-SMPs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
5.1.1. General aspects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
5.1.2. Multi-SMPs via distinct thermal phase switches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
5.1.3. Multi-SMPs via molecular switches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
5.1.4. Tunable multi-SMPs via a single broad thermal phase switch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
5.2. Extrinsic multi-SMPs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
5.3. Reprocessible multi-SMPs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
6. Two-way SMPs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
6.1. Quasi 2W-SME for liquid crystalline elastomers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
6.2. Quasi 2W-SME for semi-crystalline networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
6.3. 2W-SME for semi-crystalline networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
7. Novel SMP fabrication methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
7.1. Electrospinning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
7.2. Four dimension printing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
7.3. Photo-induced shape transformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
7.4. Origami-inspired manufacturing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
8. Modeling of thermomechanical behavior of SMPs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
8.1. Thermoviscoelastic models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
8.2. Phase evolution models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
8.3. Models for predicting shape memory performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
9. Summary and outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
1. Introduction In contrast, shape memory polymers (SMPs) are a

class of stimuli-responsive SCPs for which shape shift-
Active shape change in response to environment condi- ing behavior can be programmed. This is the single most
tions is commonplace in nature [1]. In fact, this property is important feature that distinguishes SMPs from other
essential for the survival of animals and even some plants. stimuli-responsive SCPs. Specifically, a typical SMP can
Among numerous interesting examples are sunflowers that be programmed to fix one temporary shape and subse-
move with the sun to maximize light exposure, the open- quently recover to its permanent shape upon stimulation
ing and closing of Venus flytrap to capture preys, and the (typically heating), as illustrated in Fig. 1a. Here, the
microscopic shape changes in the skin layers of cattlefish temporary shape is usually defined by the applied force
that enables its macroscopic color change. during the shape fixing step. Accordingly, this step is
Polymer materials, in their early days, were mostly also called a programming step. Since the shape mem-
studied for use as static structural parts. In the modern ory cycle in Fig. 1a involved a total of two shapes (one
days, however, advanced polymer materials that exhibit temporary and one permanent), the associated behavior
special functions in response to external conditions have is called a dual-shape memory effect (dual-SME), repre-
attracted more and more attention. Such behavior is similar senting the simplest and most well known behavior for
to the biological intelligence observed in nature. Accord- SMPs.
ingly, such polymers are also called smart polymers, or From a molecular structure standpoint, the dual-SME
in a more scientific term, stimuli-responsive polymers. is enabled by the combination of a reversible switching
Whereas macroscopic changes in polymer functions are mechanism and a network structure [4]. Thus, the simplest
often accompanied by chain conformational changes (i.e., (and very common) polymer systems, such as a crosslinked
shape changes at the molecular scale), stimuli-responsive (chemically or physically) amorphous (or crystalline) poly-
shape changing polymers (SCPs) commonly refer to those mer would exhibit such a behavior [5–10]. Indeed, the
for which shape changes are either macroscopic or at dual-SME has been known for quite a long time and
least visible under microscopes. Examples of stimuli- is the basis of numerous technological innovations. Best
responsive SCPs are numerous, ranging from the well known examples of commercial applications include heat-
known swelling/deswelling of hydrogels [2] to the intrigu- shrinkable tubes used in cable industry, heat-shrinkable
ing reversible surface morphological changes of liquid labels used for packaging, and shrink-dink toys. Other
crystalline elastomers (LCEs) [3]. Of importance here is that, high value-added applications, mostly in the biomedi-
for the preceding two examples, after the materials are fab- cal [11–15] and aerospace fields [16–18], are also being
ricated, the shape change can only occur between a fixed actively explored. The vast potential to enable technologi-
number of equilibrium shapes (often two). That is, their cal innovations highly relevant for the human society (e.g.,
shape shifting cannot be manipulated externally after fab- health) has been the driving force for active research on
rication, a non-programmable shape changing behavior. SMPs.
Q. Zhao et al. / Progress in Polymer Science 49–50 (2015) 79–120 81
Fig. 1. Schematic illustrations of (a) Dual-shape memory effect; and (b) Triple-shape memory effect.
Broadly speaking, current SMP research can be classified subsequently recovering in a highly controllable manner.
into the following focus areas: Notably, this definition of multi-SMPs makes no distinction
between the so-called triple-SMP (Fig. 1b) and an SMP with
(1) exploration of new and exciting applications; more than two temporary shapes. It is critical to point out
(2) tuning SMP molecular structures to meet application that, in both Fig. 1a and b, the shape changing occurs only in
specific requirements; one direction, as the arrows indicated. That is, temperature
(3) alternative stimulation mechanisms that are relevant change alone would not drive the recovered shape(s) back
for the intended application; to the temporary shape(s). Importantly, this “irreversibil-
(4) new molecular designs beyond the simplest SMP sys- ity” should be distinguished from reprogramming in that
tem mentioned in the preceding; change in both temperature and external stress condition
(5) discovery of new shape memory behavior (or effects). would enable the switch from the recovered shape to the
temporary shape. Thus, both the dual- and multi-SMEs
Among these focus areas, the first one is perhaps most discussed in the preceding belong to the so-called one-way
important over the long term. This is simply due to the (or irreversible) shape memory effect (1W-SME). In con-
proven fact that any research field can only thrive with trast, a 2W-SMP can exhibit programmable and reversible
promises in impacting technological innovations. Within shape switching between two (or more) distinct shapes.
this big umbrella (focus area 1), focus areas 2–4 have been Here again, “programmable” is the key that distinguishes
the routine parts of SMP research. Such a statement does a 2W-SMP from many common examples of reversible
not discredit their importance. In fact, without 2 and 3, any SCPs. Giving the importance of the programmability, we
potential application ideas would not advance beyond the will elaborate it more further in the following.
conceptual stage. Focus area 4 is not just scientifically inter- It is noted here that many excellent reviews have been
esting, but could contribute to all the other focus areas. published by various groups active in SMP research. These
For instance, physical doping of crystalline small molecules reviews cover general aspects of SMPs [4–10], multi-
into an elastomer network opens up a simple way to tune functional SMPs [20], SMP composites [21–26], SMP foams
the transition temperature of SMPs [19]. The discovery of [27], SMP fibers [28–30], SMP characterization [31], and
new shape memory behavior (focus area 5), by contrast, SMPs for biomedical [12–15] and aerospace applications
happens very rarely. Yet, new shape memory behavior [16–18]. Readers interested in specific topics are referred to
could lead to new application opportunities in ways that these reviews accordingly. Although some general aspects
are much more drastic than the others. As such, there are of SMPs will be necessarily covered, the main objective of
two notable achievements during the past decade in SMPs: this review is to highlight notable progress made, mostly
multi-shape memory polymers (multi-SMPs) and two-way in the last five years, clarify some conceptual confusion
shape memory polymers (2W-SMPs). within the field, and point out unanswered questions
A multi-SMP refers to a polymer that is capable of (opportunities). The structure of the review is organized
“memorizing” more than one temporary shape and as following: general aspects of SMPs, shape memory
switches and triggers, variants in SMP forms, multi-SMPs, shape B and reprogrammed to switch between shape A to
2W-SMPs, SMP fabrication methods, and SMP modeling. shape B , in a reversible fashion [33]. This “programmabil-
ity” clearly distinguishes SMPs from other forms of SCP.
As an example of a non-programmable SCP, a thermo-
2. General aspects of SMPs
sensitive hydrogel such as poly(N-isopropylacrylamide)
can undergo expansion-shrinking shape transformation
2.1. Programming
(Fig. 2c) upon swelling and deswelling induced by an
environmental change [2,34]. Another example is the tem-
From the literal meaning of the term “shape mem-
perature induced surface morphological change of a liquid
ory polymer”, most researchers would agree that an
crystal elastomer (LCE) (Fig. 2d) [3]. In both cases, the shape
SMP should possess the ability to transform from any
shifting is between two fixed states defined in the mate-
deformed temporary shape back into a “memorized” shape
rial fabrication step. Adding a bit confusion to the issue
(i.e., permanent or original shape). This seemingly simple
is that although the two examples in Fig. 2c and d did
description, however, may cause confusion in the larger
not show programmability, hydrogels and LCEs can both
context of SCPs. As stated earlier, SMPs are a sub-class
be designed to introduce programmability, thus qualifying
within SCPs for which shape shifting is programmable.
them as SMPs. Examples of such can be found later in this
Thus, the fundamental issue lies in the definition of pro-
review.
gramming (or programmability). Programming refers to
It is worth pointing out that the term “program (or
the external and physical manipulation process that defines
programming)” has been used in a broader sense in the
the shape shifting pathway. This process should be inde-
context of SCPs. For instance, Hayward’s group developed
pendent of material synthesis (or fabrication). Using a
a half tone lithography method to introduce material het-
dual-SMP shown in Fig. 1a as an example, it is the thermo-
erogeneity into a hydrogel sheet that can undergo highly
mechanical shape fixing (programming) that determines
complex shape changing [35]. The term “program” could be
the temporary shape, thus the shape shifting pathway goes
used there to describe the lithography process that defines
from this particular shape to its original shape. Importantly,
the shape shifting. Since such programming is an integral
since programming is external and physical, for the same
part of the fabrication process, the material after fabrica-
material, countless temporary shapes (thus pathways) are
tion can only switch between fixed shapes, no matter how
possible by changing the programming conditions (e.g., the
sophisticated they are. Likewise, our group has recently
deformation force) (shown in Fig. 2a) [32]. Of course, within
reported a “Lego” hydrogel system for which shape-
each shape memory cycle, one has to pick one particular
changing behavior could be mechanically programmed by
shape shifting pathway for the SMP to perform, but differ-
prestretching hydrogel building blocks during the material
ent pathways can be implemented in other shape memory
assembly step [36]. But again, the “programming” occurs in
cycles. Similarly, for a 2W-SMP system, the same sample
the material fabrication step, fundamentally different from
(Fig. 2b) can be programmed to change from shape A and
Fig. 2. Classification of shape changing polymers: (a) One way shape memory polymers [32], Copyright 2009. Reproduced with permission from Elsevier
Ltd; (b) Two way shape memory polymers [33], Copyright 2013. Reproduced with permission from John Wiley and Sons; (c) Hydrogels [34], Copyright
2012. Reproduced with permission from Royal Society of Chemistry; (d) Liquid crystalline elastomers [3], Copyright 2006. Reproduced with permission
from the American Chemical Society.
Fig. 3. The molecular mechanism of the dual-SME. Black dots: netpoints; blue lines: molecular chains of low mobility below Ttrans ; red lines: molecular
chains of high mobility above Ttrans . (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this
article.)
the external and physical nature of the programming for polymer form such as a hydrogel or a composite [6,21]. As
SMPs. shall be seen, these variations from the simplest form of
With the clear definition of “programming”, SCPs can SMP provide the basic logic for this review.
be readily classified as programmable SCPs (i.e., SMPs) Traditionally, SMPs have been categorized into chemi-
and non-programmable SCPs (Fig. 2). We should note that cally (or physically) crosslinked glassy (or semi-crystalline)
Lendlein’s group uses the same definition for SMPs, how- polymers [4,5]. This classification emphasizes chemical
ever, they use “actively moving polymers” to describe the structures and morphologies that are more closely related
general class of SCPs used in our definition. This allows to the molecular designs of SMP. We note that such
them to reserve the term “SCPs” for the non-programmable a classification is only valid for conventional SMP rely-
SCPs in our definition. This difference in terminology is ing on thermal phase changes, called phase switches in
purely a matter of preference and, in our case, we simply this review. Increasingly, SMP enabling mechanisms called
elect to use the literal meaning of shape changing polymers molecular switches [9], originate from reversible molecu-
to describe SCPs. lar interactions instead of reversible phases are becoming
more popular. Examples are discussed extensively in a
dedicated section in this review. Accordingly, SMPs operat-
2.2. Classification
ing via various molecular switching mechanisms represent
nonconventional molecular design of SMP.
The simplest and also the most common SMP is a ther-
mally induced one-way dual-SMP in its neat polymer form.
Among the attributes, “thermally induced” refers to the 2.3. Basic molecular requirements and mechanism
external stimulation method (or so-called trigger) that is
most often employed to induce shape recovery. “One-way” Here we take the baseline case, i.e., a thermally induced
implies that the shape recovery is irreversible. That is, one-way dual-SMP with phase switches, for the discus-
shape shifting during recovery can only follow the path- sion of the basic molecular requirements and general
way from a temporary shape to the permanent shape, mechanism for SMPs. The mechanism for other more
but not vice versa. “Dual-shape” means that such a shape sophisticated forms of SMPs will be deduced accordingly
memory cycle involves only two shapes: a metastable and presented in the coming sections. Fig. 3 shows a typical
temporary shape and an equilibrium permanent shape. thermally induced one-way dual-shape memory cycle. At
Both the terms “one-way” and “dual-shape” refer to the room temperature, the SMP is typically a rigid plastic. Upon
shape shifting behavior of the materials. Accordingly, heating above its shape memory transition temperature
non-traditional SMPs include non-thermally induced SMP, Ttrans , the material becomes a soft rubber (or elastomer)
2W-SMPs, and multi-SMPs. Any combination of these three due to the increased mobility of the molecular chains.
defining attributes can be used to the categorize SMPs. Here, Ttrans can be a glass transition temperature Tg , a
An example of such is a non-thermally induced one-way melting temperature Tm , or less commonly liquid crystal
multi-SMP. In addition, an SMP may also be in a non-neat clearing temperature Tcl . Heating allows easy deformation
when an external force is applied. The associated elevated As long as the network is stable, any macroscopic defor-
temperature is called Td (deformation temperature). In an mation imposed on the permanent shape would change
ideal situation, this macroscopic deformation should cor- the conformation of the polymer chains and raise the
respond to changes in the molecular chain conformation, energy state of the materials. At this point, it is the cooling
or entropy. When cooled to a temperature below the Ttrans , induced depression of the molecular mobility that serves
the deformation imposed onto the SMP will be maintained as the kinetic trap to maintain the system in the high
even after the removal of the external force. At the molecu- energy state. Once this kinetic trap is removed by reactivat-
lar level, this is simply due to the freezing of the molecular ing the segmental mobility (e.g., by heating), the entropic
chain (i.e., reduction of chain mobility) that locks in the energy is released, driving the molecular chains back to
deformed chain conformation, or enables the storage of their lowest energy (highest entropy) state, thus the mate-
entropic energy in the system. The corresponding tempera- rials would accordingly recover the original shape. Here,
ture at this stage is called Tf (shape fixing temperature). This the netpoints are either chemical or physical crosslinking
temporary shape remains stable unless when it is heated points or phases. They are the mechanism that ensures
above the Ttrans again in the absence of external stress. storage and release of entropic energy that is responsible
Under this condition, the SMP will recover to its perma- for the shape fixing and recovery of an SMP. The netpoints
nent shape. Again at the molecular scale, heating activates should be sufficiently stable to withstand the thermo-
the molecular mobility, which releases the entropic energy, mechanical conditions encountered in the shape memory
thus driving the molecular chains back into its highest process without breakage; otherwise the permanent shape
entropic state corresponding to the permanent shape. would not fully recover. The extent of shape recovery is
The preceding discussion suggests that polymer shape thus a direct reflection of the stability of the network estab-
memory effect is an entropic phenomenon. From the lished by the netpoints under the relevant conditions.
molecular requirement standpoint, what is readily clear The netpoints can be made up of either chemical cova-
is the need for molecular chains for which mobility can lent bonds or physical interactions (e.g., crystallites, glassy
be activated and depressed, typically by a temperature hard domains [5], hydrogen bonding [37], ionic assem-
change. Given its switchable nature, we called these switch bly [38–40], and chain entanglements [41]). Whereas we
segments. Here, the term segments instead of chains are use the term “netpoints” here, it should be noted that
used to signify that the switching does not necessarily the netpoints are not necessarily points per se, but the
have to occur to the entire chain, but can happen at the term is used mostly for lack of better words. For chemi-
finer segmental level. In Fig. 3, the color change for all the cally crosslinked system, the network is indeed established
chains illustrates the mobility switching, but it is impor- by the atomic netpoints each of which connects more
tant to understand this is done for ease of demonstration. than two chains together. For many physically crosslinked
In reality, the switching may only happen within certain SMP systems such as thermoplastic polyurethane SMPs,
segments along the chain. A point that needs to be empha- the netpoints are microscopic phases or domains corre-
sized is that, whereas the schematics shown in Fig. 3 shows sponding to the hard segments. Interesting variants are
free space between chains, it is for ease of demonstra- the so-called nanoparticle tethered SMP systems with lin-
tion only. In reality, this type of free space (volume) is ear polymer chains grafted onto nanoparticles [37,42–44].
not necessary since chain conformation change typically For this type of systems, the nanoparticles alone do not
occurs in a highly coordinate manner to satisfy the spa- establish the necessary network since the other ends of
tial requirements. That is, the chains or segments that are the grafted chains remain free. When combined with
not responsible for conformation change would move at chemical crosslinks or physical entanglement, however, a
the system level to accommodate conformation changes network structure is formed with the nanoparticles serving
within the switching segments, even though they them- as the netpoints in conjunctions with chemical crosslinking
selves do not necessarily undergo conformation changes. points or entanglement. Here, of course, the nanoparti-
The segmental switching alone is not sufficient for poly- cle netpoints are neither atomic points nor microscopic
mer shape memory effect. Another mechanism has to be in phases. What is critical here is not the scale of the net-
place to ensure that any macroscopic deformation is the points, but that they establish a stable network within the
results of molecular chain conformation change, that is, polymer.
deformation should lead to an entropic change that is the Since the network (or netpoints) defines the permanent
necessary driving force for recovery. To satisfy this require- shape, often time the term “permanent” is also extended to
ment, the system has to possess a mechanism to ensure that describe the network. Such a “permanent” network is only
there is no slippage of entire polymer chains that would of relative meaning. That is, it is ideal that it is permanent
lead to macroscopic deformation but no entropic change. (non-breakable) within a particular shape memory cycle.
Such a mechanism is provided by the network established Under conditions different from shape memory cycles, the
by the black netpoints shown in Fig. 3. capability of reversibly forming the “permanent” netpoints
Overall, an SMP should meet two structural require- implies the SMPs can be re-processed into a new perma-
ments: one is the netpoints that defines the permanent nent shape. Physical permanent netpoints usually endow
shape and the other is the switch segments that can enable the polymers with reprocessability. However, many exam-
fixation of a temporary shape. The netpoints signify that ples of physical netpoints are not as robust as chemical
polymer chains within an SMP form a network structure. In netpoints. That is, they are more likely to undergo unin-
the permanent shape, the molecular chains within the SMP tended changes in a shape memory cycle, which would lead
network are at the lowest energy (highest entropy) state. to incomplete shape recovery. In contrast, the robustness
of most chemical netpoints is responsible for the superior standard variations inherent in this type of measurements.
shape recovery performance for those systems, but they are Traditionally, a εmax value much larger than that of shape
typically non-reprocessable. An interesting approach is the memory alloys has been perceived as one of the major
use of reversible covalent bonds such as Diels–Alder moi- advantages of SMPs. However, many SMP applications do
eties as the netpoints for SMPs, leading to reprocessable not require very high εmax . Despite such, it is important
SMPs with good shape recovery performance [45–47]. to note that SMPs with εmax above 800% are known in the
Typical switch segments are those related to reversible state of art [41,50].
thermal phase transitions (e.g., glass transition, melting- Precise measurement of Rf and Rr , determination of
crystallization transition, liquid crystal clearing transition). Tsw and maximum recovery stress max , and evaluation
Since most polymers possess either a glass transition or of shape recovery rates, are best accomplished by run-
a melting transition, these two transitions serve as the ning a quantitative shape memory cycle (SMC) using a
main backbones for designing SMP. Several differences thermo-mechanical analyzer. This type of test records pre-
between glassy SMP and crystalline SMP are notewor- cisely the time evolution of temperature, stress, and strain
thy: (1) Tunability of the transition temperature: Tg for a in a SMC. An example of a typical dual-SMC is shown
given SMP class is more readily tunable in a large tem- in Fig. 4a. Except its more quantitative nature, this dual-
perature range via the introduction of comonomers. This SMC is identical to the qualitative demonstration shown in
has been demonstrated nicely for epoxy SMP [32,48] and Fig. 1a. The sample is first heated to a deformation temper-
meth(acrylate) SMP [49] systems. In contrast, tuning a ature Td , normally (but not always) above Ttrans , a force is
crystallization temperature in a similar fashion is much then applied as reflected in the stress curve. Upon subse-
harder as the introduction of too much comonomers would quent cooling under the load to Tf , usually below Ttrans , the
destroy the crystallizability of the system all together; (2) sample reaches the so-called strain under load εload . The
Shape memory cycle time: since crystallization is a much stress is then removed with the sample strain responding
slower process than vertrification, a Tm -type SMP may accordingly to a fixed strainε. This completes the fixing (or
require longer time for fixation; (3) Shape memory tem- programming step). Under a stress free condition, recovery
perature range: there is often a ca. 20 degree difference is accomplished by simply heat the sample to a recovery
between Tm and Tc (crystallization temperature), the tem- temperature Tr , leading to a recovered strain εrec . Rf and Rr
perature range required for a Tm -type SMP is often time can be readily calculated y using the following equations:
larger; and (4) Diversity of the shape memory behavior: ε
as will be discussed later in more detail, a Tm -type SMP, Rf = 100% × (1)
εload
along with a liquid crystalline-type SMP, is capable of a
2W-SME. This is a significant advantage over Tg -type SMPs. ε − εrec
Rr = 100% × (2)
Besides the two general types of thermal reversible mech- ε
anisms, other switch mechanisms have also been explored. Alternatively, an SMC may also be presented in a 3D
The alternative switch mechanisms, while less general diagram shown as Fig. 4b using the same set of data. The
for polymers, are important enablers for non-traditional three axes here are temperature, stress, and stress. Notably
stimulation methods. This will be elaborated further in a absent is the time information. This absence does not pre-
dedicated section. clude the determination of Rf and Rr . As such, the use of this
type of 3D shape memory diagram is quite popular in the
2.4. SMP characterization literature. We caution, however, that the loss of time infor-
mation may be quite detrimental for more sophisticated
Performance parameters for an SMP are diverse; with SMCs. For instance, if the sample is, at any stage, annealed
the most essential ones being the shape memory transi- under a constant stress, the annealing time and whether or
tion temperature Ttrans , shape stability ratio Rf and shape not the strain reaches equilibrium during annealing would
recovery ratio Rr . Depending on the circumstance (e.g., not be known.
the intended applications), maximum recoverable strain The strain curve in the recovery portion of the SMC
εmax and maximum recovery stress max may also be rele- shown in Fig. 4a can be used to calculate instantaneous
vant. When the recovery speed is of concern, average and strain recovery rate Vr , the time derivative of the strain
instantaneous recovery rates can also be calculated. The as defined in Eq. (3). Plotting Vr against temperature as is
switching temperature Tsw , closely related to, but different shown in Fig. 5a, one can obtain Tsw , defined as the tem-
from, Ttrans is also sometimes reported. perature corresponding to the maximum Vr .
Here, Ttrans is commonly equal to Tg for an amorphous
∂ε
SMP or Tm for a crystalline SMP, that can be measured Vr = × 100% (3)
using standard thermal analysis methods such as DSC or ∂t
DMA. Rf and Rr can be conveniently determined by manu- We note that the SMC in Fig. 4a is obtained under
ally measuring sample dimension changes by performing a stress controlled condition in a tensile mode. This
a shape memory cycle using a thermal oven according to is a typical way for constructing an SMC. The type of
the steps outlined in Fig. 3. Maximum recoverable strain deformation mode may vary and is not critical in most
εmax , on the other hand, may be best evaluated using a cases. For instance, a bending mode may also be employed
mechanical tester equipped with a thermal chamber by particularly for samples for which εmax is relatively small.
elongating the sample to its failure at the relevant tempera- Under such a circumstance, the bending angles at the
ture. Multiple measurements are necessary due to the large corresponding stages are used in place of tensile strains
Fig. 4. Quantitative dual-shape memory cycles: (a) 2D diagram [163], Copyright 2013. Reproduced with permission from Nature Publishing Group; and
(b) 3D diagram [5], Copyright 2010. Reprinted with permission from John Wiley and Sons.
Fig. 5. Consecutive 1W-SMCs of crosslinked EVA with: (a) A low gel fraction; and (b) A high gel fraction. Solid line: strain, dash line: temperature, dot line:
stress [51], Copyright 2011. Reproduced with permission from Elsevier Ltd.
to calculate Rf and Rr in a similar fashion. Alternative to comparisons for both samples reveals important differ-
the stress controlled condition, the SMC may also be run ence, that is, there is a notable cycle to cycle strain shift
under a strain controlled condition. That is, during the pro- for sample A, whereas such a shift is notably absent for
gramming stage, the tester imposes a certain strain onto sample B. Clearly, this difference in cycling performance is
the sample while the stress changes accordingly. Another critical for device applications involving multiple cycles.
variation lies in the recovery step. Whereas the recovery Other performance parameters or variations to the SMC
in Fig. 4a is run under zero-stress for strain recovery (i.e., conditions are possible, particularly for shape memory
stress-free strain recovery), it may also be operated under experiments that are more sophisticated than the simple
an iso-strain condition. Under such a condition, the tester dual-SMC in Fig. 4a. While it is impossible to discuss all the
holds the strain constant, but the tendency of the sample possibilities, one only needs to keep in mind that the princi-
to recover generates a stress called a recovery stress . ples provided in the preceding remain the same, or at least
This is called an iso-strain stress recovery, run to obtain similar. Here, proper selection of SMC conditions requires
the max value. The temperature corresponding to max is good understanding of polymer shape memory behavior
called Tmax , in a way similar to Tsw . and what performance parameters a researcher intends to
Besides a single cycle evaluation, often time it is also evaluate. In the following section, we shift our focus to dis-
necessary to run a cycling experiment to probe the robust- cuss the switching and triggering mechanisms that enable
ness of the shape memory performance over multiple polymer shape memory behavior.
consecutive cycles. This type of cycling experiment is
particularly relevant for SMPs intended for device applica- 3. Shape memory switches and triggers
tions. The essence and necessity of the cycling experiment
are illustrated in Fig. 5. Fig. 5a and b correspond, respec- 3.1. Phase versus molecular switches and thermal versus
tively, to two crosslinked EVA (ethylene-vinyl acetate non-thermal triggers
copolymer) samples [51]. The only difference between the
two samples lies in the degree of crosslinking, thus the gel Early discussion on the molecular origin of polymer
fraction. Despite the low gel fraction (48%) for sample A shape memory effect revealed that a conventional SMP
in Fig. 5a, its Rf and Rr are about 95% as determined from relies on a reversible thermal phase transition to switch
the first SMC. Although sample B in Fig. 5b (gel fraction of off/on the molecular mobility so as to freeze/release the
76%) shows higher Rf and Rr values (98% for both), such temporary shape. Naturally, the associated SME is typi-
a difference may appear trivial. However, cycle to cycle cally thermally activated via direct or indirect heating. One
notable exception is the water triggered SMP in which searching the library of stimuli-reversible bonds, selecting
water molecules plastize the amorphous phase, thus reduc- a desired one as the molecular switch and synthesizing
ing the Tg to a level below the ambient temperature to a permanent network containing the reversible bonds.
trigger the recovery [52]. The typical need for tempera- Possible molecular switches may include reversible
ture changes to realize an SME can be fulfilled in many covalent bonds (e.g., disulfide bonds, Diels–Alder adduct,
SMP applications, but this also represents a hurdle to real- and acylhydrazone bonds) and strong non-covalent
ize its full practical potential for instance in situations for physical interactions (e.g., multiple hydrogen bonding,
which heating is not feasible. On this point, new switching metal–ligand coordination, and supramolecular recogni-
mechanisms that do not rely on thermal phase transitions tion). Reviews on reversible chemical and physical bonds
may offer attractive solutions. Thanks to the development are particularly helpful [70–74]. It is important to note that
of reversible covalent and non-covalent chemistry, oppor- the target network should possess sufficient chain mobility
tunities for designing switch segments at the molecular to facilitate the reversible chemical reactions (or physical
level (i.e., molecular switches) have emerged. While phase interactions). Here, the reversible interactions simply form
transitions are by and large related to a thermal effect, temporary crosslinking points while a set of permanent
molecular switches bring more options in terms of trig- crosslinks should be in place to memorize the permanent
gering mechanisms (e.g., thermal, light, redox, and pH). shape. Selective molecular switches are presented in Fig. 6.
A prime example of a non-thermal trigger was reported Several examples of SMP following the preceding design
in 2005 by Lendlein et al. [53]. Photo-reversible [2 + 2] principle are presented in the following. A disulfide bond
cycloaddition of cinnamate (or coumarin) groups were is a well known reversible covalent bond. To utilize such,
introduced into a polymer as the switch mechanism. A thiol units were introduced as side groups onto cellulose
deformed sample was exposed to UV light ( > 260 nm) acetate chains [75]. At the oxidized state, most of the thiol
to promote cycloaddition of the cinnamate groups to units tuned into disulfide crosslinks, enabling temporary
form crosslinks to lock in the temporary shape. Subse- shape fixing. Recovery was triggered by reducing the disul-
quently, shape recovery could be triggered upon irradiation fide bonds into thiol units. Multiple fixing/recovery cycles
of UV with < 260 nm, as the result of the cleavage of could be realized. However, the SME decayed progressively
the crosslinks. The non-thermal triggering method has as the result of redox side reactions. Moreover, the shape
attracted attention, but some practical issues have kept memory cycle time is very long (four days) due to the low
it from becoming popular. The cinnamate-based SMP reaction rate.
showed limited Rf (30–50%) and slow respond speed (∼1 h). Strong physical interactions can also serve as the
Subsequent efforts by other groups have improved the enabling molecular switches. For instance, ureidopyrim-
shape memory performance to a certain extent [54–56]. idinone (UPy) units form self-complementary hydrogen
By contrast, Lee and coworkers [57–59] reported bonding (SCHB) interactions that are much stronger than
a photo-responsive SMP system with an entirely dif- typical hydrogen bonds [76–79]. The SCHB interactions
ferent operating mechanism. Instead of relying on possess a dynamic equilibrium feature that had been
photo-reversible covalent chemistry, isomerization of explored as molecular switches. In such an example, UPy
azobenzene was explored as the switching mechanism. units were introduced as the side groups onto a chemically
This particular SMP is a crosslinked glassy liquid crystalline crosslinked butylacrylate based elastomer network [77].
polymer containing azobenzene moieties. The sample At a low temperature, two UPy units trended to strongly
could be deformed and fixed when exposed to visible associate, establishing the temporary netpoints necessary
light with wavelength of 442 nm. The authors observed to fix a deformed shape. Such a strong association corre-
that the fixed shape was stable for more than a year. sponds to a very slow exchange between the associated
Particularly intriguing was the fact that light at 442 nm UPy units. While at elevated temperatures, the equilibrium
induced repeated trans–cis and cis–trans photoisomeriza- would shift towards the dissociated states (fast exchange
tion, unlike typically used 365 nm light that would only mode), allowing recovery of the fixed shape. Perhaps due
cause the trans- to cis-isomerization. As the authors spec- to the dynamic nature of the association, the deformation
ulated, this repeated isomerization induced adjustments fixed by the SCHB at low temperatures would sponta-
to the polymer chains in the glassy matrix, analogous neously recover by nearly 20% over one hour. Guan’s group
to the typical thermal fixation. When the fixed sample also reported a thermoplastic SMP containing cyclic UPy
was exposed to circularly polarized light of the same dimmers [80]. While the molecular design is quite dif-
wavelength, photoreconfigured polymer network was ran- ferent and in fact intriguing (i.e., the lack of permanent
domized, leading to recovery. crosslinks), the material also shows a similar shape mem-
We note that light penetration deep into the SMP bulk ory function and relaxation behavior. In contrast, the strain
is intrinsically difficult, limiting the effectiveness of light fixed by the reversible cinnamate cycloaddition exhibited
triggering to thin films and/or surfaces. On the positive side, hardly any relaxation over four hours in Lendlein’s study
many SMP applications that utilize only the surface shape [53]. As such, there are some small yet notable distinc-
memory effect have emerged recently [60–69]. It is thus tions between these two types of molecular switches. For
foreseeable that the preceding SMP systems may find its the dynamic UPy switches, the crosslinks would break and
niche due to the spatio-selectivity of the light triggering reform at any time (i.e., at equilibrium). An external stimu-
method. lation only changes the kinetics. In contrast, the cinnamate
Expanding the concept of molecular switches further, molecular switches, while reversible, are non-dynamic (at
one can in principle design a variety of new SMPs by least within the relevant timescale) and can only break or
Fig. 6. Selective examples of molecular switches: (a) UPy coupling; (b) Metal–ligand coordination bonds [73], Copyright 2011. Reproduced with permission
from Nature Publishing Group; (c) Coumarin [2 + 2] cycloaddition; (d) Disulfide bonds.
reform under two specific conditions (UV lights of different kinetics between the metal ions and the ligands. Similar
wavelengths). This comparison implies that, an SMP based to the UPy based SMP, Rowan’s SMP also showed 10–15%
on dynamic molecular switches may intrinsically exhibit strain relaxation one hour after the initial unloading. Again,
the undesirable strain relaxation behavior; while such a this behavior is most likely due to the dynamic nature of
behavior may not be expected for non-dynamic molecular the metal–chelating ligands.
switches. This particular point is somewhat speculative in The concept of molecular switches is not limited to
nature and further studies are required to go deep into the the molecular interactions within the polymer matrix. A
operating mechanisms of reversible molecular switches. recent collaboration between Weder’s group and Rowan’s
In another study, Rowan’s group synthesized a group led to a water triggered SMP composite system
poly(butadiene) permanent network with metal–ligand in which the molecular switching mechanism was the
coordination bonds as molecular switches [81]. Owing to hydrogen bonding interactions between the polymer
the multi-responsiveness of the metal–ligand complex, matrix and the nanocrystalline cellulose fillers [82,83]. In
shape recovery can be triggered by various methods includ- this case, water molecules diffused into the polymer-filler
ing heating, UV irradiation, and specific chemical exposure interface to dissociate the interfacial hydrogen bonding.
that led to either decomplexation or increase the exchange At this state, a temporary shape was created and fixed by
removal of water (thus reforming the interfacial hydrogen these SMP composites can be triggered by Joules heating.
bonding) via drying. When the deformed sample was Via induction heating, alternating magnetic fields can also
immersed in water, the interfacial hydrogen bonding was induce internal heating within SMP composites containing
again dissociated, leading to the recovery. Here the water magnetic nanoparticles (Fe2 O3 , Fe3 O4 or Ni, etc.) [86–89].
triggered recovery behavior is identical to Huang’s system The so-called photo-thermal effect has also been found
mentioned earlier, but the underlying mechanism is quite effective by irradiating light of certain wavelength onto
different. The former relies on water sensitive molecular SMP composites containing carbon or gold nanofillers
switches whereas the latter operates by the water reduced [90–94].
phase transition temperature. High intensity focused ultrasound (HIFU) was applied
The preceding SMPs are all solids. The concept of to remotely and locally heat SMPs [95]. HIFU, that can be
molecular switches can be expanded to shape-memory concentrated onto a focal spot, was originally developed to
hydrogels. In an aqueous environment, the choice of treat tumors due to its thermal effect. The acoustic energy
reversible interactions is even more readily available. How- can convert into mechanical energy in the form of vis-
ever, we leave this discussion to a dedicated section on cous shearing oscillation, which would subsequently be
shape-memory hydrogels (Section 4). absorbed by an SMP in the form of temperature increases.
In comparison to the phase switches, Rf of an SMP Although the HIFU heating usually requires a liquid trans-
enabled by molecular switches is typically lower, at least fer media to ensure the generation of sufficient thermal
for all the reported examples. From the molecular stand- energy, the ability to remotely heat a neat SMP is a unique
point, a phase transition can freeze a very large portion advantage compared to the other indirect heating methods
(if not all) of the chain segments in a network, whereas that can only heat SMP composites.
molecular switches can only add a quite limited number The concept of utilizing indirect heating has signif-
of temporary netpoints to fix the deformed chains. A per- icantly expanded the scope of the recovery triggering
haps related macroscopic phenomenon refers to the elastic methods. Unlike the non-thermal triggering methods
modulus change of an SMP between its activated state and enabled by the various molecular switches discussed in
the deactivated state. For a conventional Tg (or Tm )-type the last section, indirect heating rely on the same phase
SMP, its elastic modulus at temperatures below its Ttrans switches behind the conventional direct heating triggered
(deactivated state) is usually 3–4 orders larger than that at SME. However, indirect heating triggers provide unparallel
temperatures above its Ttrans (activated state). In contrast, technological advantages due to their highly controllable
typical molecular switches are only able to enlarge the nature (i.e., remote, local, and momentary). Electromag-
modulus by at most 10 times due to the additionally formed netic wave, infrared light, and ultrasound can all penetrate
crosslinking points. Despite the relatively low Rf , molec- through human skin. When an SMP is embedded into a
ular switches have greatly expanded the approaches for human body as a biomedical device, the remote heating
achieving alternative triggering mechanisms. This may be process can locally increase the temperature of the device.
particularly helpful in the exploration of applications that Due to the low heat transfer coefficient of polymers, it
require isothermal SMEs. Because of this unique advantage, is possible not to spread too much of the heat into the
one can envision that more examples of SMPs with novel surrounding body environment. Moreover, endogenous
and attractive triggering methods will emerge in the near heating is usually much faster than exogenous heating due
future. to its shorter heat diffusion length, important for polymers
for which thermal conductivity is typically low. Of course,
3.2. Direct/indirect trigger methods the ability to instantaneously turn the heating on/off in an
on-demand fashion is also beneficial.
Despite a few available molecular switches, most SMPs The endogenous heating could be coupled with exoge-
are based on reversible thermal phase transition that is nous heating to affect the apparent switching temperature
intrinsically triggered by heating. The most common and Tsw,app of an SMP [96],defined as the actual environmental
convenient way to heat a material is directly increasing temperature at which the recovery occurs. In this instance,
its surrounding temperature using hot gas or liquid due to the extra heat provided by the endogenous induction
(i.e., exogenous heating). However, such direct heating heating, the SMP can be triggered at an environment tem-
methods are not applicable in many cases encountered in perature that is lower than that without the endogenous
applications. For example, when SMPs are used as mor- heating. The extent of endogenous heat can be conve-
phing devices in aerospace applications, it is impossible to niently controlled by adjusting the endogenous heating
heat the devices by increasing the outer space temperature power. As a result, Tsw,app can be tuned without changing
or blowing hot air onto the device. One solution to this is the actual material and its intrinsic Ttrans .
endogenous heating, namely, generating heat from inside Instead of heating, reducing Ttrans of an SMP to a
the material. Electrical, magnetic, optical, acoustic and temperature below the environmental temperature can
chemical energies can all be converted into thermal energy also induce a recovery process. In 2005, Huang’s group
and cause heating within materials. The endogenous heat- discovered that Tg of some commercial polyurethane SMPs
ing normally requires the incorporation of functional fillers could be greatly reduced using water as the plasticizer [52].
into an SMP matrix to achieve the energy transformation. When the SMP was immersed in water, water molecules
Conductive fillers such as carbon fibers, carbon nanotubes, penetrated into the SMP bulk, broke the hydrogen bonding
carbon nanofibers, metal powders, etc. were incorporated within the polyurethane segments, and consequently
into SMPs [84,85]. When electric currents pass through, increased the chain mobility. As a result, Tg of the SMP
could be decreased below the ambient temperature, extended the application potential for SMPs. For examples,
thus the recovery could be triggered without heating. a porous SMP allows extremely large volume changes rela-
In addition to polyurethane, SMP composites such as tive to SMPs in its solid dense form. Since porous SMPs have
polylactide/cellulose composite [97] and polyvinyl alco- been extensively reviewed in the literature, we hereafter
hol/graphene oxide composite [98] were also reported to concentrate on shape-memory surfaces, shape-memory
show water triggered SME. In both cases, water molecules hydrogels, and shape-memory microparticles.
destroyed the hydrogen bonding at the polymer-filler
interfaces to trigger recovery. The water-induced SME can 4.1. Shape memory surfaces
be also extended to Tm -type SMP that contains hydrophilic
crystalline poly(ethylene glycol) chains as the switches Surface microstructures on polymers govern a vari-
[99]. The shape recovery occurred when water molecules ety of material properties such as adhesion, wetting,
reduced the Tm of poly(ethylene glycol) to sub-ambient and friction, to name just a few. Consequently, fabri-
temperature. Although the water-induced SME achieved cating and controlling microstructures on polymers are
through reducing Ttrans cannot be universally applied, it critical for numerous technologies. Extending the SME
is particularly relevant for biomedical applications due effect on the macroscopic bulk scale to microstruc-
to the natural presence of water in a human body and tured surfaces thus offers great potential in terms of
the difficulty to apply direct heating in those instances. technological innovations. We note here that the molec-
An issue to consider here is that water induced recovery ular principle of shape memory surfaces is not different
may take long times (e.g., days) due to the slow diffusion from bulk SMP. The mere difference lies in the form
process. However, a recent effort by Mather’s group led to of deformation (surface versus bulk). However, as shall
an extremely fast (1.3 s) water responsive SMP by greatly be illustrated below, this difference is non-trivial as it
reducing the diffusion length using microfibers to con- brings significantly more opportunities in terms of tech-
struct the SMP [100]. Another issue of the water-induced nological innovations. Overall, there are three scenario in
SME is that the shape recovery process would always be which a surface SME is particularly relevant: (1) Creating
accompanied with a certain volume expansion of the SMPs erasable temporary surface microstructures; (2) Utiliz-
due to the water absorption. This may or may not be an ing SMEs as a way for microfabrication of permanent
issue depending on the targeted application, but is a factor surface microstructures; (3) Reversible tuning of perma-
to keep in mind nevertheless. nent surface microstructures fabricated using conventional
Following the same principle of the water-induced SME, microfabrication methods.
organic molecules such as DMF [101] and ethanol [102] Erasable temporary microstructures can be introduced
can also be used to induce the SME. Unlike water nat- on a featureless surface by micro-deformation. Nelson et al.
urally present in many environments SMP applications [104] demonstrated an SME at a nanoscopic scale. In their
encounter, organic solvents do not naturally exist in mostly work, nano-indents with depth less than 30 nm were cre-
any cases. In fact, the use of organic solvents is quite unde- ated on a flat surface by pressing an AFM tip onto a
sirable for most applications. Thus, it is not clear what thermoset SMP surface. The imprinted feature can fully
practical benefits such a triggering method may offer. recover upon heating. More important than the experi-
A particularly intriguing example is to use light to mental demonstration itself is that, despite the microscopic
reduce the Ttrans of an SMP. In this example, spiropyran heterogeneity inherent to a thermoset network, the SME
was doped into an ethylene-vinyl acetate copolymer (EVA) appeared to remain valid at the nanoscopic scale. Whereas
with a crystalline phase as the netpoints and a glass tran- Nelson’s work demonstrated the material capability, Alte-
sition as the switching phase [103]. Upon UV irradiation baeumer et al. [105] extended this concept one step further.
at 365 nm, spiropyran molecules isomerize to change from Relying on a massively parallel operation of scanning
a closed spiro form to an open zwitterionic merocyanine probes, thousands of nanoimprints were created onto a
form. The latter plasticized the amorphous phase to trigger shape memory surface, this was viewed as a data writing
the recovery. Thus, a UV-induced SME was realized. Such a process. These data (nanoimprints) can be erased and new
mechanism is rather conducive to develop new SMEs. data can be rewritten onto the surface by virtue of the SME.
Intriguingly, the surface could sustain thousands of writ-
4. Variants in shape memory forms ing/erasing cycles without noticeable fatigue, implying the
possibility of rewritable data storage. The surface SME is
In the early days, SMPs were known to exist mostly in not just limited to the conventional dual-SMPs. Utilizing
the form of dense bulk polymers. Increasingly, SMP in other the multi-SME, discussed in more detail later in this review,
forms were developed in efforts to meet various require- three sets of micro-patterns could be imprinted onto a
ments in diverse potential applications. On this front, multi-SMP at three different temperatures [106]. Inter-
shape-memory surfaces and shape-memory microparti- estingly, the individual sets of surface patterns could be
cles that exhibit SMEs at the microscopic scales have independently recovered at three recovery temperatures.
attracted increasing attention. In addition, porous SMPs This demonstration has the potential to greatly expand the
and shape-memory hydrogels have been developed to real- scope of the surface SME.
ize additional functions not offered by solid dense SMPs. SMPs have also found unique uses as a nonconventional
The underlying molecular principle of the various forms microfabrication tool. An example of this is the use of an
of SMPs is identical to that of dense bulk SMPs. However, SMP to create surface nano-wrinkles. Surface wrinkles are
the diverse forms have greatly improved properties and often created using a rigid thin film (e.g., metal) supported
on a soft thick substrate. Due to the modulus mismatch and the QR code became readable again. The authors
between the two layers, deformation to the substrate may anticipated that tagging products based on this demon-
lead to lateral compression to the rigid thin film. This stration could be useful for one-time secure identification.
would lead to mechanical instability and the formation of Cell growth is known to be sensitive to the topographic
nanoscopic wrinkles. Typically a soft rubber is used as the features of the substrate. Typically, only static surface
wrinkle substrate. Because the deformation of rubber can topographic conditions can be realized. Recent work by
only be conveniently created on a global level, it is difficult several groups has demonstrated that, when a suitable
to realize spatio-control of wrinkle formation. In contrast, SMP is applied as the cell culture substrate, the substrate
the surface deformation of an SMP can be easily controlled can be programmed to change its topography under cell
at any localized fashion via indentation. Following this culture conditions. As such, different cell behavior can
principle, we demonstrated a new way to create spatially be observed in response to the surface micro-structure
localized wrinkle (i.e., surface nano-structures) utilizing changing [111–114]. This ability to control the cell growth
an SMP as the wrinkle substrate [60,61]. This principle behavior via switchable surface topography adds a new
can be potentially expanded to create other types of “weapon to the arsenal”.
micro-structures and the methodology may open up new Switchable surface topography enabled by an SMP can
opportunities for microfabrication. also be used to control surface wetting. Via replica mold-
Surface microstructures fabricated on an SMP may ing, Yang’s group fabricated high aspect ratio pillar arrays
be reversibly tuned after they are formed. Permanent on an SMP surface (Fig. 7c) [115]. When the pillars were
microstructures can be directly fabricated onto an SMP deformed into a bended state, water droplet on the sur-
surface using traditional microfabrication methods. Any face exhibited a Wenzel state wetting with an anisotropic
deformation occurs at the macroscopic level, easy to intro- droplet shape, depending on the spacing between the pil-
duce, translates directly into changes to the microstructure. lars (Fig. 7d). Upon recovery of the pillars to the original
An excellent example along this line is the use of an SMP to straight state, the droplet displayed a Cassie-Baxter state
enable so-called programmable soft lithography [107]. In wetting with a large contact angle and a small sliding angle.
soft lithography, micro-molding is often time used. Typ- Such reconfigurable wettability may be potentially useful
ically, each elastomeric master mold can only be used for water collection.
to create one set of microstructures, the corresponding Micro-optics, which relies on surface microstructures
negative structures. However, when an SMP is used in to control the optical properties, plays an important role
place of the elastomer mold, multiple master molds may in modern technologies. Conventional micro-optics is fab-
be created via macroscopic deformation of the original ricated on non-deformable materials such as glass. Once
mold as schematically shown in Fig. 7a. This technique fabricated, the surface structure, thus the corresponding
introduces flexibility for nanoimprint lithography or other optical property, cannot be altered. In a recent collabora-
micro-molding methods. Similarly, surface micro-features tion the concept of shape memory surfaces was extended
created by other methods such as a Breath Figure method to a new generation of deformable and programmable
can also be altered via the use of an SMP supporting sub- micro-optics [69]. Via replica molding, the optically rel-
strate [108]. evant surface microstructures were introduced onto an
Tuning surface microstructure via an SMP can also SMP. Macroscopic deformation via stretching or compres-
be utilized to fabricate multifunctional material systems sion led to changes in the surface microstructure, thus the
(or devices). One strategy has been to couple surface dry related optical properties (e.g., light transmission) (Fig. 8).
adhesion property with an SMP to make reversible dry This concept is not limited to any particular micro-optics,
adhesives [62–68], e.g., molding micro-tips onto an SMP but can be extended to a variety of optical devices includ-
[68]. Despite the adhesion properties of the particular SMP ing retro-reflective surfaces, holograms, grating surfaces,
used, the sharp micro-tips on its surface prohibit physical and micro-lenses. In particular, the demonstration on holo-
contact with potential bonding substrates. As a result, the grams showed that the surface SME remains valid at a scale
micro-tipped SMP is in a non-adhesive state (Fig. 7b, left). below 10 nm.
However, the surface micro-tips can be deformed via the Most of the concepts involving shape memory surfaces
SME, thus achieving conformal contact and bonding with employ direct heating methods. In principle, all other forms
the substrate (Fig. 7b, right). Upon heating, the micro-tips of external stimulation methods can be readily applied
returned to their protruded state, causing debonding. Such [114,116]. Of particular interests are those that offer spatio-
an SMP reversible adhesive possesses great potential for selectivity. They would permit valuable localized property
use in transfer printing, a key step in the manufacture manipulation, as can be imagined for the examples shown
of flexible electronics that is gaining increasing attention in the preceding. Whereas light stimulation is an obvious
[109]. choice, an electronically programmed recovery is also an
Pretsch et al. [110] demonstrated a concept using an interesting alternative. This new form of stimulation relies
SMP as a switchable information carrier. The surface of an on a micro-heater array installed on the backside of a shape
SMP was first colored with a dye. Subsequently, the col- memory surface [69]. The resistive heating through the
ored surface was patterned with machine-readable quick micro-heater arrays can be controlled locally by electronic
response (QR) codes via laser ablation. Through shape programs. Notably, the sequence of the spatio-recovery of
programming, the tag prototype was deformed into tem- the surface microstructures (and the associated functions)
porary shapes with non-decodable QR code information. can be pre-defined by computer programs, a capability not
Upon heating, the original surface features were recovered offered by any other means of stimulation.
Fig. 7. Examples of shape memory surfaces: (a) Schematics of the programmable soft lithography [107], Copyright 2011. Reproduced with permission from
the American Chemical Society; (b) SEM photos of the reversible dry adhesive (scale bar = 50 mm) [68], Copyright 2013. Reproduced with permission from
the American Chemical Society; (c) Schematics of the fabrication and deformation process of the shape memory pillar array; and (d) Anisotropic wetting on
the deformed sample (c), from left to right: top-view of the deformed pillar array; water droplet viewed perpendicularly to the deformed pillars; droplet
viewed in parallel to the deformed pillars; the droplet on the original surface [115], Copyright 2013. Reproduced with permission from John Wiley and
Sons.
4.2. Shape memory hydrogels of acrylic acid, stearyl acrylate, and methylenebisacry-
lamide (the crosslinker). Crystalline/melting transition of
Stimuli-sensitive hydrogels can change their shapes in the stearyl side chains (Tm = 50 ◦ C) serves as the switching
response to external stimuli such as temperature, pH, light, segments and the crosslinker sets the permanent shape.
electric signal, or chemicals [2]. A typical stimuli-sensitive As such, a temporary shape can be fixed upon deformation
hydrogel is a crosslinked water-swollen network with via cooling from above its melting temperature to room
environment sensitive groups. Usually, these groups can temperature. This, and other early examples of shape mem-
change their affinity with water rather than form additional ory hydrogels rely on thermal reversible phase changes as
crosslinking points upon stimulation. Thus, most stimuli- the phase switches and their recovery is typically induced
sensitive hydrogels show only non-programmable shape thermally [121–124]. As has been laid out in an earlier
transformation between swollen states and shrunken section, the concept of molecular switches has unparal-
states. Nevertheless, stimuli-sensitive hydrogels have been lel advantages in designing solid SMPs. The presence of
extensively studied for drug delivery, cell culture, fluid con- water in hydrogels, on the other hand, creates even more
trol, separation, and sensing applications not only because opportunities for designing shape memory hydrogels using
of their stimuli sensitivity, but also their unique softness, molecular switches. Theoretically, the same principle can
high hydrophilicity, biocompatibility, and ability for rapid be applied to organo-gels although not much has been
diffusion of molecules [117,118]. reported in the literature. Examples of designing shape
The programmable nature of shape memory hydrogels memory hydrogels using molecular switches are presented
set them apart from typical stimuli-sensitive hydrogels. below.
The SME of hydrogels was first demonstrated by Osada Hydrophobic interactions, unique in an aqueous envi-
and coworkers in the mid-1990s [119,120]. A thermally ronment, have been demonstrated as a switch mechanism
triggered shape memory hydrogel was formed by water for shape memory. Inomata et al. [125] synthesized
swelling of a polymer network obtained by copolymerizing polypeptide hydrogel networks with hydrophobic alkyl
Fig. 8. Schematics of the programmable retro-reflective surface: SEM images of (a) Compressed (50% strain) and (b) Recovered microprism array; (c)
Schematic illustration of the total reflection effect of the microprism array; (d) Transmittance spectra of the shape memory film with microprism arrays
under a shape memory cycle; (e) Optical photographs showing the optical response of (c) [69], Copyright 2013. Reproduced with permission from John
Wiley and Sons.
side chains. At low molar fractions, the side chains form complex formed between ferric ions (Fe3+ ) and phosphate
temperature dependent molecular segregations. A tempo- groups was effective in fixing a temporary shape (Fig. 9a)
rary shape was introduced at a high temperature associated [127]. Upon reduction of Fe3+ to ferrous ions (Fe2+ ), the
with weakened segregation and fixed upon cooling due original shape would be recovered due to the disassoci-
to the formation of stronger segregation. This tempo- ation of the ion-phosphate physical crosslinking points.
rary shape recovered gradually upon heating, due to the Similarly, the complex between copper ions and pyri-
re-weakening of the segregation. In contrast, Huang and dine ligands was also applied as molecular switches to
coworkers [126] developed an intriguing shape mem- enable a redox-induced SME since temporary crosslinking
ory hybrid hydrogel that showed cooling-induced SME. complexes could be reversibly formed through controlling
Aqueous solution of poly(ethylene oxide)–poly(propylene oxidative states of copper ions (Cu+ versus Cu2+ ) (Fig. 9b)
oxide)–poly(ethylene oxide) [PEO–PPO–PEO] tri-block [128]. Also of relevance is that the oxidative states of ions
copolymer was incorporated into an elastic sponge to form can be tuned both chemically and electrochemically.
the hybrid. Upon heating from room temperature to 35 ◦ C, Supramolecular physical linkage based on a host-guest
the PEO–PPO–PEO solution resulted in a hydrogel due to complex is another catalog of molecular switches that
hydrophobic interaction of the PPO block. In other words, has been applied to enable an SME in water swollen
heating led to additional temporary crosslinks that could polymer systems. Cyclodextrins (CDs) are well known
fix the hybrid into a temporary shape. Shape recovery host compounds in supramolecular chemistry [129,130].
occurred when the PEO–PPO–PEO hydrogel melted upon The molecular structure of CDs consists of a hydrophilic
cooling. rim and a hydrophobic cavity. The cavity can capture
In addition to thermal stimulation, molecular switches a hydrophobic guest molecule with a proper size that
have been increasingly applied to construct shape memory is a little smaller than the cavity, forming a host-guest
hydrogels that can be triggered by various other stimuli. complex. When the hydrophobicity or size of the guest
Molecular switches made of metal ions–organics com- molecule is changed upon exposure to an external stimu-
plexation can enable SMEs by controlling the oxidative lus such as light, redox and pH, the complexes would show
state of the metal ions [127,128]. It was found that the a reversible assembly-disassembly behavior. An example
Fig. 9. Molecular mechanism of some shape memory hydrogels: (a) Iron ion-induced SME [127], Copyright 2014. Reproduced with permission from Royal
Society of Chemistry; (b) Copper ion-induced SME [128], Copyright 2013. Reproduced with permission from the American Chemical Society; (c) pH-Induced
SME [131],Copyright 2012. Reproduced with permission from John Wiley and Sons; and (d)Antigen-induced SME [135], Copyright 1999. Reproduced with
permission from Nature Publishing Group.
using the host-guest complex as the molecular switches 50%, much lower than common hydrogels. Its versa-
was recently reported by Zhang and coworkers [131]. tility is quite notable nevertheless. Another report by
They synthesized crosslinked alginate networks contain- Ritter and coworker [133] concentrated on a perma-
ing both ␤-CD and diethylenetriamine (DETA) as the side nently crosslinked poly(N-isopropylacrylamide) hydrogel
groups (Fig. 9c). In a basic medium (pH = 11.5), the DETA with ␤-CD-adamantane temporary linkages. Adding a
molecules showed a hydrophobic nature and the ␤-CD- sufficient amount of free ␤-CD molecules to the sys-
DETA complexes were formed, that enabled the fixation of tem would disassociate the supramolecular temporary
a temporary shape. While in a neutral medium, the DETA netpoints. Other permanently crosslinked hydrogels con-
molecules became hydrophilic due to the partial ionization taining reversible supramolecular crosslinks (hydrogen
of the amine groups. In this case, the ␤-CD-DETA com- bonds [134] or antibody–antigen interactions [135]) have
plexes disassociated, resulting in the macroscopic shape been already developed. These hydrogels are intrinsically
recovery. In a later study, the same group demonstrated pH or antigen-induced (Fig. 9d) shape memory hydrogels.
that the redox-sensitive ␤-CD-ferrocene complexes could However, only non-programmable shape changing behav-
also be applied to enable an oxidation-induced SME due ior (not the SME) had been investigated in these studies.
to the hydrophobic-hydrophilic transition occurred to the
ferrocene groups at different oxidative states [132]. 4.3. Shape memory microparticles
More intriguingly, the SME can be induced by glu-
cose molecules if glucose oxidase was immobilized Multifunctional microparticles have been the subject
within the polymer network. In the presence of glu- of numerous studies due to their technological relevance
cose, hydrogen peroxide was produced, that oxidized in the biomedical field [136]. Despite their technological
ferrocene. As a result, a glucose-induced SME was real- potential, shape changing microparicles had been limited
ized. It should be noted here that, the equilibrium to those exhibiting only non-programmable shape chang-
water contents for the preceding systems were around ing behavior [137,138].
In 2014, Lendlein’s group reported the first shape 5.1. Intrinsic multi-SMPs
memory microparticle system [139,140]. In their work,
microparticles with diameters around ∼30 ␮m were pre- 5.1.1. General aspects
pared using an oil-in-water (o/w) emulsification method. A multi-SMP is not fundamentally different from a
The polymer used was a multi-block copolymer of poly(␧- dual-SMP. In fact, the molecular requirements for intrin-
caprolactone) [PCL] and poly(␻-pentadecalactone) [PPDL]. sic multi-SMPs are quite similar to that of dual-SMPs. The
Crystalline domains of PPDL segments (Tm = 83 ◦ C) and PCL network structure that is responsible for memorizing the
segments (Tm = 49 ◦ C) served as the “permanent” netpoints permanent shape for a dual-SMP should also be present in a
and the switching segments, respectively. For program- multi-SMP. The difference is in the switches, which should
ming, the microparticles in a permanent spherical shape be more versatile for a multi-SMP. A typical triple-SMP
were first embedded in a polyvinyl alcohol film (Tg = 66 ◦ C, should possess two thermal transitions that can be utilized
Tm = 163 ◦ C), followed by stretching the polyvinyl alcohol to fix two independent temporary shapes. In addition, the
film at 70 ◦ C and subsequent cooling. After dissolution of realization of the triple-SME typically requires a two-step
PVA in water, microparticles with the temporary ellipsoidal programming instead of one-step for the dual-SME. This
shapes were obtained. Shape recovery of the microparti- two-step programming was called a triple-shape-memory
cles could be triggered upon heating. In the same work, the creation process by Lendlein’s group. Without the two-step
application of sonication enabled the preparation of SMP programming, an SMP system with two Ttrans s would not
nanoparticles with much sizes around 250 nm. Although exhibit the triple-SME [144]. It should also be noted that the
crystallinity of the nanoparticles (≈31%) was smaller than multiple temporary shape recoveries differ from interme-
that of the microparticles (≈39%), the SME was not notice- diate shapes occurring along a single recovery pathway for
ably affected since the minimum average crystallite size a dual-SMP. Although intermediate shapes can be obtained
is approximately 16 nm. Because the specific surface area via on and off application of the stimulus, they are not
of an ellipsoid was much larger than that of a sphere with independent to each other. Fig. 10 illustrates the recovery
the same volume, the authors envisioned that such SMP of a triple-SMP [145]. By comparing the three shapes in
microparticles may provide switchable diffusion interfaces Fig. 10, it is quite obvious that the temporary shape B def-
that could be potentially beneficial for controlled drug initely altered the recovery pathway from the temporary
release. shape A to the permanent shape C, in a discontinuous fash-
ion. Intermediate shapes, by comparison, can only be found
5. Multi-SMPs along the same recovery pathway. Thus, triple-SMPs pro-
vide much more controllable and complex shape morphing
A dual-SMP can only switch between one temporary ability.
shape and its permanent shape in each shape memory A typical quantitative triple-shape cycle is presented in
cycle. In contrast, a multi-SMP can be broadly defined as Fig. 11b [146]. In comparison to a dual-shape cycle with
a polymer that can be programmed to exhibit more than a single fixing and a single recovery steps (Fig. 4a), the
one distinctive shape change in the recovery event. This triple-shape cycle contains two programming (shape fix-
definition notably does not place any emphasis on funda- ing) steps and two recovery steps. The polymer has two
mental enabling mechanisms. Instead, it focuses solely on thermal transition temperatures, Ttrans ,B and Ttrans ,A . At
the phenomelogical material behavior that is relevant for the programming stage, the permanent shape A (εA ) is
technological innovations. first deformed to εBload at Thigh (above both Ttrans s) and
A triple-SMP system (with two temporary shapes) was then cooled to Tmid (Ttrans ,B < Tmid < Ttrans ,A ), resulting in the
first reported by Lendlein’s group in 2006 [141]. In the temporary shape B (εB ) after releasing the stress. In the sec-
following 2–3 years, it was the only group working on ond programming step, the shape B is further deformed to
this subject. For this reason, when they wrote a progress εCload . A second temporary shape C (εC ) can be fixed upon
report on the subject in 2010, the focus was mostly on their cooling to Tlow < Ttrans ,B and subsequent removal of the sec-
own systems [142]. In 2011, at the time we published a ond stress. The shape recovery is typically conducted under
review on the updated status of the general SMP field, the a stress free condition. Heating shape C from Tlow to Tmid
examples of multi-SMPs have bloomed to a point that we leads to the recovery of the shape B (εBrec ) while further
dedicated two sessions on the subject [4]. Few years have heating to Thigh results in the recovered shape A (εArec ).
passed since then, most groups active in SMPs have now The respective shape stability (Rf ) and shape recovery (Rr )
published their own multi-SMP systems. It is fair to say can be calculated using the following equations and strain
that the speed at which multi-SMPs have gained traction data obtained from the triple-shape cycle:
is not short of phenomenal [143]. We hereafter introduce εY − εX
the development in this very active research area. Rf (X → Y) = (4)
εYload − εX
As will be revealed in this section, the diverse strate-
εY − εXrec
gies to design the broadly defined multi-SMPs are either Rr (Y → X) = (5)
strongly tied to the chemistry and physics intrinsic to the εY − εX
polymer matrix or extrinsic factors that are more inde- where X and Y denote two different shapes, respectively.
pendent of the polymer matrix (e.g., innovative actuation The preceding triple-shape cycle is obtained under a
methods). To emphasize this mechanistic difference, we stress controlled programming condition and a stress free
classify multi-SMPs into two categories: intrinsic multi- recovery condition. Alternatively, the programming can
SMPs and extrinsic multi-SMPs. also be conducted under a strain-controlled mode, while
Fig. 10. Photographs showing the sequential recovery of a triple-SMP from temporary shape C (a), to temporary shape B (b), and to permanent shape A (c)
[145], Copyright 2010. Reproduced with permission from John Wiley and Sons.
the recovery can be performed under iso-strain condi- systems were both covalently crosslinked networks each
tion. This situation is quite similar to a dual-shape cycle of which has two switch phases [141]. The first exam-
introduced in Section 2. In addition to the commonly used ple was obtained by UV curing of cyclohexyl methacrylate
two-step programming procedure (2SPP), a one-step pro- (CHMA) monomer and poly(caprolactone)dimethacrylate
gramming procedure (1SPP) may also be employed. The [PCLdimethacrylate] crosslinker. This yielded a perma-
1SPP can be conducted at either Thigh or Tlow (i.e., cold nent network with two thermal transitions from the
drawing). A two-step recovery can still be observed during crystalline PCL with Tm = 50 ◦ C and glassy poly(cyclohexyl
reheating despite the one-step fixation. The 1SPP tech- methacrylate) [PCHMA] with Tg = 140 ◦ C (Fig. 12a). The sec-
nique represents a simpler way to for achieve a triple-SME. ond triple-SMP network was synthesized similarly except
However, the technique cannot program two independent that the CHMA monomer was replaced by a polyethy-
strains into the SMP. Therefore, the 2SPP is still the pre- lene glycol (PEG) monoacrylate, this yielded a crosslinked
ferred method for creating complex shape transformations. PCL (Tm = 50 ◦ C) network with PEG (Tm = 34 ◦ C) grafted
From the preceding description, it can be seen that the chains (Fig. 12b). Despite the difference in the network
triple-SME is a natural extension of the dual-SME. Likewise, architecture and the nature of thermal switches (one amor-
triple-SME can be further expanded to quadruple-SME (and phous plus one crystalline versus two crystalline phases),
beyond) following the same principle. From a practical both networks possessed two distinct thermal transi-
sense, multi-SMPs should provide properties and functions tions and showed a triple-SME. In fact, ample examples
that can be tuned to meet the requirements for potential showed below suggest that combination of any two ther-
applications, similar to dual-SMPs discussed in Section 2.4. mal reversible phase transitions (glass transition, melting
To this end, various strategies to design multi-SMPs have transition, or liquid crystalline transition) would enable a
been reported in the literature, as discussed in the follow- triple-SMP.
ing. In 2009, Mather’s group reported a liquid crystalline
homopolymer (a polyester) with a glass transition (80 ◦ C)
5.1.2. Multi-SMPs via distinct thermal phase switches and isotropic–nematic transition (150 ◦ C) [147]. The poly-
In its simplest form, a triple-SMP should have two mer exhibited two recovery events that are characteristic
distinct thermal transitions and one set of netpoints. of a triple-SMP, although the authors did not call this
Copolymerization is the most natural choice for introduc- a triple-SME. Because both the transitions are linked to
ing multiple thermal transitions. The first two triple-SMP the same monomer, this homopolymerization approach
Fig. 11. Triple-shape memory cycles: (a) Visual demonstration; and (b) Quantitative themomechanical cycle [146], Copyright 2009. Reproduced with
permission from John Wiley and Sons.
Fig. 12. Molecular structures of selective triple-SMPs [158], Copyright 2013. Reproduced with permission from John Wiley and Sons.
does not allow easy independent tuning of two transition switches, similar to what was possible with Lendlein’s
phases. Ahn and coworkers [148,149], on the other hand, copolymerization approach.
synthesized a crosslinked glassy network containing liquid A “grafting to” technique is another way to design
crystalline side chains via ring-opening methathesis poly- a triple-SMP. Polybutylene succinate (PBS) was
merization (ROMP) followed by radical initiated thermal grafted onto maleated-polystyrene-b-poly(ethylene-
crosslinking (Fig. 12c). The polymer also exhibited triple- co-butylene)-b-polystyrene (m-SEBS) [150]. The
shape functions. Unlike the preceding case, however, the poly(ethylene-co-butylene) segments within the m-
Tg and the Tcl of this system correspond to the main net- SEBS backbone and the grafted PBS segments provided
work and the side chains, respectively. This should allow two Tm s (60 ◦ C and 110 ◦ C, respectively), whereas the
independent tuning of the two responsible thermal phase high glass transition of the amorphous polystyrene phase
provided the physical netpoints. Notably, since there was macroscopically heterogeneous, it met the basic structural
no permanent crosslinking, the triple-SMP system could requirement for a triple-SMP. When the interface between
be reprocessed. the layers was sufficiently strong, the composite readily
A triple-SMP can also be designed via an interpen- exhibited the triple-SME. Similarly, a triple-SMP compos-
etrating polymer network (IPN) approach [151]. UPy ite was prepared by incorporating two shape-memory
end-capped star-shaped PCL and hydroxyl groups termi- polyurethane layers that were permanently crosslinked by
nated poly(p-dioxanone) [PPDO] were used. Crosslinking silica nanoparticles [156]. Mather’s group pursued a differ-
PDDO via its reaction with a diisocyanate led to a PPDO ent approach to create a triple-SMP composite [145]. They
network while the PCL naturally formed a supramolecu- fabricated non-woven thermoplastic PCL nanofibers (Tm
lar network via the self-complementary hydrogen bonding around 55 ◦ C) via electrospinning. The fiber mats were then
between UPy units. The two crystalline phases (Tm s of embedded into an epoxy dual-SMP matrix with Tg around
40 ◦ C and 95 ◦ C, respectively) provided by PCL network and 35 ◦ C. The composite also displayed good triple-SME. One
PPDO network formed the basis for the triple-SME. They common characteristic for the three composite examples
further demonstrated that the ratio between the PCL and shown here was that the individual contributing compo-
PPDO within the IPN could be easily tuned without affect- nents could be independently adjusted both in terms of
ing much the two transition temperatures. One interesting transition temperatures and their fractions. This is quite
benefit of the IPN design was that the crystallizability of advantageous in tuning triple-shape functions.
both the PCL and PPDO segments was much better than
that in a co-network. 5.1.3. Multi-SMPs via molecular switches
Blending two (or more) polymers with different tran- Section 3.1 described molecular switches for dual-SMPs.
sition temperatures followed by crosslinking is a versatile We show in this section that various molecular switch-
approach to prepare a multi-SMP [152,153]. For instance, ing mechanisms can also be utilized to enable triple-SMPs,
a linear high density polyethylene (Tm = 132 ◦ C) and two either by themselves or in combination with a thermal
ethylene-1-octene copolymers with Tm s of 59 ◦ C and 96 ◦ C phase switch.
were blended and then thermally crosslinked with a perox- We described earlier in this review that supramolecular
ide. This yielded a crosslinked network with three distinct self-complimentary hydrogen bonding by UPy is an effec-
melting events, resulting in a quadruple-SMP. Choosing tive molecular switch for elastomeric dual-SMPs. Voit’s
various other polyolefins with different Tm s would allow group synthesized a UPy containing chemically crosslinked
tuning the multi-SMP. (meth)acrylate network with a Tg tunable between 0 and
Doping small molecules into a dual-SMP network rep- 60 ◦ C (Fig. 12d) [157]. When the Tg does not overlap with
resents another interesting approach to introduce an the UPy thermal activation temperature, they found that
additional transition phase [154,155]. In one example, a both the amorphous phase and the supramolecular molec-
thermoplastic polyurethane was synthesized by polymer- ular switch could be used to record two independent
ization of PCL-diol, 4,4 -methylenediphenyl diisocyanate shapes, thus achieving a triple-SME.
(MDI), pentaerythritol, and 2,2-bis(hydroxymethyl) pro- A photo-responsive molecular switch can also be incor-
pionic acid [154]. The PCL and the glassy hard domains porated into a polymer with a thermal phase switch to
provided a Tm around 50 ◦ C and a Tg around 70 ◦ C. enable a multi-SME [158]. The polymer showing this fea-
More importantly, the presence of carboxylic acid groups ture is a hyperbranched coumarate polyester, synthesized
provided a mechanism for further doping. Thus, liquid crys- by the polycondensation between a glycol and a diester
talline small molecules, cholesteryl isonicotinate (INCh), monomer containing cinnamic double bonds (Fig. 12e).
were doped into the polyurethane via the ionic hydro- With its Tg between 36 and 64 ◦ C, the polymer showed
gen bonding between pyridine on INCh and carboxylic acid a thermally induced dual-SME. In addition, reversible
groups. With a high transition temperature (above 110 ◦ C), photo-deformation capability was also observed. Upon UV
INCh formed the basis for physical netpoints. Thus, a ther- ( = 280–450 nm) light exposure to the polymer film from
moplastic triple-SMP was obtained. Similarly, Chen et al. the upper side, it transformed to a concave shape (Fig. 13)
[155] doped 4-hexadecyloxybenzoic acid (HOBA) into a due to the uneven formation of cyclobutane crosslinks
thermoplastic dual-shape memory polyurethane (SMPU) through the film thickness. The cyclobutane structure was
with a Tg of 90 ◦ C. HOBA is a liquid crystal mesogen stable after the UV light was turned off, so that the concave
with a transition temperature of 100 ◦ C. Within the net- shape could act as the permanent shape for a thermally
work, it also acted as a plasticizer, lowing the Tg of the induced SME (Fig. 13). When needed, the concave shape
SMPU to around 50 ◦ C, while not destroying the H-bonding could be erased upon UV exposure of = 250 nm. In short,
crosslinks inherent to the SMPU network. The liquid crystal the hyperbranched polyester provided photo and thermal
doped SMPU showed not only a triple-shape effect due to switching mechanisms that contribute to its higher ver-
the two thermal transitions, but also a nematic phase upon satile multi-SME. Notably, the photo-induced deformation
heated to a temperature between 100 ◦ C and 130 ◦ C. could be sorted as a special SME in which the corresponding
The intrinsic multi-material nature of composites natu- programming was achieved by the internal stress gener-
rally points to a highly attractive and versatile option to created via the uneven light exposure, in a hands free fashion.
ate multi-shape functions. In fact, one of the first few exam- A water sensitive molecular switch was also used in
ples of a triple-SMP was a composite laminate consisting combination with a thermal phase switch to design a
of two layers of thermoset epoxy polymers with well- triple-SMP. To this end, Hu’s group fabricated a triple-SMP
separated Tg s [146]. Although the composite laminate was nanocomposite using cellulose-whiskers as reinforcement
content. That is, the hydrogel displayed both the SME

and a non-programmable shape change at the same time.
In this case, it is more appropriate to evaluate its shape
memory properties such as Rf and Rr via bending tests
instead of tensile tests, since the bending angles are
less impacted by the swelling/shrinking of the material.
The combination of non-programmable shape changing
behavior and the SME could be used to achieve very
sophisticated shape-morphing behavior. Indeed, when the
hydrogel was patterned with stripes on one side of the
surface, by controlling the directions of the stripes, various
curling shapes were realized upon exposure to zinc ions
Fig. 13. Series of photographs of the hyperbranched coumarate polyester
[161]. This is shown in Fig. 14b. The curling was attributed
sample showing a photo-induced multi-SME. The permanent shape A to heterogeneous volume change within the hydrogel.
was irradiated by UV light at = 280–450 nm for 20 s to form shape B. Due to its larger surface area, the volume change on the
Upon being deformed by an external load to the screw shape at 70 ◦ C surface-patterned side was faster than the non-patterned
and cooled/fixed at room temperature, shape C was fixed. Shape D, the
side. In addition to its intriguing shape memory behavior,
same as shape B, was recovered by heating again to 70 ◦ C. Shape A was
recovered by irradiation at = 250 nm for 180 minutes. The size of shape the hydrogel possessed low cytotoxicity and might be
A is 25 mm × 10 mm × 0.2 mm [158], Copyright 2013. Reproduced with potentially useful as a tissue engineering scaffold [162].
permission from John Wiley and Sons.
5.1.4. Tunable multi-SMPs via a single broad thermal
and a segmented polyurethane SMP as the matrix [159]. In phase switch
addition to the thermal phase switch inherent in the poly- The extension from a dual-SME to a triple-SME seemed
mer matrix, the hydrogen bonding within the percolation to imply that the number of temporary shapes that a poly-
network of cellulose-whiskers formed an additional set of mer can fix is determined by the number of distinct phase
water sensitive switch. The composite deformed in warm transitions. Thus, any additional distinct phase transitions
water can be fixed via cooling and drying to yield a tem- would theoretically bring in additional temporary shapes.
porary shape A. Via heating under a dry condition, shape A However, synthesizing a polymer network with multiple
recovered to a shape B. By immersing in warm water, the (>2) discrete thermal transitions, while possible, becomes
shape B recovered further to the original shape C. increasingly more challenging as the number of transitions
A single set of molecular switches exhibiting two goes up. In 2010, we discovered a rather different approach
different molecular states represents a highly intriguing to achieve a multi-SME [163].
mechanism to achieve a non-thermal triple-SME. Liu’s Nafion, a thermoplastic perfluorosulfonic acid ionomer
group synthesized crosslinked hydrogels by copoly- (Fig. 15a), has a single broad thermal phase transition rang-
merization of acrylonitrile and 2-methacryloyloxyethyl ing from 55 ◦ C to 130 ◦ C (Fig. 15b). We discovered that,
phosphorylcholine [160]. In pure water and a low concen- within this single phase transition, one could program
tration of ZnCl2 solution, the cyano groups formed strong two temporary shapes into the material at two deforma-
CN–CN dipole–dipole interactions and Zn(CN)2 complex, tion temperatures (Td s). Subsequent heating (stress free)
respectively. When the ZnCl2 concentration was increased to the corresponding recovery temperatures (with Tr = Td )
to a high level (>38%), each individual cyano group would led to two distinct recovery events. In essence, a triple-SME
associate with one zinc ion, thus destroying both inter- was realized without the need for two discrete transitions.
actions. At this latter state, the hydrogel possessed the In addition, a quadruple-SME (possibly beyond) could be
permanent shape. Upon subsequent immersion in pure realized with this single transition (Fig. 15c). This type of
water, a deformed temporary shape could be fixed due multi-SME is different from a triple-SMP with two distinct
to the formation of strong dipole–dipole interactions that transitions. For the latter, the two recovery events always
acted as the temporary crosslinks. A second type of addi- occur at two temperatures in between and above both tran-
tional crosslinks formed when the hydrogel was soaked in sition temperatures, respectively. Thus, if one wishes to
a relatively low concentration (<30%) ZnCl2 solution, due move the recovery temperatures, as may be demanded
to the formation of Zn(CN)2 complexes. These two types for different applications, the transition temperatures will
of molecular switches were utilized to independently fix have to be altered. This would require synthesis of a new
two temporary shapes and subsequently recover when polymer. For Nafion, however, the multiple temporary
the solution conditions were reversed. That is, the hydro- shapes would recover fully at the recovery temperatures
gel displayed a triple-SME (Fig. 14a). In addition to the corresponding to its deformation temperatures (Tr = Td ).
triple-SME, the dipole–dipole interactions and the Zn(CN)2 Therefore, the recovery temperatures for the temporary
complex are quite beneficial for the mechanical properties shapes could be altered via the choice of Td , instead of hav-
of the hydrogel. At an equilibrium water contents higher ing to synthesizing new polymers. We call this behavior the
than 75%, the hydrogel possessed tensile strength ranging tunable multi-SME, emphasizing that the recovery temper-
from 1.5 MPa to 4.3 MPa and high Young’s modulus up to ature can be physically tuned (or manipulated) within one
39 MPa. An additional feature is that increasing/decreasing polymer.
additional crosslinks also impacted the volume of the Mechanistically, we proposed a theory to explain this
hydrogel due to the change in the equilibrium water behavior. The critical point for this theory is that a single
Fig. 14. Zn ion induced shape memory effect in hydrogels. (a) Schematic illustration of the triple-SME [160], Copyright 2012. Reproduced with permission
from John Wiley and Sons. (b) A schematic diagram illustrating automatic shape changing of the hydrogel with surface patterned stripes of different angles
[161], Copyright 2012. Reproduced with permission from Royal Society of Chemistry.
broad thermal transition can be regarded as the collective would always activate two different groups of EMUs for
contribution of numerous infinitely sharp transitions con- fixing. More importantly, these two EMU groups can be
tinuously distributed in the broad temperature range. Each distinctively activated during reheating for two recovery
of these sharp transitions can be viewed as an elemental events. Overall, this led to a tunable triple-SME. The same
memory unit (EMU) corresponding to an infinitely sharp principle could be extended to a tunable multi-SME beyond
transition temperature. Based on this theory, at any Td , only triple.
the EMUs with their Ttrans s below this particular Td are acti- Closely related to the tunable multi-SME is the so-called
vated for the shape memory function. For a single broad temperature memory effect (TME). The TME was first dis-
transition, the distribution of the EMUs is continuous. Thus, covered in 2007 for a polyvinyl alcohol (PVA)-carbon fiber
any Td within this broad transition would correspond to a composite exhibiting a broad glass transition [164]. The
particular group of activated EMUs. Performing a triple- composite was first deformed at a Td within the glass tran-
shape programming at any two deformation temperatures sition. After cooling to fix the strain, Td could be traced back
Fig. 15. The molecular structure, dynamic mechanical curve, and multi-SME of Nafion. (a) Molecular structure; (b) Dynamic mechanical analysis curve;
and (c) Visual demonstration of a quadruple-SME [163], Copyright 2010. Reproduced with permission from Nature Publishing Group.
as it was equivalent to Tmax observed in the correspond- single broad transition, the controlled polymerization
ing recovery event run under an iso-strain condition. In technique could be used to precisely manipulate the glass
essence, the material memorized Td . Thus, Tmax ≈ Td estab- transition(s) in an arbitrary fashion through controlling
lishes the basis for the TME. More precisely, this can be comonomer feedings. Thus, the methodology represents a
called stress based TME since the TME was reflected in the highly attractive platform for synthesizing multi-SMPs.
stress recovery event. The same effect was later confirmed Generality of the enabling mechanism allowed the
by us and Lendlein’s group for different polymers also tunable multi-SME to be realized in various other ways
with a broad thermal transition, confirming its generality for a variety of polymers. In one instance, hydrophobic
[165,166]. While intriguing, the stress based TME may not polyurethane (PU) chains were grafted onto hydrophilic
be particularly relevant since most SMP applications rely- PVA [173]. The hydrophobic PU domains acted as effec-
ing on strain recovery instead of stress recovery. With this tive physical crosslinking points while the PVA–graft–PU
in mind, we designed a series of experiments in an attempt retained the broad Tg of PVA from 25 ◦ C to 75 ◦ C, enabling
to establish a strain based TME for Nafion. The result of a triple-SME with stability and recovery ratios all higher
our investigation showed that, when the programming was than 89%. Moreover, the SME of the material could be trig-
conducted in a strain controlled condition, Tsw obtained gered by water, since the Tg could be decreased when water
from the corresponding strain recovery curve can indeed be molecules diffused into the polymer matrix. Other Tg -type
quantitatively related to Td . Accordingly, the strain based tunable multi-SMPs have also been developed. Linear PEG
TME was defined as Tsw ≈ Td . While the strain based TME was immobilized in a crosslinked poly (methyl methacry-
was established for Nafion in our case, Lendlein’s group late) [PMMA] network, forming a PMMA/PEG semi-IPN
observed the same behavior for a semi-crystalline network [174]. The incorporation of PEG in the network not only
with a broad melting temperature [167,168]. Very recently, introduced an additional Ttrans (Tm of PEG) but also broad-
Zhou’s group extended the TME to a Tm -based biocom- ened the Tg of PMMA, resulting in a quintuple-SME. In
patible and biodegradable polymer [169]. In their work, another example, norbornene cholic acid (NCA) and nor-
four-arm PCL and linear PEG terminated with cinnamon bornene triethylene glycol monomethyl ether (NTEG) were
groups were photo-crosslinked to yield a PCL-PEG network. copolymerized through a ring-opening metathesis poly-
By adjusting the ratio of PCL/PEG and the photo-irradiation merization [175]. Variation of the composition of NCA and
time, a broad melting transition from 20 to 55 ◦ C could be NTEG within the copolymer allowed facile tune of mate-
obtained. Although such a transition range was not partic- rial properties including the breadth of the Tg transition,
ularly broad, the authors were able to use small distinction hydrophilicity, solubility, and mechanical properties. The
of deformation/transition temperature (e.g., 37 and 45 ◦ C) resulted norbornene-based copolymer displayed a broad
to achieve a TME. Tg , enabling a quadruple-SME.
While different from the tunable multi-SME, the TME The preceding examples of tunable multi-SMPs rely
can be readily explained by the same EMU theory. That is, on broad amorphous transitions. A broad melting transi-
a Td corresponds to a particular group of activated EMUs tion was also found equally effective. An ethylene-vinyl
for strain fixing. At the strain recovery event, this group of acetate copolymer (EVA) typically possesses a relatively
EMUs were all re-activated at the corresponding Td , lead- broad melting transition, with the Tm value determined by
ing to a maximum instantaneous strain recovery rate at the comonomer ratio [165]. By blending two EVAs of dif-
this particular temperature Tsw . Such a mechanistic dis- ferent comonomer ratios, the breadth of the Tm transition
cussion suggests that, although the tunable multi-SME and could be extended to 100 ◦ C. This was found sufficient to
TME are not identical phenomenologically, they share the realize the tunable multi-SME. Weiss’s group attempted to
same molecular principle. This was supported by two inde- use the broad melting transition (60–100 ◦ C) of a commer-
pendent modeling efforts by Huang’s group and Qi’s group cial poly(ethylene-co-methacrylic acid) [PEMA] ionomer
[170,171]. For this reason, we chose to discuss them jointly to achieve the multi-SME [40]. The PEMA were partially
in this section. zinc-neutralized, forming ionic nanodomains as the phys-
Because of Nafion’s sophisticated ionomeric nature, it ical crosslinks. Multiple temporary shapes could be well
was not readily clear in the beginning if the mechanism fixed with high Rf s, however, the Rr values were not
proposed for the tunable multi-SME could be generalized particularly high probably due to the inefficiency of the
for more common polymers. Collaborating with Luo’s ionic crosslinking points. When the ionomer was fur-
group, we synthesized a symmetric gradient copolymer of ther crosslinked with electron beam, Rr was successfully
styrene (St) and methyl acrylate (MA) using a reversible improved to values approaching 100%. In another exam-
addition–fragmentation chain transfer polymerization ple, ethyl cellulose was grafted with two PCLs of different
technique [172]. The gradient nature led to a broad glass molecular weights (7600 and 1400) [176]. The graft copoly-
transition between 20 ◦ C (Tg of PMA homopolymer) and mers were mixed and subsequently crosslinked by MDI
100 ◦ C (Tg of PSt homopolymer). With the high Tg St to form a permanent network. Instead of showing two
segments at both ends of a chain and the low Tg MA separated Tm , the material exhibited a broad Tm (from
segments at the chain center, nano-domains with the 34.4 ◦ C to 92.7 ◦ C), enabling a quadruple-SME. Intriguingly,
highest Tg were formed. These domains acted as the a broad melting transition could also be achieved by dop-
physical netpoints. With this polymer, tunable multi-SMEs ing with small molecules. Two paraffins with different Tm s
including quadruple- and quintuple-SMEs were observed. (32 ◦ C and 60 ◦ C, respectively) were incorporated into a
The work thus verified the generality of the tunable physically crosslinked styrene-b-(ethylene-co-butylene)-
multi-SME and the underlying mechanism. Besides a b-styrene network. Within this network, the two paraffins
exhibited a co-crystallization effect, yielding a broad Tm and the multi-shape capability by combining selective light
a tunable multi-SME [177]. triggering and non-selective direct triggering. Their simple
yet elegant approach started by printing black ink on
5.2. Extrinsic multi-SMPs selective areas of the flat Shrinky-Dinks film cut into a
particular shape (Fig. 16b). Upon exposure to infrared light,
Practical constraints for many device applications limit localized absorption of the light triggered heating and thus
access to direct heating. As a result, non-direct heating shrinking only on the printed areas, leading to the self-
based triggering methods have attracted ever increasing folding of the flat sheet into a three dimensional box. When
attention in the general area of SMPs. As an added bonus, the 3D box was further exposed to uniform direct heating,
spatio-selective actuation is feasible for some non-direct it unfolded back into a flat sheet that is proportionally
heating based actuations (e.g., light). This enables tempo- smaller than the starting Shrinky-Dinks film.
ral control of the recovery of a dual-SMP in multiple steps, The photothermal effect of nanoparticles has also been
resulting in an overall multi-shape behavior. Such behavior explored as the basis for light triggered shape recov-
would fit into the broad definition of multi-SMPs, yet the ery. Within the current context, Zhao and coworkers
molecular principle remains the same as dual-SMPs. We [90] fabricated an SMP nanocomposite by incorpo-
hereby suggest calling this type of systems extrinsic multi- rating Au nanoparticles into a chemically crosslinked
SMPs, in distinction to the intrinsic multi-SMP for which poly(caprolactone) network. Through the use of masks,
the multi-shape behavior finds its roots in their more ver- laser light (wavelength: 532 nm) was shined onto des-
satile molecular switch mechanism. Strictly speaking, it can ignated areas of the sample, triggering corresponding
be argued that these types of systems are not multi-SMPs. recovery. The spatio-selective nature of the light trigger-
We will however illustrate below that extrinsic multi- ing allowed multi-stage recovery in a highly controllable
SMPs are of equal (perhaps more) importance than intrinsic manner (Fig. 16c). Along a similar line, they [95] demon-
multi-SMPs, at least from the standpoint of potential device strated that high intensity focused ultrasound (HIFU) can
applications. also be utilized to induce heating in focused areas. As such,
Iron oxide particles are known to induce heating when the recovery of the first temporary shape to the permanent
exposed to an AC magnetic field. Incorporating them into shape can be manipulated step-wise to achieve multi-
an SMP is a known practice to fabricate a SMP composite for shape capability (Fig. 16d). As an added benefit, when drug
which the shape recovery can be actuated remotely via AC molecules are incorporated in this SMP system, the multi-
magnetic field. In a recent collaboration with Hilt’s group, shape changes can be synchronized with controlled release
we discovered that carbon nanotubes can also induce heat- profile of the drug.
ing, importantly at a much higher radiofrequency than iron Another unique extrinsic multi-SMP is the so-called
oxide [178]. The principle that different nanofillers respond functionally graded SMP, referring to a material that pos-
to an AC magnetic field (or radiofrequency (RF) wave) of sesses a range of transition temperatures that are spatially
different frequencies can be utilized to achieve multi-shape distributed in a gradient fashion [180]. This material system
behavior. The proof of concept is illustrated in Fig. 16a with was synthesized by a UV pre-curing at room temperature,
a multi-composite containing three segments of an iron followed by postcuring on a temperature gradient hot plate
oxide SMP composite, a carbon nanotube SMP composite, under the same UV source. Owing to the diffusion con-
and a neat SMP. Here, the SMP matrix in the three seg- trolled reaction kinetics typical for a UV curing system, the
ments does not need to be different. The multi-composite Tg of the material is strongly tied to the curing tempera-
sample in a straight strip form (permanent shape) was ture. As such, the temperature gradient on the hot plate
deformed into a “” shape in which all three segments resulted in the functionally graded SMP. When a deformed
were deformed. When the sample was exposed to a high RF, sample of the functionally graded SMP was exposed to
the carbon nanotube composite segment was recovered. At continuous linear heating, spatially sequential recoveries
a low frequency, iron oxide composite segment was recov- were observed as the environmental temperature progres-
ered. The neat polymer segment was recovered by direct sively exceeded the Tg ’s. In a similar example, local Tg s of a
heating. Following this particular actuation sequence, a loosely crosslinked polymer film were spatially changed by
total of four shapes were observed in a single shape mem- immersing its different parts in a crosslinker solution with
ory cycle. Conceptually, one can alter the sequence of the various concentrations [181]. The higher concentration of
actuations. As a result, the recovery of the temporary “” the crosslinker was applied, the higher local Tg could be
shape to the permanent shape can follow five different obtained.
routes involving a total of five shapes. Overall, the behav- Instead of spatially gradient chemical structure, spa-
ior of this extrinsic multi-SMP is drastically more versatile tially distributed apparent Ttrans can be realized within a
since the recovery of an intrinsic multi-SMP always follows magnetic nanoparticles/SMP composite. As revealed in Sec-
a particular order predetermined in the programming step tion 3.2, endogenous heating allows adjustment of Tsw,app
(i.e., from the low temperature shape to the high tempera- [96]. For samples with an identical chemical composition,
ture shape, and to the permanent shape). smaller samples with higher specific surface areas provide
Shrinky-Dinks film is a well known toy that shrinks higher Tsw,apps and vice versa. In this case, a sample with a
upon microwave heating. This behavior is a typical heterogeneous geometry provided locally distinct Tsw,apps ,
dual-SME and the Shrinky-Dinks film is essentially is a enabling the extrinsic multi-SME [182].
dual-SMP film in its temporary (biaxially stretched) state. As stated in the preceding, the multi-shape behav-
Utilizing the Shrinky-Dinks film, Liu et al. [179] achieved ior of an intrinsic multi-SMP can only be achieved via
Fig. 16. Extrinsic multi-SMP systems. (a) RF selective actuation of a multi-composite [178], Copyright 2011. Reprinted with permission from John Wiley
and Sons; (b) Local actuation by light followed by uniform heating of an SMP printed with black inks [179], Copyright 2012. Reprinted with permission
from Royal Society of Chemistry; (c) Selective actuation by near IR of a gold SMP nanocomposite [90], Copyright 2012. Reprinted with permission from
Royal Society of Chemistry; (d) Local actuation by focused high intensity ultrasound [95], Copyright 2012. Reprinted with permission from Royal Society
of Chemistry.
controllable heating methods (either staged heating or permanent shape of a triple-SMP via dynamically
slow linear heating). As such, the fact that most of exchangeable covalent bonds [183]. In their work, a
the extrinsic multi-SMP systems were realized by easily liquid crystal elastomer (LCE) with ample ester linkages in
controllable remote triggering methods is advantageous. the network was synthesized. The LCE possessed a liquid-
Despite the versatility of the various extrinsic multi-SMP crystal isotropic transition temperature (Ti ≈ 100 ◦ C) and a
concepts, however, one has to be mindful of their limi- glass transition (Tg ≈ 55 ◦ C). With the addition of triazobicy-
tations. For instance, a common limitation of an extrinsic clodecene (a transesterification catalyst) in the LCE, a trans-
multi-SMP lies in its spatio-selective nature. That is, they esterification reaction could be activated at a vitrification
are not suitable for devices that require multi-recovery on temperature (Tv , typically higher than 160 ◦ C). Below the
the identical spots in the sample. In addition, each extrin- Tv , the LCEs were able to exhibit a triple-SME utilizing the Tg
sic multi-SMP concept has its own practical limitation(s). and Ti phases. Above Tv , the ester linkages would break and
Potential devices based on the selective radiofrequency reform in a dynamic fashion. Under this condition, when an
actuation may require access to bulky induction heating external force was applied, the network topology was rear-
equipment. Light triggering, on the other hand, is not par- ranged to release the stress. In essence, such a stress free
ticularly suitable for three dimensional bulk devices due deformed shape would represent a new permanent shape.
to the requirement for direct exposure. For HIFU trigger- In a more conventional way, thermal reversible cova-
ing, it may not have sufficient spatial resolution needed for lent bonds such as the Diels–Alder reaction were also
miniature devices. applied to enable reprocessability for a triple-SMP. This
As is always true for the SMP field, there is not a single was demonstrated by Yang and coworkers [184]. Crys-
system suitable for all applications as each system has its talline poly(tetramethylene oxide) (PTMEG, Tm = 25 ◦ C) and
prons and cons. The diversity of the various extrinsic and poly(p-dioxanone) (PPDO, Tm = 105 ◦ C) with furan groups
intrinsic multi-SMP concepts, on the other hand, supplies on both ends were first synthesized. The two polymers
different options that can be picked upon for a particular were subsequently crosslinked to form a single network
application. To this end, one particularly interesting area of via the Diels–Alder reaction between the furan groups
future development may lie in new systems that combine and the maleimide groups from a tri-functional crosslinker
the intrinsic and extrinsic multi-SMP concept. This may [tris(2-maleimide ethyl)amine]. Whereas the two Tm s from
overcome some of the limitations while providing enor- PTMEG and PPDO provided the basis for a triple-SME, heat-
mous opportunities for designing easily controlled highly ing to a higher temperature allowed decrosslinking for
versatile SMPs. reprocessing via the reverse Diels–Alder reaction.
Photo-reversible covalent bonds in principle should also
5.3. Reprocessible multi-SMPs allow reprocessing of a triple-SMP. In an effort by Zhou and
coworkers [169], PCL-diol and N,N-bis (2-hydroxyethyl)
Whereas thermoplastic multi-SMP naturally allow cinnamamide was copolymerized with MDI to yield a lin-
reprocessing to create new permanent shapes [163,172], ear polyurethane with a Tm around 40 ◦ C and a Tg around
a number of other approaches based on reversible cova- 70 ◦ C. The presence of the cinnamon groups allowed UV
lent and non-covalent bonds have also been explored to crosslinking (=356 nm) to introduce the netpoints, result-
achieve the same goal. ing in a triple-SMP. In theory, the photo-reversibility of the
A recent collaboration between Ji and Terentjev netpoints should permit resetting the permanent shape,
explored a highly intriguing idea of resetting the although the author did not report such an attempt.
6. Two-way SMPs nature, this formed the basis for the subsequent develop-
ment on 2W-SMPs.
The various SMEs described so far in this review belong Mondomains can be formed either chemically or physi-
to the one-way SME (1W-SME). Within a particular one- cally. During the synthesis of a LCE, a two-step crosslinking
way shape-memory cycle, the shape recovery step can only method could be employed [187–191]. In the first step,
go from the temporary shape(s) to the permanent shape. an isotropic network is established via partial crosslink-
That is, unless the SMP is mechanically deformed again ing. The network is then deformed to induce anisotropy,
via a re-programming process, the recovered permanent subsequently fixed by further crosslinking reactions in the
shape cannot revert back to the temporary shape by alter- second step. Alternatively, a one-step crosslinking method
ing the external stimulus only. Thus, 1W-SME is often time can also be used. Small molecular or polymeric liquid crys-
also referred to as an irreversible SME. Within the broad tal precursors are macroscopically oriented by applying an
examples of SCPs, many polymers exhibit reversible shape external field (e.g., mechanical, magnetic, or electric field)
changing behavior. Typically however, once the geome- [187–189]. Afterwards, the aligned precursors are poly-
try is fabricated, an SCP can only switch between two merized/crosslinked to form a macroscopically anisotropic
fixed equilibrium shapes. We call this type of polymers LCE. Notably, physical methods can also be employed to
as non-programmable reversible SCPs, in contrast to the induce monodomain formation after the LCE polydomains
1W-SMP for which the temporary shape can be arbitrarily are synthesized. Applying an external stress via hanging a
defined via programming. While most SMEs are one-way, weight is the easiest and most commonly used method to
it is of no surprise that scientists working in this field have physically induce the monodomain formation [192,193].
been trying to develop polymers that can exhibit two-way A critical difference between the chemically induced and
SME (2W-SME) or reversible SME. In contrast to the non- the physically induced anisotropy is that while the former
programmable reversible SCPs and the programmable but cannot be altered after the sample fabrication, the latter
irreversible 1W-SMPs, a 2W-SMP takes best attributes of can be tuned by changing the external stress (e.g., weight).
both and is defined as a polymer that can reversibly switch Thus, the degree of anisotropy and the corresponding shape
between two shapes that can be thermomechanically pro- changing induced by the physical method can be tuned. In
grammed and reprogrammed. Thus, a 2W-SMP may also be other words, the shape changing becomes programmable.
referred to as a programmable reversible SCP. The history We note here that, due to the need of an external force,
and the latest development on 2W-SMPs are reviewed in the loose definition of “programmable” is used as will be
this section. clarified latter in this section. Nevertheless, the associated
behavior can be called quasi 2W-SME. Not surprisingly, the
combination of chemical and physical domain orientation
6.1. Quasi 2W-SME for liquid crystalline elastomers methods (i.e., applying an external stress onto a chemically
oriented LCE) led to high anisotropy and a reversible strain
As early as 1975, de Gennes [185] predicted liquid crys- change as high as 400% [194,195].
talline elastomers (LCEs), that were not synthesized until
1981 by Finkelmann’s group [186]. LCEs are elastic poly- 6.2. Quasi 2W-SME for semi-crystalline networks
mer networks containing main-chain or side-chain liquid
crystal units that are capable of undergoing reversible Despite the intriguing shape changing behavior
mesomorphic-to-isotropic phase transitions. The liquid observed for LCEs, they are difficult to synthesize and
crystalline domains are typically randomly oriented with the transition temperatures are not easy to tune. In 2008,
respect to each other, they are thus called liquid crystalline Mather’s group [196] reported an important discovery,
polydomains. The liquid crystalline domains can be aligned demonstrating that a semi-crystalline polymer network
in a certain direction when an external field (e.g., magnetic exhibited a similar behavior. In the presence of a constant
field, stretching force) is applied during the formation of tensile load, a crosslinked poly(cyclooctene) [PCO] semi-
a LCE network [187–190]. This results in the so-called liq- crystalline network underwent elongation when they
uid crystalline monodomains in which the polymer chains were cooled across the Tm (i.e., crystallization induced
are aligned and exhibit an extended chain conformation. elongation or CIE) as schematically shown in Fig. 17a.
Macroscopically, it corresponds to an elongated shape. When heated above the Tm under the same load, the
When heated to a temperature above the liquid crys- elongation was reversed (i.e., melting induced contraction
tal clearing temperature (Tcl ), the polymer chains would or MIC). Similar to LCEs, the extent of reversible strain
reduce its anisotropy, leading to macroscopic contraction change was also tunable via the adjustment of the constant
in the sample length. Upon cooling below Tcl , the sample external stress. Subsequent investigations by other groups
would revert to the original anisotropic state and elongates. suggested that this behavior is generic for semi-crystalline
Between the elongated state and the contracted state, a networks [197–202]. In particular, Lendlein’s group in
large strain change of c.a. 90% can be observed [191]. Very 2010 reported a two-way reversible triple-shape system
importantly, this cooling induced elongation and heating in which a semi-crystalline network with two distinct
induced contraction is fully reversible. This is an early melting transitions could reversibly switch among three
example in which the reversible shape change originated shapes, again, with the presence of a constant external
from a thermal phase transition. However, there was no stress [203].
demonstration of manipulating the shape changing event The achievement of the CIE–MIC transformation for
via a programming step. Despite its non-programmable the preceding semi-crystalline networks all required the
Fig. 17. Schematic illustration of various 2W-SMEs: (a) crystallization induced elongation (CIE) and melting induced contraction (MIC) under a constant
stress; (b) Reversible bending action of a bilayer composite under zero external stress: the black layer represents a pre-stretched SMP and the gray layer
represents an elastomer; (c) Programmable 2W-SME.
presence of an external force, albeit constant. Looking poly(cyclooctene) as the active SMP layer and silicon rub-
back to the mechanisms behind LCEs, however, perhaps ber as the inactive elastomer layer [206,207]. Despite the
the anisotropic alignment of the polymer chains was the difference in material constituents, the bilayer showed a
true underlying mechanism for semi-crystalline polymer similar reversible bending behavior. For both bilayer sys-
networks. Nevertheless, it was the external stress that tems, the shape changing was reversible, but the shapes
allowed tuning the anisotropy and the strain change. involved cannot be programmed since the internal stress
However, the requirement for an external stress is a was determined in the fabrication process and cannot be
serious limitation for the potential device applications altered afterwards. The big questions were: can a mecha-
based on this type of behavior. The search for alternative nism be perceived to enable the alternation of the internal
mechanisms and materials to enable a 2W-SME without stress after the fabrication; if so, will it lead to a pro-
the need for an external stress (i.e., true 2W-SME) has grammable reversible shape changing behavior without
been a constant pursuit in the SMP community. Along the the need for an external stress, which can be defined as
way, several contributions are particularly noteworthy. a 2W-SME in a strict sense?
Hu’s group fabricated an SMP bilayer by bonding a
pre-stretched Tm -type shape memory polyurethane with 6.3. 2W-SME for semi-crystalline networks
a non-stretched polyurethane elastomer [204,205]. Upon
heating above the Tm of the SMP layer, the bilayer bended In 2013, Lendlein’s group devised an elegant approach
towards the direction of the SMP. The bending angle that answered both questions posed in the preceding sec-
was significantly reduced when the sample was cooled tion in a positive way. In their work [33], a polyester
below the Tm . The procedure is schematically shown in urethane (PEU) network with poly(␻-pentadecalactone)
Fig. 17b. Importantly, the change in the bending angle [PPDL] and PCL segments was synthesized. The two
was fully reversible. Notable differences from the exam- polyesters provided a high Tm (Tm,high ) around 64 ◦ C and a
ples discussed in the last paragraph are: the lack of an low Tm (Tm,low ) around 34 ◦ C, respectively. Whereas the PEU
external stress; reversible bending versus reversible elon- network is an intrinsic triple-SMP due to the two Tm s, this
gation/contraction. Analyzing the details a bit deeper, one time they illustrated a 2W-SME following the procedure
might realized that the shape mismatch between the elas- outlined in Fig. 2b. The original PEU sample (permanent
tomer layer and the SMP might have served the role of shape C) was first deformed at a Treset > Tm,high by the appli-
the external stress. Specifically, the SMP tended to shrink cation of an external force. The deformation was then
upon heating while the elastomer provided an internal fixed at a Tlow < Tm,low . Upon stress removal, a shape B was
force to resist its shrinking. In essence, an internal tensile obtained. This completed the programming step, which
stress was generated on the SMP. Because of the macro- was actually identical to the programming step encoun-
scopic bilayer construction, the direction of this internal tered in a typical one way shape memory cycle. However,
stress was such that it caused bending instead of elon- it was at this stage that the rest of shape memory cycle
gation/contraction. Here, it was this internal stress that became different. The PEU sample was reheated to a Thigh
induced the network anisotropy for the semi-crystalline (Tm,low < Thigh < Tm,high ), leading to another shape A. What
SMP, thus enabling a CIE effect. Due to the elongation was truly intriguing was the discovery that the sample
of the SMP layer upon cooling, the bilayer reduced its exhibited a reversible shape transformation between shape
bending angle. More recently, Qi and Mather fabricated A and shape B when the temperature was cycled between
their own bilayer system with crosslinked semi-crystalline Thigh and Tlow . This established the reversible shape
associated with both crystalline phases were changed in

shape A. Reheating to Thigh removed only partially the
orientation in the PCL chains. The deformation fixed by
the PPDL domains was untouched, setting the network
anisotropy for the PCL domains without external force,
i.e. the PCL chains are in oriented conformation due to
the constraint by the deformed PPDL domains. In this
case, macroscopic CIE-to-MIC transformation of the PCL
domains could be induced without external force. There-
fore, the PEU network exhibits the 2W-SME. Overall, it
is this internally created network anisotropy that was
responsible for the reversible shape shifting. Since this
network anisotropy was determined by the deformation
during the programming step, the reversible shape switch-
ing can be reprogrammed. In a slightly different way, one
may also view the high temperature PPDL phase as the
phase that determines the internal stress that can be tuned
via programming. Thus, the mechanism around network
anisotropy or an internal stress unifies the different types of
reversible shape switching behavior for the various mate-
rials discussed throughout this section.
In a different design, Lendlein’s group extended the pre-
ceding material concept to a network exhibiting a broad
Tm transition, instead of two distinct Tm transitions [208].
A network was established by blending EVA copolymers
with different Tm followed by thermal crosslinking. This
led to a single broad melting transition from 30 ◦ C to 90 ◦ C.
Whereas such a network was showed to exhibit a TME
as mentioned before, it was found that the principle of
Fig. 18. The molecular mechanism for 2W-SME of a crosslinked crys-
the 2W-SME could also be applied here. According to the
talline polymer with an isotropic or anisotropic network structure. EMU theory in Section 6.1.3, a broad Tm could be regarded
as the continuous distribution of a succession of count-
less and infinitely sharp Tm s. For the PEU network, Thigh
changing behavior without any external stress. What was should be between Tm,high and Tm,low . Thigh for this EVA sys-
equally important was the programmable aspect of the tem, however, could be randomly chosen within its broad
observed behavior. Upon heating to Treset , the sample melting transition. Once the Thigh is chosen, the group of
would recover to the original shape C. At this stage, a EMUs with melting temperatures below this temperature
new programming step could start, through which the would play a role similar to the low melting temperature
reversible switch between two new shapes B and A phase in the PEU network. Similarly, the rest of the EMUs
could be realized. As shown in Fig. 2b, a sample can with melting temperatures above Thigh served the role of
be programmed to undergo multiple reversible shape the high melting temperature phase. Thus, the preceding
switching between two shapes. It could be reprogrammed principle for the 2W-SME observed for the PEU network
to reversibly switch between two completely different could be readily applied here for EVA network, except that
shapes, without any external stress present during the Thigh could be any temperature within the broad melting
reversible shape switching. This was the first realization transition. Experimental results confirmed that the 2W-
of a 2W-SME in its strict definition. SME could indeed be realized in this fashion. Depending on
Mechanistically, the authors suggested that the PPDL the chosen Thigh , however, the fractions of the low melt-
domains associated with the Tm,high were used to ing temperature EMUs versus high melting temperature
determine the shape shifting geometry while the low tem- EMUs would change accordingly. This would quantitatively
perature melting PCL domains were responsible for the impact the 2W-SME. Along the same line, Sheiko’s and
reversible shape shifting. The deformed PPDL domains coworkers [209] synthesized a poly(octylene adipate) net-
built a skeleton of crystallized geometry to orient the PCL work that also exhibited a broad melting transition. Via an
domains. Conformational orientation of the PCL segments identical mechanism but a different deformation mode, a
resulted in reversible geometric change of the PCL domains surface 2W-SME was successfully demonstrated [210].
during crystallization and melting. Consistent with the Pretsch and coworker [211], on the other hand, demon-
preceding description, we hereby proposed a further mech- strated a 2W-SME for a polyurethane elastomer. Upon
anistic picture with the concept of network anisotropy deformation to a large strain, strain-induced crystalliza-
discussed throughout this section. This is outlined in Fig. 18. tion occurred. Upon unloading, the authors discovered that
Beginning from the original shape C at Treset , the network the sample could undergo thermo-reversible elongation
chains adopt random chain conformation. Upon deforma- and contraction. Progressive growth of oriented crystallites
tion during the programming step, the chain conformations during cooling was believed to be the mechanism behind
such a behavior, as verified an in situ wide-angle X-ray a triple-SMP composite by replacing the rubber encapsu-
scattering study. Besides linear elongation and contraction, lating phase with a dual-SMP. This has been described in
geometrically more complex reversible shape switching details in Section 5.
(e.g., twisting) could also be realized by manipulation of
the crystalline orders in the first deformation step. This ini- 7.2. Four dimension printing
tial deformation step was thus called a training process,
the term commonly used for shape memory alloys. Being 3D printing has attracted significant interest lately
a different term, “training” was essentially “programming” due to its unlimited flexibility in terms of the geometric
used for 2W-SMPs in this section. complexity of the fabricated parts [216,217]. This unique
Despite differences in programming and the associ- capability is quite complementary to shape manipulation
ated materials for the various 2W-SMPs, the fundamental via the shape memory programming. Thus, combining
underlying mechanism is the same. The 2W-SME requires shape memory properties with 3D printing offers great
an internal way to create network anisotropy or an internal potential in two aspects: producing SMP devices with prac-
stress. This network anisotropy (or internal stress) should tically relevant complex geometries that cannot be easily
be erasable to allow reprogramming. made by traditional processing methods; more shape vari-
ants can be realized for a 3D printed SMP part via shape
7. Novel SMP fabrication methods memory programming. The time dependent SME offers an
additional dimension (i.e., time), leading to the so-called
Processing polymers into certain shapes is a neces- four dimensional 4D printing. In principle, 4D printing
sary step towards their eventual uses. Fabrication of SMP can be realized in two ways: a thermoplastic SMP can be
devices in the relevant and potentially complex shapes is 3D printed directly using fusion deposition modeling; UV
particularly important given the fact that their functions radiation curing of liquid SMP precursors. Qi, Dunn, and
are much more tightly and uniquely bound to the shapes. coworkers, however, reported an entirely different method
In many device applications of SMPs, the involved mate- [218,219]. Using a 3D multimaterial polymer printer, an
rials cannot provide the intended functions without been active composite was constructed with SMP fibers embed-
processed into proper macroscopic three dimensional (3D) ded in an elastomeric matrix. The orientation of the fibers
shapes or microstructured surfaces. In addition to tradi- was spatially controlled. Since the composite was printed
tional polymer processing methods (e.g., injection molding in 2D, its fabrication was fast relative to direct printing
and liquid molding) that can be used to produce rela- of multiple layers in 3D. Simple programming can lead to
tively simple shapes, several methods that lead to new complex 3D configurations with spatially nonuniform cur-
SMP design, capability, and highly complex shapes have vature due to the controllable internal stress (Fig. 19). The
emerged. These methods are introduced in this section. curvature can be manipulated by adjustment of the stretch-
ing strain. Upon heating, the 2D sheet can be recovered.
7.1. Electrospinning Overall, the most intriguing aspect of this technique is that
it reduces significantly the fabrication time by printing in
Electrospinning is a versatile method for preparing com- 2D, followed by its conversion to 3D.
plex two dimensional (2D) or 3D construction of fibrous
assemblies [212]. Thermoplastic SMP can be processed via 7.3. Photo-induced shape transformation
electrospinning and the method offers two benefits: fast
recovery kinetics due to the porous structures that facili- Permanent shapes of chemically crosslinked polymers
tate diffusion; the porous structures can be tuned via SME are defined by the topology of molecular network struc-
[41,213]. Mather’s group extended the method to make ture. By manipulating network topology, bulk shapes or
a composite with shape memory functions by combin- surface morphologies of the polymers can be accordingly
ing two non-SMP components [214]. A nonwoven fiber controlled. Bowman’s group developed two methods that
mat was formed by electrospinning of a thermoplastic can locally vary network topology induced by photo irra-
crystalline polymer. The mat was then encapsulated with diation, including photo-induced plasticity and two-stage
silicone rubber. The resulting composite exhibited shape curing. These two methods have been used to create com-
memory functions with the polymer fibers as the switch- plex shapes utilizing light induced and spatially defined
ing phase, while the encapsulating rubber phase ensured internal stress.
the permanent shape. Such an approach is highly versa- Covalent networks have been synthesized
tile in that it allows individual tuning of the properties incorporating allyl sulfide groups, providing the
of the two components, thus the overall shape memory addition–fragmentation chain transfer capability (Fig. 20a)
behavior. For instance, when the fibers were aligned dur- [220,221]. Due to the presence of a photo-initiator, light
ing the electrospinning process, the composite exhibited irradiation created radicals, which induced the chain
strong anisotropy in Young’s modulus, strain-to-failure, transfer within the network. This introduced a dynamic
yield stress, and shape fixing ratio (Rf ) [215]. In addition character, potentially allowing alteration of the topology of
to the predefined fiber alignment, isotropic SMP fibers can the networks without changing the network composition
be temporarily oriented through shape memory program- and connectivity. When a sample was strained under a
ming. This enabled an isotropic/anisotropic transition that certain force and subsequently irradiated with spatially
may be useful as dynamic tissue engineering scaffolds. The defined light through a photomask, the exposed regions
versatility of the method is also reflected in the making underwent network rearrangement, that led to localized
Fig. 19. Complex low-temperature shapes of 4D-printed laminates. (a) A two-layer laminate with one layer consisting of oriented SMP fibers embedded
in an elastomer and the other layer being a pure elastomer matrix. The sample was heated, stretched, cooled, and released. Upon release of the stress it
assumes a complex shape, depending on the laminate architecture. When reheating it assumes its original shape, a flat rectangular strip. (b) Shows an
actual strip in its original shape and (c)–(h) show results of this process with differing fiber architectures [218], Copyright 2013. Reprinted with permission
from AIP Publishing LLC.
stress relief (or photo-induced plasticity) (Fig. 20b). As a [225], a stoichiometric thiol-isocyanate system with excess
result, the polymer chains in the exposed regions were in reactive acrylate groups was used to construct a first net-
an unstrained state while those in the unexposed regions work at stage I. By controlling the light irradiation time
were still strained. After cessation of the external force, and intensity at stage II, intricate property gradients can be
buckling features emerged due to the recovery in the incorporated into the material. Due to the light attenuation
unexposed regions (Fig. 20c). These features were deter- along the penetration depth, the crosslinking density at
mined by the mask design. In a similar demonstration, stage II was significantly different between the surface and
photo-reversible disulfide bonds were incorporated in a the interior of the material. Therefore, gradient modulus
hydrogel network, resulting in a similar photo-induced and Tg along the irradiation direction were endowed within
rearrangement of network topology [222]. We note that the network. Due to the internal stress mismatch between
the preceding two examples did not involve SMP as the the exposed and unexposed regions upon straining and
subject of study. However, the same principle can be subsequent heating induced stress relief of the bulk mate-
readily applied to SMPs. Indeed, Rowan group synthesized rial, surface buckling or bulk geometric transformations
a polydisulfide network that exhibited both thermally- could be achieved. Similarly, electron beam irradiation was
induced SME and photo-induced plasticity [223], the also applied to manipulate the crosslinking density in order
latter allowing manipulation of the permanent shapes. to finely tune material properties such as rubbery modulus,
In their work, bisthiol oligomer synthesized via thiol-ene strain capacity, ultimate tensile strength, recovery stress,
chemistry between 1,6-hexanedithiol and 1,5-hexadiene and glass transition temperature.
was crosslinked by tetrathiol in the form of disulfide bonds
(Fig. 20d). The network exhibited a Tm around 61 ◦ C that 7.4. Origami-inspired manufacturing
enabled a thermally-induced SME. More importantly, the
permanent shape could be repeatedly changed under UV Complex 3D structures may also be created via folding
light exposure. This allowed complex permanent shapes a 2D sheet, the essence of the origami art. Locally-induced
to be obtained via deformation under photo-irradiation, shape recovery of a SMP sheet can lead to a self-folding
instead of traditional molding or remolding (Fig. 20e). effect. This technique, so-called origami-inspired manufac-
Besides photo-induced plasticity, a two-stage-curing turing, is attractive for a number of applications such as
technique can also be applied to manipulate SMP per- packaging and mechanical actuation.
manent shapes [224]. It refers to dual curing reactions Dickey’s group demonstrated the self-folding of an SMP
that can selectively occur to form two types of netpoints sheet triggered by infrared heating [179]. The biaxially
within a polymer network. From a non-stoichiometric stretched sheet in its neat SMP form was optically trans-
thiol-acrylate system possessing excess acrylate groups, a parent and displayed in-plane shrinkage upon uniform
first polymer network can be formed via a base catalyzed heating. When black ink was locally patterned on one
stoichiometric thiol-acrylate Michael addition (stage I). The side of the sheet, the patterned regions could absorb opti-
excess acrylate groups can be further polymerized upon cal energy to heat the underlying SMP. The light induced
UV irradiation (stage II). In stage II, if the UV irradiation heating is mostly limited to the patterned regions on one
was locally applied on a strained first network, the strained side of the sheet. As a result, the patterned side shrunk
network topology in the exposed regions would be perma- while the opposite side did not, leading to an overall
nently fixed by the newly formed netpoints. In contrast, self-folding effect. This intriguing approach is simple and
the unexposed regions would recover to its initial state efficient, however, the folding in different regions can
upon relief of the external force. As a result, light-defined only be simultaneously triggered. In contrast, Felton et al.
buckling 3D features could be formed. In another work [226] developed a method that could achieve sequential
Fig. 20. Photo-induced complex shape transformations. (a) Monomers and reaction mechanism for the addition–fragmentation chain transfer within the
polymer backbone; (b) Mechanism of the mechanophotopatterning: the material is uniaxially stretched and UV irradiated though a mask; (c) The hexagonal
pattern obtained after process (b) (scale bar = 1 mm) [221], Copyright 2011. Reprinted with permission from John Wiley and Sons; (e) Synthesis of the
semicrystalline polydisulfide network; (f) Visual demonstrations showing the reprogrammable shape memory properties of the polydisulfide network: (1
and 2). A flat specimen is fixed into a helical shape and subsequently recovered to the permanent flat shape; (3). The permanent shape of the film can be
photo-reprogrammed by holding it in the helical shape while exposing it to UV light; (4 and 5). Programming and recovery of the new permanent shape
[223], Copyright 2013. Reprinted with permission from the American Chemical Society.
self-folding. Composite sheets were constructed by lami- 8. Modeling of thermomechanical behavior of SMPs
nating three layers: a pre-stretched SMP, a piece of scored
paper, and resistive copper traces (as shown in Fig. 21). As being realized by many researchers, the SME is
Heating induced shrinkage of the SMP layer caused the lam- a response of materials under a special set of thermo-
inate to bend. The bending behavior would only occur at mechanical loading conditions, rather than a material
the scored location via joule heating of the copper traces, property. Therefore, SME is determined not only by mate-
leading to localized folding. Moreover, the fold could be rial properties, but also by complicated thermomechanical
triggered simultaneously or sequentially upon appropriate loading conditions. In many cases, the latter complicates
on-off control of the copper traces. Through the origami- the evaluation of shape memory performance as well as
inspired approach, 3D complex structures can be created material design. For example, material properties, such
by an efficient 2D printing technique. Parameters within as glass transition temperature and the width of glass
the 2D fabrication can be modeled and controlled, because transition, affect the shape stability and shape recovery;
the folding behavior (the maximum fold angle) is strongly in addition, these features may be affected by the temper-
affected by the composite thickness, the score depth, and atures at which the material is deformed and the shape
the trace width. is fixed [227–232]. Therefore, it is well recognized that
Fig. 21. The construction and operating mechanism of a self-folding composite. (a) an SMP layer, (b) silicone tapes, (c) a polyimide sheet with copper trace,
and (d) a paper substrate scored along the hinge in the middle; (e and f) The composite bends at the hinge due to the contraction of the SMP layer upon
activation; (g–j) A locking slot and tab assembly. This fold is only possible with sequential folding [226], Copyright 2013. Reprinted with permission from
Royal Society of Chemistry.
SME modeling should play an important role in guiding 8.1. Thermoviscoelastic models
the design of SMP applications as well as the design of
materials for specific applications. Amorphous polymer based SMPs consist of a large por-
In SME modeling, one approach is to develop constitu- tion of SMPs currently studied, and are therefore heavily
tive models (or thermomechanical constitutive models) for investigated. The molecular mechanism of this group of
the polymer. The expectation is that one can then use such SMPs is the large change in chain mobility during the glass
models, together with simulation tools, such as finite ele- transition [5]. From polymer physics or mechanics point of
ment methods, to simulate the shape memory behavior. view, this chain mobility change is reflected in the change
Another approach is to directly predict shape memory per- of relaxation time or viscosity [238–241], encompassed in
formance. Here, although the model development is based viscoelastic behavior. Some good discussions on how shape
on certain assumption of constitutive models for polymers, memory effect is an outcome of temperature-dependent
the end result is an analytical solution that can directly pre- viscosity can be found in a few recent works [171,236,237].
dict the shape memory performance, such as shape stability It is therefore natural to use viscoelastic models where the
or shape recovery. To date, the majority of modeling efforts viscosity is a function of temperature, or the so-called ther-
have been on the development of constitutive models for moviscoelasticity, to capture the shape memory effect. In
SMPs; very few models have been developed in the second the early models developed by Tobushi and co-workers
group. [242–244], viscosity and relaxation time were not explic-
In continuum mechanics, constitutive models are itly written as functions of temperature. Later, the concept
defined as equations that link state functions, such as of thermoviscoelasticity was used by Nguyen et al. [245]
internal energy, entropy, heat flux, and Cauchy stress, and Castro et al. [227] where the relaxation time was
to state variables, such as deformation gradient, tem- considered to follow the principle of time-temperature
perature, etc., sometimes with the help of internal superposition [239,246] and the temporary shape is a direct
variables. From such a definition, it is clear shape mem- result of frozen viscous strains in nonequilibrium elements.
ory effects are an exhibition of constitutive behavior It should be noted that the phenomenological model frame
under complicated thermomechancial conditions. There- developed by Liu and coworkers [247,248] where the glass
fore, constitutive models of shape memory polymers can transition is considered as a phase transition served as the
sometimes also be found in those for polymers going basis for many modeling works in recent years [249–256].
through thermomechanical transitions. For example, in In many of these models, a storage strain (initially proposed
a recent work by Srivastava et al. [233], shape mem- by Liu et al. [247]) was also used as a mechanism to cap-
ory behavior are naturally captured under the framework ture the shape memory effect. In addition, the slip element
of a constitutive model of polymers for finite deforma- for internal friction for shape memory effect by Tobushi
tion behavior spanning across glass transition temperature and co-workers [242–244] was recently used for 3D models
[234]. [257–264].
A large number of constitutive models have been devel- One advantage of thermoviscoelastic models is that it
oped for SMPs and several excellent reviews are available closely follows the underlying polymer physics for shape
for these models [9,235–237]. It is therefore not necessary memory effects in amorphous SMPs and thus can be easily
to repeat the review of these models again. Below, we will used to consider other effects, such as aging and solution
focus on describing the prevailing concepts in constitutive driven SME, into constitutive model. For example, when
modeling as well as some efforts in using simple models to both time-temperature dependent structural relaxation (or
estimate shape memory performance. physical aging) and stress relaxation were considered, Choi
et al. [265] showed that the initial shape recover speed can used their thermo-mechanically coupled finite deforma-
be improved. Li and co-workers [266] used the thermovis- tion constitutive model for polymers to study the shape
coealstic model to investigate a SMP based syntactic form. memory effects [233]. There, they argued that there should
It is known that water (or solution) can be used as a means M-micromechanisms that contribute to the relaxation
for shape memory effect [52,267], because small molecules, processes in the polymer; each micromechanism can be
such as water, can serve as plasticizers to decrease the glass modeled by a nonlinear Maxwell element, for which flow
transition temperature, or effectively increase the relax- behavior depends on the specific micromechanism and
ation time, effects included in models by Xiao and Nguyen may require a large number of material parameters. In
[268] and Lu and coworkers [269,270]. For example, in Xiao order to reduce the number of material parameters, they
and Nguyen, the structural relaxation time can be written chose to use only three micromechanisms, one for inter-
as molecular interaction mechanism, and two for molecular
network resistance. In the model by Srivastava and cowork-
Rk (T, Tf , ) = a(T, Tf , )Rk (T, Tf , 0), (6) ers [232,233], although each micromechanism can find its
where is the solution concentration, T is the tempera- own physical explanation, identifying these micromecha-
ture, Tf is the fictive temperature of the nonequilibrium nisms for polymers can be a challenge.
structure, a(T, Tf , ) is a temperature-dependent solution- stress relax-
In a simplified viscoelastic representation,
shifting factor, Rk (T, Tf , 0) is the dry-state relaxation time, ation is represented by the function exp − t , where t is
time and is the relaxation time. It is easy to see that this
t = 0.1,
factor starts at 1 when t = 0, decays to 0.9 when
a(T, Tf , 0) = a−1 (T, Tf , 0) and to 4.5 × 10−5 when t = 10. Therefore, exp − t decays
from 0.9 (a noticeable change) to almost zero within two
−1
B decades, too fast to be realistic for most engineering poly-
exp ,
T C −1 T2−1 − Tf−1 − [ln () + (1 − )ln (1 − )] mers, as it usually take five or more decades for a stress to
(7a) be relaxed in a typical polymer. For a simple mechanical
loading, this is sometimes approximately accommodated
by using the so-called KWW (Kohlrausch–Williams–Watts)
⎧ ⎫ stretched exponential, i.e.,
⎨ ⎬
a(T, Tf , 0) = exp
B
, (7b) t
ˇ
⎩ T C T −1 − T −1 ⎭ exp −

, 0 < ˇ ≤ 1. (8)
2 f
The simple exponential form is recovered for ˇ = 1,

Rk (T, Tf , 0)
with a smaller value of ˇ required to approximate the
g
slower relaxation process characteristic of polymers over
g
g C1 C2 (T − Tf ) + Tr − Tgref
= R exp − g
, (7c) the time scale of interest, e.g., values for ˇ are often
k log (e) T C2 + Tr − Tgref around 0.5 [273]. One disadvantage of Eq. (8) is that it
is difficult to be convert into a differential form, pre-
g
where R is the relaxation time at an equilibrium reference ferred for developing constitutive models for finite element
k
temperature at dry state, T2 is the Kauzmann temperature based simulations. It is known that the relaxation behavior
for an ideal glass. Note the exponential term in Eq. (7c) described by Eq. (8) can also be captured by a multi-
is the shifting factor for time-temperature superposition, branch model (or Prony series, or generalized standard
g g
where C1 and C2 are two parameters in the WLF equation linear solid model) where multiple Maxwell elements
[246]. More recently, a similar approach was used by Lu are arranged in parallel. Therefore, it becomes a natural
and coworkers [269,270] to consider effects of moisture on choice to use multi-branch models for thermoviscoelas-
the shape memory effect. ticity. In fact, as demonstrated by Diani et al. [274] and
Early thermoviscoelastic models [242–245] for SMP Yu et al. [171], the multi-branch model, when being used
assumed stress relaxation is a relatively simple process that together with the time-temperature superposition, can
can be described by one relaxation process (or one Maxwell adequately capture the shape memory behavior. In both
or Kelvin-Viogt element in one dimensional rheological works, since the focuses were the shape memory behav-
representation). However, real polymers have multiple ior (instead of stress–strain behavior at high and low
relaxation modes, due to the generally complicated macro- temperatures), linear Maxwell elements were used in indi-
molecular structures, therefore requiring more branches vidual nonequilibrium branches. For the comprehensive
in the constitutive model. For example, to model poly- constitutive models for finite deformation thermomechan-
mer stress–strain behavior from low to high strain rates, ical behavior, Westbrook et al. [275] used one nonlinear
Mulliken and Boyce [271] used two Maxwell branches to Maxwell branch to capture the yielding behavior at tem-
capture the ␣- and ␤-transitions, respectively, that can perature below glass transition temperature and multiple
be excited at different strain rates. In the modeling of linear Maxwell branches for time-dependent stress–strain
finite deformation behavior of amorphous polymers in behavior at temperatures above glass transition; Xiao et al.
and above the glass transition, Dupaix and Boyce [272] [276] used multiple nonlinear nonequilibrium Maxwell
used two branches to represent intermolecular resistance elements. As shown in these models, when multiple non-
and molecular network resistance. Srivastava et al. [234] linear branches are used, the models show improved
capability of capturing stress–strain behavior at differ- phases as well as the external load applied. In addition,
ent temperatures as well as shape recovery behavior in as it can be seen from Fig. 22d, even though the sample
a shape memory cycle. However, the trade-off is more has no external load after unloading, the material is not
material parameters involved, especially when nonlinear stress-free.
Maxwell elements are used. In additional, the multibranch If the phase formation and mechanical deformation
approach was shown to be effective in capture the multi- occur simultaneously, the stress evolution becomes rather
shape memory effects due to the broad temperature range complicated. For example, Long et al. [305] derived the
of thermomechanical transition [171]. Cauchy stress tensor for an incompressible elastomer that
undergoes thermally induced bond exchange reaction,
8.2. Phase evolution models
t
In semi-crystalline polymers, polymer crystallization COR (, t)
=− 1+ ˝[I(t)]Bet + ˝[H(, t)]Be→t d
can be used as an effective mechanism for shape mem- Jt Jt C 0 Jt C 0
ory effects. This is because the newly crystallized portion 0T −1 (9)
of the material (or phase) is stress-free and thus takes Be→t = F e→t F e→t , F e→t = F et (F e ) ,
the current configuration as its undeformed configuration.
The stress-free state of newly formed crystal phase was
T
e
H(, t) = tr B→t , I(t) = tr F et F et ,
referred as a natural configuration by Rajagopal [277,278].
Such concept was widely used by many researchers to
study rubbers with thermally induced scissoring and ref- where COR and (, t) are chain densities of original chains
and reformed chains, F→te is the mechanical deformation
ormation [279–282], and crystallizing polymers [283–286].
Since polymer crystallization is a time consuming process, gradient measured from time to time t, Fte is the mechan-
the crystallization process may be coupled with mechanical deformation gradient measured at time t, (y) is the
ical deformation, such as in the recent study of two-way inverse Langevin function scaled by material parameters, p
shape memory effect in a semi-crystalline polymer where is the Langrange multiplier to enforce isothermal incom-
a constant force was applied during crystallization pro- pressibility, Jt is the total volume change. It should be
cess [196]. In this case, small polymer crystals formed a noted that the integral term in Eq. (9) has the nature of
different moment have different natural (or reference) con- convolutional integrals and therefore is computationally
figures. Therefore, the evolution of stress-free state is also very expensive. A potential solution is to use a so-called
called multi-natural configuration approach [9], or phase effective phase to homogenize different evolving phases
evolution approach [287]. Such a concept was also the [286]; its 3D form, however, still needs to be developed and
foundation for some of the recent constitutive models for verified.
shape memory polymers [288–291] and two-way shape
memory effects [206,292]. In addition, if one takes “phase”
as a more general representation of a small collection of 8.3. Models for predicting shape memory performance
material fraction that has distinct deformation history, one
will find that many active materials can achieve shape While most of the constitutive models developed so
memory effects or its active motion through the phase far target capturing the comprehensive thermomechani-
evolution mechanism, such as light activated polymers cal behavior, they are also very complicated and not so
developed by Lendlein et al. [53] and Scott and cowork- easy to be used. In recognition of this drawback, there are
ers [220,293], covalent adaptable network polymers with some modeling efforts that targets providing simple meth-
bond exchange reaction [294,295], Diels–Alder network ods to predict shape memory behavior of a polymer. Diani
[296,297], ionic gels [298]. Several constitutive models et al. [274] used Prony series to predict the shape mem-
were developed in the recent years to model these material ory behavior and achieved excellent agreement. To predict
systems, such as light activate polymers [299–301], shape the shape recovery time, Bonner et al. [307] proposed to use
memory elastomeric composites [302] and its triple shape the recovery half life (similar to relaxation time). Since each
memory behavior [303,304], polymers with temperature- recovery half life time recovers 50% of the remaining defor-
dependent bond exchange reactions [305], and Diels–Alder mation, seven times of recovery half life time give 99.2% of
networks [306]. The concept of employing phase formation shape recovery. For SMPs deformed and programmed at
to achieve shape memory effect can be easily illustrated by different temperatures, they found that the recovery half
Fig. 22. A material in its homogeneous phase state (Fig. 22a) time can be calculated as the ratio between the viscosity
is stretched (Fig. 22b). While the load is held, due to envi- measured at the recovery temperature and the initial mod-
ronmental change or external stimulus, new phases (such ulus at the deformation temperature. Under the testing
as crystals) are formed (Fig. 22c). Note that these new conditions in their paper, good agreement was achieved.
phases are in a stress-free state. Upon unloading, how- Since it is well known that shape recovery of a SMP depends
ever, the material cannot return to its original shape due strongly on the recovery temperature, Ge et al. [308] devel-
to the requirement of force balance. Instead, it takes a new oped an analytical approach to predict shape recovery. By
temporary shape (Fig. 22d). Upon another environmen- using the KWW stretch exponential to describe the vis-
tal change or external stimulus, the material recovers its coelastic behavior of the SMP, for a temperature above Tr ,
original shape. Here it should be noted that the temporary they found that the recovery ratio ␨as a function of time
shape depends the volume fraction and distribution of new can be given by
Fig. 22. Phase formation and shape memory effect.
⎧ ⎫
⎪ L )/Q
⎨(Tr −T t ⎪
⎬
ˇ AFc 1 1
ln =− exp − + 10(C1 (TL +Qt−Tr ))/(C2 +(TL +Qt−Tr )) t ˇ−1 dt, (8)
ˇ
0 1 + E1 /E0 ⎪
⎩ kb TL + Qt Tr ⎪
⎭
0 (Tr −TL )/Q
where E0 and E1 are the equilibrium and nonequilibrium

1. “2W-SME. Can the phenomenon be extended beyond
moduli, respectively, ˇ is the parameter in KWW exponen-
crosslinked semi-crystalline polymers? Besides
tial, Q is the heating rate, 0 is the relaxation time measured
reversible bending, can a 2W-SME be realized that
at a reference temperature Tr , kb is Boltzmann’s constant,
allows reversible shape switching in a non-linear
AFc , C1 , and C2 are the parameters in time-temperature
fashion under zero external stress? Shape memory
superposition principle. The advantage of this solution is
alloys can be trained to exhibit two-way reversible
that all the parameters in the model can be measured by
surface topographic changes. Can this be realized with
a few simple rheological experiments, such as DMA tests
polymers?”
and stress relaxation tests. However, this model cannot
2. Multi-SME. Can we as a research community come up
consider the effects of programming temperature on the
with the new and exciting application concepts that best
recovery, that can significantly affect the recovery behavior
utilize the versatile multi-SME?
[229–231,309,310]. More recently, Yu et al. [311] systemat-
3. “Recovery stress. Can an SMP be developed that exhibits
ically studied the factors affecting shape stability and shape
recovery stress comparable to shape memory alloys,
recovery rate. They found that shape stability is uniquely
without compromising shape fixing and recovery behav-
determined by the relaxation modulus before unloading.
ior?”
According to this observation, a longer holding time after
loading can give higher shape stability, but loading at a
Looking back at these questions, clearly the biggest
lower temperature will give lower shape stability. In addi-
advance and exciting surprise within a short three years has
tion, the recovery is determined by the shape stability
been the discovery of the true 2W-SME for semi-crystalline
and lower shape stability normally gives a fast recovery.
polymers. Intriguingly, recent study has shown that poly-
Finally, the shape memory behavior under different the-
mer crystals formed via strain-induced crystallization can
momechanical loading conditions can be unified by using
be manipulated to precisely tune the melting tempera-
the reduced time. For example, by using reduced time, a
ture [312–314]. This may open up new ways to induce
master recovery curve can be constructed.
2W-SME. Although the 2W-SME has only been demon-
strated for semi-crystalline polymers, indications that this
can be extended to other types of polymers were also
9. Summary and outlook reported such as the quasi 2W-SME for the ionomeric
Nafion [315]. As for the multi-SME, materials exhibiting
In this review, we presented notable recent progress such a behavior have expanded significantly and molec-
in the field of SMPs. Although conventional SMEs were ular (instead of phase) multi-switching mechanisms have
briefly introduced, the main emphasis was placed on also been demonstrated, yet concepts on new applications
new enabling mechanisms such as molecular switches, have not advanced much. Not such exciting development
new shape memory behavior (multi-SME and 2W-SME), has been reported in terms of the recovery stress. This is
and unconventional SMP fabrication methods. Molecular perhaps due to the lack of sufficient interest. The SMP com-
enabling mechanisms were presented from the standpoint munity has increasingly realized that SMPs as soft materials
of underlying polymer physics. Various modeling meth- would be best used in ways that are different from shape
ods, providing further understanding from a different angle memory alloys. Thus, an ultra-high recovery stress may not
(i.e., solid mechanics), were also reviewed. On closing, we be urgently needed.
hereby go back to a set of questions, challenges, and oppor- Other notable advances include the enrichment of
tunities that we laid out in a review [4] published in 2011: molecular switching mechanisms, which have naturally led
to novel non-thermal based triggering methods that may [14] Huang WM, Song CL, Fu YQ, Wang CC, Zhao Y, Purnawali H, et al.
be critical for some end applications. The various applica- Shaping tissue with shape memory materials. Adv Drug Delivery
Rev 2013;65:515–35.
tion concepts based on the surface SME are particularly [15] Wischke C, Lendlein A. Shape-memory polymers as drug carriers-a
noteworthy. Some of these concepts have the potential multifunctional system. Pharm Res 2010;27:527–9.
to become commercialized in the near term. Also on SMP [16] Kuder IK, Arrieta AF, Raither WE, Ermanni P. Variable stiffness mate-
rial and structural concepts for morphing applications. Prog Aerosp
applications, a notable development at the bulk device level Sci 2013;63:33–55.
is the use of SMPs as substrates for electronic components, [17] Liu YJ, Du HY, Liu LW, Leng JS. Shape memory polymers and their
leading to electronics with complex three dimensional composites in aerospace applications: a review. Smart Mater Struct
2014;23:023001/1–23001.
shapes that are otherwise difficult to fabricate without
[18] Zhang L, Du HY, Liu LW, Liu YJ, Leng JS. Analysis and design of
using the shape memory properties [316–319]. Mechanical smart mandrels using shape memory polymers. Composites, B
design is very important to achieve versatile shape mor- 2014;59:230–7.
[19] Weiss RA, Izzo E, Mandelbaum S. New design of shape memory
phing. On this front, the unconventional SMP fabrication
polymers: mixtures of an elastomeric ionomer and low molar mass
methods summarized in this review have been valuable fatty acids and their salts. Macromolecules 2008;41:2978–80.
additions. With all the exciting progress within such a short [20] Behl M, Razzaq MY, Lendlein A. Multifunctional shape-memory
time, we anticipate the SMP research to continue on its polymers. Adv Mater 2010;22:3388–410.
[21] Leng JS, Lan X, Liu YJ, Du SY. Shape-memory polymers and their
explosive pace in the near future. composites: stimulus methods and applications. Prog Mater Sci
2011;56:1077–135.
[22] Hsu L, Weder C, Rowan SJ. Stimuli-responsive, mechanically-
Acknowledgments adaptive polymer nanocomposites. J Mater Chem
2011;21:2812–22.
[23] Ratna D, Karger-Kocsis J. Recent advances in shape memory poly-
TX thanks the financial support from the following mers and composites: a review. J Mater Sci 2008;43:254–69.
programs: the National Key Basic Research Program of [24] Meng QH, Hu JL. A review of shape memory polymer composites
and blends. Composites, A 2009;40:1661–72.
China (Grant no. 2015CB351903), National Natural Sci-
[25] Liu YJ, Lv HB, Lan X, Leng JS, Du SY. Review of electro-active shape
ence Foundation of China (Grant no. 21474084); the memory polymer composite. Compos Sci Technol 2009;69:2064–8.
Chinese central government’s Recruitment Program of [26] Huang WM, Yang B, Zhao Y, Ding Z. Thermo-moisture responsive
Global Experts; and the 985 program for the startup polyurethane shape-memory polymer and composites: a review. J
Mater Chem 2010;20:3367–81.
funding. QZ acknowledges the support from the State [27] Hearon K, Singhal P, Horn J, Small W, Olsovsky C, Maitland KC, et al.
Key Laboratory of Chemical Engineering (No. SKL-ChE- Porous shape-memory polymers. Polym Rev 2013;53:41–75.
14D04) and Fundamental Research Funding of Zhejiang [28] Hu JL, Lu J, Zhu Y. New developments in elastic fibers. Polym Rev
2008;48:275–301.
University (No. 2014QNA4021). HJQ acknowledges sup- [29] Kotek R. Recent advances in polymer fibers. Polym Rev
port from US the Air Force Office of Scientific Research 2008;48:221–9.
(FA9550-13-1-088, Dr. B.-L. “Les” Lee, Program Manager), [30] Hu JL, Chen SJ. A review of actively moving polymers in textile
applications. J Mate Chem 2010;20:3346–55.
US National Science Foundation (CMMI-1404621, CMMI- [31] Wagermaier W, Kratz K, Heuchel M, Lendlein A. Characteri-
1404627, EFRI-1435452, EFRI-1038305). zation methods for shape-memory polymers. Adv Polym Sci
2010;226:97–145.
[32] Xie T, Rousseau IA. Facile tailoring of thermal transition tem-
References peratures of epoxy shape memory polymers. Polymer 2009;50:
1852–6.
[33] Behl M, Kratz K, Zotamann J, Nöchel U, Lendlein A. Reversible bidi-
[1] Liu F, Urban MW. Recent advances and challenges in designing rectional shape-memory polymers. Adv Mater 2013;25:4466–9.
stimuli-responsive polymers. Prog Polym Sci 2010;35:3–23. [34] Wu JJ, Lin YT, Sun JZ. Anisotropic volume change of poly(N-
[2] Qiu Y, Park K. Environment-sensitive hydrogels for drug delivery. isopropylacrylamide)-based hydrogels with an aligned dual-
Adv Drug Delivery Rev 2001;53:321–39. network microstructure. J Mater Chem 2012;22:17449–51.
[3] Yang Z, Herd GA, Clarke SM, Tajbakhsh AR, Terentjev EM, Huck WTS. [35] Kim J, Hanna JA, Byun M, Santangelo CD, Hayward RC. Designing
Thermal and UV shape shifting of surface topography. J Am Chem responsive buckled surfaces by halftone gel lithography. Science
Soc 2006;128:1074–5. 2012;335:1201–5.
[4] Xie T. Recent advances in polymer shape memory. Polymer [36] Ma CX, Li TF, Zhao Q, Yang XX, Wu JJ, Luo YW, et al. Supramolec-
2011;52:4985–5000. ular Lego assembly towards three-dimensional multi-responsive
[5] Lendlein A, Kelch S. Shape-memory polymers. Angew Chem Int Ed hydrogels. Adv Mater 2014;26:5565–9.
2002;41:2034–57. [37] Zhou SB, Zheng XT, Yu XJ, Wang JX, Weng J, Li XH, et al. Hydrogen
[6] Behl M, Zotzmann J, Lendlein A. Shape-memory polymers and bonding interaction of poly(d,l-lactide)/hydroxyapatite nanocom-
shape changing polymers. Adv Polym Sci 2010;226:1–40. posites. Chem Mater 2007;19:247–53.
[7] Liu C, Qin H, Mather PT. Review of progress in shape-memory poly- [38] Dong J, Weiss RA. Shape memory behavior of zinc oleate-filled elas-
mers. J Mater Chem 2007;17:1543–58. tomeric ionomers. Macromolecules 2011;4:8871–9.
[8] Mather PT, Luo XF, Rousseau IA. Shape memory polymer research. [39] Shi Y, Yoonessi M, Weiss RA. High temperature shape memory
Annu Rev Mater Res 2009;39:445–71. polymers. Macromolecules 2013;46:4160–7.
[9] Hu JL, Zhu Y, Huang HH, Lu J. Recent advances in shape-memory [40] Dolog R, Weiss RA. Shape memory behavior of a polyethylene-
polymers: structure, mechanism, functionality, modeling and based carboxylate ionomer. Macromolecules 2013;46:7845–52.
applications. Prog Polym Sci 2012;37:1720–63. [41] Gu XZ, Mather PT. Entanglement-based shape memory
[10] Meng H, Mohamadian H, Stubblefield M, Jerro D, Ibekwe S, Pang polyurethanes: synthesis and characterization. Polymer
SS, et al. Various shape memory effects of stimuli-responsive shape 2012;53:5924–34.
memory polymers. Smart Mater Struct 2013;22:093001/1–93001. [42] Agarwal P, Chopra M, Archer LA. Nanoparticle netpoints for shape-
[11] Lendlein A, Behl M, Hiebl B, Wischke C. Shape-memory polymers memory polymers. Angew Chem Int Ed 2011;50:8670–3.
as a technology platform for biomedical applications. Expert Rev [43] Zhang YM, Wang QH, Wang C, Wang TM. High-strain shape
Med Devices 2010;7:357–79. memory polymer networks crosslinked by SiO2 . J Mater Chem
[12] El Feninat F, Laroche G, Fiset M, Mantovani D. Shape memory mate- 2011;21:9073–8.
rials for biomedical applications. Adv Eng Mater 2002;4:91–104. [44] Du K, Gan ZH. Shape memory behaviour of HA-g-PDLLA nanocom-
[13] Safranski DL, Smith KE, Gall K. Mechanical requirements of posites prepared via in situ polymerization. J Mater Chem B
shape-memory polymers in biomedical devices. Polym Rev 2014;2:3340–8.
2013;53:76–91.
[45] Ishida K, Yoshie N. Two-way conversion between hard and [73] Wojtecki RJ, Meador MA, Rowan SJ. Using the dynamic bond to
soft properties of semicrystalline cross-linked polymer. Macro- access macroscopically responsive structurally dynamic polymers.
molecules 2008;41:4753–7. Nat Mater 2011;10:14–27.
[46] Inoue K, Yamashiro M, Iji M. Recyclable shape-memory polymer: [74] Yang Y, Urban MW. Self-healing polymeric materials. Chem Soc Rev
poly(lactic acid) crosslinked by a thermoreversible Diels–Alder 2013;42:7446–67.
reaction. J Appl Polym Sci 2009;112:876–85. [75] Aoki D, Teramoto Y, Nishio Y. SH-containing cellulose acetate
[47] Defize T, Riva R, Raquez JM, Dubois P, Jerome C, Alexandre derivatives: preparation and characterization as a shape memory
M. Thermoreversibly crosslinked poly(e-caprolactone) as recy- recovery material. Biomacromolecules 2007;8:3749–57.
clable shape-memory polymer network. Macromol Rapid Commun [76] Yamauchi K, Lizotte JR, Long TE. Thermoreversible poly(alkyl
2011;32:1264–9. acrylates) consisting of self-complementary multiple hydrogen
[48] Rousseau IA, Xie T. Shape memory epoxy: composition, struc- bonding. Macromolecules 2003;36:1083–8.
ture, properties and shape memory performances. J Mater Chem [77] Li JH, Viveros JA, Wrue MH, Anthamatten M. Shape-memory effects
2010;20:3431–41. in polymer networks containing reversibly associating side-groups.
[49] Yakacki CM, Shandas R, Safranski D, Ortega AM, Sassaman K, Gall K. Adv Mater 2007;19:2851–5.
Strong, tailored, biocompatible shape-memory polymer networks. [78] Li JH, Lewis CL, Chen DL, Anthamatten M. Dynamic mechanical
Adv Funct Mater 2008;18:2428–35. behavior of photo-cross-linked shape-memory elastomers. Macro-
[50] Voit W, Ware T, Dasari RR, Smith P, Danz L, Simon D, et al. High- molecules 2011;44:5336–43.
strain shape-memory polymers. Adv Funct Mater 2010;20:167–71. [79] Zhu Y, Hu JL, Liu YJ. Shape memory effect of thermoplastic
[51] Li JJ, Rodgers WR, Xie T. Semi-crystalline two-way shape memory segmented polyurethanes with self-complementary quadruple
elastomer. Polymer 2011;52:5320–5. hydrogen bonding in soft segments. Eur Phys J E 2009;28:3–10.
[52] Huang WM, Yang B, An L, Li C, Chan YS. Water-driven pro- [80] Kushner AM, Vossler JD, Williams GA, Guan ZB. A biomimetic mod-
grammable polyurethane shape memory polymer: demonstration ular polymer with tough and adaptive properties. J Am Chem Soc
and mechanism. Appl Phys Lett 2005;86:114105/1–114105. 2009;131:8766–8.
[53] Lendlein A, Jiang HY, Junger O, Langer R. Light-induced shape- [81] Kumpfer JR, Rowan SJ. Thermo-, photo-, and chemo-responsive
memory polymers. Nature 2005;434:879–82. shape-memory properties from photo-cross-linked metallo-
[54] Wu LB, Jin CL, Sun XY. Synthesis, properties, and light-induced supramolecular polymers. J Am Chem Soc 2011;133:
shape memory effect of multiblock polyesterurethanes contain- 12866–74.
ing biodegradable segments and pendant cinnamamide groups. [82] Capadona JR, Shanmuganathan K, Tyler DJ, Rowan SJ, Weder C.
Biomacromolecules 2011;12:235–41. Stimuli-responsive polymer nanocomposites inspired by thesea
[55] Rochette JM, Ashby VS. Photoresponsive polyesters for tai- cucumber dermis. Science 2008;319:1370–4.
lorable shape memory biomaterials. Macromolecules 2013;46: [83] Mendez J, Annamalai PK, Eichhorn SJ, Rusli R, Rowan SJ, Foster
2134–40. EJ, et al. Bioinspired mechanically adaptive polymer nanocompos-
[56] Zhao XH, Dang Y, Deng JG, Zhang JH. Photoinduced shape fixity ites with water-activated shape-memory effect. Macromolecules
and thermal-induced shape recovery properties based on polyvinyl 2011;44:6827–35.
alcohol bearing coumarin. Colloid Polym Sci 2014;292:85–95. [84] Koerner H, Price G, Pearce NA, Alexander M, Vaia RA.
[57] Lee KM, Wang DH, Koerner H, Vaia RA, Tan LS, White TJ. Remotely actuated polymer nanocomposites-stress-recovery
Enhancement of photogenerated mechanical force in azobenzene- of carbon-nanotube-filled thermoplastic elastomers. Nat Mater
functionalized polyimides. Angew Chem Int Ed 2012;51:4117–21. 2004;3:115–20.
[58] Lee KM, Tabiryan NV, Bunning TJ, White TJ. Photomechanical mech- [85] Fei GX, Li G, Wu LS, Xia HS. A spatially and temporally controlled
anism and structure-property considerations in the generation of shape memory process for electrically conductive polymer-carbon
photomechanical work in glassy, azobenzene liquid crystal poly- nanotube composites. Soft Matter 2012;8:5123–6.
mer networks. J Mater Chem 2012;22:691–8. [86] Mohr R, Kratz K, Weigel T, Lucka-Gabor M, Moneke M, Lendlein A.
[59] Lee KM, Koerner H, Vaia RA, Bunning TJ, White TJ. Light-activated Initiation of shape-memory effect by inductive heating of magnetic
shape memory of glassy, azobenzene liquid crystalline polymer nanoparticles in thermoplastic polymers. Proc Natl Acad Sci USA
networks. Soft Matter 2011;7:4318–24. 2006;103:3540–5.
[60] Xie T, Xiao XC, Li JJ, Wang RM. Encoding localized strain [87] Leng JS, Lv HB, Liu YJ, Du SY. Electroactivate shape-memory polymer
history through wrinkle based structural colors. Adv Mater filled with nanocarbon particles and short carbon fibers. Appl Phys
2010;22:4390–4. Lett 2007;91:144105/1–144105.
[61] Li JJ, An YH, Huang R, Jiang HQ, Xie T. Unique aspects of a shape [88] Leng JS, Lan X, Liu YJ, Du SY, Huang WM, Liu N, et al. Elec-
memory polymer as the substrate for surface wrinkling. ACS Appl trical conductivity of thermoresponsive shape-memory polymer
Mater Interfaces 2012;4:598–603. with embedded micron sized Ni powder chains. Appl Phys Lett
[62] Reddy S, Arzt E, del Campo A. Bioinspired surfaces with switchable 2007;92:014104/1–14104.
adhesion. Adv Mater 2007;19:3833–7. [89] Luo X, Mather PT. Conductive shape memory nanocompos-
[63] Xie T, Xiao XC. Self-peeling reversible dry adhesive system. Chem ites for high speed electrical actuation. Soft Matter 2010;6:
Mater 2008;20:2866–8. 2146–9.
[64] Kim S, Sitti M, Xie T, Xiao XC. Reversible dry micro-fibrillar [90] Zhang HJ, Xia HS, Zhao Y. Optically triggered and spatially con-
adhesives with thermally controllable adhesion. Soft Matter trollable shape-memory polymer-gold nanoparticle composite
2009;5:3689–93. materials. J Mater Chem 2012;22:845–9.
[65] Wang RM, Xiao XC, Xie T. Viscoelastic behavior and force nature [91] Zhang HJ, Zhao Y. Polymers with dual light-triggered functions
of thermo-reversible dry adhesives. Macromol Rapid Commun of shape memory and healing using gold nanoparticles. ACS Appl
2010;31:295–9. Mater Interfaces 2013;5:13069–75.
[66] Wang RM, Xie T. Shape memory- and hydrogen bonding- based [92] Zhang HJ, Zhang JM, Tong X, Ma DL, Zhao Y. Light polarization-
strong reversible adhesives. Langmuir 2010;26:2999–3002. controlled shape-memory polymer/gold nanorod composite.
[67] Chen CM, Chiang CL, Lai CL, Xie T, Yang S. Buckling-based strong Macromol Rapid Commun 2013;34:1575–9.
dry adhesives via interlocking. Adv Funct Mater 2013;23:3813–23. [93] Kohlmeyer RR, Lor M, Chen J. Remote, local, and chemical program-
[68] Eisenhaure J, Xie T, Varghess S, Kim S. Microstructured shape mem- ming of healable multishape memory polymer nanocomposites.
ory polymer surfaces with reversible dry adhesion. ACS Appl Mater Nano Lett 2012;12:2757–62.
Interfaces 2013;5:7714–7. [94] Kohlmeyer RR, Chen J. Wavelength-selective, IR light-driven hinges
[69] Xu H, Yu C, Wang S, Malyarchuk V, Xie T, Rogers J. Deformable, pro- based on liquid crystalline elastomer composites. Angew Chem Int
grammable, and shape-memorizing micro-Optics. Adv Funct Mater Ed 2013;52:9234–7.
2013;23:3299–306. [95] Li G, Fei GX, Xia HS, Han JJ, Zhao Y. Spatial and temporal control of
[70] Lehn JM. Dynamers: dynamic molecular and supramolecular poly- shape memory polymers and simultaneous drug release using high
mers. Prog Polym Sci 2005;30:814–31. intensity focused ultrasound. J Mater Chem 2012;22:7692–6.
[71] Maeda T, Otsuka H, Takahara A. Dynamic covalent polymers: reor- [96] Kumar UN, Kratz K, Heuchel M, Behl M, Lendlein A. Shape-memory
ganizable polymers with dynamic covalent bonds. Prog Polym Sci nanocomposites with magnetically adjustable apparent switching
2009;34:581–604. temperatures. Adv Mater 2011;23:4157–62.
[72] Kloxin CJ, Scott TF, Adzima BJ, Bowman CN. Covalent adaptable [97] Liu Y, Li Y, Chen HM, Yang G, Zheng XT, Zhou SB. Water-induced
networks (CANs): a unique paradigm in cross-linked polymers. shape-memory poly(d,l-lactide)/microcrystalline cellulose com-
Macromolecules 2010;43:2643–53. posites. Carbohydr Polym 2014;104:101–8.
[98] Qi XD, Yao XL, Deng S, Zhou TN, Fu Q. Water-induced shape memory [125] Inomata K, Terahama T, Sekoguchi R, Sugimoto H, Nakanishi
effect of grapheme oxide reinforced polyvinyl alcohol nanocom- E. Shape memory properties of polypeptide hydrogels having
posites. J Mater Chem A 2014;2:2240–9. hydrophobic alkyl side chains. Polymer 2012;53:3281–6.
[99] Jung YC, So HH, Cho JW. Water-responsive shape mem- [126] Wang CC, Huang WM, Ding Z, Zhao Y, Purnawali H. Cooling-
ory polyurethane block copolymer modified with polyhedral /water-responsive shape memory hybrids. Compos Sci Technol
oligomeric silsesquioxane. J Macromol Sci Phys B 2006;45: 2012;72:1178–82.
453–61. [127] Yasin A, Li HZ, Lu Z, Rehman S, Siddiq M, Yang HY. A shape mem-
[100] Gu XZ, Mather PT. Water-triggered shape memory of multiblock ory hydrogel induced by the interactions between metal ions and
thermoplastic polyurethanes (TPUs). RSC Adv 2013;3:15783–91. phosphate. Soft Matter 2014;10:972–7.
[101] Lv HB, Leng JS, Liu YJ, Du SY. Shape-memory polymer in response [128] Harris RD, Auletta JT, Motlagh SAM, Lawless MJ, Perri NM, Saxena
to solution. Adv Eng Mater 2008;10:592–5. S, et al. Chemical and electrochemical manipulation of mechani-
[102] Zhao Y, Wang CC, Huang WM, Purnawali H. Buckling of poly(methyl cal properties in stimuli-responsive copper-cross-linked hydrogels.
methacrylate) in stimulus-responsive shape recovery. Appl Phys ACS Macro Lett 2013;2:1095–9.
Lett 2011;99:131911/1–131911. [129] Sangeetha NM, Maitra U. Supramolecular gels: functions and uses.
[103] Zhang XZ, Zhou QQ, Liu HR, Liu HW. UV light induced plasticization Chem Soc Rev 2005;34:821–36.
and light activated shape memory of spiropyran doped ethylene- [130] Yan XZ, Wang F, Zheng B, Huang FH. Stimuli-responsive
vinyl acetate copolymers. Soft Matter 2014;10:3748–54. supramolecular polymeric materials. Chem Soc Rev
[104] Nelson BA, King WP, Gall K. Shape recovery of nanoscale 2012;41:6042–65.
imprints in a thermoset “shape memory” polymer. Appl Phys Lett [131] Han XJ, Dong ZQ, Fan MM, Liu Y, Li JH, Wang YF, et al. pH-
2005;86:103108/1–103108. Induced shape-memory polymers. Macromol Rapid Commun
[105] Altebaeumer T, Gotsmann B, Pozidis H, Knoll A, Duerig U. Nanoscale 2012;33:1055–60.
shape-memory function in highly cross-linked polymers. Nano Lett [132] Dong ZQ, Cao Y, Yuan QJ, Wang YF, Li JH, Li BJ, et al. Redox- and
2008;8:4398–403. glucose-induced shape-memory polymers. Macromol Rapid Com-
[106] Wang Z, Hansen C, Ge Q, Maruf SH, Ahn DU, Qi HJ, et al. mun 2013;34:867–72.
Programmable, pattern-memorizing polymer surface. Adv Mater [133] Peters O, Ritter H. Supramolecular controlled water uptake of
2011;23:3669–73. macroscopic materials by a cyclodextrin-induced hydrophobic-to-
[107] Lee MH, Huntington MD, Zhou W, Yang JC, Odom TW. Pro- hydrophilic transition. Angew Chem Int Ed 2013;52:8961–3.
grammable soft lithography: solvent-assisted nanoscale emboss- [134] Myung D, Koh WU, Ko JM, Hu Y, Carrasco M, Noolandi J, et al.
ing. Nano Lett 2011;11:311–5. Biomimetic strain hardening in interpenetrating polymer network
[108] Yabu H, Jia RK, Matsuo Y, Ijiro K, Yamamoto S, Nishino F, et al. Prepa- hydrogels. Polymer 2007;48:5376–87.
ration of highly oriented nano-pit arrays by thermal shrinking of [135] Miyata T, Asami N, Uragami T. A reversibly antigen-responsive
honeycomb-patterned polymer films. Adv Mater 2008;20:4200–4. hydrogel. Nature 1999;399:766–9.
[109] Carlson A, Bowen AM, Huang YG, Nuzzo RG, Rogers JA. Transfer [136] Xu J, Wong DHC, Byrne JD, Chen K, Bowerman C, DeSimone JM.
printing techniques for materials assembly and micro/nanodevice Future of the particle replication in nonwetting templates (PRINT)
fabrication. Adv Mater 2012;24:5284–318. technology. Angew Chem Int Ed 2013;52:6580–9.
[110] Pretsch T, Ecker M, Schildhauer M, Maskos M. Switchable infor- [137] Marshall JE, Gallagher S, Terentjev EM, Smoukov SK. Anisotropic
mation carriers based on shape memory polymer. J Mater Chem colloidal micromuscles from liquid crystal elastomers. J Am Chem
2012;22:7757–66. Soc 2014;136:474–9.
[111] Ebara M, Uto K, Idota N, Hoffman JM, Aoyagi T. Shape-memory sur- [138] Saha S, Copic D, Bhaskar S, Clay N, Donini A, Hart AJ, et al. Chemically
face with dynamically tunable nanogeometry activated by body controlled bending of compositionally anisotropic microcylinders.
heat. Adv Mater 2012;24:273–8. Angew Chem Int Ed 2012;51:660–5.
[112] Davis KA, Burke KA, Mather PT, Henderson JH. Dynamic cell [139] Wischke C, Schossig M, Lendlein A. Shape-memory effect of micro-
behavior on shape memory polymer substrates. Biomaterials /nanoparticles from thermoplastic multiblock copolymers. Small
2011;32:2285–93. 2014;10:83–7.
[113] Tseng LF, Mather PT, Henderson JH. Shape-memory-actuated [140] Wischke C, Lendlein A. Method for preparation, programming, and
change in scaffold fiber alignment directs stem cell morphology. characterization of miniaturized particulate shape-memory poly-
Acta Biomater 2013;9:8790–801. mer matrices. Langmuir 2014;30:2820–7.
[114] Li WB, Gong T, Chen HM, Wang L, Li JR, Zhou SB. Tuning surface [141] Bellin I, Kelch S, Langer R, Lendlein A. Polymeric triple-shape mate-
micropattern features using a shape memory functional polymer. rials. Proc Natl Acad Sci USA 2006;103:18043–7.
RSC Adv 2013;3:9865–74. [142] Behl M, Lendlein A. Triple-shape polymers. J Mater Chem
[115] Chen CM, Yang S. Directed water shedding on high-aspect- 2010;20:3335–45.
ratio shape memory polymer micropillar arrays. Adv Mater [143] Zhao Q, Behl M, Lendlein A. Shape-memory polymers with mul-
2014;26:1283–8. tiple transitions: complex actively moving polymers. Soft Matter
[116] Fei GX, Tuinea-Bobe G, Li DX, Li G, Whiteside B, Coates P, et al. 2013;9:1744–55.
Electro-activated surface micropattern tuning for microinjection [144] Liu GQ, Ding XB, Cao YP, Zheng ZH, Peng YX. Novel shape-memory
molded electrically conductive shape memory polyurethane com- polymer with two transition temperatures. Macromol Rapid Com-
posites. RSC Adv 2013;3:24132–9. mun 2005;26:649–52.
[117] Drury JL, Mooney DJ. Hydrogels for tissue engineering: scaffold [145] Luo XF, Mather PT. Triple-shape polymeric composites (TSPCs). Adv
design variables and applications. Biomaterials 2003;24:4337–51. Funct Mater 2010;20:2649–56.
[118] Chung BG, Lee KH, Khademhosseini A, Lee SH. Microfluidic fabrica- [146] Xie T, Xiao XC, Cheng YT. Revealing triple-shape memory effect by
tion of microengineered hydrogels and their application in tissue polymer bilayers. Macromol Rapid Commun 2009;30:1823–7.
engineering. Lab Chip 2012;12:45–59. [147] Qin H, Mather P. Combined one-way and two-way shape
[119] Osada Y, Matsuda A. Shape memory in hydrogels. Nature memory in a glass-forming nematic network. Macromolecules
1995;376:219. 2009;42:273–80.
[120] Kagami Y, Gong JP, Osada Y. Shape memory behaviors of crosslinked [148] Ahn SK, Deshmukh P, Kasi RM. Shape memory behavior of
copolymers containing stearyl acrylate. Macromol Rapid Commun side-chain liquid crystalline polymer networks triggered by dual
1996;17:539–43. transition temperatures. Macromolecules 2010;43:7330–40.
[121] Merline JD, Nair CPR, Gouri C, Shrisudha T, Ninan KN. Shape [149] Ahn SK, Kasi RM. Exploiting microphase-separated morphologies
memory characterization of polytetra methylene oxide/poly of side-chain liquid crystalline polymer networks for triple shape
(acrylic acid-co-acrylonitrile) complexed gel. J Mater Sci memory properties. Adv Funct Mater 2011;21:4543–9.
2007;42:5897–902. [150] Suchao-in K, Chirachanchai S. Grafting to” as a novel and simple
[122] Izawa H, Kadokawa J. Preparation and characterizations of func- approach for triple-shape memory polymers. ACS Appl Mater Inter-
tional ionic liquid–gel and hydrogel materials of xanthan gum. J faces 2013;5:6850–3.
Mater Chem 2010;20:5235–41. [151] Xiao LP, Wei M, Zhan MQ, Zhang JJ, Xie H, Deng XY, et al. Novel
[123] Skrzeszewska PJ, Jong LN, de Wolf FA, Stuart MAC, van der Gucht J. triple-shape PCU/PPDO interpenetrating polymer networks con-
Shape-memory effects in biopolymer networks with collagen-like structed by selfcomplementary quadruple hydrogen bonding and
transient nodes. Biomacromolecules 2011;12:2285–92. covalent bonding. Polym Chem 2014;5:2231–41.
[124] Hao JK, Weiss RA. Mechanically tough, thermally activated shape [152] Kolesov IS, Radusch HJ. Multiple shape-memory behavior and
memory hydrogels. ACS Macro Lett 2013;2:86–9. thermal-mechanical properties of peroxide cross-linked blends of
linear and short-chain branched polyethylenes. Express Polym Lett [177] Zhang QL, Song SJ, Feng JC, Wu PY. A new strategy to prepare poly-
2008;2:461–73. mer composites with versatile shape memory properties. J Mater
[153] Hoeher R, Raidt T, Krumm C, Meuris M, Katzenberg F, Tiller JC. Chem 2012;22:24776–82.
Tunable multiple-shape memory polyethylene blends. Macromol [178] He ZW, Satarkar N, Xie T, Cheng YT, Hilt JZ. Remote controlled mul-
Chem Phys 2013;214:2725–32. tishape polymer nanocomposites with selective radiofrequency
[154] Chen HM, Liu Y, Gong T, Wang L, Zhao KQ, Zhou SB. actuations. Adv Mater 2011;23:3192–6.
Use of intermolecular hydrogen bonding to synthesize triple- [179] Liu Y, Boyles JK, Genzer J, Dickey MD. Self-folding of polymer sheets
shape memory supermolecular composites. RSC Adv 2013;3: using local light absorption. Soft Matter 2012;8:1764–9.
7048–56. [180] DiOrio AM, Luo XF, Lee KM, Mather PT. A functionally graded shape
[155] Chen SJ, Yuan HM, Ge ZC, Chen SG, Zhuo HT, Liu JH. Insights into memory polymer. Soft Matter 2011;7:68–74.
liquid-crystalline shape-memory polyurethane composites based [181] Zeng C, Seino H, Ren J, Yoshie N. Polymers with multishape memory
on an amorphous reversible phase and hexadecyloxybenzoic acid. controlled by local glass transition temperature. ACS Appl Mater
J Mater Chem C 2014;2:1041–9. Interfaces 2014;6:2753–8.
[156] Bae CY, Park JH, Kim EY, Kang YS, Kim BK. Organic–inorganic [182] Razzaq MY, Behl M, Kratz K, Lendlein A. Triple-shape effect
nanocomposite bilayers with triple shape memory effect. J Mater in polymer-based composites by cleverly matching geome-
Chem 2011;21:11288–95. try of active component with heating method. Adv Mater
[157] Ware T, Hearon K, Lonnecker A, Wooley KL, Maitland DJ, Voit 2013;25:5514–8.
W. Triple-shape memory polymers based on self-complementary [183] Pei ZQ, Yang Y, Chen QM, Terentjev EM, Wei Y, Ji Y. Mouldable
hydrogen bonding. Macromolecules 2012;45:1062–9. liquid-crystalline elastomer actuators with exchangeable covalent
[158] Wang SQ, Kaneko D, Okajima M, Yasaki K, Tateyama S, Kaneko bonds. Nat Mater 2014;13:36–41.
T. Hyperbranched polycoumarates with photofunctional multiple [184] Zhang JJ, Niu Y, Huang CL, Xiao LP, Chen ZT, Yang KK, et al. Self-
shape memory. Angew Chem Int Ed 2013;52:11143–8. healable and recyclable triple-shape PPDO–PTMEG co-network
[159] Luo HS, Hu JL, Zhu Y. Polymeric shape memory nanocompos- constructed through thermoreversible Diels–Alder reaction. Polym
ites with heterogeneous twin switches. Macromol Chem Phys Chem 2012;3:1390–3.
2011;212:1981–6. [185] de Gennes PG. Réflexions sur un type de polymères nématiques. C
[160] Han YJ, Bai T, Liu Y, Zhai XY, Liu WG. Zinc ion uniquely induced R Seances Acad Sci Ser B 1975;281:101–3.
triple shape memory effect of dipole–dipole reinforced ultra- [186] Finkelmann H, Kock HJ, Rehage G. Liquid crystalline elastomers—a
high strength hydrogels. Macromol Rapid Commun 2012;33: new type of liquid crystalline material. Makromol Chem Rapid
225–31. Commun 1981;2:317–22.
[161] Bai T, Han YJ, Zhang P, Wang W, Liu WG. Zinc ion-triggered two-way [187] Krause S, Zander F, Bergmann G, Brandt H, Wertmer H, Finkelmann
macro-/microscopic shape changing and memory effects in high H. Nematic main-chain elastomers: coupling and orientational
strength hydrogels with pre-programmed unilateral patterned sur- behavior. C R Chim 2009;12:85–104.
faces. Soft Matter 2012;8:6846–52. [188] Broemmel F, Kramer D, Finkelmann H. Preparation of liquid crys-
[162] Nan WJ, Wang W, Gao H, Liu WG. Fabrication of a shape memory talline elastomers. Adv Polym Sci 2012;250:1–48.
hydrogel based on imidazole–zinc ion coordination for poten- [189] Ohm C, Brehmer M, Zentel R. Liquid crystalline elastomers as actu-
tial cellencapsulating tubular scaffold application. Soft Matter ators and sensors. Adv Mater 2010;22:3366–87.
2013;9:132–7. [190] Ikeda T, Mamiya J, Yu YL. Photomechanics of liquid-crystalline elas-
[163] Xie T. Tunable polymer multi-shape memory effect. Nature tomers and other polymers. Angew Chem Int Ed 2007;46:506–28.
2010;464:267–70. [191] Kupfer J, Finkelmann H. Nematic liquid single crystal elastomers.
[164] Miaudet P, Derré A, Maugey M, Zakri C, Piccione PM, Inoubli R, et al. Makromol Chem Rapid Commun 1991;12:717–26.
Shape and temperature memory of nanocomposites with broad- [192] Schatzle J, Kaufhold W, Finkelmann H. Nematic elastomers: the
ened glass transition. Science 2007;318:1294–6. influence of external mechanical stress on the liquid-crystalline
[165] Kratz K, Madbouly SA, Wagermaier W, Lendlein A. Temperature- phase behavior. Makromol Chem 1989;190:3269–84.
memory polymer networks with crystallizable controlling units. [193] Kaufhold W, Finkelmann H. Nematic elastomers, 1 Effect
Adv Mater 2011;23:4058–62. of the spacer length on the mechanical coupling between
[166] Xie T, Page KA, Eastman SA. Strain-based temperature mem- network anisotropy and nematic order. Makromol Chem
ory effect for nafion and its molecular origins. Adv Funct Mater 1991;192:2555–79.
2011;21:2057–66. [194] Tajbakhsh AR, Terentjeva EM. Spontaneous thermal expansion of
[167] Kratz K, Voigt U, Lendlein A. Temperature-memory effect of nematic elastomers. Eur Phys J E 2001;6:181–8.
copolyesterurethanes and their application potential in minimally [195] Ahir SV, Tajbakhsh AR, Terentjev EM. Self-assembled shape-
invasive medical technologies. Adv Funct Mater 2012;22:3057–65. memory fibers of triblock liquid-crystal polymers. Adv Funct Mater
[168] Razzaq MY, Behl M, Lendlein A. Magnetic memory effect of 2006;16:556–60.
nanocomposites. Adv Funct Mater 2012;22:184–91. [196] Chung T, Rorno-Uribe A, Mather PT. Two-way reversible
[169] Wang L, Di SB, Wang WX, Chen HM, Yang XF, Gong T, et al. Tunable shape memory in a semicrystalline network. Macromolecules
temperature memory effect of photo-cross-linked star PCL–PEG 2008;41:184–92.
networks. Macromolecules 2014;47:1828–36. [197] Raquez JM, Vanderstappen S, Meyer F, Verge P, Alexandre
[170] Sun L, Huang WM. Mechanisms of the multi-shape memory effect M, Thomassin JM, et al. Design of cross-linked semicrystalline
and temperature memory effect in shape memory polymers. Soft poly(epsilon-caprolactone)-based networks with one-way and
Matter 2010;6:4403–6. two-way shape-memory properties through Diels–Alder reactions.
[171] Yu K, Xie T, Leng JS, Ding YF, Qi HJ. Mechanisms of multi-shape Chem Eur J 2011;17:10135–43.
memory effects and associated energy release in shape memory [198] Razzaq MY, Behl M, Nochel U, Lendlein A. Magnetically con-
polymers. Soft Matter 2012;8:5687–95. trolled shape-memory effects of hybrid nanocomposites from
[172] Luo YW, Guo YL, Gao X, Li BG, Xie T. A general approach towards oligo(omega-pentadecalactone) and covalently integrated mag-
thermoplastic multishape-memory polymers via sequence struc- netite nanoparticles. Polymer 2014;55:5953–60.
ture design. Adv Mater 2013;25:7430–8. [199] Razzaq MY, Behl M, Kratz K, Lendlein A. Multifunctional hybrid
[173] Wang L, Yang XF, Chen HM, Yang G, Gong T, Li WB, et al. nanocomposites with magnetically controlled reversible shape-
Multi-stimuli sensitive shape memory poly(vinyl alcohol)-graft- memory effect. Adv Mater 2013;25:5730–3.
polyurethane. Polym Chem 2013;4:4461–8. [200] Baker RM, Henderson JH, Mather PT. Shape memory poly(3-
[174] Li J, Liu T, Xia S, Pan Y, Zheng ZH, Ding XB, et al. A caprolactone)-co-poly(ethylene glycol) foams with body tem-
versatile approach to achieve quintuple-shape memory effect perature triggering and two-way actuation. J Mater Chem B
by semi-interpenetrating polymer networks containing broad- 2013;1:4916–20.
ened glass transition and crystalline segments. J Mater Chem [201] Bothe M, Pretsch T. Two-way shape changes of a shape-
2011;21:12213–7. memory poly(ester urethane). Macromol Chem Phys 2012;213:
[175] Shao Y, Lavigueur C, Zhu XX. Multishape memory effect of 2378–85.
norbornene-based copolymers with cholic acid pendant groups. [202] Bai YK, Zhang XR, Wang QH, Wang TM. A tough shape memory
Macromolecules 2012;45:1924–30. polymer with triple-shape memory and two-way shape memory
[176] Bai YK, Jiang C, Wang QH, Wang TM. Multi-shape-memory prop- properties. J Mater Chem A 2014;2:4771–8.
erty study of novel poly(␧-caprolactone)/ethyl cellulose polymer [203] Zotzmann J, Behl M, Hofmann D, Lendlein A. Reversible
networks. Macromol Chem Phys 2013;214:2465–72. triple-shape effect of polymer networks containing
polypentadecalactone- and poly(epsilon-caprolactone)-segments. [231] McClung AJW, Tandon GP, Baur JW. Deformation rate-, hold time-
Adv Mater 2010;22:3424–9. , and cycle-dependent shape-memory performance of Veriflex-E
[204] Chen SJ, Hu JL, Zhuo HT, Zhu Y. Two-way shape memory effect in resin. Mech Time-Depend Mater 2013;17:39–52.
polymer laminates. Mater Lett 2008;62:4088–90. [232] Tobushi H, Hara H, Yamada E, Hayashi S. Thermomechanical prop-
[205] Hu JL, Chen SJ, Zhuo HT. Properties and mechanism of two-way erties in a thin film of shape memory polymer of polyurethane
shape memory polyurethane composites. Compos Sci Technol series. Smart Mater Struct 1996;5:483–91.
2010;70:1437–43. [233] Srivastava V, Chester SA, Anand L. Thermally actuated shape-
[206] Westbrook KK, Mather PT, Parakh V, Dunn ML, Lee BM, Qi HJ. Two- memory polymers: experiments, theory, and numerical simula-
way reversible shape memory effects in a free-standing polymer tions. J Mech Phys Solids 2010;58:1100–24.
composite. Smart Mater Struct 2011;20:065010/1–65010. [234] Srivastava V, Chester SA, Ames NM, Anand L. A thermo-
[207] Ge Q, Westbrook KK, Mather PT, Dunn ML, Qi HJ. Thermomechanical mechanically-coupled large-deformation theory for amorphous
behavior of a two-way shape memory composite actuator. Smart polymers in a temperature range which spans their glass transition.
Mater Struct 2013;22:055009/1–55009. Int J Plast 2010;26:1138–82.
[208] Behl M, Kratz K, Nöchel U, Sauter T, Lendlein A. Temperature- [235] Nguyen TD. Modeling shape-memory behavior of polymers. Polym
memory polymer actuators. Proc Natl Acad Sci USA Rev 2013;53:130–52.
2013;110:12555–9. [236] Heuchel M, Sauter T, Kratz K, Lendlein A. Thermally induced shape-
[209] Zhou J, Turner SA, Brosnan SM, Li QX, Carrillo JMY, Nykypanchuk memory effects in polymers: quantification and related modeling
D, et al. Shapeshifting: reversible shape memory in semicrystalline approaches. J Polym Sci, B: Polym Phys 2013;51:621–37.
elastomers. Macromolecules 2014;47:1768–76. [237] Qi HJ, Dunn ML. Thermomechanical behavior and modeling
[210] Turner SA, Zhou J, Sheiko SS, Ashby VS. Switchable micropatterned approaches. In: Leng JS, Du SY, editors. Shape-memory polymers
surface topographies mediated by reversible shape memory. ACS and multifunctional composites. Boca Raton, FL: CRC Press; 2010.
Appl Mater Interfaces 2014;6:8017–21. p. 65–90.
[211] Bothe M, Pretsch T. Bidirectional actuation of a thermoplastic [238] Flory PJ. Principles of polymer chemistry. Ithaca, NY: Cornell Uni-
polyurethane elastomer. J Mater Chem A 2013;1:14491–7. versity Press; 1953. p. 432–88.
[212] Sun B, Long YZ, Zhang HD, Li MM, Duvail JL, Jiang XY, et al. Advances [239] Ferry JD. Viscoelastic properties of polymers. 3rd ed. New York, NY:
in three-dimensional nanofibrous macrostructures via electrospin- Wiley; 1980. p. 1–20.
ning. Prog Polym Sci 2014;39:862–90. [240] McKenna GB. Glass formation and glassy behavior. In: Booth C,
[213] Chen HL, Cao XY, Zhang JN, Zhang JJ, Ma YM, Shi GQ, et al. Electro- Price C, editors. comprehensive polymer science: the synthesis,
spun shape memory film with reversible fibrous structure. J Mater characterization, reactions & applications of polymers. Oxford, UK:
Chem 2012;22:22387–91. Pergamon; 1989. p. 311–62.
[214] Luo XF, Mather PT. Preparation and characterization of shape mem- [241] Hutchinson JM. Physical aging of polymers. Prog Polym Sci
ory elastomeric composites. Macromolecules 2009;42:7251–3. 1995;20:703–60.
[215] Rodriguez ED, Weed DC, Mather PT. Anisotropic shape-memory [242] Tobushi H, Hashimoto T, Hayashi S, Yamada E. Thermomechanical
elastomeric composites: fabrication and testing. Macromol Chem constitutive modeling in shape memory polymer of polyurethane
Phys 2013;214:1247–57. series. J Intell Mater Syst Struct 1997;8:711–8.
[216] Ahn BY, Shoji D, Hansen CJ, Hong E, Dunand DC, Lewis JA. Printed [243] Tobushi H, Ito N, Takata K, Hayashi S. Thermomechanical consti-
origami structures. Adv Mater 2010;22:2251–4. tutive modeling of polyurethane-series shape memory polymer.
[217] Derby B. Printing and prototyping of tissues and scaffolds. Science Shape Mem Mater 2000;327–3:343–6.
2012;338:921–6. [244] Tobushi H, Okumura K, Hayashi S, Ito N. Thermomechani-
[218] Ge Q, Qi HJ, Dunn ML. Active materials by four-dimension printing. cal constitutive model of shape memory polymer. Mech Mater
Appl Phys Lett 2013;103:131901/1–131901. 2001;33:545–54.
[219] Ge Q, Dunn C, Qi HJ, Dunn ML. Active origami by 4D printing. Smart [245] Nguyen TD, Qi HJ, Castro F, Long KN. A thermoviscoelastic model
Mater Struct 2014;23:094007/1–94007. for amorphous shape memory polymers: Incorporating structural
[220] Scott TF, Schneider AD, Cook WD, Bowman CN. Photoinduced plas- and stress relaxation. J Mech Phys Solids 2008;56:2792–814.
ticity in cross-linked polymers. Science 2005;308:1615–7. [246] Williams ML, Landel RF, Ferry JD. Temperature dependence of
[221] Kloxin CJ, Scott TF, Park HY, Bowman CN. Mechanophotopat- relaxation mechanisms in amorphous polymers and other glass-
terning on a photoresponsive elastomer. Adv Mater 2011;23: forming liquids. J Am Chem Soc 1955;77:3701–7.
1977–81. [247] Liu YP, Gall K, Dunn ML, Greenberg AR, Diani J. Thermomechanics
[222] Fairbanks BD, Singh SP, Bowman CN, Anseth KS. Photodegradable, of shape memory polymers: uniaxial experiments and constitutive
photoadaptable hydrogels via radical-mediated disulfide fragmen- modeling. Int J Plast 2006;22:279–313.
tation reaction. Macromolecules 2011;44:2444–50. [248] Diani J, Liu YP, Gall K. Finite strain 3D thermoviscoelastic con-
[223] Michal BT, Jaye CA, Spencer EJ, Rowan SJ. Inherently photohealable stitutive model for shape memory polymers. Polym Eng Sci
and thermal shape-memory polydisulfide networks. ACS Macro 2006;46:486–92.
Lett 2013;2:694–9. [249] Chen YC, Lagoudas DC. A constitutive theory for shape mem-
[224] Nair DP, Cramer NB, Gaipa JC, McBride MK, Matherly EM, McLeod ory polymers. Part I—Large deformations. J Mech Phys Solids
RR, et al. Two-stage reactive polymer network forming systems. 2008;56:1752–65.
Adv Funct Mater 2012;22:1502–10. [250] Chen YC, Lagoudas DC. A constitutive theory for shape memory
[225] Podgoı́rski M, Nair DP, Chatani S, Berg G, Bowman CN. Pro- polymers. Part II—A linearized model for small deformations. J
grammable mechanically assisted geometric deformations of Mech Phys Solids 2008;56:1766–78.
glassy two-stage reactive polymeric materials. ACS Appl Mater [251] Kafka V. Shape memory polymers: a mesoscale model of the
Interfaces 2014;6:6111–9. internal mechanism leading to the SM phenomena. Int J Plast
[226] Felton SM, Tolley MT, Shin BH, Onal CD, Demaine ED, Rus D, 2008;24:1533–48.
et al. Self-folding with shape memory composites. Soft Matter [252] Bol M, Reese S. Computer simulations of temperature induced
2013;9:7688–94. shape memory polymers by means of the finite element method. In:
[227] Castro F, Westbrook KK, Long KN, Shandas R, Qi HJ. Effects of Boukamel A, Laiarinandrasana L, Meo S, Verron E, editors. Consti-
thermal rates on the thermomechanical behaviors of amorphous tutive models for rubber V. London: Taylor & Francis Group; 2008.
shape memory polymers. Mech Time-Depend Mater 2010;14: p. 93–9.
219–41. [253] Bol M, Reese S. Micromechanical modelling of shape memory poly-
[228] Heuchel M, Cui J, Kratz K, Kosmella H, Lendlein A. Relaxation mers. Adv Sci Tech 2009;54:137–42.
based modeling of tunable shape recovery kinetics observed under [254] Xu W, Li GQ. Constitutive modeling of shape memory poly-
isothermal conditions for amorphous shape-memory polymers. mer based self-healing syntactic foam. Int J Solids Struct
Polymer 2010;51:6212–8. 2010;47:1306–16.
[229] Chen X, Nguyen TD. Influence of thermoviscoelastic properties and [255] Baghani M, Naghdabadi R, Arghavani J, Sohrabpour S. A
loading conditions on the recovery performance of shape memory thermodynamically-consistent 3D constitutive model for shape
polymers. Mech Mater 2011;43:127–38. memory polymers. Int J Plast 2012;35:13–30.
[230] McClung AJW, Tandon GP, Baur JW. Strain rate- and temperature- [256] Baghani M, Naghdabadi R, Arghavani J. A large deformation frame-
dependent tensile properties of an epoxy-based, thermosetting, work for shape memory polymers: constitutive modeling and
shape memory polymer (Veriflex-E). Mech Time-Depend Mater finite element implementation. J Intell Mater Syst Struct 2013;24:
2012;16:205–21. 21–32.
[257] Srinivasa AR, Gosh P. A simple, Gibbs potential based multi- [283] Negahban M, Wineman AS. Modeling the mechanical response of
network model for shape memory polymers. AIP Conf Proc a material undergoing continuous isothermal crystallization. Int J
2008;1029:58–74. Eng Sci 1992;30:953–62.
[258] Ghosh P, Srinivasa AR. A Two-network thermomechanical model [284] Negahban M. Simulation of mechanical response in crystallizing
of a shape memory polymer. Int J Eng Sci 2011;49:823–38. polymers—crystallization under constant shearing deformations.
[259] Ghosh P, Reddy JN, Srinivasa AR. Development and implementation Mech Mater 1993;16:379–99.
of a beam theory model for shape memory polymers. Int J Solids [285] Ma R, Negahban M. Simulation of mechanical response in crystal-
Struct 2013;50:595–608. lizing polymers—crystallization under a constant shear force. Acta
[260] Ghosh P, Srinivasa AR. A two-network thermomechanical model Mech 1995;112:59–76.
and parametric study of the response of shape memory polymers. [286] Ma RJ, Negahban M. Simulation of mechanical response
Mech Mater 2013;60:1–17. during polymer crystallization around rigid inclusions and
[261] Shi GH, Yang QS, He XQ. Analysis of intelligent hinged shell struc- voids—homogeneous crystallization. Mech Mater 1995;21:
tures: deployable deformation and shape memory effect. Smart 25–50.
Mater Struct 2013;22:125018/1–125018. [287] Long KN, Dunn ML, Qi HJ. Mechanics of soft active materials with
[262] Shi GH, Yang QS, He XQ, Liew KM. A three-dimensional con- phase evolution. Int J Plast 2010;26:603–16.
stitutive equation and finite element method implementation [288] Barot G, Rao IJ. Constitutive modeling of the mechanics associated
for shape memory polymers. CMES—Comp Model Eng 2013;90: with crystallizable shape memory polymers. Z Angew Math Phys
339–58. 2006;57:652–81.
[263] Balogun O, Mo C. Shape memory polymers: three-dimensional [289] Barot G, Rao IJ, Rajagopal KR. A thermodynamic framework for the
isotropic modeling. Smart Mater Struct 2014;23:045008/1–45008. modeling of crystallizable shape memory polymers. Int J Eng Sci
[264] Ghosh P, Srinivasa AR. Development of a finite strain two-network 2008;46:325–51.
model for shape memory polymers using QR decomposition. Int J [290] Qi HJ, Nguyen TD, Castroa F, Yakacki CM, ShandaSa R. Finite defor-
Eng Sci 2014;81:177–91. mation thermo-mechanical behavior of thermally induced shape
[265] Choi J, Ortega AM, Xiao R, Yakacki CM, Nguyen TD. Effect of physi- memory polymers. J Mech Phys Solids 2008;56:1730–51.
cal aging on the shape-memory behavior of amorphous networks. [291] Matveenko VP, Smetannikov OY, Trufanov NA, Shardakov IN.
Polymer 2012;53:2453–64. Models of thermomechanical behavior of polymeric materials
[266] Xu W, Li GQ. Thermoviscoplastic modeling and testing of shape undergoing glass transition. Acta Mech 2012;223:1261–84.
memory polymer based self-healing syntactic foam programmed [292] Westbrook KK, Parakh V, Chung T, Mather PT, Wan LC, Dunn ML,
at glassy temperature. J Appl Mech 2011;78:061017/1–61017. et al. Constitutive modeling of shape memory effects in semicrys-
[267] Huang WM, Yang B, Li C, Chan YS, An L. Response to “Comment on talline polymers with stretch induced crystallization. J Eng Mater
‘Water-driven programmable polyurethane shape memory poly- Technol 2010;132:041010/1–41010.
mer: Demonstration and mechanism [Appl Phys Lett 97, 056101, [293] Scott TF, Draughon RB, Bowman CN. Actuation in crosslinked
(2010)]¨’. Appl Phys Lett 2010;97:056102/1. polymers via photoinduced stress relaxation. Adv Mater
[268] Xiao R, Nguyen TD. Modeling the solvent-induced shape-memory 2006;18:2128–32.
behavior of glassy polymers. Soft Matter 2013;9:9455–64. [294] Bowman CN, Kloxin CJ. Covalent adaptable networks: reversible
[269] Lu HB, Huang WM, Leng JS. A phenomenological model for bond structures incorporated in polymer networks. Angew Chem
simulating the chemo-responsive shape memory effect in poly- Int Ed 2012;51:4272–4.
mers undergoing a permeation transition. Smart Mater Struct [295] Montarnal D, Capelot M, Tournilhac F, Leibler L. Silica-like
2014;23:045038/1–45038. malleable materials from permanent organic networks. Science
[270] Lu HB, Du SY. A phenomenological thermodynamic model 2011;334:965–8.
for the chemo-responsive shape memory effect in polymers [296] Fringuelli F, Taticchi A, Wenkert E. Diels–Alder reactions of
based on Flory–Huggins solution theory. Polym Chem 2014;5: cycloalkenones in organic-synthesis—a review. Org Prep Proced Int
1155–62. 1990;22:131–65.
[271] Mulliken AD, Boyce MC. Mechanics of the rate-dependent [297] Oh T, Reilly M. Reagent-controlled asymmetric Diels–Alder
elastic–plastic deformation of glassy polymers from low to high reactions—a review. Org Prep Proced Int 1994;26:129–58.
strain rates. Int J Solids Struct 2006;43:1331–56. [298] Hui CY, Long R. A constitutive model for the large deformation of a
[272] Dupaix RB, Boyce MC. Constitutive modeling of the finite strain self-healing gel. Soft Matter 2012;8:8209–16.
behavior of amorphous polymers in and above the glass transition. [299] Long KN, Scott TF, Qi HJ, Bowman CN, Dunn ML. Photomechanics
Mech Mater 2007;39:39–52. of light-activated polymers. J Mech Phys Solids 2009;57:1103–21.
[273] Strobl GR. The physics of polymers: concepts for understanding [300] Sodhi JS, Rao IJ. Modeling the mechanics of light activated shape
their structures and behavior. Berlin/New York, NY: Springer; 1996. memory polymers. Int J Eng Sci 2010;48:1576–89.
p. 255–6. [301] Long R, Qi HJ, Dunn ML. Thermodynamics and mechanics
[274] Diani J, Gilormini P, Fredy C, Rousseau I. Predicting thermal shape of photochemcially reacting polymers. J Mech Phys Solids
memory of crosslinked polymer networks from linear viscoelastic- 2013;61:2212–39.
ity. Int J Solids Struct 2012;49:793–9. [302] Ge Q, Luo XF, Rodriguez ED, Zhang X, Mather PT, Dunn ML, et al.
[275] Westbrook KK, Kao PH, Castro F, Ding YF, Qi HJ. A 3D finite deforma- Thermomechanical behavior of shape memory elastomeric com-
tion constitutive model for amorphous shape memory polymers: posites. J Mech Phys Solids 2012;60:67–83.
a multi-branch modeling approach for nonequilibrium relaxation [303] Ge Q, Luo XF, Iversen CB, Mather PT, Dunn ML, Qi HJ. Mechanisms of
processes. Mech Mater 2011;43:853–69. triple-shape polymeric composites due to dual thermal transitions.
[276] Xiao R, Choi JW, Lakhera N, Yakacki CM, Frick CP, Nguyen TD. Model- Soft Matter 2013;9:2212–23.
ing the glass transition of amorphous networks for shape-memory [304] Ge Q, Luo XF, Iversen CB, Nejad HB, Mather PT, Dunn ML, et al. A
behavior. J Mech Phys Solids 2013;61:1612–35. finite deformation thermomechanical constitutive model for triple
[277] Rajagopal KR, Srinivasa AR. Mechanics of the inelastic behav- shape polymeric composites based on dual thermal transitions. Int
ior of materials—Part 1, Theoretical underpinnings. Int J Plast J Solids Struct 2014;51:2777–90.
1998;14:945–67. [305] Long R, Qi HJ, Dunn ML. Modeling the mechanics of covalently
[278] Rajagopal KR, Srinivasa AR. Mechanics of the inelastic behavior of adaptable polymer networks with temperature-dependent bond
materials. Part II: Inelastic response. Int J Plast 1998;14:969–95. exchange reactions. Soft Matter 2013;9:4083–96.
[279] Rajagopal KR, Wineman AS. A constitutive equation for nonlin- [306] Long KN. The mechanics of network polymers with thermally
ear solids which undergo deformation induced microstructural reversible linkages. J Mech Phys Solids 2014;63:386–411.
changes. Int J Plast 1992;8:385–95. [307] Bonner M, de Oca HM, Brown M, Ward IM. A novel approach to
[280] Huntley HE, Wineman AS, Rajagopal KR. Stress softening, strain predict the recovery time of shape memory polymers. Polymer
localization and permanent set in the circumferential shear 2010;51:1432–6.
of an incompressible elastomeric cylinder. Ima J Appl Math [308] Ge Q, Yu K, Ding YF, Qi HJ. Prediction of temperature-dependent
1997;59:309–38. free recovery behaviors of amorphous shape memory polymers.
[281] Wineman A, Min JH. Viscoelastic hinge formation in beams. Acta Soft Matter 2012;8:11098–105.
Mech 2000;140:183–205. [309] Castro F, Westbrook KK, Hermiller J, Ahn DU, Ding YF, Qi
[282] Wineman A, Min JH. Time dependent scission and cross-linking in HJ. Time and temperature dependent recovery of epoxy-based
an elastomeric cylinder undergoing circular shear and heat con- shape memory polymers. J Eng Mater Technol 2011;133:
duction. Int J Nonlinear Mech 2003;38:969–83. 021025/1–21025.
[310] Yu K, McClung AJW, Tandon GP, Baur JW, Qi HJ. A thermome- [315] Xie T, Li JJ, Zhao Q. Hidden thermoreversible actuation behav-
chanical constitutive model for an epoxy based shape memory ior of nafion and its morphological origin. Macromolecules
polymer and its parameter identifications. Mech Time-Depend 2014;47:1085–9.
Mater 2014;18:453–74. [316] Sekitani T, Zschieschang U, Klauk H, Someya T. Flexible organic
[311] Yu K, Ge Q, Qi HJ. Reduced time as a unified parameter determin- transistors and circuits with extreme bending stability. Nat Mater
ing fixity and free recovery of shape memory polymers. Nature 2010;9:1015–22.
Communications 2014;5:3066:1–13066:. [317] Yu ZB, Zhang QW, Li L, Chen Q, Niu XF, Liu J, et al. Highly flexible sil-
[312] Katzenberg F, Heuwers B, Tiller JC. Superheated rubber for cold ver nanowire electrodes for shape-memory polymer light-emitting
storage. Adv Mater 2011;23:1909–11. diodes. Adv Mater 2011;23:664–8.
[313] Heuwers B, Quitmann D, Katzenberg F, Tiller JC. Stress-induced [318] Avendano-Bolivar A, Ware T, Arreaga-Salas D, Simon D, Voit W.
melting of crystals in natural rubber: a new way to tailor the tran- Mechanical cycling stability of organic thin film transistors on
sition temperature of shape memory polymers. Macromol Rapid shape memory polymers. Adv Mater 2013;25:3095–9.
Commun 2012;33:1517–22. [319] Reeder J, Kaltenbrunner M, Ware T, Arreaga-Salas D, Avendano-
[314] Quitmann D, Gushterov N, Sadowski G, Katzenberg F, Tiller JC. Envi- Bolivar A, Yokota T, et al. Mechanically adaptive organic transistors
ronmental memory of polymer networks under stress. Adv Mater for implantable electronics. Adv Mater 2014;26:4967–73.
2014;26:3441–4.

Shrinky Dink Molecula

Uploaded by

Copyright:

Available Formats

Shrinky Dink Molecula

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Shrinky Dink Molecula

Uploaded by

Copyright:

Available Formats

Progress in Polymer Science 49–50 (2015) 79–120

Contents lists available at ScienceDirect

Progress in Polymer Science

Recent progress in shape memory polymer: New behavior,

4.2. Shape memory hydrogels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92

1. Introduction In contrast, shape memory polymers (SMPs) are a

content. That is, the hydrogel displayed both the SME

associated with both crystalline phases were changed in

The simple exponential form is recovered for ˇ = 1,

Fig. 22. Phase formation and shape memory effect.

where E0 and E1 are the equilibrium and nonequilibrium

You might also like