Cohades 2018

PROGRESS REPORT
Self-Healing Materials
www.advmatinterfaces.de
Progress in Self-Healing Fiber-Reinforced Polymer

Composites
Amaël Cohades, Callum Branfoot, Steven Rae, Ian Bond,* and Véronique Michaud*
With the potential for small-scale damage

This paper sets out to review the current state of the art in applying self- to grow or coalesce under subsequent
healing/self-repair to high-performing advanced fiber-reinforced polymer loading, FRP structures incorporate high
composite materials (FRPs). A significant proportion of self-healing studies safety factors to account for this defect
sensitivity to ensure they will remain load
have focused so far on developing and assessing healing efficiency of bulk
bearing throughout their service life. This
polymer systems, applied to particulate composites or low-volume fraction paradigm has resulted in many research
fiber-reinforced materials. Only limited research is undertaken on self-healing groups worldwide exploring and devel-
in advanced structural FRP composite materials. This review focuses on what oping the concept of self-healing or self-
is achieved to date, the ongoing challenges which have arisen in imple- repair being applied to FRPs, with the
menting self-healing into FRPs, how considerations for industrialization and aim of achieving autonomous structural
recovery via an embedded functionality,
large-scale manufacture must be considered from the outset, where self-
to address damage formation in its early
healing may provide most benefits, and how a functionality like self-healing stages before the integrity of the structure
can be validated for application in real structures. Systems are compared in is compromised. In principle, this could
terms of process parameters, resulting mechanical properties, methods of reduce the structural redundancy and fully
healing assessment, as well as values of healing quantification. Guidelines realize the mass saving associated with
using FRP composite materials.
are further given for a concerted effort to drive toward standardization of tests
This paper sets out to review the
and the use of specific reinforcement architectures in order to allow reliable current state of the art in applying
comparison between the available healing systems in structural composites. self-healing/self-repair to high-performing
advanced fiber-reinforced polymer com-
posite materials. Our definition of an
1. Introduction advanced fiber-reinforced polymer composite material for this
work is as follows:
The intrinsically poor interlaminar properties of laminated
• A nominal fiber volume fraction, Vf > 30%.
fiber-reinforced polymers (FRPs) may lead to delamination
• A nominal matrix modulus, E > 500 MPa.
and crack growth when subjected to flexural and compressive
loading. This behavior has resulted in a reticence regarding
their design, which led to their conservative use in real struc- Our review discusses what has been achieved to date, the
tures. This conservatism negates the potential weight-saving ongoing challenges which have arisen in implementing self-
benefits that can be realized with fully optimized structures. healing into FRPs, how considerations for industrialization
and large-scale manufacture must be considered from the
outset, where self-healing may provide the most benefits (with
Dr. A. Cohades, Prof. V. Michaud respect to current FRP approaches to design), and how a func-
Laboratory for Processing of Advanced Composites (LPAC)
tionality like self-healing can be validated for application in real
Institute of Materials (IMX)
Faculty of Engineering (STI) structures.
Ecole Polytechnique Fédérale de Lausanne (EPFL) A significant proportion of self-healing studies have focused
Station 12, CH 1015 Lausanne, Switzerland on assessing the healing efficiency of the system in bulk poly-
E-mail: [email protected] mers, liquid (or low fiber volume) polymers, particulate com-
C. Branfoot, Dr. S. I. Rae, Prof. I. Bond posites, or low-volume fraction fiber-reinforced materials with
Bristol Composites Institute (ACCIS)
School of Civil inferior mechanical properties.[1,2] Comparatively, limited
Aerospace and Mechanical Engineering (CAME) research has been undertaken on self-healing in advanced
Faculty of Engineering FRP composite materials, specifically based on systems com-
University of Bristol monly used in industry or those that possess high-performance
BS8 1TR Bristol, UK
characteristics.
E-mail: [email protected]
Two broad categories of self-healing systems can be found in
The ORCID identification number(s) for the author(s) of this article
can be found under https://doi.org/10.1002/admi.201800177. the available literature:[3] i) intrinsic self-healing systems, which
can heal a crack through interactions (physical or chemical)
DOI: 10.1002/admi.201800177 at the molecular level; ii) extrinsic self-healing systems, which
Adv. Mater. Interfaces 2018, 1800177 1800177 (1 of 20) © 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
utilize pre-embedded healing agents (via microcapsules or vas-

cules) to fill the crack and reunite its faces (Figure 1). Section 2 Ian Bond is currently the
reviews the main systems integrated so far into structural com- Head of the School of Civil,
posites. It is worth making clear at this juncture that, to date, Aerospace and Mechanical
only repair of the polymer matrix has been demonstrated. If Engineering at the University
fibers are fractured, there is no remedial action available at this of Bristol, UK. His research
time. The integration of a self-healing/repair function within interests include developing,
a high fiber volume fraction FRP presents some major chal- characterizing, and opti-
lenges as the embedded function must compete for volumetric mizing a variety of innovative
space with the reinforcing fibers (Figure 1). and ingenious approaches
Naturally, the effectiveness of healing in composites that provide functionality
is predicated on the question of how healing/recovery is to fiber-reinforced polymer
characterized, which is discussed in Section 3. Numerous composite materials and take
methods to quantify healing performance have been them beyond their structural
reported, but a standard approach is yet to be decided. Cur- role. This includes bioinspired and biomimetic approaches.
rently, the variety of tests and performance metrics serves to Functionalities such as self-healing, electromagnetic
hinder any direct comparisons and wider evaluation. More- response, and shape change (morphing) within fiber rein-
over, reported healing efficiencies often conflate multiple forced composites are currently being developed, along-
strengthening or toughening mechanisms within FRPs, by side research into creating novel hierarchical architectures
the inclusion of entities such as microcapsules or vascules, and improving damage tolerance via innovative means.
or serve to diminish innate properties of the host structure
due to the creation of defects, fiber architecture disruption,
Véronique Michaud is
or alteration to load transfer capabilities.[4] Furthermore, any
currently the Head of the
demonstrated loading scenarios must be representative and
Laboratory for Processing
repeatable, and not serve to enhance the effectiveness of the
of Advanced Composites
healing agent.[2]
and Associate Dean of
Engineering for Education,
at the Ecole Polytechnique
Fédérale de Lausanne, in
Switzerland. Her research
interests are the fundamen-
tals of composite mate-
rials processing, in view of
cost-effective and functional materials and structures.
Functions include self-healing, shape and vibration control,
controlled heat transfer, tailored damping properties, as
well as nonlinear mechanical behavior, as encountered
in biological structures, while the focus is on developing
adapted processing routes to integrate these into struc-
tural composite materials, for applications in transport,
sport, or industrial sectors.
Any viable self-healing approach must be able to cope with

the adverse environments to which it is exposed. This includes
the processes encountered in manufacturing, as discussed
in Section 4. In many cases, the underpinning chemistry
behind the healing mechanism is either thermally activated or
enhanced. In the small proportion of published research that
investigates thermal stability, most inactive self-healing systems
show a significant decrease in subsequent healing efficiency
when exposed to temperatures above ambient.
Considering the process of manufacture in more general
terms, incorporating an extrinsic additional functionality into
an already highly architectured structure may affect the process
Figure 1. General concept to produce, test and design a self-healing as well as the quality of the produced composite. For example,
system within structural polymer composites (components not neces- achieving a dispersal of microcapsules, vascules, and/or cata-
sarily to scale). lyst must be achieved in addition to the reinforcing fibers,
which already typically occupy >50% of the volume available self-healing and articulates what might be the future need to
(see Figure 1).[5–7] Furthermore, the presence of such secondary see its implementation at an industrial scale.
features can be detrimental to the manufacturing process, Self-healing polymeric materials are generally classified into
e.g., compressibility and permeability of the fiber preform[8] two distinct groups, intrinsic and extrinsic (further classified as
(Figure 1). The challenges associated with incorporation of self- capsules and vascules), according to their method of incorpo-
healing systems are evident in simple laboratory scale experi- ration. Those categories include further subcategories, which
ments: this is undoubtedly a factor in the slow progression of are detailed with some examples in Figure 2. In practice, some
the technology to larger structures. However, it is quite apparent systems blur the boundary between these groupings (notably
from a design perspective that imparting a self-healing func- thermoplastic-derived systems) and there are cases where both
tionality to an entire structure may not be necessary. Instead, have been employed within one host structure. Indeed, there
specifically targeting critical features and locations (such as is an argument for doing the latter since extrinsic and intrinsic
joints, holes, edges, and other discontinuities) may be a more approaches are complementary in terms of the length scales
viable and rewarding application of self-healing technologies as over which they are effective. The effect of healing (or recovery)
discussed in Section 5. performance is commonly described in terms of a healing effi-
In most studies to date, the repair is not strictly autono- ciency (η). This dimensionless value has been defined in sev-
mous; however, perhaps it is imprudent to consider nonau- eral ways, by multiple authors; however, it is most commonly
tonomy as a barrier to maturation of the technology. Although expressed as a percentage in terms of strength, stiffness, or
fully autonomous self-healing capability is the long-term goal, toughness and is typically derived in two forms
current challenges in conventional repair of FRPs means that
there is considerable value in developing and implementing  healed property 
an embedded repair function that still requires intervention η = 100 *   (1)
 pristine property 
or initiation. Indeed, the necessity for external intervention
could be embraced as an opportunity to demonstrate a step  healed property − damaged property 
change in improving repair strategies that can be employed η = 100 *   (2)
 pristine property − damaged property 
in FRP structures. At present, one of the most effective and
commonly used FRP repair strategies is the use of exter- For simplicity and because Equation (1) is derived from
nally bonded patches.[9] A self-healing system, which is com- Equation (2) with the damaged property set to zero, Equation (2)
petitive with such an approach in terms of either i) the time is exclusively used in this article and where possible, published
required to effect the repair, ii) repair effectiveness (in terms values reported herein have been modified for consistency. The
of operational lifetime, stiffness, strength, or toughness), iii) nature of healing efficiency is discussed in detail in Section 3.
repair simplicity (e.g., simply heating the damaged area), or
iv) repair cost, could represent an achievable short-term target
for existing self-healing technologies. Furthermore, manual 2.1. Extrinsic Self-Healing
intervention also presents a considerable advantage in terms
of providing assurance and validation via inspection or moni- Extrinsic self-healing in FRPs is implemented by the inclusion
toring, particularly where the application is safety critical of additional functional materials or architecture within the
or highly loaded. A major hurdle to the implementation of laminate. Upon damage formation, distributed microcapsules,
self-healing FRPs remains the challenge of manufacturing, which act as embedded healing agent reservoirs, or a network
characterization, and validation: processing method, integra- of microvascules, which form a circulatory network, are rup-
tion to the FRPs, performance, repeatability, monitoring, and tured or breached, allowing the entrained liquid(s) to infuse
quality assurance. Such metrics are common barriers to the the fracture surfaces. Following subsequent transformation of
adoption of new technologies into commercial environments. the liquid healing agent to a solid, structural integrity can be
Research is needed to demonstrate that self-healing function- restored and the damage ameliorated.
ality can remain effective for extended periods and in nonop- Different approaches to incorporating extrinsic self-healing
timized, realistic environments; these aspects are evaluated in systems engender a range of advantages and disadvantages in
Section 6. terms of damage volume, stoichiometry, repeated healing, gap
filling, stability, and long-term reactivity, host compatibility,
ease of incorporation/manufacture, cost of constituents, and
2. Self-Healing Fiber-Reinforced Polymer healing efficiency achievable.
Composites: The State of the Art
Self-healing of polymeric materials has been extensively 2.1.1. Microencapsulation
reviewed in recent years and the reader is directed to the fol-
lowing comprehensive articles describing systems, characteri- This approach typically employs either a single or a dual micro-
zation, and utilization of self-healing polymeric materials.[3,10–12] capsule archetype: the former predominantly relies upon acti-
Much of this literature does not relate to or is not compatible vation postrelease by a catalyst codistributed throughout the
with structural FRP composites. The following review adheres matrix material to initiate cure; in the latter system, an appro-
to our definition of what constitutes an FRP composite, as out- priate hardener material is sequestered into separate capsules
lined in Section 1, and highlights the current state of the art for and also distributed throughout the matrix material, relying on
Figure 2. Classes of self-healing systems within structural polymer composites: i) capsule-based systems including DCPD–Grubbs’ (Reproduced with
permission.[13] Copyright 2010, Elsevier.) and ethyl phenyl acetate (EPA) solvent capsules;[14] ii) vascular systems including hollow fibers (Reproduced
with permission.[15] Copyright 2007, Elsevier.) and 3D vascular network (scale bar = 5 mm) (Reproduced with permission.[16] Copyright 2007, Nature
Publishing.); iii) intrinsic healing systems including reversible covalent bonds (Reproduced with permission.[17] Copyright 2017, Elsevier.), supramo-
lecular resins (Reproduced with permission.[18] Copyright 2017, IOP Publishing.) and polymer blends (Reproduced with permission.[19] Copyright 2017,
Elsevier.).
mixing of the components to effect repair. Much of the seminal strength of pristine nonfunctionalized laminates was reported,
research exploited the ring-opening metathesis polymerization no RCS data at increasing impact energy levels was available
(ROMP) of dicyclopentadiene (DCPD) using Grubbs’ catalyst. for comparison; one has to be cautious therefore to consider if
However, research was mainly limited to neat resin systems the improvement in RCS in healed specimens is only relative
or where the matrix did not meet the selection criteria speci- to the functionalized specimens with compromised damage
fied above, or where fiber volume was not reported or below resistance and therefore lower localized buckling loads of the
the threshold amount.[1,20–26] Furthermore, the DCPD–Grubbs’ sublaminates. The exacerbation of damage due to inclusion of
system has several limitations such as deactivation during cure the wax particles may be less prominent at lower energy levels,
of the host matrix, dispersal, cost, toxicity, and low temperature thus facilitating higher true healing efficiency, but this needs to
resistance. Although it has been shown that some of these con- be investigated and reported if true healing efficiencies are to
cerns can be addressed by protecting the catalyst in a paraffin be established. Consistency, and validity, of calculating healing
wax shell, doing so can have a significantly detrimental effect efficiency in a way that reflects the benefits of incorporating
on the host material’s mechanical properties.[2] self-healing systems is discussed in Section 3.
Patel et al.[13] incorporated the DCPD–Grubbs’ system into Yin et al. published a number of papers investigating the
woven glass fiber–reinforced polymer composites (GFRPs) use of epoxy-loaded capsules with a copper complex as the
with the aim of improving compression after impact (CAI) embedded catalyst in FRP specimens.[27–30] Again, the fiber
strength. However, inclusion of the wax-encapsulated particles volume reported (27–30%) was lower than our selection cri-
resulted in significantly longer crack lengths following impact teria; however, the seminal nature of the research should not be
than observed in control specimens. This could be attributed overlooked and the reader is directed for further reading. Coope
to matrix plasticization, or to the increased area/number of et al. investigated metal triflate–based Lewis acids as catalysts
inclusions introduced, thus reducing interlaminar mechanical to circumvent the stability issues associated with Grubbs’ cata-
properties. Although Patel et al.[13] report respectable healing lyst but only demonstrated moderate healing capability in neat
efficiencies, these are inferred from the recovery of compres- resin systems.[31]
sive strength (RCS) of the control panels containing microcap- Dual microcapsule systems facilitate separate integration
sules and wax particles (but no catalyst) which, as previously of an epoxy healing agent and hardener, thus overcoming
observed, increased delamination. Although the compressive the issues associated with metal-based catalysts such as cost,
toxicity, reactivity, dispersal, etc. Encapsulation of amine-based of the capsules reduced the compressive strength of virgin
curing agents is notoriously difficult and requires more com- laminates by up to 53% due to the attraction of propagating
plex synthesis procedures and techniques than that required for microcracks. Furthermore, no healing was observed in CAI
encapsulation of epoxy materials due to the difficulty in creating testing. Several reasons were proposed: loss of healing agent
an amine suspension.[32,33] To circumvent the difficulties in due to processing and aging, unsuitability of the solvent for
stabilizing an amine suspension, attempts have been made to the resin system used, and the small scale over which polymer
infiltrate hollow microcapsules and porous glass beads with an reptation acts to induce healing compared to the macroscopic
amine solution.[34–36] The latter method in particular enhanced parameters associated with CAI testing. The solvent micro-
the long term stability of the system even at elevated tempera- capsules were also tested with respect to recovery of fracture
tures. Microfluidic devices have also been shown to success- toughness under Mode I loading.[43] However, no healing was
fully produce well-controlled distributions of double emulsions observed despite lightly clamping the fracture surfaces together
using a photocurable middle phase to encapsulate an amine to minimize interfacial distance and ensure full wetting of the
curing agent.[37,38] Alternative chemistries utilizing mercaptan, fracture surface. This is particularly noteworthy considering
thiol, and imidazole-based curing agents have also been suc- the complete recovery of stiffness achieved in the bulk polymer
cessfully encapsulated and demonstrated in curing epoxies, but as reported previously.[48] Where DCB specimens were pre-
as with previous systems discussed, they should be explored pared by bonding two separate FRP plates with a functional-
further to determine their capabilities in FRP samples.[39–41] ized intermediate layer of resin, healing efficiencies between
To the best of the authors’ knowledge, very few single or 4% and 35% were achieved. When completely immersed, FRP
double capsule systems have been demonstrated in FRPs samples exhibited an 80–95% lower uptake of EPA solvent
that meet the selection criteria specified. One exception pub- compared to a fiber-free polymer. This may be attributed to
lished by Bolimowski et al.[42] utilizes the same chemistry as both the obvious lower relative volume of polymer present, but
Coope et al.[31] to functionalize unidirectional (UD) composite also the constraints imposed by the fibers on swelling, inhib-
specimens. Although Bolimowski et al.[42] did not report fiber iting the uptake of solvent. Since the main fracture mecha-
volume fraction, it can be inferred from the cure procedure nism of FRPs in Mode I loading is dominated by interfacial
reported, and the manufacturer’s data, that the sample fiber debonding during crack propagation, rather than propagation
volume should be between 52% and 56%. Double-cantilever through matrix rich regions, it can be concluded that solvent-
beam (DCB) specimens demonstrated a maximum healing effi- based healing is not suitable in such domains (see Figure 3).
ciency of 44% with a catalyst loading of 10 wt%; however, even Conversely, however, solvent-based healing may prove benefi-
this modest recovery is amplified by the significant detrimental cial in areas that are typically resin rich, such as resin joints and
effect the inclusion of the microcapsules and catalyst particles areas with ply drops. However, healing imparted via entropic
has on initial fracture toughness: the strain energy release matrix swelling is reliant upon unreacted functionalities in the
rate of the virgin-functionalized specimens only averaged 22% matrix. As the degree of cure can increase over time, healing
of that of the nonfunctionalized control specimens. Virgin efficiency decreases with aging time.[45,49]
fracture toughness inversely correlated with catalyst loading; The findings of Bolimowski et al.[42] and Manfredi et al.[14,43]
however, decreasing catalyst loading resulted in lower healing hint at the disparate outcomes achieved when self-healing sys-
efficiencies, presumably due to incomplete cure of the healing tems are applied to materials that either meet or fall short of the
resin. Varying the concentration of microcapsules was found structural parameters defined at the outset. Future work needs
to have no effect on healing efficiency, indicating the lowest to focus on the ability of self-healing to address the greater
concentration in the study provided sufficient resin to cover
the fracture surface. Although it would be imprudent to draw
conclusions about the validity of microcapsule/catalyst systems
from a single study, and only based upon recovery of fracture
toughness, it nevertheless raises questions about the effective-
ness of this approach when employed in an FRP compared
with a bulk polymer. It should also be noted that Bolimowski
et al.[42] observed an average of 14% recovery in peak load in
samples with no catalyst present, suggesting a significant con-
tribution from the solvent-induced polymer reptation between
the host matrix and the healing agent, further questioning the
effectiveness of the microcapsule/catalyst system.
Manfredi et al.[14,43] provided further insight into the effects
of solvent inclusion and the disparity that exists between the
performance of self-healing in bulk and fiber-reinforced poly-
mers. Based upon previous work investigating solvent-induced
reptation for healing applications,[44–47] EPA microcapsules
were dispersed onto dry woven glass fabrics before infusion via
vacuum-assisted resin infusion molding (VARIM, an industri- Figure 3. SEM image of the crack face of a healing specimen, showing
ally relevant manufacturing route) to produce samples with a traces of capsules and fiber bridging. Reproduced with permission.[43]
nominal fiber volume fraction of 50%. However, the presence Copyright 2015, IOP Publishing.
loading that will typically be imparted to high-performance

FRPs, which are not present in low fiber volume or bulk
polymer materials.
2.1.2. Hollow-Fiber Systems
An alternative to the use of discrete microcapsules, and their

inherent geometric incompatibility with a fibrous architecture, Figure 4. The release of UV-fluorescent “healing agent” after pseudo
is to exploit the fibers themselves. The introduction of hollow impact damage. Reproduced with permission.[55] Copyright 2005, IOP
fiber elements allows sequestration of healing agents within Publishing.
the fiber reinforcement. Damage events see the rupture of the
hollow filaments, resulting in the release of the healing agents damage, which is not repairable by these recovery mechanisms.
stored within. This allows adoption of similar healing agent A 24 h healing cycle under ambient conditions was employed
approaches, as previously described for microencapsulation, in each case, following imparting of the damage. A healing
where effusion of the entrained liquids can undergo reaction efficiency of 73% was recorded; however, the research revealed
(polymerization) to restore surrounding matrix integrity.[50] a marked decrease in healing performance when the samples
Note that although a variety of fiber materials have been con- were stored for 3–9 weeks.
sidered, the majority of work has focused on hollow glass fibers Trask and Bond[55] reported a similar system of hybrid solid/
(HGFs). Importantly, a modest number of such hollow fibers, hollow glass fiber-reinforced epoxy. Interestingly, in attempt to
dispersed among the conventional reinforcing filaments, do not maximize efficiency, the researchers conducted a study to iden-
significantly degrade the global mechanical properties of the tify the most critical interfaces with their specimens and posi-
FRP.[51] tioned the healing fibers only at these positions. Again, damage
Bleay et al.[52] reported the first example of an HGF/epoxy was imparted using a quasistatic indentation method (Figure 4).
FRP which demonstrated (nonautonomous) self-healing However, in this work the researchers elected to use much
behavior. Continuous glass fibers of 15 µm outer and 5 µm shorter curing cycles, with test specimens subject to 100 °C for
inner diameter were impregnated with epoxy resin and cured 2 h before testing via four-point bending to ascertain flexural
into flat laminated panels containing hollow fibers orientated strength and stiffness. In contrast to previous work, some of
orthogonally. The hollow fibers within the laminate panels were these tested specimens were infused with healing agent prior to
then filled with healing agents; hardener in one fiber direc- autoclave curing. On average, the healable laminates exhibited
tion and epoxy in the other(s)—both epoxy and hardener were ≈84% strength of the nonhealable equivalents and were shown
diluted with acetone to facilitate this. This dilution was critical to recover 87% of the flexural strength of the undamaged non-
to lower the healing agent viscosity to a sufficient level to enable healable baseline laminate.
its efficient release and distribution from the hollow fibres. This Williams et al.[5] reported the first example of a healable
is of further note due to solvent-induced matrix swelling, which carbon fiber–reinforced polymer (CFRP). In this article, the
serves to increase the mobility of the surrounding polymer researchers targeted the repair of a specific damage category
chains. The researchers then used compression after impact which is frequently observed in FRP parts: barely visible
tests (CAI, 80 J) to quantify healing performance in terms of impact damage (BVID). HGFs were incorporated into two
strength recovery. After a postdamage treatment of vacuum- ply interfaces within a laminate of carbon fiber/epoxy and
assisted heating, modest strength recoveries of ≈5% healing after cure they were infused with a premixed two-part epoxy
efficiency were achieved (in accordance with Equation (2)). The healing agent. Quasistatic indentation of peak loads 1.7 and
researchers concluded that larger diameter fibers could help 2.0 kN was used to simulate BVID in each specimen prior
to increase the volume of healing agent availability and thus to healing (45 min at 70 °C, followed by 75 min at 125 °C)
overall healing performance. However, this modest recovery and being subject to four-point bend flexural testing. Two
may also be attributed to the high-energy nature of the impact; different HGF distributions were investigated (Figure 5).
unrepairable fiber-damage may have occurred. Although the higher density of HGFs resulted in a greater
Accordingly, Pang and Bond[53,54] reported a similar hybrid degradation of baseline mechanical strength (8% reduction
HGF/solid E-glass fiber epoxy composite with comparatively relative to unmodified CFRP vs 2% for lower density), these
large hollow fibers (60 µm external diameter, 50% hollowness). specimens exhibited the best healing performance following
Using a resin film infusion process, preimpregnated uncured the 2.0 kN indentation, reporting an 83% healing efficiency.
tapes were prepared containing the hollow fibers. Similar to
the work reported by Bleay et al.,[52] the acetone-diluted epoxy
healing agent was infused within the laminates postcure. These
researchers proposed an alternative method of healing quanti-
fication in terms of flexural strength recovery under four-point
bending. To create a controlled damage region within the test
specimens—and rupture the HGFs—a quasistatic indentation
process was used to simulate low-energy impact, in contrast Figure 5. The distribution of HGFs throughout a carbon fiber laminate.
to the previous research. High-energy impacts lead to fiber Reproduced with permission.[56] Copyright 2007, Elsevier.
This recovery represented a significant improvement in flex- two methods, hand lay-up (HLU) and a vacuum-assisted resin
ural strength relative to the damaged unmodified CFRP. The transfer molding (VARTM)-type process. After cure, the hollow
damage created by the 1.7 kN indentation was not sufficient fibers were filled with a catalyzed polyester healing agent
to investigate recovery. Evidently, there is a limited window (and accelerant). The healing performance was characterized
of damage in which it is practical to repair: too little damage, via three-point bend flexural testing before and after impact
where the recovery mechanism cannot be reliably initiated damage and a 12-h 60 °C healing cycle. Recovery of flexural
(and the recovery is inadequate compensation for the degrada- stiffness was observed for both VARTM and HLU prepared
tion of baseline performance), and too much damage, where samples although the HLU samples exhibited better recovery:
irreparable fiber breakage occurs. healing efficiency of ≈35%. This work concluded that the
Williams et al.[56] prepared, in a subsequent study, self- polyester healing agent was more robust than the epoxy agents
healing CFRP with embedded HGFs as described previously. reported previously due to the former’s insensitivity to mixing
However, here the authors came back to the compression after ratio (stoichiometry). The fiber-volume (reinforcement) fraction
impact method of assessing healing performance reported by (Vf) was not reported; however, simple calculations using the
Bleay et al.[52] This method was adjudged to be more repre- laminate thickness and density suggest that the Vf is lower than
sentative of realistic damage and a more stringent assessment the threshold set for this review. Nevertheless, this is a com-
of healing performance (than flexural testing) due to the sen- plex matter due to all of the fiber reinforcement being hollow.
sitivity of compressive strength to BVID. However, in contrast Consequently, the net density of the reinforcement is much
to the early CAI tests, these researchers used lower energy lower than with typical glass fibers although the ratio of fiber
impacts (<10 J), preventing significant fiber damage. The incor- to matrix may be representative of a structural composite, the
poration of HGFs at five interfaces within CFRP test specimens fiber volume fraction itself does not appear to be.
saw a healing efficiency of 63%. The authors concluded that Zhu et al.[59] presented an alternative to glass-fiber encap-
this hybrid system performed well though conceded that the sulation by preparing a woven E-glass epoxy composite (via
healing agent and method by which the HGF are infused are HLU), which incorporated a healing ply comprised of hollow
far from ideal. In essence, at this point, meaningful recovery polypropylene (PP) fibers (filled with epoxy/mercaptan). To pre-
of realistic composite damage had been clearly demonstrated; serve the healing agent in the hollow fibers through the cure
however, this was limited by idealized cure, manufacture, and process, test laminates were cured at either 25 or 40 °C. The
testing environments. samples were then subject to the indentation method of simu-
Kousourakis and Mouritz[57] sought to expand understanding lating impact damage prior to healing at 70 °C for either 4 or
of the influence of HGF inclusion on the baseline mechanical 8 h. Recovery was characterized using four-point bend flexure.
performance of FRP laminates using a similar CFRP system The novel healing agent represented a key part of this study:
to that reported by Williams et al.[56] Studies were made of the upon healing initiation, gas formation within the healing agent
influence of HGF diameter on the static stiffness, strength, was intended to facilitate an internal pressurization, leading
toughness, and fatigue performance of the host CFRP. These to its improved distribution across damaged areas. Impressive
experiments were performed with empty HGFs under the recoveries in flexural strength of up to 62% were reported for
hypothesis that any entrained liquid healing agent would not all specimen types containing healing agent.
significantly contribute to baseline mechanical performance. In Despite the success achieved by HGF-based healing systems
stark contrast to the microcapsules discussed in Section 2.1.1, in both carbon-reinforced and glass fiber–reinforced plastics,
this work concluded that there was little influence of HGF on several outstanding issues remain which have thus far pre-
stiffness or strength (compressive or tensile) when aligned vented their uptake in industrial applications. The studies by
with the load-bearing fibers (parallel to the loading force). Any Williams et al.[5,56] and Kousourakis and Mouritz[57] suggest
small decrease could be attributed to a reduction in the overall that inclusion of HGFs parallel to load-bearing fibers does not
volume fraction of higher performing fibers. In contrast, there severely degrade the baseline mechanical properties of the
is a small but significant degradation in both tensile and com- host laminate. Although there is a reduction in stiffness and
pressive strength (but not stiffness) with loading transverse to strength transverse to the HGF direction, this is usually modest
the HGF orientation for larger HGF diameters (≈200 µm). No and may be adequately offset by an increase in toughness. In
significant influence on fatigue performance was noted from addition, it is worth noting that in many of the aforementioned
Mode I testing though there was some improvement in tough- cases, the fiber volume fraction was not reported. This makes
ness. A linear increase in G1C (strain energy release rate) with it difficult to contextualize the quality of results: where the
respect to HGF diameter was observed. This effect was attrib- volume fractions are not representative of realistic high-perfor-
uted to the HGFs acting as “crack stoppers,” pinning any cracks mance FRPs, the results have less relevance.
in place. The investigation into fatigue performance here serves With a hollow-fiber approach, the most pressing challenges
to highlight its absence elsewhere, raising questions as to the relate to the manufacturing complexity and the nature of the
viability of HGF healing over extended periods. healing agent. Accordingly, researchers must consider the
In contrast to some of the earlier work by Williams et al.[56] position of the hollow fibres (e.g., critical interfaces or uni-
and Trask and Bond,[55] Kling and Czigány[58] reported an formly distributed) and the method by which they are incor-
H-glass/epoxy GFRP in which all of the reinforcement fibers porated (e.g., as separate plies or otherwise); when the healing
were hollow. Importantly, these HGFs were comparable in size agent should be incorporated (e.g., pre/postcure); the extent
to typical solid glass reinforcement, at ≈10–12 µm outer diameter of autonomy (e.g., thermally initiated or ambient curing); the
(≈50% hollowness). This FRP was cured into laminates via required volume of healing agent (i.e., Vf of the hollow fibers
and their size/hollowness); healing agent viscosity (i.e., is

it infusible); and stoichiometric mixing (where the agent is
multicomponent).
2.1.3. Vascular Systems
A third method of effecting self-healing using an extrinsic

approach is to utilize embedded vascular networks. Inspired
by the transport networks found in both animals and plants,
this approach uses networks of microchannels, or vascules, to
distribute and deliver healing agent to areas of damage. As for
the two previous mechanisms, the principle relies on crack for-
mation after a damage event propagating into an adjacent vas-
cule. In turn, this rupturing then triggers the localized release
of healing agent into the crack(s), initiating a repair.[60] This
approach shares many similarities to that described for HGF,
the primary difference being the replacement of discrete enti-
ties with a connected network. This difference has profound
consequences on the associated methods of manufacture and Figure 6. The delamination-repair cycle reported by Patrick et al. Repro-
the extent of reusability since in principle vascules may be duced with permission.[60] Copyright 2014, Wiley-VCH.
replenished to provide an ongoing healing capability.[16]
Williams et al.[61] developed a GFRP-skin/foam-core FRP Presenting an alternative method of network manufacture,
sandwich panel capable of repair via a vascular network Patrick et al.[60] created microvascular networks via interweaving
embedded within the foam core. A two-part low viscosity sacrificial polylactic acid (PLA) fibers (300 µm), infused with
(<1000 cP) epoxy resin was chosen as the healing agent. Both tin oxalate, into glass-fiber fabric preforms. The modified pre-
syringes and a peristaltic pump were tested as a means of forms were then infused using a VARTM process with epoxy
providing healing agent pressure, which was delivered either resin and cured into laminates. When exposed to raised tem-
premixed or as separate epoxy/hardener networks. Sandwich perature (200–240 °C), the tin oxalate catalyzed a depolymeri-
beam healing and general mechanical performance were zation of the PLA fibers to create a separate pair of embedded
assessed via edgewise compression (ASTM C364) after drop vascular networks. The researchers then investigated recovery
weight impact testing. All specimens were cured postimpact at in terms of delamination repair via DCB Mode I fracture tough-
ambient temperature for at least 48 h, prior to a final 1 h at ness testing. During the fracture toughness testing, the dual
60 °C. Of the three categories of self-healing specimens tested, networks of vascules were supplied with either hardener or
the “premixed resin” healing and the “high-pressure nonmixed” epoxy. The crack was allowed to propagate until the rupture of
healing performed well, with healing efficiencies of 143% and the vasculature, after which point the specimens were unloaded
150%, respectively. This apparent increase above the virgin and healing took place (48 h at 30 °C). The specimens were
state can be rationalized in terms of the local densification of reloaded and healed multiple times (Figure 6). In both of the
the damaged foam core. The third category of healable speci- vascule morphologies tested, an increase in fracture toughness
mens, “low-pressure nonmixed,” did not perform well, 39% was reported after healing, when compared with the undam-
healing efficiency. Post-test inspection revealed insufficient aged laminate. This was attributed to the improved toughness
resin-hardener mixing and consequently a failure of the healing of the healing epoxy relative to the matrix epoxy. Increasing
agent to cure. This is a perennial challenge associated with fracture toughness (in excess of 100% healing efficiency) was
multipart agents. Moreover, this particular result serves as an reported for consecutive healing events. Although excellent
important reminder that in almost all literature examples sys- recovery was reported, it is certainly debatable if the 48-h curing
tems that are intended to function “autonomously” do not typi- regime is viable. Clearly, this would be dependent on the parts
cally result in significant property recovery. in which healing is desired.
Norris et al.[7] prepared CFRP laminates incorporating The self-healing performance of cross-ply laminates after
a simple vascular array via a “lost wax” process using low static and fatigue loading was investigated by Luterbacher
melting point solder which could be readily removed post- et al.[62] In this study, a 1D vascular network was created via
cure. The laminates were tested via compression after impact polytetrafluoroethylene (PTFE)-coated nickel chromium wires
(low velocity, 10 J), where the healing was achieved via infu- (0.56 mm diameter). The wires were placed at discrete intervals
sion of epoxy resin through the network followed by a cure during the lay-up and removed after cure. The self-healing per-
time of 7 d. Different vascule sizes and specimen configura- formance of the specimens was characterized in both static and
tions were tested, in all cases healing efficiencies of ≈95% fatigue (4 Hz) tensile testing. Two epoxies (RT151, Resintech,
were observed. Evidently, useful recoveries were achievable UK and an in-house formulation) were investigated as healing
using the vascular method; however, as with HGF systems, agents by injecting in a premixed form (to avoid stoichiometric
questions remained about the viability of manufacture and limitations) with a syringe pump, after being subject to impact
healing agent inclusion. damage. The RT151 system was cured for 1 h at 65 °C followed
by 1 h at 45 °C, and the in-house epoxy was cured for 1 h at nature of the healing agent is also critical in determining the
45 °C prior to a further 3 d under ambient conditions. In the healing regime; it is important to remember what conditions
static testing, there was no significant difference in stiffness or are achievable in service (i.e., are high temperatures and 24 h
strength between the vascularized and baseline laminates. In cure cycles realistic?).
the fatigue experiments there was some observed degradation
in the stiffness of the vascularized specimens (compared to the
baseline), and this was more significant at high loading intensi- 2.2. Intrinsic Self-Healing
ties. Static healing efficiencies (stiffness) were reported as 74%
for RT151 and 67% for the in-house formulation. The differ- Intrinsic self-healing systems are based on the intrinsic func-
ence being attributed to the lower viscosity of RT151, facili- tion of a polymeric material and offer the prospect of effecting
tating improved infusion into damaged regions. Under fatigue repair via reversible bonding. These systems are often defined
loading, stiffness was seen to decay by 7% over 50 000 cycles, as “mendable” polymers as many of the developed systems heal
which was initially fully recovered by healing, only to fail again damage events under an external stimuli (typically heat).[64] In
within 2500 cycles of further loading. This relatively rapid FRP applications, the development of intrinsic healing matrices
degradation after healing is attributed to incomplete infusion can be divided into three main approaches: i) systems based
and poor adhesion. This work further highlights the need for on the reversibility of covalent bonds, ii) systems based on
research into the dynamic performance of self-healing FRPs, supramolecular interactions, and iii) systems based on polymer
good performance under static loading not necessarily being blends.
indicative of dynamic loading.
Despite the potential advantages of vascular networks over
other healing mechanisms, their inclusion in FRPs tends to sig- 2.2.1. Reversible Covalent Bonds
nificantly complicate manufacturing. Moreover, the inclusion
of 2D and 3D networks particularly disrupts fiber architecture The growth in use of thermoset matrices with or without fiber
due to the presence of vasculature transverse to the fiber direc- reinforcement for many applications has highlighted the dif-
tion. The inclusion of such networks results in void formation ficulty in repairing and/or recycling these covalently bonded,
and the presence of resin rich and resin poor pockets within crosslinked systems. Sastri and Tesoro[65,66] introduced cova-
the laminate. Vascule preforms also exert compression in adja- lently bonded disulfide groups within an epoxy network
cent plies up to a distance equivalent to 6 times the diameter that provided reversible crosslinks. These crosslinks could
of the vascule, affecting the stress field in the laminate when be activated upon heating while maintaining relatively high
subject to load.[63] These factors contribute to a degradation in mechanical performance, notably in terms of stiffness (Young’s
baseline mechanical performance which should be considered modulus in the range of 20–200 MPa). These systems further
when assessing healing efficiency. In general, these effects can demonstrated recovery of tensile strength in excess of 90%
be mitigated by reducing the diameter of the vasculature; how- after a thermal treatment for 2 h at 80 °C.[67] A recent study
ever, this necessitates higher pressures to achieve equivalent produced GFRP based on disulfide-containing organic–inor-
healing efficiencies.[6] ganic thermoset matrices with fiber volume fractions of about
Vascular healing has been demonstrated in structurally rel- 50%.[17] A postdamage thermal treatment for 16 h at 85 °C (with
evant FRPs in a variety of ways. As for HGF healing, efficien- an applied pressure up to 20 bar) demonstrated full recovery of
cies close to (or in excess of) 100% are achievable. However, interlaminar fracture toughness and 80% recovery in strength
these studies usually consider highly idealized arrangements. after an 8 J impact. The application of high pressure throws
The inclusion of a useful vasculature which does not sig- into question the action of healing.
nificantly degrade baseline mechanical properties has been The selective thermal cleavage and reformation of (the same)
achieved although none of these methods are fully optimized. covalent bonds represents a versatile strategy for repairing
Furthermore, excellent morphological control has also been thermoset polymers and does so in a manner akin to ther-
demonstrated, and repeated healing of macroscale damage moplastic welding. Moreover, this strategy has been demon-
(delamination) within a structural FRP has also been achieved. strated in polymers which afford mechanical properties close
Nevertheless, these achievements have not been made in any to those of commercial epoxies and unsaturated polyesters.
one single system, hence there are several challenges currently The most widely used reaction for these remendable systems
preventing the advancement of vascular healing as a concept. is based on the Diels–Alder (DA) reaction, where a high den-
In terms of manufacture, rate and reliability should be con- sity of crosslinks are generated by the reaction of electron-rich
sidered; some of these methods are more suitable than others dienes (such as furans) and electron-poor dienophiles (typically
for scaling-up. There remain significant question marks over maleimides). At temperatures above 120 °C, ≈30% of monomer
how healing agents are delivered in vascular systems: at pre- linkages disconnect, and they then reconnect upon cooling to
sent, manual pumping is required but this is unlikely to be restore both the mechanical and visual properties (i.e., erase
feasible in many applications and further raises the question scratches) of the undamaged polymer. The first material uti-
of autonomy. As with HGF and capsular systems, the nature lizing this strategy was reported by Chen et al.[68] and showed
of the healing is also important. No one agent has been opti- recovery of 57% in terms of fracture load and of 41% in frac-
mized for extrinsic self-healing though researchers should take ture toughness, as determined by compact tension tests after
care to consider viscosity and the different practicalities associ- thermal mending at 120 °C. Furthermore, the researchers
ated with single- and multicomponents systems. Ultimately, the observed only a 20% reduction in the load to failure between
the second and third healing cycles, showing the ability of these exploit such chemistries is a major impediment to realizing
systems to heal repeatedly. However, these systems have a low self-healing in the near term. A self-healing strategy exploiting
temperature resistance (glass transition temperature is about or adapting existing commercial resin systems is more readily
30–40 °C), which limits their application.[69] The chemistry and applicable. Thus, the idea of thermoplastic blending, where a
the wide variety of the different existing DA reaction systems new thermoplastic phase is dispersed within a thermoset host,
was reviewed by Liu and Chuo[70] and Bergman and Wudl.[71] can lead to an ability for the thermoplastic phase to separate
Park et al.[72] proved the utility of remendable DA matrices out under heating to facilitate crack healing. It is evidently dis-
within FRPs. They processed a CFRP using this matrix and putable whether the application of temperature is technologi-
achieved a fiber volume fraction of 40%. They assessed damage cally appropriate for providing crack healing on an in-service
healing efficiency through three-point bend flexural test and structural FRP. However, the systems are included here as they
showed recovery in strain energy of around 90% in up to three demonstrated large degrees of healing while keeping struc-
successive damaging cycles when followed by a 150 °C resistive tural integrity during the healing treatment, as further detailed
heating treatment. It is worth noting that in systems such as below.
these healing is performed above the glass transition tempera- Healing through the use of miscible polymer blends was
ture. Consequently, this method of repair may be inappropriate demonstrated by Hayes et al.[79,80] where a solid-state epoxy
for some structures under residual loading because this could resin was created by dissolving different weight fractions of
cause a loss of structural integrity. Moreover, no quantification thermoplastic (polybisphenol-A-co-epichlorohydrin) into epoxy
of the properties in the damaged state was performed. Impor- resin (Epikote 828). Upon curing, the thermoplastic remained
tantly, the hardness of these systems was comparable to that of dissolved within the thermoset. After damage, healing could be
traditional Bisphenol A diglycidyl ether (DGEBA) epoxy resin achieved through heating at 130 °C, enabling the thermoplastic
(e.g., EPON 862 from Momentive). Heo and Sadano[73] suc- to diffuse through the thermoset and allowing chain re-entan-
cessfully produced short-beam-shear specimens with a similar glement at the crack faces. Results showed recovery in load
resin system to that of Chen et al.[68] using a VARTM process. to failure of these blended samples up to 70% with compact
Respective short beam strength healing efficiencies of 85% and tension tests, and a recovery in Charpy impact energy of 50%
73% for the first and second healing cycles were demonstrated for the blends containing the highest (20 wt%) thermoplastic
while keeping laminate mechanical performance in the range content.[79,80] By tailoring the thermoplastic content as well as
of conventional carbon fiber/epoxy systems. the healing temperature, recoveries of up to 66% in terms of
reduction in damage area could be achieved after a 2.7 J impact;
however, no mechanical recovery was quantified.
2.2.2. Supramolecular Interactions For immiscible thermoplastic/thermoset blends, the healing
phenomenon is related to i) melting and subsequent volume
In order to avoid a thermal trigger, systems based on nonco- expansion of the thermoplastic phase, ii) flow of the melt into
valent interactions have been developed.[74] However, these the damage volume, and iii) characteristic physical or chemical
have resulted in relatively poor mechanical performance and phenomena taking place at the molecular level.[81] This last
are less appropriate for integration in FRPs. To overcome mechanism can consist of either chain re-entanglement of
this issue, while keeping the healing process at ambient tem- the thermoplastic melt, based on thermally enhanced chain
perature (despite improved healing at higher temperatures), mobility, or the formation of reversible noncovalent (e.g.,
hybrid systems combining reformable covalent and noncova- hydrogen or ionic) bonds in the thermoplastic phase. Two lim-
lent bonds were developed.[75–77] Sordo et al.[77] observed that iting morphologies are usually found, for healing purposes,
by combining bifunctional and tetrafunctional epoxy resins, depending on the constituent concentrations: i) a particulate
the resulting polymer had the ability to fully recover (24 h at thermoset phase embedded in a thermoplastic matrix (e.g.,
ambient temperature) tensile properties after being cleaved. refs. [81,82]), obtained by polymerization-induced phase sepa-
Sordo and Michaud[18] then demonstrated that these systems ration or ii) a particulate thermoplastic phase embedded in a
could be included within FRPs (with 50% fiber volume frac- thermoset matrix (e.g., ref. [83]), obtained by simple mixing of
tion) and could be produced using conventional manufacturing the thermoplastic material within the thermoset.
techniques. Moreover, after ambient temperature healing, Systems consisting of a particulate thermoplastic phase
recoveries of 65% and 72% of flexural strength and bending embedded in a thermoset matrix have been studied for thermal
stiffness, respectively, were reported[18] in addition to a recovery healing applications in neat resins[83,84] as well as when inte-
in impact force and dissipated energy of 55% and 76%, respec- grated into FRPs.[85,86] These studies mainly concern the iono-
tively.[78] Even though these systems have shown high healing meric copolymer ethylene methacrylic acid (EMAA) blended
and damping properties, at present, the low stiffness of any with epoxy. The interest of this ionomeric copolymer resides
resulting FRP limits their use to semistructural applications. in (i) the strong adhesion between epoxy and EMAA as cova-
lent bonds are formed during curing[87] and (ii) the thermal
expansion of EMAA that is 7 times greater than that of epoxy.[83]
2.2.3. Polymer Blends Meure et al.[83] showed that by including 15 vol% EMAA parti-
cles within an epoxy resin, the strong adhesion between the two
Self-healing via reversible covalent bonding has been shown to components increased the load to failure, measured by Mode
demonstrate efficient crack healing (Section 2.2.1). However, I DCB crack opening, by up to 25%. After thermal healing
the prerequisite to formulate completely new resin systems that of the cracked samples for 30 min at 150 °C, they further
compared to unmodified systems; however, toughness was

decreased by 40%. After Mode I DCB testing, up to 45% tough-
ness recovery and over 100% stiffness recovery were observed
over several cycles when the blended matrix composite sam-
ples were retested after thermal mending.[19] Damage recovery
after low-velocity impact was further assessed and composites
with epoxy-PCL blends demonstrated similar energy absorp-
tion capacity to pure epoxy composites even though the extent
of damage was higher. Ultimate compressive residual strength
of the modified composites was 20–30% lower as compared to
Figure 7. Delamination cracks in composites containing EMAA. Repro- unmodified composites; however, these modified matrix com-
duced with permission.[85] Copyright 2013, RSC Publishing. posites were able to fully recover low-velocity impact damage
after impacts of up to 8.5 J.[89]
demonstrated recovery in fracture toughness of up to 85%. This
amount of recovery was achieved thanks to the high expansi-
bility of EMAA, which in addition creates gas bubbles during 3. Healing Characterization
reaction with epoxy, creating pressure gradients that allow the
EMAA to flow and fill the cracks. Those systems were further Due to their heterogeneous nature, FRP composite materials
integrated into FRPs made by hand lay-up,[85] which were char- can lead to a multiplicity of failure modes depending on the
acterized via Mode I DCB testing. This testing demonstrated, intrinsic properties of the reinforcement and the matrix, but
as compared to a relatively brittle epoxy composite, a 63% also on the interaction between these phases and thus their
improvement in fracture toughness for EMAA-modified sys- interfacial and interlaminar properties. This article focuses
tems as well as the ability to recover this property by 156% after on composites with long fibers and thermoset matrices, thus
thermal healing thanks to the faculty of EMAA to expand and delamination, intralaminar, and translaminar fracture are the
bridge the crack faces (see Figure 7). The potential of EMAA for main matrix failure mechanisms along with fiber fracture. In
thermal healing in FRPs has been further demonstrated when theory, damage repair (autonomous or via manual intervention)
integrated in the form of layers[88] or stitches.[64] Even though can recover any degraded matrix property following any type of
those systems fully demonstrated the crack healing potential damage. To quantify this recovery, the repair/healing efficiency
in damaged FRPs, EMAA is highly sensitive to humidity (data concept was introduced and can be rewritten from Section 2
usually provided by manufacturers but no detailed research), as[3]
which prohibits for now its application development.
Systems consisting of a particulate thermoset phase f healed − f damaged
η= (3)
embedded in a thermoplastic matrix have also been studied f virgin − f damaged
for thermal healing applications in neat resins[81,82] and when
integrated into FRPs.[19,89] In these systems, the thermoset where f is the property of interest. However, caution must be
resin forms a network of interconnected particles, whose taken when calculating healing efficiency:[14]
interstices contain a continuous thermoplastic phase having
• f damaged, the property in the damage state, is often set to zero.
a significant self-diffusion (reptation) at moderate tempera-
This setting is true when measuring fracture properties in
tures, but adequate thermal stability under service conditions.
Mode I where the fracture resistance, after testing, is null.
Luo et al.[82] first proposed creation of an epoxy-polycaprolac-
However, for other fracture modes and loading conditions,
tone (PCL) biphasic morphology containing 15.5 wt% PCL
the property in the damage state is not null and thus needs
content in the matrix; the microstructure consisted of a PCL
to be measured.
matrix with particulate epoxy. Epoxy particles were thus inter-
• f virgin, the property in the virgin state, can be that of the un-
connected and acted as “bricks” for load bearing, while the
modified material (without healing system) or that of the
expansion of PCL (which expands at least 10 times more than
modified material (with healing system). Indeed, the healing
epoxy at elevated temperatures) facilitates a differential expan-
system usually influences the initial material properties, and
sive bleeding mechanism (comparable to EMAA blends). In
depending on which virgin property is used, the healing ef-
a recent study, Cohades et al.[81] investigated a similar system
ficiency can be completely different. Thus, considering both
using lower healing temperature (150 °C, well below epoxy
of these cases is of critical importance.
Tg of 200 °C vs 190 °C for Luo et al.[82]) in order to maintain
• f healed, the property in the healed state, can reach higher values
structural integrity. Blends with 25 vol% PCL exhibited a
than the property in the virgin state, in some cases. This is
healing efficiency in excess of 70%, while retaining suitable
because many crack healing mechanisms create an interface
ambient-temperature mechanical properties (E > 1.5 GPa and
of a new material and this material may have improved per-
KIC > 0.5 MPa m1/2) and were concluded to be promising can-
formance relative to the original matrix.
didates to be integrated as a matrix for healable FRPs. Woven
E-glass FRPs with the epoxy-PCL blends were then manufac-
tured through a VARIM process at high temperature.[19,89] The Section 2 provides a review of healing systems that were
resulting laminate showed similar storage and bending mod- used to repair damage into structural FRP composites. These
ulus, bending strength, and interfacial shear strength when systems are compiled in Table 1, which defines the processes
Table 1. Self-healing structural composites: mechanical properties and healing efficiency. Values indicated with “*” were recalculated from data pro-
vided in the reference.
Healing Type of Type of Process Vf [%] Strength Strength Toughness Healing Healing Healing Ref.
agent matrix reinforcement method [MPa] baseline assessment treatment efficiency
[MPa] [%]
Microcap- DCPD– Epon 828 Plain weave HLU, 37.5* 2.79–2.85 WTDCB RT for 48 h 38% [25]
sules Grubbs’ carbon fiber compression MPa m1/2 80 °C for 48 h 66%
DCPD– Epon 862 Plain weave HLU, 33 82 (com- 80 (com- CAI at 45.1 J RT for 48 h, 68% [13]
Grubbs’ S2-glass fabric compression pression) pression) 1077 kPa 78%
DCPD– Epon 862 Plain weave E-glass HLU, 36 Indentation Overnight at 100% [96]
Grubbs’ compression 30 °C
EPA solvent Epon 828 Weave 2 × 2 E-glass VARIM 54 160 (com- 260 (com- 796.8 J m−2 Mode I DCB 1–5 d at RT 0% [43]
pression) pression)
Hollow Epoxy Epoxy HGF (Hollex) Prepreg CAI at 80 J 60 °C for 1 h 5%* [52]
Fibers
MY750 Epoxy Hybrid HGF/E-glass Prepreg 61.5 624 668 4PBT after ≈0.5 24 h at RT 73%* [54]
epoxy fibers (bending) (bending) J impact
Cycom 823 Epoxy Hybrid HGF/E-glass Prepreg 57 559 668 4PBT after ≈0.6 2 h at 100 °C 129%* [55]
fibers (bending) (bending) J impact
Discrete Epoxy Hexcel T300/914 Prepreg 60 535 583 4PBT after ≈0.7 70, 125 °C 83.2%* [5]
HGF (bending) (bending) J impact
Discrete Epoxy Hexcel T300/914 Prepreg 60 373 (com- 330 (com- CAI at 3 J 70, 125 °C 62.6%* [56]
HGF pression) pression)
Discrete Epoxy Hexcel T300/914 Prepreg 60 580 (com- 610 (com- 600–800 [57]
HGF pression) pression) J m−2
Discrete Epoxy H-glass weave VARTM 25.1* 270 3PBT after RT for 120 h 19.3%* [58]
HGF reinforcement (bending) impact at 9.3 J
H(PP)F Epon 828 Woven E-glass HLU 41 500 530 4PBT after RT 62%* [59]
(bending) (bending) ≈0.5 J impact
Vascular Epoxy 913 Epoxy E-glass (Hexcel) Prepreg 58 184 (com- CAI at 3 J RT, 60 °C for 40%* [61]
pression) 1h
Epoxy with Epoxy Hexcel IM7/8552 Prepreg 58 350–370 390 (com- 1000 J m−2 CAI at 10 J RT for 7 d 96–99% [7]
EPA (compres- pression)
sion)
Epoxy Epoxy 8H satin weave VARTM 20.5* 480 J m−2 Mode I DCB 30 °C 48 h 90% [60]
E-glass
Epoxy Epoxy E-glass (Hexcel) Prepreg 57 462 (tensile) 471 (tensile) Static tensile 1 h at 45–65 °C, 67–74%* [62]
or blend testing at 15 kN RT 3 d
Reversible Disulfide Same as Weave 2 × 2 E-glass VARIM 51 75 (bending) 389 650 J m−2 Visual after 8.5 16 h at 85 °C 85.3% [17]
covalent thermoset healing fabric (bending) J impact with 0.2 bar
bonds
Mendomer Same as AS4 carbon fibers Pressure 40 Recovery in 150 °C for 1 92–94% [72]
(DA rx) healing holding strain energy min
Polyure- Same as Unidirectional VARTM 55.7 (short 65.2 (short SBS 6.9 bar, 135, 87.8% [73]
thanes healing carbon fabric beam) beam) 90, 70 °C
(DA rx)
Supramo- Reverlink Same as Weave 2 × 2 E-glass VARIM 48.9 33.5 389 Visual after 20 J 28 d at RT 100% [18,78]
lecular HR-NR healing (bending) (bending) impact
Polymer Polybi- Epon 828 Cross-ply E-glass VARIM Visual after 2.7 130 °C for 1 h 30% [79,80]
Blends sphenol J impact
EMAA Epoxy Plain-weave carbon HLU 52 410 J m−2 Mode I DCB 150 °C 30 min 156% [64,85]
PCL Epon 828 Weave 2 × 2 E-glass VARIM 48 349.9 389 620.3 J m−2 Visual after 150 °C 30 min 94.9% [19,89]
(bending) (bending) 8.5 J impact
in terms of the same healing categories, while highlighting Healing assessment in terms of strength recovery after
process parameters, resultant mechanical properties, and the impact damage can also be performed through the CAI test
methods and values of healing assessment and quantification. (i.e., measuring the recovery of compressive strength after
The fiber volume fraction achieved is, in general, a function of impact). Notice that Equation (3) is necessary to quantify
the manufacturing method used and is observed to vary sub- healing efficiency in this case as the specimens are only “par-
stantially between the different studies. This parameter is a first tially damaged.”
indication of the extent of variability that is observed when com- Flexural testing is commonly used to measure flexural mod-
paring mechanical properties of different systems. This topic ulus and strength. In structural composites, these tests can be
is further discussed in Section 4. Mechanical properties of the used to measure the flexural properties of specimens that have
self-healing FRPs are reported in Table 1 in terms of a variety been previously impacted at a given energy. To quantify healing
of metrics: first, strength, which is usually measured either in efficiency, this test requires again three states to be measured:
flexure or compression. This strength can be compared to the the virgin state, the healed state and the damaged state.
strength of a baseline FRP with the same reinforcement, but Excluding quasistatic, Mode I fracture testing, healing
without healing functionality. Second, resistance to crack prop- efficiency quantification is typically performed after impact
agation (i.e., interlaminar fracture toughness) is also reported. damage. The impact energies can vary between 1 and 80 J
This is typically compared to a baseline woven E-glass/epoxy (see Table 1). Presently, all FRP self-healing systems have been
laminate of interlaminar fracture toughness of 1000 J m−2. developed to heal matrix damage only, i.e., no fiber damage. It
Although stiffness (flexural, compressive, and tensile) has is, therefore, debatable whether characterization of specimens
been reported as a means of characterizing healing, Table 1 that have been impacted at high energies, where fiber damage
is limited to considering strength for simplicity, and because has been induced, is relevant or not.
it is more representative of crack propagation and healing. Once damage has been created in a part, healing is not
Finally, the quantification method, the healing treatment, and instantaneous. Healing treatments vary in nature, but all
the reported healing efficiency are given in Table 1. Again, the depend on three critical parameters: time, temperature, and
diversity of tests and healing treatments prevent direct com- pressure. Generally, as the value of these parameters increases,
parison between systems, consequently comparisons should be healing is observed to be more effective. Again, it is disput-
considered as only qualitative. able whether the application of temperature and pressure is
The recovery of mechanical properties in healable FRP technologically appropriate for providing crack healing on an
composites has been measured via four main methods: in-service structural FRP. While pressure is complicated to
Mode I crack opening, impact, compression after impact, and apply to in-service parts, localized heating can be compara-
bending. Quasistatic fracture experiments (e.g., Mode I with tively easily applied to damaged regions. Although thermal
the double cantilever beam or the width-tapered double can- resolution is poor in general because of the facile conduction
tilever beam (WTDCB) geometries) are commonly performed of heat, the relatively poor thermal conductivity of common
when crack healing needs to be assessed directly because they FRP matrices enables reasonably good spatial control. The use
allow controlled and predictable crack propagation under ten- of elevated temperature to induce/improve repair needs to be
sile loading. Even though Mode I crack opening is convenient carefully considered. Heating must be assessed for compat-
and provides controlled crack propagation, it implies large ibility with the material’s thermal transitions as some healing
damage volumes which are not representative of realistic temperatures can be above the matrix Tg and thus could com-
damage experienced by structural parts during service. Tests promise structural integrity. However, the thermal repair of an
generating smaller damage volumes are thus of more prac- unloaded structure, assuming sufficient spatial control, should
tical interest in terms of industrial applicability. In contrast be possible without structural compromise. Finally, even with
to Mode I experiments, low velocity impact testing is highly fully autonomous systems (where repair is affected without
representative of in-service damage and has been extensively heat, pressure, or any other additional stimulus), the duration
reviewed.[90,91] Low velocity impact damage is a critical con- of repair must also be considered; in many cases it may be of
cern in the aerospace sector where imperceptible damage greater technological concern to leave a part to heal for several
events, attributed to operational or maintenance activities, can days at ambient, than to apply raised temperature.
greatly decrease strength, durability, and stability of the struc- Table 1 shows the maximum healing efficiency for each
ture.[92] For example, in carbon–epoxy systems, BVID (arising reported study. This has been recalculated from the data pro-
from low velocity impact of blunt objects, typically 5–10 J) can vided in the corresponding publications, in accordance with
lead to tensile and compressive strength losses of up to 50% Equation (3). As some data were taken directly from graphical
and 60%, respectively.[93] Higher-energy impacts can easily representations, there may be a discrepancy between values
occur through such events, thus there is a significant interest originally reported and those recalculated herein. As expected
in the autonomous healing of impact damage of up to 20 J. from the variation seen with fiber volume fractions, mechan-
Healing efficiency can be quantified through this test in terms ical properties and assessment methods, the reported healing
of recovery of impact force and absorbed energy; however, efficiencies vary substantially. However, these values are a
the obtained values were demonstrated to be only indicative function of the damage volume to be healed, which is typi-
of general trends (due to large statistical uncertainty).[17,78,89] cally a function of the impact energy imparted to the object
A more common way to measure healing efficiency is to (see Figure 8). As a general trend, for each healing system,
measure the recovery in impact damage area by means of healing efficiency decreases as the impact energy increases.
optical or C-scan observations.[17,78,89] This is because delamination becomes more important, but
in each reported study, without any healing functionality). For

the case of intrinsic systems, these reference materials simply
employ a commercially available thermosetting polymer matrix.
This comparison shows that a maximum level of healing can
be reached in all healing systems (assessed as a function of
the damage volume, see Figure 8). However, the normalized
strength leads to a sharp divergence between intrinsic and
extrinsic systems. Even though extrinsic healing systems can
be quite intrusive, a tailored insertion (as has been studied by
many research groups) provides strength values close to base-
line systems, which is not the case for intrinsic, where new
resin formulations are necessitated and consequently it is more
difficult to reach the strength of commercial systems. Note that
the solvent filled microcapsule systems of Manfredi et al.[94] are
not linked to extrinsic systems in Figure 9 due to their non-
healing behavior.
Obviously, the comparison of the different healing systems
found for structural FRP composites in the available literature
reveals that small damage volumes can be fully healed. How-
Figure 8. Healing efficiency as a function of the impact energy used to ever, the reliability of further comparison is questioned because
provide damage, for the studies reported in Table 1. Some studies proof i) the multicomponent nature (reinforcement, matrix,
vided additional healing data, which have been included here.
interface, and healing system) and ii) the many different tests
used to characterize healing. Now that self-healing is increas-
also because above a critical threshold fiber rupture is initiated. ingly studied in the literature, there is a need for a concerted
The rate at which the efficiency decreases when impact energy effort to drive toward standardization of testing, and the use of
is increased differs for each system due to various parameters specific reinforcement architectures in order to allow reliable
and cannot be readily identified from Figure 8 as work is usu- comparison between the available healing systems in struc-
ally only reported at one impact energy. In that respect, two tural FRP composites—both in terms of the initial and the
studies reported healing efficiency values for three different recovered properties (i.e., healing capability). Such an effort
impact energies,[17,89] and the tendency for the repair system has been made to quantify the healing efficiency of thermoset
to be less efficient as damage increases was confirmed accord- polymers;[95] however, structural composites have an additional
ingly, but the rate of this decrease was not consistent. A second dimension of complexity where the variation in reinforcement
representation of healing efficiency values is given in Figure 9 type, architecture, and stacking sequence, all complicate the
as a function of the strength of the healing FRP in the undam- overall comparison. This natural complexity of FRPs means
aged state normalized by the strength of a baseline (provided that it is of paramount importance to not further complicate
analysis. Consequently, some form of standardization, or at
least uniformity in the reinforcements and testing methods is
necessary. Self-healing systems have been broadly designed to
repair microcracks, and tests which characterize healing need
to be designed to impart appropriate levels of damage to the
specimens to reflect this. The Mode I DCB test is certainly a
Healing Efficiency [%]
useful tool in many cases for measuring the interlaminar frac-

ture toughness of composite laminates; however, it induces
large damage volume relative to the specimen dimensions.
Therefore, to prevent this, specimens need to be clamped to
provide smaller damage volumes and thus test predetermined
levels of healing. This clamping procedure is not representa-
tive of what would occur in real-life structures. Thus, mode
I DCB testing needs to be complemented by testing which
Vascular investigates response to impact damage, and allow healing of
representative damage volumes. Through impact testing, the
extent of specimen damage can be readily varied (by the impact
energy used) and thus the range of healing effectiveness can be
(Ststem Strength)/(Baseline Strength) quantified. Evidently, there is no one perfect solution, without
a consensus of the materials community, to the question of
Figure 9. Performance map of self-healing structural composites, from
which reinforcement and standardized test to use in order to
data provided in Table 1. Maximum healing efficiency is shown as a func-
tion of system strength normalized by the baseline strength (i.e., without characterize healing. However, the authors suggest that new
healing functionality). Data are divided as intrinsic or extrinsic, and fur- studies in this field should, at the first instance, model their
ther subdivided as microcapsules, hollow fibers, vascular, and intrinsic. experiments on established studies (in terms of reinforcement,
techniques and conditions) in order to facilitate comparison

with at least one existing system. Also, if impact testing is to
be adopted, the above discussion leads to the recommendation
that several impact energies should be used in order to test
different extents of damage. An attempt to reliably compare
three different systems was reported for the intrinsic healing
of composites.[17,78,89] In these studies, similar impact energies,
reinforcements, architecture and ply stacking sequences were
used. These investigations, combined with the fact that in two
of these studies[17,89] three different damage volumes (impact
energies) were assessed, allowed a fairer comparison and thus
yielded a more reliable conclusion of the effectiveness of those
systems. Note that even though Mode I DCB tests were first
used to prove the efficacy of the materials, they extended the
characterization to impact damage in order to obtain damage
representative of real life structures.[17,19,89]
4. Manufacturing Compatibility
As reported in the previous section, high volume fraction
fiber composites are considered herein; thus, any potential
healing system or matrix will have to be compatible with the
geometrical and physical requirements linked with composite Figure 10. Schematic description of fiber-reinforced composite pro-
processing practices. When considering thermoset composite cessing techniques. Panel (a) is a manual process whereby resin is forced
processing, a few methods are currently economically viable for by a compaction roller into the reinforcement; panel (b) is an infiltration
generic parts (depending on the desired volume of production): process whereby liquid resin is forced to flow (under pressure or vacuum
or both) into the reinforcement held within a mold (rigid, or with a flex-
hand lay-up, liquid composite molding techniques (vacuum ible top); panel (c) is a consolidation process whereby preimpregnated
infusion, resin transfer molding), molding of preimpregnated reinforcement is pressed in a closing mold; and panel (d) is a process
fabrics or of prepreg materials under vacuum only or under whereby a preimpregnated reinforcement is sealed under vacuum, then
pressure imposed by gas (autoclave) or by a press, and finally consolidated and cured in an oven or an autoclave. Automated tape
automated tape lay-up or placement. These are schematically lay-up or placement is not shown here, but consists in robotic placement
presented in Figure 10. In addition, the reinforcement can be of preimpregnated reinforcement tapes followed by consolidation as in
(c) or (d).
under the form of unidirectional plies or noncrimp fabrics, or
of textiles.[97]
When integrating extrinsic healing systems with FRPs, the 0.3 bars as pressure difference, instead of a full vacuum as usu-
architecture of the reinforcement is of critical importance. ally employed. The microcapsules used for self-healing studies
Indeed, woven reinforcements have been used in many healing in FRPs were typically of 200 µm diameter, which is around
studies, rather than unidirectional arrays, because resin rich 20 times larger than single glass or carbon fibers (8–16 µm).
areas are found between crossing warp and fill yarns where The thickness of the stacks of plies is therefore increased,
the capsules can be naturally stored without affecting the which results in a decrease of the fiber volume content and of
intrinsic waviness of the fiber reinforcement, if their size is virgin interlaminar fracture toughness. As these microcapsules
lower than the resin rich gaps.[13] The introduction of extrinsic may have a toughening effect (by crack arrest), a compromise
systems, which have a dimension comparable to or larger than can be found to maintain similar interlaminar properties to
the natural void space in the reinforcement will generate dis- nonhealable FRPs. Another route would be to vary the aspect
tortions, and in turn modify the achievable Vf as well as ini- ratio of the spherical particles in order to keep the same volume
tial (mostly compressive and interlaminar, as well as fatigue) of liquid healing agent, while keeping the same FRP thickness.
properties.[6,8,15,26,43,54,98,99] Recent work[8] also showed that a compromise can be found
In addition, processing must be adapted to ensure that the by an adapted match of the capsule dimensions and the textile
capsules or hollow fibers do not rupture during part manufac- microstructure, allowing capsules to nest without deforming
turing. This is a major limitation of these extrinsic systems for the structure too much. Nonetheless, scale-up possibilities
their application to high volume fraction fiber composites. In remain limited, also considering that capsule production is so
general, low-pressure consolidation techniques are required, so far often proposed in laboratory scale batch process, although
as to preserve the microcapsules or hollow fibers from breaking scale-up can be envisaged using continuous parallel microflu-
during processing. As an example, in the work of Manfredi idics lines, or techniques borrowed from industrial microcap-
et al.,[43,98] urea–formaldehyde capsules, which were shown to sule production, as found in laundry detergents.
rupture under a force of 4–6 mN could be successfully intro- Systems relying on the introduction of fugitive vascular
duced into glass fiber composites (with fiber volume frac- systems are somewhat more easily adapted, as they are well
tions around 40%) only with an adapted VARIM process, with integrated into the reinforcement and can sustain higher
compaction pressure. As a result, vascular systems and hollow 5. Self-Healing by Design: Targeted Applications
fibers are in general more compatible with unidirectional rein-
forcements (as often used with preimpregnated plies), as long To date, the pursuit of self-healing FRPs has focused almost
as they align well with the fiber direction, in a similar fashion exclusively on laboratory or small-scale studies, investigating
as integrated optical fiber networks.[100,101] uniform and regular coupons, with few examples of attempts to
Concerning the resin properties, in general, the resin work at the larger scale.[88,104,105] Furthermore, it has generally
needs to flow a certain amount, in particular in liquid com- been the case that workers have sought to confer a self-healing
posite molding processes. For intrinsic systems, the resin vis- capability as a ubiquitous function throughout the entire FRP
cosity is usually higher than that of conventional resins and host. While this could result in a fully self-healing material, as
flow is thus limited, leading to slower processes, or to the has been shown above, it raises questions of manufacturability,
need for higher processing temperature. For extrinsic sys- cost and performance. While this approach has served its pur-
tems containing microcapsules, the viscosity of the polymer pose most effectively, it has perhaps missed an opportunity to
is greatly increased if the capsules are mixed into the flowing demonstrate self-healing FRPs as a solution to specific complex
polymer; aggregation of the particles might severely limit the and difficult loading or damage scenarios, often very localized
flow of liquid through the reinforcement. One solution to this as a result of design, and thereby justify the further investment
problem, which does not change the resin permeability (and in their realization.
can even increase it), is to functionalize the reinforcement For example, skin-stiffened FRP panel structures are widely
prior to infusion as demonstrated by Manfredi et al.[98] How- used for lightweight applications in aerospace, and increas-
ever, in that case, the choice of epoxy resin system remains ingly finding applications in other sectors; marine, automotive,
limited to resins with relatively low cure temperature, so as renewable energy. For a variety of reasons, such as access holes
to remain within the stability range of the capsules (in terms or changes in section or loading, stringers have to be termi-
of capsule shell stability and breaking strength above room nated. These so-called “stringer run-outs” are sensitive to local-
temperature). In general, room temperature or low cure ized damage initiation as the loads from the stringer end have
temperature (around 35 °C) are selected to demonstrate the to be transferred into the skin, leading to the need for localized
principle. design solutions or the use of additional fasteners to assure fail
Experimental studies reported above often rely on a hand safe design criteria can be met.[106] Upon excessive loading, the
lay-up manufacturing technique to process FRPs. However, stringer tends to delaminate at the skin–stringer interface and
this process leads to low fiber volume content (usually 30%) as then delaminations propagate into the skin. This mechanism
well as the high void content of the final part. Mechanical prop- has been described and covered by several authors.[107–110]
erties in the virgin state may therefore be lower than expected Other examples of local design features where self-healing
for a structural composite; as a result, healing assessment may offer benefits include, T-joints, changes in panel/section
may be biased. Higher reinforcement contents and improved thickness (ply drops), holes (either for access or for fasteners),
quality are obtained using vacuum infusion, as reported in adhesive joints, and free edges (Figure 11).[9,111] By their nature,
refs. [17,18,78,89] for example. Although the host FRPs inves- all of these design features can give rise to damage cases as a
tigated thus far have typically been prepared via simple hand result of their in-service loading. This raises the prospect of
lay-up or infusion techniques, several additional techniques “design for self-healing” where consideration can be given to
have been utilized to incorporate the self-healing function- the inclusion of self-healing functionality from the outset, as
ality.[7,17,19,43,53,62,85,96,102,103] Autoclave compaction of preimpreg- a means of avoiding implementation of more complex or less
nated plies has also been explored with capsules, as reported in satisfactory remedial measures, e.g., mechanical fasteners,
Bolimowski et al.[42] Adaptation in the cure and postcure tem- Z-pinning.[112]
perature may however be required. As an example, for micro- The judicious selection, combination and tailoring of the var-
encapsulated healing systems, the temperature needs to be ious intrinsic and extrinsic self-healing approaches, to address
carefully chosen to avoid damaging the microcapsules, whereas the likely type and extent of damage morphologies commonly
in latent functionality systems, appropriate cure schedules need associated with such practical design features, provides an
to be defined in order to achieve the right amount of latent opportunity to demonstrate the value and capability of self-
functionality. healing within FRPs. Furthermore, the inherent manufacturing
Overall, all processes are in principle applicable as long techniques, necessary for the creation of these design features,
as pressure and temperature are tailored to the healing provide opportunities to incorporate self-healing functionality
system, and as long as the size of healing reservoirs is com- without major detriment or disruption to the host structure,
patible with the porosity of the reinforcement network, e.g., microcapsule inclusion in void region associated with ply
which is sometimes difficult to achieve. Many of the self- drop, vascular network localized to edges of access/fastener
healing embodiments or chemistries still cannot readily be hole.
scaled and/or are incompatible with industrial processing
techniques such as autoclave cure, resin transfer molding,
automated tape lay-up or automated fiber placement. These 6. Validation
incompatibilities may be attributed to a variety of factors, the
main ones remaining the geometrical, thermal and mechan- To find application in commercial products, self-healing FRPs
ical constraints imposed by the reinforcement and the pro- must be trusted by both the original manufacturers and end
cess requirements. users. Broadly, this confidence in their ability to perform
Figure 11. Typical localized design features suitable for targeted application of self-healing FRPs: a) T-joint, b) open hole, and c) stringer run-out (scale
bar = 1 mm).
reliably and repeatedly is a product of validation. Hereafter electromagnetically active healing agent. Upon release/cure
we define validation as evidence that the technology works, of the agent, the electromagnetic profile would locally change
is robust and reliable, and is demonstrable via inspection. In around the damaged area. Regular inspection during routine
much of the literature, including most of the state of the art maintenance would thus reveal the action of the healing mech-
presented herein, the work has successfully proven that the anism. Techniques for embedded sensing associated with SHM
technology can recover some degree of structural performance. have been the focus of a considerable amount of research, one
Whether this has been demonstrated to an extent that means of the most developed techniques is described by the research
the technology is cost-effective is another matter. Regardless, of Kirkby who has used optical fibers to detect the occurrence of
in most of the examples given, the work has not proven that impact damage.[114] It is very likely that only a self-healing
the technology is robust and/or reliable. With the advent of a functionality in an FRP which can be uncompromisingly vali-
new technology, proof of concept is paramount, whereas reli- dated will attract the confidence of both manufacturers and
ability and robustness are secondary considerations. However, end-users.
it is now suggested that in order for self-healing technologies to
progress to the next stage of development, it is perhaps time to
initiate research to address these concerns. 7. Conclusion
Evidently, validation is intimately related to the nature by
which recovery is quantified (see Section 3). After an initial Although numerous self-healing systems have been tested in
period of testing under laboratory optimized conditions, fur- FRPs, there remain no examples of their industrial/commer-
ther work needs to consider more realistic load cases (both cial application. In this report, a multitude of factors have been
static and dynamic), the effect of environment (pressure/tem- identified that have contributed to this lack of commercializa-
perature/humidity) and the effect of extended service/longevity. tion, some factors are common to all self-healing systems,
The ability to validate self-healing (either in its latent potency or whereas others are system specific. The specific advantages and
final completion) by nondestructive methods is urgently needed perhaps more pertinently disadvantages of each self-healing
to avoid the current test to destruction approaches. Computa- category are detailed in Table 2.
tional science, in the context of structural fiber-reinforced com- Further to the system-specific limitations addressed in
posites, may also bring large improvements in the validation of Table 2, there are more general considerations that apply in
such systems for a potential implementation, as has been seen almost all cases of self-healing composites. In general, the
with self-healing ceramic matrix composites for example,[113] majority of self-healing studies—even those citing structural
but has not been deeply studied for self-healing FRPs yet. The composites as a target application—do not involve the testing
combination of self-healing with a sensory network to create of FRPs, which are structurally useful. That is, self-healing is
a responsive form of structural health monitoring (SHM) infrequently tested in materials that approximate the mechan-
has been proposed; however, this is likely to compromise the ical performance of industrially relevant materials. A vast
goal of autonomy currently sought. Perhaps more simply, one number of healing chemistries and architectures have been
might imagine an extrinsic self-healing system which uses an investigated, but more focus is required on researching the
Table 2. The system-specific advantages and limitations of self-healing studies that have led such investigations, healing performance
categories, note that healing efficiencies approaching (or exceeding) has been observed to significantly deteriorate with time.[53,62]
100% have been reported for each category as seen in Table 1. Note that
more general limitations which apply to all (or almost all) self-healing
More of these investigations are required, as is analysis of the
systems are not tabulated but are discussed below. root cause of the instability: only by investigating this phenom-
enon can effective solutions be proposed.
Self-healing system Advantages Disadvantages/limitations Similarly, few systems have demonstrated fully autonomous
healing and those that have usually suffer from low healing
Microcapsules Autonomous activation Increases manufacturing
has been demonstrated complexity
efficiencies. In multicomponent (extrinsic) systems this is often
due to incomplete cure of the healing agent, which in turn is
Catalyst activation Moderate reinforcement
a product of poor mixing and/or uneven reagent distributions.
eliminates requirement disruption: significant degra-
for stoichiometric
Overall, in both intrinsic and extrinsic classes, heat is usually
dation of baseline mechanical
mixing of healing agent performance
required to affect a repair within practicable timescales. This
is true of the vast majority of successful systems and imme-
Limited damage volumes
diately characterizes them as semiautonomous at best. How-
Single use at any one loca- ever, the suggestion from this report is that semiautonomous
tion, with residual voidage
healing is, at present, a much more achievable target than fully
after rupture
autonomous healing and importantly may be just as industri-
Dual capsules: stoichiometric ally relevant.
mixing required
As illustrated by Table 1, there exists a variety of ways to
Healing agent leakage characterize recovery and calculate healing efficiency. However,
Challenging encapsulation there is a clear need to test self-healing composites in less-ideal-
and dispersion ized conditions, and with more standardized and representative
Hollow fibers Little degradation of Increases manufacturing tests. This will lead to a more objective analysis of “healing per-
baseline mechanical complexity formance.” The key recommendation from the analysis in this
properties article is that future research should consider using impact-
Moderate healing Limited reuse based experiments to characterize healing at the first instance.
volumes This is primarily due to these tests being more representative of
Autonomous activation Dual fibers: stoichiometric realistic damage states. Furthermore, to most effectively charac-
has been demonstrated mixing required terize recovery, at least three different impact energies should
Healing agent inclusion is
be used: healing efficiency tends to be inversely proportional to
challenging the extent of damage. Evidently, the sensitivity of this relation-
ship is a metric which deserves attention.
Vascules Large healing volumes Severely increases manufac-
In general, the field of self-healing is plagued by the com-
turing complexity
plications it imparts on manufacturing. In composites this
Extensive reusability Limited autonomy (manual
is compounded by the variety and complexity of processing
pumping of healing agents)
methods. For all cases, the temperature should be carefully
External reservoirs of 2D/3D networks can sig- controlled during manufacture to strike a balance between the
healing agent reduce nificantly degrade baseline
optimized mechanical performance of the host composite and
stability issues mechanical performance
the optimized healing performance of the embedded function-
Pressurized delivery Dual networks: stoichiometric ality. In the case of intrinsic systems, this means controlling the
promotes mixing mixing required (remains
temperature to retain enough latent functionality. In the case
poor even when pressurized)
of extrinsic systems, this means curing the composite but not
Hydrodynamic curing the healing agent (during manufacture). For vascular
shielding of crack tips (and hollow-fiber) systems, the healing agent can be added
Intrinsic Extensive reusability Very small healing volumes after cure; however, in doing so the overall manufacture may
Comparatively simple Requires intimate contact of be greatly complicated. Consequently, it is necessary to assess
(conventional) crack faces the viability of doing this for the specific application in mind.
manufacturing If these systems are to progress, considerable optimization of
Nonintrusive: does Can severely degrade baseline their manufacture is required. Ultimately, as out-of-autoclave
not deform fiber mechanical performance processes mature (e.g., resin transfer molding, compres-
architecture (compared to traditional sion molding, and automated tape lay-up) researchers should
matrices) consider attempting to incorporate their systems with these
methods (which are all readily automated).
One of the most important conclusions from this article
implications of self-healing in composites rather than devel- is that in the short-term, self-healing is unlikely to be viable
oping new systems which have not been optimized for such when incorporated throughout complete composite struc-
usage. tures. Although this may represent a desirable long-term goal,
A very few studies have addressed the performance of aged in reality self-healing is most likely to be financially viable
(healable) composites or even their fatigue response. In the few when used selectively in critical locations within critical parts:
consequently, researchers should consider targeted application. [8] B. Caglar, L. Orgéas, S. Rolland du Roscoat, E. M. Sozer,
At its simplest, this may take the form of identifying where self- V. Michaud, Composites, Part A 2017, 99, 1.
healing can offer greatest value. [9] K. B. Katnam, L. F. M. Da Silva, T. M. Young, Prog. Aerosp. Sci.
Few self-healing studies have discussed the subject of vali- 2013, 61, 26.
[10] M. Scheiner, T. J. Dickens, O. Okoli, Polymer 2016, 83, 260.
dation, but due to the nature of this technology, self-healing is
[11] N. Zhong, W. Post, Composites, Part A 2015, 69, 226.
unlikely to be employed in commercial materials without con- [12] K. M. Wiggins, J. N. Brantley, C. W. Bielawski, Chem. Soc. Rev.
sumer confidence and reliable validation methods. Therefore, it 2013, 42, 7130.
is critically important that this is considered when developing [13] A. J. Patel, N. R. Sottos, E. D. Wetzel, S. R. White, Composites, Part
healing systems, old and new alike. Combining self-healing A 2010, 41, 360.
with structural health monitoring may present one solution to [14] E. Manfredi, PhD Thesis, EPFL, Switzerland 2015.
this. [15] G. Williams, R. S. Trask, I. P. Bond, Composites, Part A 2007, 38,
Although self-healing technology has greatly advanced in 1525.
recent years, there is yet much research to come before self- [16] K. S. Toohey, N. R. Sottos, J. a Lewis, J. S. Moore, S. R. White, Nat.
healing composites will be found within the structures of cars, Mater. 2007, 6, 581.
[17] W. Post, A. Cohades, V. Michaud, S. van der Zwaag, S. J. Garcia,
aircraft, watercraft, or wind turbines. To take the next steps
Compos. Sci. Technol. 2017, 152, 85.
toward commercialization, the target applications of self-healing [18] F. Sordo, V. Michaud, Smart Mater. Struct. 2016, 25, 1.
composites as well as their means of manufacture, characteri- [19] A. Cohades, V. Michaud, Composites, Part A 2017, 99, 129.
zation and validation will all require careful consideration. [20] S. R. White, N. R. Sottos, P. H. Geubelle, J. S. Moore, M. R. Kessler,
S. R. Sriram, E. N. Brown, S. Viswanathan, Nature 2001,
409, 794.
[21] E. N. Brown, N. R. Sottos, S. R. White, Exp. Mech. 2002, 42,
Acknowledgements 372.
A.C. and C.B. contributed equally to this work. We would like to [22] J. D. Rule, N. R. Sottos, S. R. White, Polymer. 2007, 48, 3520.
acknowledge support from the European Union Seventh Framework [23] E. L. Kirkby, J. D. Rule, V. J. Michaud, N. R. Sottos, S. R. White,
Program (FP7/2007-2013) under grant agreement no. 290308 (Marie J. A. E. Månson, Adv. Funct. Mater. 2008, 18, 2253.
Curie ITN SHeMat “Self-Healing Materials: from Concepts to Market”). [24] E. L. Kirkby, V. J. Michaud, J. A. E. Månson, N. R. Sottos,
A. Cohades and V. Michaud acknowledge support from the Swiss S. R. White, Polymer. 2009, 50, 5533.
National Science Foundation (SNF 200020-150007-1/2, 200020_124899, [25] M. R. Kessler, N. R. Sottos, S. R. White, Composites, Part A 2003,
200020-105169 and 113343). C. Branfoot and I. P. Bond acknowledge 34, 743.
support from the UK Engineering and Physical Sciences Research [26] M. R. Kessler, S. R. White, Composites, Part A 2001, 32, 683.
Council under agreement no. EP/L016028/1. [27] T. Yin, M. Z. Rong, M. Q. Zhang, G. C. Yang, Compos. Sci. Technol.
2007, 67, 201.
[28] T. Yin, L. Zhou, M. Z. Rong, M. Q. Zhang, Smart Mater. Struct.
Conflict of Interest 2008, 17, 015019.
[29] T. Yin, M. Z. Rong, M. Q. Zhang, J. Q. Zhao, Smart Mater. Struct.
The authors declare no conflict of interest. 2009, 18, 074001.
[30] T. Yin, M. Z. Rong, J. Wu, H. Chen, M. Q. Zhang, Composites, Part
A 2008, 39, 1479.
[31] T. S. Coope, U. F. J. Mayer, D. F. Wass, R. S. Trask, I. P. Bond, Adv.
Keywords Funct. Mater. 2011, 21, 4624.
commercialization, damage, fiber-reinforced polymer composites, [32] D. A. McIlroy, B. J. Blaiszik, M. M. Caruso, S. R. White, J. S. Moore,
manufacturing, self-healing materials N. R. Sottos, Macromolecules 2010, 43, 1855.
[33] J. Li, A. D. Hughes, T. H. Kalantar, I. J. Drake, C. J. Tucker,
J. S. Moore, ACS Macro Lett. 2014, 3, 976.
Received: January 31, 2018 [34] H. Jin, C. L. Mangun, D. S. Stradley, J. S. Moore, N. R. Sottos,
Revised: March 1, 2018 S. R. White, Polymer. 2012, 53, 581.
Published online: [35] H. Zhang, J. Yang, J. Mater. Chem. A 2013, 1, 12715.
[36] H. Zhang, J. Yang, Smart Mater. Struct. 2014, 23, 065003.
[37] S. Neuser, P. W. Chen, A. R. Studart, V. Michaud, Adv. Eng. Mater.
2014, 16, 581.
[1] E. N. Brown, S. R. White, N. R. Sottos, Compos. Sci. Technol. 2005, [38] P. W. Chen, S. Neuser, V. Michaud, A. R. Studart, in Proc. Int.
65, 2474. Conf. on Self-Healing Materials, ICSHM2013 (Eds.: N. DeBelie,
[2] J. D. Rule, E. N. Brown, N. R. Sottos, S. R. White, J. S. Moore, Adv. S. van der Zwaag, E. Gruyaert, K. VanTittelboom, B. Debbaut),
Mater. 2005, 17, 205. Ghent, Belgium 2013.
[3] B. J. Blaiszik, S. L. B. Kramer, S. C. Olugebefola, J. S. Moore, [39] Y. C. Yuan, M. Z. Rong, M. Q. Zhang, J. Chen, G. C. Yang, X. M. Li,
N. R. Sottos, S. R. White, Annu. Rev. Mater. Res. 2010, 40, 179. Macromolecules 2008, 41, 5197.
[4] E. Brown, N. R. Sottos, S. R. White, Exp. Mech. 2002, 42, 372. [40] Y. C. Yuan, M. Z. Rong, M. Q. Zhang, Polymer. 2008, 49, 2531.
[5] G. Williams, R. Trask, I. Bond, Composites, Part A 2007, [41] M. Z. Rong, M. Q. Zhang, W. Zhang, Adv. Compos. Lett. 2007, 16,
38, 1525. 167.
[6] C. J. Norris, I. P. Bond, R. S. Trask, Composites, Part A 2013, [42] P. A. Bolimowski, D. F. Wass, I. P. Bond, Smart Mater. Struct.
44, 78. 2016, 25, 084009.
[7] C. J. Norris, G. J. Meadway, M. J. O’Sullivan, I. P. Bond, R. S. Trask, [43] E. Manfredi, A. Cohades, I. Richard, V. Michaud, Smart Mater.
Adv. Funct. Mater. 2011, 21, 3624. Struct. 2015, 24, 015019.
[44] M. M. Caruso, D. A. Delafuente, V. Ho, N. R. Sottos, J. S. Moore, [80] S. A. Hayes, W. Zhang, M. Branthwaite, F. R. Jones, J. R. Soc. Inter-
S. R. White, Macromolecules 2007, 40, 8830. face 2007, 4, 381.
[45] S. Neuser, V. Michaud, Polym. Chem. 2013, 4, 4993. [81] A. Cohades, E. Manfredi, J.-C. Plummer, V. Michaud, Eur. Polym. J.
[46] S. Neuser, V. Michaud, Exp. Mech. 2014, 54, 293. 2016, 81, 114.
[47] S. Neuser, V. Michaud, in Int. Conf. on Self-Healing Materials., [82] X. Luo, R. Ou, D. E. Eberly, A. Singhal, W. Viratyaporn, P. T. Mather,
(Eds.: N. DeBelie, S. van der Zwaag, E. Gruyaert, K. VanTittelboom, ACS Appl. Mater. Interfaces 2009, 1, 612.
B. Debbaut), Ghent, Belgium 2013, pp. 43–46. [83] S. Meure, D. Y. Wu, S. Furman, Acta Mater. 2009, 57, 4312.
[48] S. Neuser, V. Michaud, S. R. White, Polymer 2012, 53, 370. [84] S. Meure, R. J. Varley, D. Y. Wu, S. Mayo, K. Nairn, S. Furman, Eur.
[49] M. M. Caruso, Solvent-Based Self-Healing Polymeric Materials, Uni- Polym. J. 2012, 48, 524.
versity of Illinois, IL 2010. [85] K. Pingkarawat, T. Bhat, D. A. Craze, C. H. Wang, R. J. Varley,
[50] C. Dry, Compos. Struct. 1996, 35, 263. A. P. Mouritz, Polym. Chem. 2013, 4, 5007.
[51] M. Motuku, U. K. Vaidya, G. M. Janowski, Smart Mater. Struct. [86] K. Pingkarawat, C. Dell’Olio, R. J. Varley, A. P. Mouritz, Polymer
1999, 8, 623. 2016, 92, 153.
[52] S. M. Bleay, C. B. Loader, V. J. Hawyes, L. Humberstone, [87] S. Meure, D.-Y. Wu, S. Furman, Vib. Spectrosc. 2010, 52, 10.
P. T. Curtis, Composites, Part A 2001, 32, 1767. [88] C. H. Wang, K. Sidhu, T. Yang, J. Zhang, R. Shanks, Composites,
[53] J. W. C. Pang, I. P. Bond, Compos. Sci. Technol. 2005, 65, 1791. Part A 2012, 43, 512.
[54] J. W. C. Pang, I. P. Bond, Composites, Part A 2005, 36, 183. [89] A. Cohades, V. Michaud, Compos. Struct. 2017, 180, 439.
[55] R. S. Trask, I. P. Bond, Smart Mater. Struct. 2006, 15, 704. [90] M. O. W. Richardson, M. J. Wisheart, Composites, Part A 1996, 27,
[56] G. J. Williams, I. P. Bond, R. S. Trask, Composites, Part A 2009, 40, 1123.
1399. [91] J. Bayandor, R. S. Thomson, M. L. Scott, M. Q. Nguyen, D. J. Elder,
[57] A. Kousourakis, A. P. Mouritz, Smart Mater. Struct. 2010, 19, Int. J. Crashworthiness 2003, 8, 297.
85021. [92] G. A. Schoeppner, S. Abrate, Composites, Part A 2000, 31, 903.
[58] S. Kling, T. Czigány, Compos. Sci. Technol. 2014, 99, 82. [93] Composite Materials Handbook, Military Handbook MIL-HDBK-17,
[59] Y. Zhu, X. J. Ye, M. Z. Rong, M. Q. Zhang, Compos. Sci. Technol. Revision F, Department of Defense, USA 2002.
2016, 135, 146. [94] E. Manfredi, A. Cohades, I. Richard, V. Michaud, Smart Mater.
[60] J. F. Patrick, K. R. Hart, B. P. Krull, C. E. Diesendruck, J. S. Moore, Struct. 2015, 24, 1.
S. R. White, N. R. Sottos, Adv. Mater. 2014, 26, 4302. [95] E. Tsangouri, D. Aggelis, D. Van Hemelrijck, Prog. Polym. Sci.
[61] H. R. Williams, R. S. Trask, I. P. Bond, Compos. Sci. Technol. 2008, 2015, 49–50, 154.
68, 3171. [96] J. L. Moll, S. R. White, N. R. Sottos, J. Compos. Mater. 2010, 44,
[62] R. Luterbacher, R. S. Trask, I. P. Bond, Smart Mater. Struct. 2016, 2573.
25, 15003. [97] V. Michaud, in Wiley Encyclopedia of Composites (Eds.: L. Nicolais,
[63] R. S. Trask, I. P. Bond, J. R. Soc., Interface 2010, 7, 921. A. Borzacchiello), John Wiley & Sons, Hoboken, NJ 2012.
[64] K. Pingkarawat, C. H. Wang, R. J. Varley, A. P. Mouritz, Composites, [98] E. Manfredi, V. Michaud, Composites, Part A 2014, 66, 94.
Part A 2013, 50, 22. [99] C. Huang, R. S. Trask, I. P. Bond, J. R. Soc. Interface 2010, 7, 1229.
[65] G. C. Tesoro, V. Sastri, J. Appl. Polym. Sci. 1990, 39, 1425. [100] J. A. Balta, F. Bosia, V. Michaud, G. Dunkel, J. Botsis, J. A. Manson,
[66] V. Sastri, G. C. Tesoro, J. Appl. Polym. Sci. 1990, 39, 1439. Smart Mater. Struct. 2005, 14, 457.
[67] J. Canadell, H. Goossens, B. Klumperman, Macromolecules 2011, [101] R. de Oliveira, S. Lavanchy, R. Chatton, D. Costantini, V. Michaud,
44, 2536. R. Salathé, J. A. E. Manson, Composites, Part A 2008, 39, 1083.
[68] X. Chen, M. a Dam, K. Ono, A. Mal, H. Shen, S. R. Nutt, K. Sheran, [102] J. F. Patrick, N. R. Sottos, S. R. White, Polymer 2012, 53, 4231.
F. Wudl, Science 2002, 295, 1698. [103] H. Zhang, P. Wang, J. Yang, Compos. Sci. Technol. 2014, 94, 23.
[69] X. Chen, F. Wudl, A. K. Mal, H. Shen, S. R. Nutt, Macromolecules [104] N. Sakurayama, S. Minakuchi, N. Takeda, Compos. Struct. 2015,
2003, 36, 1802. 132, 833.
[70] Y.-L. Liu, T.-W. Chuo, Polym. Chem. 2013, 4, 2194. [105] S. Minakuchi, D. Sun, N. Takeda, Smart Mater. Struct. 2014, 23,
[71] S. D. Bergman, F. Wudl, J. Mater. Chem. 2008, 18, 41. 115014.
[72] J. S. Park, H. S. Kim, H. Thomas Hahn, Compos. Sci. Technol. 2009, [106] S. Psarras, S. T. Pinho, B. G. Falzon, Composites, Part B 2013, 45, 70.
69, 1082. [107] P. Minguet, T. K. O’Brien, in Composite Materials: Testing And
[73] Y. Heo, H. A. Sodano, Compos. Sci. Technol. 2015, 118, 244. Design, (Eds.: R. B. Deo, C. R. Saff) STP 1274, Vol. 12, ASTM
[74] F. Herbst, D. Döhler, P. Michael, W. H. Binder, Macromol. Rapid International, PA, USA 1996, p. 105.
Commun. 2013, 34, 203. [108] E. Greenhalgh, M. Huertas Garcia, Composites, Part A 2004, 35, 1447.
[75] P. Cordier, F. Tournilhac, C. Soulié-Ziakovic, L. Leibler, Nature [109] C. Meeks, E. Greenhalgh, B. G. Falzon, Composites, Part A 2005,
2008, 451, 977. 36, 934.
[76] D. Montarnal, O. I. S. Tournilhac, M. Hidalgo, L. Leibler, J. Polym. [110] R. Krueger, M. Cvitkovich, T. O’Brien, P. Minguet, J. Compos.
Sci., Part A: Polym. Chem. 2010, 48, 1133. Mater. 2000, 34, 1263.
[77] F. Sordo, S. J. Mougnier, N. Loureiro, F. Tournilhac, V. Michaud, [111] F. Hélénon, M. R. Wisnom, S. R. Hallett, R. S. Trask, Composites,
Macromolecules 2015, 48, 4394. Part A 2013, 54, 182.
[78] F. Sordo, L. Berret, B. Caglar, V. Michaud, presented at 21st Int. [112] D. D. R. Cartié, G. Dell’Anno, E. Poulin, I. K. Partridge, Eng. Fract.
Conf. on Compos. Materials, ICCM21, Xi’an, China, Vol. 2216-3, Mech. 2006, 73, 2532.
August 2017. [113] M. Genet, L. Marcin, E. Baranger, C. Cluzel, P. Ladevèze,
[79] S. A. Hayes, F. R. Jones, K. Marshiya, W. Zhang, Composites, Part A. Mouret, Composites, Part A 2012, 43, 294.
A 2007, 38, 1116. [114] E. Kirkby, PhD Thesis, EPFL, Switzerland 2009.

Cohades 2018

Uploaded by

Copyright:

Available Formats

Cohades 2018

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

Cohades 2018

Uploaded by

Copyright:

Available Formats

PROGRESS REPORT

Progress in Self-Healing Fiber-Reinforced Polymer

With the potential for small-scale damage

utilize pre-embedded healing agents (via microcapsules or vas-

Any viable self-healing approach must be able to cope with

loading that will typically be imparted to high-performance

2.1.2. Hollow-Fiber Systems

An alternative to the use of discrete microcapsules, and their

and their size/hollowness); healing agent viscosity (i.e., is

2.1.3. Vascular Systems

A third method of effecting self-healing using an extrinsic

compared to unmodified systems; however, toughness was

in each reported study, without any healing functionality). For

useful tool in many cases for measuring the interlaminar frac-

techniques and conditions) in order to facilitate comparison

You might also like