Article

pubs.acs.org/JACS

Polymer Composition and Substrate Influences on the Adhesive

Bonding of a Biomimetic, Cross-Linking Polymer
Cristina R. Matos-Pérez,† James D. White,† and Jonathan J. Wilker*,†,‡
†
Department of Chemistry, Purdue University, 560 Oval Drive, West Lafayette, Indiana 47907-2084, United States
‡
School of Materials Engineering, Purdue University, Neil Armstrong Hall of Engineering, 701 West Stadium Avenue, West Lafayette,
Indiana 47907-2045, United States

ABSTRACT: Hierarchical biological materials such as bone,

sea shells, and marine bioadhesives are providing inspiration
for the assembly of synthetic molecules into complex
structures. The adhesive system of marine mussels has been
the focus of much attention in recent years. Several catechol-
containing polymers are being developed to mimic the cross-
linking of proteins containing 3,4-dihydroxyphenylalanine
(DOPA) used by shellfish for sticking to rocks. Many of these biomimetic polymer systems have been shown to form
surface coatings or hydrogels; however, bulk adhesion is demonstrated less often. Developing adhesives requires addressing
design issues including finding a good balance between cohesive and adhesive bonding interactions. Despite the growing number
of mussel-mimicking polymers, there has been little effort to generate structure−property relations and gain insights on what
chemical traits give rise to the best glues. In this report, we examine the simplest of these biomimetic polymers, poly[(3,4-
dihydroxystyrene)-co-styrene]. Pendant catechol groups (i.e., 3,4-dihydroxystyrene) are distributed throughout a polystyrene
backbone. Several polymer derivatives were prepared, each with a different 3,4-dihyroxystyrene content. Bulk adhesion testing
showed where the optimal middle ground of cohesive and adhesive bonding resides. Adhesive performance was benchmarked
against commercial glues as well as the genuine material produced by live mussels. In the best case, bonding was similar to that
obtained with cyanoacrylate “Krazy Glue”. Performance was also examined using low- (e.g., plastics) and high-energy (e.g.,
metals, wood) surfaces. The adhesive bonding of poly[(3,4-dihydroxystyrene)-co-styrene] may be the strongest of reported
mussel protein mimics. These insights should help us to design future biomimetic systems, thereby bringing us closer to
development of bone cements, dental composites, and surgical glues.

■ INTRODUCTION
Adhesives play a prominent role in everyday life, being used in
to high-energy surfaces via metal chelation,14−18 individual
metal−ligand bonds,16,19 nonspecific adsorption,18 or hydro-
many industries including aerospace, automobile manufactur- gen-bonding.18,20 Oxidative21,22 and enzymatic21−23 cross-
ing, housing construction, wood products, packaging, and linking may also be involved.
labeling.1,2 Worldwide revenue generated by adhesives topped Incorporating DOPA and analogous reactive groups such as
$40 billion in 2010.3 New roles for specialty adhesives will be catechol (i.e., 1,2-dihydroxybenzene) into polymers is being
found once we can develop the materials in demand for pursued for a variety of applications. This field is expanding
applications such as surgical adhesives, orthopedic cements, and rapidly, especially in the past 5 years, with many laboratories
dental glues. Marine biology can provide inspiration for the contributing.24 Mussel mimetic polymers are being generated
design of such materials. The natural adhesive system of marine from polypeptides,25−27 polyamides,28 polyacrylates,17,29−35
mussels is receiving growing interest in the context of polyethylene glycols,36−52 polystyrenes,53−59 and polyur-
biomimetics. These shellfish affix themselves to wet rocks by ethanes.60 These polymers are enabling the development of
assembling a cross-linked matrix of proteins.4,5 Essential to the imaging agents,48 nanoparticle shells,44,48,61 elastomers,30,33,59
cross-linking chemistry of these proteins is the 3,4-dihydrox- resins,58,62 coacervates,31 hydrogels,36−38,42,43 surface treat-
yphenylalanine (DOPA) residue.4,5 Several proteins have been ments,27,40,49,52 antibacterial coverings,51,63 and antifouling
isolated from mussel adhesive plaques, each with DOPA coatings.34,35,45−47,50,51 A subset have shown the ability to
comprising between 3 and 30% of the total amino acid bond two substrates together.25,26,29−33,36−42,53,54,60
content.4,5 A mechanism we have proposed for the formation of Whereas a coating requires only adhesive bonding to the
mussel adhesive involves Fe3+ templating DOPA residues surface of interest, bulk glues also need the presence of cohesive
followed by redox chemistry to generate radicals.6−13 Reactivity forces. These cohesive interactions are required to form the
of these radicals may bring about protein−protein coupling for majority of the material and reach between substrates to yield a
cohesive bonding within the bulk material and protein−
substrate linkages for surface adhesive bonding.12,13 Alter- Received: April 8, 2012
natively, or perhaps complementary, is direct binding of DOPA Published: May 14, 2012

© 2012 American Chemical Society 9498 dx.doi.org/10.1021/ja303369p | J. Am. Chem. Soc. 2012, 134, 9498−9505
Journal of the American Chemical Society Article

functional glue. Too much cohesion, however, will result in a dihydroxystyrene)-co-styrene] found lap shear bulk adhesion at
hardened material without significant affinity for a surface. up to 1.2 ± 0.5 MPa.53 Over 1 MPa (∼145 pounds per square
Likewise, too much adhesive bonding will come at the expense inch (psi)) can be considered in the realm of high-strength
of cohesion, and the bulk material will not exist. This balance of bonding and, once achieved, will enable development of
cohesion and adhesion can be elusive, with no way to predict applications in several fields.1,2 Of course, even stronger
where an optimal interplay may reside. bonding is often desired.
Despite the growing number of synthetic systems mimicking Several factors influence the performance of an adhesive,
aspects of mussel adhesive proteins, there have been few including the substrate type, surface preparation (e.g., rough-
detailed and systematic studies to illustrate which aspects of the ness), cure conditions (e.g., temperature, time, humidity),
polymers give rise to the greatest bulk adhesion. In particular, solvent, concentration, and viscosity.2 Beyond such formulation
performance enhancements will arise from understanding how issues, an appealing chemical aspect to explore is that of
the polymer composition dictates function. In other words: polymer composition. By varying the ratio of 3,4-dihydrox-
How much pendant catechol should a polymer contain in order ystyrene:styrene within poly[(3,4-dihydroxystyrene)-co-styr-
to achieve the strongest bulk bonding? To answer this question, ene], we can gain access to a family of adhesive copolymers
we embarked upon a structure−property study in which the with varied degrees of cross-linking. This type of systematic
relative contributions of cohesion and adhesion could be study has not been carried out in detail with any other mussel
changed systematically by altering the polymer composition. mimetic polymer system. Bonding performance described
The resulting insights will show where one might find the below was examined on an array of low- to high-energy
highest-performing biomimetic material. surfaces: poly(tetrafluoroethylene) (PTFE, common name for
In an effort to gain straightforward chemical insights and also the DuPont product Teflon), poly(vinyl chloride) (PVC),
to keep future scale-up in mind, our mimics of mussel adhesive polished aluminum, sanded steel, and wood. Polymer
proteins are kept as simple as possible. The DOPA amino acid composition turns out to be a major factor dictating bonding
can be stripped down to only a catechol group pendant from a performance. This study presents the synthesis, character-
polymerizable olefin, hence the choice of 3,4-dihydroxystyrene ization, and bulk adhesion of several polymers. We are excited
(Figure 1). To minimize structural and thermal perturbations to report that the strongest bonding of these polymers displays
to the host polymer resulting from this monomer, polystyrene adhesion on par with that of commercial products such as
“Krazy Glue”, albeit with very different adhesion chemistry.

■
was chosen to represent a protein backbone (Figure 1). Styrene
is commercially available and easy to polymerize on large scales.
A further advantage for these studies is that polystyrene alone EXPERIMENTAL SECTION
does not exhibit any appreciable bonding capability.53 The Styrene and 3,4-dimethoxystyrene monomers were purchased and
target copolymer is thus poly[(3,4-dihydroxystyrene)-co-styr- purified with alumina columns for removal of polymerization
ene], shown in Figure 1. inhibitors. Details are provided in our earlier report.53 Solvents were
commercial anhydrous grade. A Varian Inova-300 MHz spectrometer
was used to collect NMR spectra. Gel permeation chromatography
(GPC) data were obtained using a Polymer Laboratories PL-GPC20
system and THF eluent. Polystyrene GPC standards (Varian, Inc.)
were used for instrument calibration. Differential scanning calorimetry
(DSC) data were obtained with a TA Instruments DSCQ2000
calorimeter.
Synthesis of Poly[(3,4-dimethoxystyrene)-co-styrene] Co-
polymers. In a typical polymerization, 2.86 mL (24.9 mmol) of
styrene and 3.70 mL (25.0 mmol) of 3,4-dimethoxystyrene were added
to a round-bottom flask with 30 mL of anhydrous toluene. The
reaction was cooled to −78 °C, and, after 10 min, 0.17 mL of n-
butyllithium was added dropwise. The solution turned orange, was
stirred under an argon atmosphere for 8 h at −78 °C, and then was
Figure 1. Mussel adhesive is comprised of DOPA-containing proteins. allowed to warm to room temperature over 12 h of reaction.
These proteins are mimicked with synthetic polymers by placing Polymerization was quenched by addition of ∼1 mL of methanol.
pendant catechol groups along a polymer chain. One of the simplest Further addition of ∼100 mL of cold (−20 °C) methanol precipitated
possible mimics is poly[(3,4-dihydroxystyrene)-co-styrene], in which the polymer. After isolation by filtering and drying under vacuum, at
polystyrene represents the protein backbone and DOPA is represented least three rounds of dissolution in chloroform (∼15 mL) and
by 3,4-dihydroxystyrene. precipitation with methanol (∼100 mL) were used to remove
unreacted monomers. Yield of poly[(3,4-dimethoxystyrene)33-co-
styrene67] was 4.4 g, 33 mmol, 66%. 1H NMR (CDCl3): δ 0.6−2.3
Copolymers were prepared by a two-step synthetic route ppm (broad, polymer backbone), 3.4−3.8 ppm (broad, methoxy
developed in our laboratory previously.53 We have also made peaks), 6.0−7.4 ppm (broad, aromatic).
cationic versions of these cross-linking polymers.54 Polymer- Synthesis of Poly[(3,4-dihydroxystyrene)-co-styrene]. Treat-
ization of styrene and 3,4-dimethoxystyrene yielded polymers ment with BBr3 and an acidic workup yielded the catechol-containing
for which the ratio of monomers in the final polymers was polymers according to our previous methods.53 A typical deprotection
generally a reflection of the starting feed.53 The styrene and 3,4- was accomplished by dissolving poly[(3,4-dimethoxystyrene)33%-co-
dihydroxystrene monomers distribute throughout the copoly- styrene67%] (4.4 g, 33 mmol) in 50.0 mL of anhydrous dichloro-
methane (DCM) under an argon atmosphere. The reaction was
mer statistically or randomly, thereby providing a suitable cooled to 0 °C, and, after 10 min, BBr3 (1.2 mL, 13 mmol) was added
model for how DOPA residues are located within mussel dropwise over 10 min. The solution was warmed to room temperature
adhesive proteins.53 The relatively simple synthesis allows and stirred overnight (∼12 h). The polymer was treated with 1% HCl
access to large quantities of polymer, up to ∼20 g per reaction followed by an aqueous workup to obtain poly[(3,4-dihydroxystyr-
in an academic laboratory. Our initial effort with poly[(3,4- ene)33%-co-styrene67%] (3.6 g, 27 mmol, 82%). Loss of the 1H NMR

9499 dx.doi.org/10.1021/ja303369p | J. Am. Chem. Soc. 2012, 134, 9498−9505

Journal of the American Chemical Society Article

Table 1. Characterization Data for Poly[(3,4-dimethoxystyrene)-co-styrene] Copolymers

feed (%) polymer observed (%)
3,4-dimethoxystyrene styrene 3,4-dimethoxystyrene styrene Mn Mw PDI Tg (°C)
0 100 0 100 32 300 38 400 1.2 106
5 95 5 95 37 500 48 800 1.3 103
9 91 10 90 39 800 50 000 1.2 100
15 85 15 85 40 700 48 700 1.2 93
22 78 19 81 40 900 54 500 1.3 97
50 50 26 74 49 600 65 800 1.3 67
50 50 33 67 57 500 84 200 1.5 62
51 49 42 58 50 575 61 700 1.2 60
53 47 36 64 32 700 43 800 1.3 68

methoxy peaks indicated complete deprotection. 1H NMR (CDCl3): δ dimethoxystyrene content of each final polymer was similar
0.6−2.3 ppm (broad, polymer backbone) and 6.0−7.4 ppm (broad, to that placed in the feed. Table 1 provides mole percent data
aromatic). for each monomer in the feed versus that found in the isolated
Adhesion Studies. Substrates for lap shear testing were prepared polymers. For targeting low catechol polymers (e.g., <15%), the
by cutting each material into rectangular pieces, 8.89 cm long × 1.25
cm wide. A centered hole of 0.64 cm diameter was drilled into each
3,4-dimethoxystyrene monomer content in the final polymer
adherend 2.22 cm from one end. Aluminum was 0.318 cm thick, type was a reflection of that in the starting feed. When targeting
6061 T6, and mirror polished with Mibro no. 3 and Mibro no. 5 polish higher catechol derivatives, the monomer ratio in feed needed
followed by washing with hexanes, ethanol, acetone, and then to include a little more 3,4-dimethoxystyrene than the catechol
deionized water, 30 min each, and air-dried overnight. The steel mole percent desired for the final polymer (Table 1).
adherends, 0.318 cm thick, were sanded with 50 grit sandpaper prior The last four entries of Table 1 serve to illustrate the
to testing and then washed with ethanol, acetone, and hexanes. PVC variability observed when repeating a given synthesis. With 50−
(0.318 cm thick) and PTFE (0.953 cm thick) were obtained from 53% of the 3,4-dimethoxystyrene monomer in the feed,
Ridout Plastics (San Diego, CA). polymers were obtained with between 26 and 42% incorpo-
Red oak was purchased at a local hardware store and, after cutting
to 1.27 cm thick, had a surface roughness approximately equivalent to
ration. This inconsistency may be related to water content of
that of 220 grit sandpaper. The wood adherends were cut and the liquid monomer. The 3,4-dimethoxystyrene was column-
adhesion strength was measured parallel to the wood grain, running purified prior to each polymerization reaction. We observed
along the 8.89 cm edge of the adherend. Water loss from these wood that higher incorporation often resulted when the time between
substrates may have occurred during the adhesive cure. Massing of purification and polymerization was minimized.
several oak adherends before versus after a typical cure treatment of 1 GPC and DSC were also carried out in order to further
h at room temperature, 22 h at 55 °C, and 1 h at room temperature characterize the isolated copolymers. Given the cross-linking
revealed an average 4.12% decrease (e.g., from 10.1 to 9.68 g). and adhesive nature of these polymers, GPC and DSC data
Lap shear adhesion measurements were conducted on an Instron were most easily obtained from the protected poly[(3,4-
5544 materials testing system equipped with a 2000 N load cell.
Copolymer solutions in 1:1 acetone/DCM (0.3 g/mL, 22.5 μL) were
dimethoxystyrene)-co-styrene)] species. This approach pre-
added to each adherend. Next, 15 μL of cross-linking solution (or vented both adhesion onto the high surface area GPC column
solvent when not adding the cross-linker) was added to deliver 0.33 and cross-linking during the high-temperature DSC experiment.
equiv of cross-linker per catechol group. The adherends were The GPC data, shown in Table 1, provided molecular weight
overlapped at 1.25 × 1.25 cm in a lap shear configuration (Figure distributions for the copolymers. Anionic polymerization
2). Each assembly was allowed to cure for 1 h at room temperature, 22 yielded consistent number-average molecular weights in the
h at 55 °C, and then 1 h cooling at room temperature. range of ∼32 000−58 000 for each derivative. Use of a 1:35
Figure 2 shows a representative extension versus force plot used for ratio of n-BuLi initiator:monomers for all polymerizations
quantifying adhesion. The early region of the trace is flat while the helped keep molecular weights similar. Polydispersity indices
crosshead moves up to begin loading the sample. Once the bond
begins to be stressed, a rise is seen until the sudden drop, indicating
(PDIs) all fell between 1.2 and 1.5. Anionic polymerization was
bond breakage. Adherends were pulled apart at a rate of 2 mm/min. used here to achieve low PDI values. Radical polymerizations
The maximum bonding force in Newtons was recorded. Final adhesive may also be suitable for the synthesis of poly[(3,4-
force in megapascals was obtained by dividing the maximum load at dihydroxystyrene)-co-styrene]. When preparing cationic deriv-
failure, in Newtons, by the measured area of adhesive overlap in square atives, nitroxide-mediated radical polymerization worked well.54
meters. For the polymer composition studies in Figure 3, each sample DSC indicated that the glass transition temperatures (Tg)
was tested a minimum of 20 times, averaged, and reported with error shifted lower with increasing 3,4-dimethoxystyrene in the
bars showing ±1 standard deviation. The comparisons to commercial polymer and less styrene. Table 1 shows Tg = 106 °C for a
adhesives in Tables 2 and 3 were each tested a minimum of 10 times, 100% polystyrene. Introduction of 3,4-dimethoxystyrene
averaged, and reported with error bars showing ±1 standard deviation.
Tensile adhesion tests were carried out in an analogous manner using
dropped the Tg values gradually toward 60 °C for 42% 3,4-
aluminum rods of 1 cm diameter. dimethoxystyrene/58% styrene copolymer. The methoxy

■
groups of 3,4-dimethoxystyrene may disrupt polymer order,
thus resulting in these decreased Tg values relative to 100%
RESULTS AND DISCUSSION polystyrene.64 Each poly[(3,4-dimethoxystyrene)-co-styrene]
Polymer Synthesis and Characterization. Nine poly- derivative showed a Tg below that of 100% polystyrene (Tg =
mers of varied composition were prepared in order to examine 106 °C) and above that of 100% poly(3,4-dimethoxystyrene),
the influence of catechol cross-linking chemistry upon adhesion found earlier to be Tg = 53 °C.53 For every polymer only one
strength. According to 1H NMR spectroscopy, the 3,4- transition temperature was observed. These single thermal
9500 dx.doi.org/10.1021/ja303369p | J. Am. Chem. Soc. 2012, 134, 9498−9505
Journal of the American Chemical Society Article

events indicate a statistical or random copolymer in which the than Fe3+ when cross-linking poly[(3,4-dihydroxystyrene)-co-
3,4-dimethoxystyrene monomers are distributed throughout styrene].53 Mussels do not have access to reagents such as
the host polystyrene chain.65 By contrast, multiple thermal (IO4)−. Nonetheless, the goal here is to achieve strong
events would have indicated segregated phases or blocks.65 adhesion, hence the choice of (IO4)− for cross-linking.
Thermogravimetric analysis (TGA, 5 °C/min) of a typical Periodate has been used to induce cross-linking in DOPA-
deprotected polymer showed 7.5% mass loss between ∼50 and containing peptides23 and synthetic polymers containing
∼200 °C. When the same polymer was examined by TGA after DOPA.25 The tetrabutylammonium salt of (IO4)− was used
being subjected to the conditions used for adhesion experi- here for solubility in the same organic solvents as the catechol-
ments (DCM/acetone solvent, 1 h room temperature, 22 h at containing copolymers. When the copolymers were cross-
55 °C, 1 h room temperature), 8.2% mass loss was noted. This linked, a 3:1 catechol:(IO4)− ratio was employed. This ratio
result indicates that a great deal of additional solvent does not preserves that of DOPA:Fe proposed to exist in Fe(DOPA)3
appear to persist within the polymer after curing. Generally c o m p l e x e s w i t h i n m u ss el a d h e s i v e p l a q u e s a n d
speaking, this synthetic approach provides control over the threads.10,12,13,68,69
polymer composition, molecular weight, and distribution of Bulk adhesion strength was measured for each of the
pendant catechol groups throughout the polymer chain. poly[(3,4-dihydroxystyrene)-co-styrene] variants. Figure 3 plots
Bulk Adhesion Strengths. Cross-linking can often adhesion as a function of the mole percent pendant catechol
enhance the adhesive bonding of polymers. Too much cross- monomer in each polymer. At 0% catechol, the 100%
linking, however, may be counterproductive and generate a polystyrene exhibited very little bonding at 0.6 ± 0.3 MPa.
hardened material without the ability to bond surfaces well. Every catechol-containing polymer showed more adhesion than
There is no easy way to predict where this “sweet spot” of 100% polystyrene, ranging from only slightly more with the 5%
optimal cross-linking resides. Consequently, we prepared the catechol polymer at 0.8 ± 0.3 MPa to ∼3 MPa for the 33%
family of copolymers shown in Table 1, each with varied catechol and higher copolymers. In general, increasing the
pendant catechol content to bring about different degrees of catechol content brought about increased adhesion up to the
cross-linking. point of ∼33%. Further addition of catechol into the polymers
General insights on the bulk adhesive bonding of these did not enhance adhesion. Perhaps catechol in the range of 33%
polymers were gained by measuring the lap shear adhesion of maximized adhesion without reaching into the range of too
mirror polished aluminum (Figure 2). Lap shear bonding is the much cross-linking being a detriment to function.
most widely used general method for quantifying adhesion.1,2
Although standard deviations may appear to be large, other
adhesion configurations such as tensile or peel tend to be
worse. Aluminum is a high-energy surface and a common
substrate for the aviation and automotive industries.2 We
mirror-polished the aluminum to make the adhesion more
challenging.

Figure 3. Adhesion strength as a function of the pendant catechol

content in a series of poly[(3,4-dihydroxystyrene)-co-styrene]
copolymers. Performance of the polymer alone (in red) and cross-
linked with periodate (in blue) are shown. Adhesive bonding is in a lap
shear configuration with aluminum substrates.
Figure 2. Lap shear bonding. Polymer glues are placed between two
substrates and pulled to failure. Maximum adhesion is indicated by the
Reactions of the copolymers with (IO4)− often brought
peak of the extension (in millimeters) versus force (in Newtons) plot.
about significant increases in adhesion. Data in Figure 3 show
that, up to 33% catechol, every polymer bonded more strongly
Adhesion of each polymer was examined both with and with (IO4)− relative to the analogous polymer alone. With 5%
without the addition of a cross-linking agent. Iron-induced,6−13 catechol, for example, adhesion of the polymer alone at 0.8 ±
simple oxidative,21,22 and enzymatic22,23 cross-linking have been 0.3 MPa jumped several-fold to 3 ± 1 MPa with (IO4)−.
proposed to be used by marine mussels when producing their Beyond 33% catechol, however, adhesion dropped in a
adhesive plaques. Our prior studies with DOPA-containing conspicuous fashion. Adhesion of (IO4)− with 100% poly-
proteins showed that adhesion could be increased with Fe3+.66 styrene was negligible, at 0.1 ± 0.1 MPa.
Addition of iron may have enhanced adhesion of poly[(3,4- These data indicate that the “sweet spot” for optimal
dihydroxystyrene)-co-styrene]53 and a DOPA-containing pro- adhesion resides at the cross-linked polymer of ∼33% catechol
tein67 somewhat, but the effects were minimal. Periodate, and ∼67% styrene. When the catechol content goes over ∼33%
(IO4)−, is a strong oxidant and did lead to stronger adhesion and (IO4)− is added, adhesion suffers. Most likely, too much
9501 dx.doi.org/10.1021/ja303369p | J. Am. Chem. Soc. 2012, 134, 9498−9505
Journal of the American Chemical Society Article

cross-linking is biasing the system toward extra cohesion within poly[(3,4-dihydroxystyrene)-co-styrene] bonds aluminum more
the bulk material at the expense of surface adhesive attachment. strongly than white glue and as strongly as a cyanoacrylate glue,
With less catechol than ∼33%, not enough cross-linking is although epoxy adhered the most. Interestingly, even though
available and bulk adhesion has not been maximized. poly[(3,4-dihydroxystyrene)33%-co-styrene67%] adheres compa-
In our first report of poly[(3,4-dihydroxystyrene)-co-styrene] rably to cyanoacrylate, the chemistry differs dramatically.
copolymers, the maximum adhesion determined from a 3.4% Cyanoacrylate adhesives are applied to surfaces in the form
catechol copolymer with (IO4)− was 0.9 MPa.53 Data presented of a flowing monomer followed by polymerization. By contrast,
here show that significant improvements can be made to the biomimetic adhesive is deposited onto the substrate already
adhesive performance from a systematic study of polymer polymerized and is then cross-linked. Whereas cyanoacrylate
composition. Relative to 100% polystyrene at 0.6 ± 0.3 or 0.1 ± does not have any specific chemistry for binding surfaces, the
0.1 MPa with (IO4)−, incorporation of mussel mimetic catechol groups of poly[(3,4-dihydroxystyrene)-co-styrene] can
chemistry brought about adhesion to this polymer that, bring about both this needed surface adhesion and cross-linking
otherwise, does not display any strong bonding properties. within the bulk.
Comparison of Polymer Catechol Content to DOPA in Adhesion on Different Substrates. Of the myriad
Mussel Adhesive Proteins. Several DOPA-containing mussel parameters influencing adhesion, the very substrate onto
foot proteins (Mfp’s) have been isolated from the adhesive which the material bonds may be one of the most significant.
plaques of this shellfish. The DOPA content of each protein Substrates can range from low-energy plastics to high-energy
can vary with several factors, the most prominent of which is metals. The surfaces can be smooth or rough. Generally
the time of year. That said, of all the amino acids in each speaking, roughened surfaces of high energy tend to be the
protein, DOPA comprises roughly 10−15% of Mfp-1,70 2−4% easiest for strong adhesion, allowing both chemical interactions
of Mfp-2,71 25% of Mfp-3,72 5% of Mfp-4,73 30% of Mfp-5,74 and mechanical interlocking between the glue and the surface.
and 3.5% of Mfp-6.74 Mussel adhesive plaques are constructed Smooth plastics are, classically, the most challenging substrates
from a hierarchy of proteins. Contacting the surface directly are for adhesion. Once the strongest adhering variant of poly[(3,4-
Mfp-3,72 Mfp-5,74 and Mfp-6.74 A protective coating over the dihydroxystyrene)-co-styrene] was identified for bonding
whole plaque is comprised of Mfp-1.75 The bulk adhesive polished aluminum (Figure 3), performance was assessed on
plaque, above the surface and below the coating, is Mfp-271 and other substrates. In addition to aluminum, PTFE, PVC, sanded
Mfp-4.73 Our most strongly adhering biomimetic copolymer steel, and red oak adherends were machined. These substrates
contains the equivalent of ∼33% DOPA. This value most provide a range of surface energies, roughness, and industrial
closely resembles Mfp-3 (∼25% DOPA) and Mfp-5 (∼30% applications. Pairs of each substrate were joined together using
DOPA). In mussel plaques, these two proteins may only be poly[(3,4-dihydroxystyrene)33%-co-styrene67%], both with and
needed to provide adhesive interactions with the surface. The without (IO4)− cross-linking, as well as with three commercial
other proteins are available for cohesion. Compared to DOPA glues.
proteins, poly[(3,4-dihydroxystyrene)-co-styrene] is a seem- Data provided in Table 3 show how each adhesive performed
ingly simpler system, but one that must bring about both on the different surfaces. After the experiments of Table 2 with
cohesion and adhesion. Prior to seeing the data in Figure 3, we aluminum, the substrate was changed to sanded steel. Here
could not have looked at the protein sequences and predicted poly[(3,4-dihydroxystyrene)33%-co-styrene67%] performed com-
the polymer composition giving rise to the strongest bonding. parably to white glue, but the cyanoacrylate and epoxy were
Comparisons to Commercial Glues. At 7 ± 1 MPa, the strongest. Interestingly, on sanded steel the biomimetic
maximum adhesion of cross-linked poly[(3,4-dihydroxystyr- copolymer displayed similar adhesion strength both alone and
ene)33%-co-styrene67%] on aluminum is quite appreciable. We when cross-linked with (IO4)−. Here we may be seeing an effect
sought to benchmark this performance against established from Fe3+ of the steel surface introducing cross-linking
commercial adhesives under identical conditions. The substrate, chemistry to the polymers and enhancing bulk bonding. For
quantity of glue, and cure conditions (e.g., time and PVC, performance of poly[(3,4-dihydroxystyrene)33%-co-styr-
temperature) were held constant. Three of the most common ene67%] with periodate was so strong that the substrate, itself,
classes of adhesives were chosen for comparison: a poly(vinyl failed prior to the adhesive joint. Likewise, the cyanoacrylate
acetate) white glue (PVA, Elmer’s Glue-All), an ethyl also broke PVC and outperformed both white glue and epoxy.
cyanoacrylate (Krazy Glue), and a two-part epoxy. Results are When periodate was left out of the formulation, poly[(3,4-
shown in Table 2. We are excited to report that cross-linked dihydroxystyrene)33%-co-styrene67%] joined PTFE as strongly as
epoxy and more strongly than white glue but not to the same
Table 2. Adhesive Performance of Poly[(3,4- degree as cyanoacrylate.
dihydroxystyrene)33%-co-styrene67%] Copolymer Compared Oak provided the strongest bonding for poly[(3,4-dihydrox-
to Commercial Gluesa ystyrene)33%-co-styrene67%] when cross-linked, at 10 ± 1 MPa.
Both white glue and epoxy were weaker, although the
adhesion strength cyanoacrylate appeared to be the strongest. The porous nature
adhesive (MPa) of wood may allow for penetration and mechanical interlocking,
poly(vinyl acetate) (PVA, white glue, Elmer’s Glue- 4±1 thereby explaining the high adhesion for poly[(3,4-dihydrox-
All) ystyrene)33%-co-styrene67%]. This result prompted us to measure
ethyl cyanoacrylate (Krazy Glue) 7 ± 1 the adhesion of a commercial wood glue (Titebond II, Franklin
epoxy (Loctite, Henkel Corp.) 11 ± 2 International) under identical conditions. Interestingly, poly-
biomimetic polymer alone 3 ± 1 [(3,4-dihydroxystyrene)33%-co-styrene67%] appears to have
biomimetic polymer with (IO4)− 7 ± 1 performed similarly to the purchased wood glue at 9 ± 2 MPa.
a These commercial adhesives have benefitted from decades of
Bonding was carried out on aluminum substrates in a lap shear
configuration. industrial formulation studies in which parameters such cure
9502 dx.doi.org/10.1021/ja303369p | J. Am. Chem. Soc. 2012, 134, 9498−9505
Journal of the American Chemical Society Article

Table 3. Lap Shear Adhesive Bonding, in MPa, for Poly[(3,4-dihydroxystyrene)33%-co-styrene67%] and Commercial Glues on
Different Substrates
PTFE PVC polished aluminum sanded steel red oak
biomimetic polymer alone 0.7 ± 0.2 0.9 ± 0.1 4 ± 1 6 ± 2 5.1 ± 0.9
biomimetic polymer and (IO4)− 0.4 ± 0.1 >5.7a 7 ± 1 5 ± 1 10 ± 1
poly(vinyl acetate) (Elmer’s Glue-All) 0.36 ± 0.04 0.5 ± 0.1 4 ± 1 5.5 ± 0.9 5±2
cyanoacrylate (Krazy Glue) 1.5 ± 0.3 >5.7a 7 ± 1 10 ± 2 >10b
epoxy (Loctite, Henkel Corp.) 0.7 ± 0.1 3.8 ± 0.7 11 ± 2 9 ± 1 4±2
a
Substrate failed while adhesive bond remained intact. bExceeded range of the instrument.

conditions, concentration, added filler, viscosity, and addition of poly[(3,4-dihydroxystyrene)33%-co-styrene67%] on aluminum

adhesion promoters have all been examined. By contrast, rods. Pairs of tensile substrates were bonded together using
poly[(3,4-dihydroxystyrene)33%-co-styrene67%] is a relative new- 13.5 mg of dissolved poly[(3,4-dihydroxystyrene)33%-co-styr-
born and, within the scope of this academic study, already ene67%] and (IO4)− over the 1 cm diameter overlap area.
performs comparably to commercial products. Ideally, an Testing revealed that the polymeric adhesive was so strong that
adhesive should be tailored for a target substrate. The not all of the joined substrates could be broken within the 2000
poly[(3,4-dihydroxystyrene)-co-styrene] with the strongest N capacity of our materials testing system. Some bonded
bulk adhesion on aluminum is not necessarily the best polymer substrates pairs did separate and provided a lower limit of ≥9
for other substrates. Beyond changing polymers for each MPa for poly[(3,4-dihydroxystyrene)33%-co-styrene67%] on
surface, a detailed series of formulation efforts may enhance aluminum in tensile mode. The biomimetic system appears to
performance even further. bond with significantly greater force than the natural material
Comparisons to Other Biomimetic Adhesive Poly- after which the polymer was designed. Although we may be
mers. We wished to place the performance of poly[(3,4- trying to make the strongest possible glue, mussels need only
dihydroxystyrene)33%-co-styrene67%] system within the expand- adhere as strongly as their environmental conditions dictate.
ing scope of other polymeric mussel protein mimics. Many of Indeed, if these shellfish were affixed to rocks any more
these new systems are being used most often to generate strongly, detachment forces exerted by waves or predators
coatings27,34,35,40,45−47,49−52,63 or hydrogels,36−38,42,43 among might pull on the byssus to the point of damaging the soft,
several other end goals, and some have shown adhe- internal tissues to which the threads connect (Figure 1).
sion.25,26,29−33,36−42,53,54,60 Direct comparisons of adhesive
performance are difficult to make given how many variables
are present including test methods, substrate composition,
■ CONCLUSIONS
With their ability to remain affixed to rocks in the turbulent
surface preparations, solvents, viscosity, cure time, cure intertidal zone it is no wonder that mussels have inspired so
temperature, and the presence or absence of water, among much research. Here we have presented a structure−property
several other conditions. Some of the stronger mussel mimics study on the simplest mimic of mussel adhesive proteins.
reported are a polyurethane at 5.2 MPa,60 polypeptides Several copolymers were synthesized, characterized, and
bonding up to 4.7 MPa,25 and a poly(ethylene glycol)/ examined for adhesive properties. A systematic approach was
polyacrylate at 1.2 MPa.29 Fusion proteins have been expressed taken in order to determine the polymer composition giving
and modified to contain DOPA.76,77 These representations of rise to the greatest bulk adhesion. With the cross-linking and
mussel proteins can adhere up to 4 MPa.67,78 The data in Table surface bonding chemistries present in these copolymers, the
3 indicate that poly[(3,4-dihydroxystyrene)33%-co-styrene67%] is strongest adhesive is likely to provide a balance between
the strongest bonding synthetic mimic of mussel adhesive cohesive and adhesive bonding. Adhesion was quantified on
tested to date. Maximum adhesion was at 10 ± 1 MPa for the substrates ranging from low-energy, smooth plastics to high-
cross-linked polymer joining wood. Polished aluminum, sanded energy, roughened metal. Performance was benchmarked
steel, and PVC were adhered at greater than 5.7 MPa, also against common commercial glues as well as the native
stronger than that reported for other biomimetic adhesives. material produced by live mussels. Adhesive performance of the
Adhesion Strength of Synthetic Mimics Compared to biomimetic polymer was comparable, and in some cases better,
Plaques from Live Mussels. Recently we developed a than commercial products and the plaques of living shellfish.
method for quantifying adhesive performance of the glue Relative to other mussel mimetic polymers, poly[(3,4-
produced by live mussels.79 On aluminum these shellfish dihydroxystyrene)-co-styrene] appears to be the strongest
adhere at 0.3 ± 0.1 MPa.79 The byssal adhesive system of bulk adhesive. These comparisons, although interesting, are
mussels is comprised of plaques contacting the surface and difficult to make directly, given the broad variations in
threads connecting each plaque to the animal’s soft inner body conditions. Overall, these results help attest to the value of
(Figure 1). Tensile measurements were required to obtain using blueprints from biology when designing new materials.
accurate adhesion data for the byssus. We were curious to see Such a biomimetic approach may aid development of the
how the performance of our biomimetic polymers compared to adhesives needed for industrial or biomedical applications
the “real” material produced by mussels. including wood glues without toxic formaldehyde, surgical
The polymer lap shear data from above (cf. Table 3) cannot reattachment of soft tissues, and cements for connecting metal
implants to bone.

■
be compared directly to tensile measurements. In a lap shear
test, the substrates are overlapped and force is applied parallel
to the adhesive bond (Figure 2). Tensile testing is an end-to- AUTHOR INFORMATION
end butt joint, and the applied force is perpendicular to the Corresponding Author
glue. Consequently, we gathered tensile adhesive data for wilker@purdue.edu
9503 dx.doi.org/10.1021/ja303369p | J. Am. Chem. Soc. 2012, 134, 9498−9505
Journal of the American Chemical Society Article

Notes (32) Shao, H.; Stewart, R. J. Adv. Mater. 2010, 22, 729−733.
The authors declare no competing financial interest. (33) Chung, H. Y.; Glass, P.; Pothen, J. M.; Sitti, M.; Washburn, N.

■
R. Biomacromolecules 2011, 12, 342−347.
(34) Gao, C. L.; Li, G. Z.; Xue, H.; Yang, W.; Zhang, F. B.; Jiang, S.
ACKNOWLEDGMENTS Y. Biomaterials 2010, 31, 1486−1492.
This work was supported by the Office of Naval Research, the (35) Li, G. Z.; Xue, H.; Cheng, G.; Chen, S. F.; Zhang, F. B.; Jiang, S.
National Science Foundation, and a Ruth L. Kirschstein Y. J. Phys. Chem. B 2008, 112, 15269−15274.
National Research Service Award from the National Institute (36) Hu, B.; Messersmith, P. B. Orthod. Craniofacial Res. 2005, 8,
of Health to C.R.M-P. We thank Harold McCarron and Jeffrey 145−149.
Youngblood for insightful discussions, Kaumba Sakavuyi and (37) Lee, B. P.; Chao, C. Y.; Nunalee, F. N.; Motan, E.; Shull, K. R.;
Messersmith, P. B. Macromolecules 2006, 39, 1740−1748.
Matt Walters for obtaining DSC data, Allison Mattes for
(38) Burke, S. A.; Ritter-Jones, M.; Lee, B. P.; Messersmith, P. B.
assistance with GPC analysis, and Courtney Jenkins and Biomed. Mater. 2007, 2, 203−210.
Heather Meredith for help collecting the commercial adhesion (39) Brubaker, C. E.; Kissler, H.; Wang, L.-J.; Kaufman, D. B.;
data. Messersmith, P. B. Biomaterials 2010, 31, 420−427.

■ REFERENCES
(1) Pocius, A. V. Adhesion and Adhesives Technology. An Introduction,
(40) Murphy, J. L.; Vollenweider, L.; Xu, F. M.; Lee, B. P.
Biomacromolecules 2010, 11, 2976−2984.
(41) Brodie, M.; Vollenweider, L.; Murphy, J. L.; Xu, F. M.; Lyman,
A.; Lew, W. D.; Lee, B. P. Biomed. Mater. 2011, 6, 015014.
2nd ed.; Carl Hanser Verlag: Munich, 2002.
(2) Petrie, E. M. Handbook of Adhesives and Sealants; McGraw Hill: (42) Ryu, J. H.; Lee, Y.; Kong, W. H.; Kim, T. G.; Park, T. G.; Lee, H.
New York, 2007. Biomacromolecules 2011, 12, 2653−2659.
(3) Croson, M. E. Adhesives Sealants Ind. 2011, 18, 17−18. (43) Holten-Andersen, N.; Harrington, M. J.; Birkedal, H.; Lee, B. P.;
(4) Waite, J. H. Integr. Comp. Biol. 2002, 42, 1172−1180. Messersmith, P. B.; Lee, K. Y. C.; Waite, J. J. Proc. Natl. Acad. Sci.
(5) Sagert, J.; Sun, C.; Waite, J. H. In Biological Adhesives; Smith, A. U.S.A. 2011, 108, 2651−2655.
M., Callow, J. A., Eds.; Springer-Verlag: Berlin, 2006; pp 125−143. (44) Black, K. C. L.; Liu, Z. Q.; Messersmith, P. B. Chem. Mater.
(6) Hight, L. M.; Wilker, J. J. J. Mater. Sci. 2007, 42, 8934−8942. 2011, 23, 1130−1135.
(7) Loizou, E.; Weisser, J. T.; Dundigalla, A.; Porcar, L.; Schmidt, G.; (45) Finlay, A. S.; Dalsin, J.; Callow, M.; Callow, J. A.; Messersmith,
Wilker, J. J. Macromol. Biosci. 2006, 6, 711−718. P. B. Biofouling 2006, 22, 391−399.
(8) Monahan, J.; Wilker, J. J. Chem. Commun. 2003, 1672−1673. (46) Dalsin, J. L.; Lin, L. J.; Tosatti, S.; Voros, J.; Textor, M.;
(9) Monahan, J.; Wilker, J. J. Langmuir 2004, 20, 3724−3729. Messersmith, P. B. Langmuir 2005, 21, 640−646.
(10) Sever, M. J.; Weisser, J. T.; Monahan, J.; Srinivasan, S.; Wilker, J. (47) Lee, H.; Lee, K. D.; Pyo, K. B.; Park, S. Y.; Lee, H. Langmuir
J. Angew. Chem., Int. Ed. 2004, 43, 448−450. 2010, 26, 3790−3793.
(11) Weisser, J. T.; Nilges, M. J.; Sever, M. J.; Wilker, J. J. Inorg. (48) Bae, K. H.; Kim, Y. B.; Lee, Y.; Hwang, J.; Park, H.; Park, T. G.
Chem. 2006, 45, 7736−7747. Bioconjugate Chem. 2010, 21, 505−512.
(12) Wilker, J. J. Curr. Opin. Chem. Biol. 2010, 14, 276−283. (49) Chawla, K.; Lee, S.; Lee, B. P.; Dalsin, J. L.; Messersmith, P. B.;
(13) Wilker, J. J. Angew. Chem., Int. Ed. 2010, 49, 8076−8078. Spencer, N. D. J. Biomed. Mater. Res. A 2009, 90A, 742−749.
(14) Lee, H.; Scherer, N. F.; Messersmith, P. B. Proc. Natl. Acad. Sci. (50) Pechey, A.; Elwood, C. N.; Wignall, G. R.; Dalsin, J. L.; Lee, B.
U.S.A. 2006, 103, 12999−13003. P.; Vanjecek, M.; Welch, I.; Ko, R.; Razvi, H.; Cadieux, P. A. J. Urology
(15) Ooka, A. A.; Garrell, R. L. Biopolymers (Biospectroscopy) 2000, 2009, 182, 1628−1636.
57, 92−102. (51) Yuan, S.; Wan, D.; Liang, B.; Pehkonen, S. O.; Ting, Y. P.;
(16) Jankovic, I. A.; Saponjic, Z. V.; Comor, M. I.; Nedeljkovic, J. M. Neoh, K. G.; Kang, E. T. Langmuir 2011, 27, 2761−2774.
J. Phys. Chem. C 2009, 113, 12645−12652. (52) Malisova, B.; Tosatti, S.; Textor, M.; Gademann, K.; Zurcher, S.
(17) Wang, J. J.; Tahir, M. N.; Kappl, M.; Tremel, W.; Metz, N.; Barz, Langmuir 2010, 26, 4018−4026.
M.; Theato, P.; Butt, H. J. Adv. Mater. 2008, 20, 3872−3876. (53) Westwood, G.; Horton, T. N.; Wilker, J. J. Macromolecules 2007,
(18) Lana-Villarreal, T.; Rodes, A.; Pérez, J. M.; Gómez, R. J. Am. 40, 3960−3964.
Chem. Soc. 2005, 127, 12601−12611. (54) White, J. D.; Wilker, J. J. Macromolecules 2011, 44, 5085−5088.
(19) McBride, M. B.; Wesselink, L. G. Environ. Sci. Technol. 1988, 22, (55) Yang, Z.; Pelton, R. Macromol. Rapid Commun. 1998, 19, 241−
703−708. 246.
(20) Mian, S. A.; Saha, L. C.; Jang, J.; Wang, L.; Gao, X.; Nagase, S. J. (56) Daly, W. H.; Moulay, S. J. Polym. Sci. 1986, 74, 227−242.
Phys. Chem. C 2010, 114, 20793−20800. (57) Cristescu, R.; Mihailescu, I. N.; Stamatin, I.; Doraiswamy, A.;
(21) Fant, C.; Sott, K.; Elwing, H.; Hook, F. Biofouling 2000, 16, Narayan, R.; Westwood, G.; Wilker, J. J.; Stafslien, S.; Chisholm, B.;
119−132. Chrisey, D. B. Appl. Surf. Sci. 2009, 255, 5496−5498.
(22) Suci, P. A.; Geesey, G. G. J. Colloid Interface Sci. 2000, 230, (58) Bernard, J.; Branger, C.; Beurroies, I.; Denoyel, R.; Margaillan,
340−348. A. React. Funct. Polym. 2012, 72, 98−106.
(23) Burzio, L. A.; Waite, J. H. Biochemistry 2000, 39, 11147−11153. (59) Pan, X. D.; Qin, Z. Q.; Yan, Y. Y.; Sadhukhan, O. Polymer 2010,
(24) Moulay, S. C. R. Chim. 2009, 12, 577−601. 51, 3453−3461.
(25) Yu, M.; Deming, T. J. Macromolecules 1998, 31, 4739−4745. (60) Peiyu, S.; Liying, T.; Zhen, Z.; Xinling, W. Acta Polym. Sin. 2009,
(26) Wang, J.; Liu, C.; Lu, X.; Yin, M. Biomaterials 2007, 28, 3456− 803−808.
3468. (61) Adkins, C. T.; Dobish, J. N.; Brown, C. S.; Mayrsohn, B.;
(27) Saxer, S.; Portmann, C.; Tosatti, S.; Gademann, K.; Zurcher, S.; Hamilton, S. K.; Udoji, F.; Radford, K.; Yankeelov, T. E.; Gore, J. C.;
Textor, M. Macromolecules 2010, 43, 1050−1060. Harth, E. Polym. Chem. 2012, 3, 390−398.
(28) Li, L.; Li, Y.; Luo, X. F.; Deng, J. P.; Yang, W. T. React. Funct. (62) Kaneko, D.; Wang, S. Q.; Matsumoto, K.; Kinugawa, S.; Yasaki,
Polym. 2010, 70, 938−943. K.; Chi, D. H.; Kaneko, T. Polym. J. 2011, 43, 855−858.
(29) Sarbjit, K.; Weerasekare, G. M.; Stewart, R. J. ACS Appl. Mater. (63) Han, H.; Wu, J.; Avery, C. W.; Mizutani, M.; Jiang, X.;
Interfaces 2011, 3, 941−944. Kamigaito, M.; Chen, Z.; Xi, C.; Kuroda, K. Langmuir 2011, 27, 4010−
(30) Glass, P.; Chung, H. Y.; Washburn, N. R. Langmuir 2009, 25, 4019.
6607−6612. (64) Li, C.-H.; Chang, T.-C. Eur. Polym. J. 1991, 27, 35−39.
(31) Shao, H.; Bachus, K. N.; Stewart, R. J. Macromol. Biosci. 2009, 9, (65) Zhang, H.; Hong, K.; Mays, J. W. Macromolecules 2002, 25,
464−471. 5738−5741.

9504 dx.doi.org/10.1021/ja303369p | J. Am. Chem. Soc. 2012, 134, 9498−9505

Journal of the American Chemical Society Article

(66) Doraiswamy, A.; Dunaway, T. M.; Wilker, J. J.; Narayan, R. J. J.

Biomed. Mater. Res. B: Appl. Mater. 2009, 89B, 28−35.
(67) Cha, H. J.; Hwang, D. S.; Lim, S.; White, J. D.; Matos-Pérez, C.
R.; Wilker, J. J. Biofouling 2009, 25, 99−107.
(68) Zeng, H.; Hwang, D. S.; Israelachvili, J. N.; Waite, J. H. Proc.
Natl. Acad. Sci. U.S.A. 2010, 107, 12850−12853.
(69) Harrington, M. J.; Masic, A.; Holten-Andersen, N.; Waite, J. H.;
Fratzl, P. Science 2010, 328, 216−220.
(70) Sun, C.; Waite, J. H. J. Biol. Chem. 2005, 280, 39332−39336.
(71) Rzepecki, L. M.; Waite, J. H. Mol. Mar. Biol. Biotech. 1995, 4,
313−322.
(72) Papov, V. V.; Diamond, T. V.; Biemann, K.; Waite, J. H. J. Biol.
Chem. 1995, 270, 20183−20192.
(73) Vreeland, V.; Waite, J. H.; Epstein, L. J. Phycol. 1998, 34, 1−8.
(74) Zhao, H.; Waite, H. H. J. Biol. Chem. 2006, 281, 26150−26158.
(75) Rzepecki, L. M.; Hansen, K. M.; Waite, J. H. Biol. Bull. 1992,
183, 123−137.
(76) Hwang, D. S.; Gim, Y.; Cha, H. J. Biotechnol. Prog. 2005, 21,
965−970.
(77) Hwang, D. S.; Gim, Y.; Yoo, H. J.; Cha, H. J. Biomaterials 2007,
28, 3560−3568.
(78) Lim, S.; Choi, Y. S.; Kang, D. G.; Song, Y. H.; Cha, H. J.
Biomaterials 2010, 31, 3715−3722.
(79) Burkett, J. R.; Wojtas, J. L.; Cloud, J. L.; Wilker, J. J. J. Adhes.
2009, 85, 601−615.

9505 dx.doi.org/10.1021/ja303369p | J. Am. Chem. Soc. 2012, 134, 9498−9505