Molecular recognition between functionalized gold nanoparticles and healable , supramolecular polymer blends – a route to property enhancement † ‡

A new, healable, supramolecular nanocomposite material has been developed and evaluated. The material comprises a blend of three components: a pyrene-functionalized polyamide, a polydiimide and pyrene-functionalized gold nanoparticles (P-AuNPs). The polymeric components interact by forming well-defined π–π stacked complexes between π-electron rich pyrenyl residues and π-electron deficient polydiimide residues. Solution studies in the mixed solvent chloroform–hexafluoroisopropanol (6 : 1, v/v) show that mixing the three components (each of which is soluble in isolation), results in the precipitation of a supramolecular, polymer nanocomposite network. The precipitate thus formed can be re-dissolved on heating, with the thermoreversible dissolution/precipitation procedure repeatable over at least 5 cycles. Robust, self-supporting composite films containing up to 15 wt% P-AuNPs could be cast from 2,2,2-trichloroethanol. Addition of as little as 1.25 wt% P-AuNPs resulted in significantly enhanced mechanical properties compared to the supramolecular blend without nanoparticles. The nanocomposites showed a linear increase in both tensile moduli and ultimate tensile strength with increasing P-AuNP content. All compositions up to 10 wt% P-AuNPs exhibited essentially quantitative healing efficiencies. Control experiments on an analogous nanocomposite material containing dodecylamine-functionalized AuNPs (5 wt%) exhibited a tensile modulus approximately half that of the corresponding nanocomposite that incorporated 5 wt% pyrene functionalized-AuNPs, clearly demonstrating the importance of the designed interactions between the gold filler and the supramolecular polymer matrix.


Introduction
][30][31][32][33][34][35] An expanding eld of healable materials research has targeted the use of reversible supramolecular interactions [36][37][38][39] such as hydrogen bonding, 13 metal-ligand interactions 23 and p-p stacking to produce healable materials. 32We have, for example, recently developed a supramolecular healable polymer blend based on electronically complementary interactions between a low molecular weight polyamide, end-capped with p-electron rich pyrene groups, and a linear polymer with multiple p-electron decient diimide residues. 31Small-molecule spectroscopic studies have shown that pairs of diimide residues are held in a well-dened chain-fold [40][41][42][43][44][45][46] around the pyrenyl residue during complexation (Fig. 1). 47ne of the major deciencies of many supramolecular materials is that, in order to achieve healing at accessible temperatures and within usable timeframes, the materials are frequently elastomeric in nature with glass transition temperatures (T g ) well below ambient. 7These properties preclude the use of most supramolecular materials as structural components in high-value engineering applications such as in the aerospace industry. 48,49nspired by the success of polymer-composite materials in replacing metals across a range of high value products, we have investigated the possibility of producing nanoparticle-reinforced, supramolecular, healable composites. 21,22In collaboration with MacKay, Rowan and co-workers, we have previously reported supramolecular nanocomposites containing cellulose nanocrystals showing tensile moduli up to 20 times greater than the supramolecular matrix polymer, whilst still retaining good healing characteristics. 50In an effort to generate healable materials with the functionality exhibited by metal nanoparticles, [51][52][53] we have also developed a supramolecular nanocomposite containing pyrene-functionalized gold nanoparticles (P-AuNPs) and a complementary chain-folding polydiimide.However, disappointingly this non-covalently cross-linked network proved to be brittle in nature and did not show thermal-healing characteristics. 54In a related study, Tee et al. have recently reported a healable material comprising nickel microparticles embedded within an hydrogen bonded polymer matrix.In this study, the metal ller facilitated electrical conduction, suggesting the possibility of a new route to providing the energy needed for healing damage to the bulk material. 55In addition, Guan et al. have reported recently the Fig. 1 Structures of the chain-folding polydiimide 1 and the pyrenyl end-capped polymer 2. A schematic of the chain-folding p-p stacking interaction 47 between the bisdiimide units in 1 and pyrenyl end-groups of 2 as part of a small section of the proposed 29 supramolecular polymer network.
Scheme 1 Schematic representation of the formation/dissociation of a supramolecular nanocomposite from chain-folding polymer 1, pyrene end-capped polymer 2 and the incorporation of a pyrene functionalized AuNP (P-AuNP) filler.The supramolecular nanocomposite network can be disrupted upon heating (ca.60 C) and interestingly this can be completely re-formed upon cooling to ambient temperature.
][58] In the present work we demonstrate that addition of the P-AuNP ller to a supramolecular polymer blend of the electronically complementary polymers 1 and 2 (Scheme 1) results in a doubling of the strength of these materials, whilst retaining attractive thermally healable characteristics.In comparison to our previous study on supramolecular gold nanocomposites, 54 the new material incorporates the pyrene end-capped polyamide 2 which, in the light of its low T g value (À7 C) 29 was predicted to afford malleable self-supporting lms with potential healable characteristics.We proposed that healing would be achieved through thermally activated disassociation of the p-p stacking interactions between polymers 1, 2 and pyrenyl ligands of the gold nano-particlesthe supramolecular interactions can then reform upon cooling, regenerating the strength of the material.

Instrumental analyses
Ultraviolet-visible spectra were acquired using a Varian Cary 300 spectrometer.For heating-cooling cycle studies, the polymer solution was heated from 20 to 60 C at $25 C min À1 , and then allowed to cool from 60 to 20 C over 24 hours.Differential scanning calorimetry (DSC) was conducted using a TA-Q2000 instrument.Samples were heated to 170 C to remove any residual solvents, cooled, and then re-scanned from À20 to 325 C. Energy-dispersive X-ray spectroscopy (EDX) analysis was carried out on a Cambridge SEM360 instrument operating at 20 kV and 10 À5 torr, with a working distance of 24 mm.The EDX images were analysed by ImageJ soware to count the number of gold rich areas.Variable temperature healing studies using light microscopy were carried with a LEICA DM2500M instrument at a heating rate of 5 C min À1 .
Tensile testing was carried out on TA.XT.Plus texture analyser (from Stable Micro Systems) using a tensile grips probe.Polymer nanocomposite lms were cut into strips with dimensions 30.0 Â 3.5 mm.Four repeat measurements per sample were carried out, at an extension rate of 0.1 mm s À1 .Tensile moduli were calculated as the stress-strain gradient between 0.5 and 1.0% strain (linear region).

Polymer nanocomposite lm preparation
Composite lms were cast from solutions in 2,2,2-trichloroethanol.The lms were dried by a stepwise process: rst at ambient atmosphere for 24 hours, then in a vacuum oven at 50 C for 24 hours, and nally maintained at 80 C under vacuum for a further 24 hours.

Healing studies
Polymer composite strips were bisected crosswise using a razor blade and then the edges were overlapped by $5 mm and gently pressed together.The samples were then placed onto a PTFE plate and transferred into an oven, pre-heated to 50 C, for 10 minutes.The samples were then removed from the oven and cooled to room temperature before being subjected to tensile testing.

Complexation studies
As shown in Scheme 1, the target material was designed to incorporate three components: the chain-folding polymer 1, the pyrenyl end-capped polyamide (2) and the P-AuNPs 3. Prior to the healing studies it was necessary to verify that mixing the presence of the nanoparticles did not adversely affect the interpolymer p-p stacking interactions that were previously studied in detail. 47This was achieved by investigating the behaviour of the three materials in solution as single components and as blends.
Fig. 2 shows ve vials, three of which contain solutions of either (A) polydiimide 1, (B) pyrenyl end-capped polymer 2, or (C) P-AuNPs 3. It can be seen that they all exhibit distinct colours, especially the P-AuNPs that appear dark red-brown as a consequence of their characteristic surface plasmon resonance absorption.As seen previously, 29 mixing polymers 1 and 2 results in the formation of supramolecular polymer network that precipitates from solution (vial D).The red colour of the network is as a consequence of a new absorption band at $520 nm which is characteristic of an aromatic, p-p stacked, charge-transfer complex. 47,59Crucially, mixing the solutions of polymers 1, 2 and P-AuNPs also results in a reddish brown precipitate and essentially colourless supernatant (vial E).This demonstrates that the P-AuNPs are being incorporated into the supramolecular cross-linked network and thus removed from solution.
To investigate the thermo-reversible nature of the supramolecular interactions within the novel nanocomposite, the three-component vial (E) was heated to $60 C, whereupon the heterogeneous system rapidly became a homogeneous solution.Over the following 24 h, the supramolecular nanocomposite reformed and re-precipitated, leaving again an almost colourless supernatant.The process could be followed by monitoring the intensity of the plasmon resonance absorption band of the P-AuNPs at 513 nm, which was observed to reduce in intensity by approximately 90% from its maximum value on cooling from 60 C to ambient temperature, indicating that a high proportion of the P-AuNPs had been incorporated in the precipitated nanocomposite.
The rate of precipitation of the polymer blend (1+2) and the nanocomposite (1+2+3) could be followed by measuring the change in the absorbance prole of the solution as a function of time.These data are plotted in Fig. 3 [see also ESI, Fig. S1, ‡ for raw data] which shows that the absorbance intensity of the solution at 513 nm drops to around 20% of its initial value aer 1600 minutes (ca.27 h).It is striking that precipitation of the polymer blend containing the P-AuNPs is much more rapid than that observed for the simple polymer blend (1+2).For example, aer 60 min, the absorbance of the sample containing P-AuNPs had fallen by over 60%, compared to a reduction of only 20% for the sample containing only polymers 1 and 2. This observation may be accounted for on the basis that the P-AuNPs provide multiple additional crosslinking points, resulting in more rapid formation of an insoluble material.
In order to produce a thermally healable supramolecular polymer system it is important that the non-covalent interactions are reversible over an appropriate temperature range.Heating the sample containing the precipitated polymer nanocomposite to 60 C resulted in an almost instantaneous generation of the reddish brown solution.This precipitation/ dissolution procedure could be maintained over at least 5 heating and cooling cycles (Fig. 4 and ESI Fig. S2 ‡ for raw data) demonstrating that the system can be switched repeatedly between the bound and 'free' states without loss of the supramolecular interactions.

Polymer nanocomposite lmscasting and properties
We have observed previously 54 that lms cast from a chainfolding polymer with a similar structure to 1 and P-AuNPs are extremely brittle, and could not be subjected to stress-strain testing to study healable characteristics.However, lms cast from (1+2) and varying quantities of P-AuNPs (up to 15 wt%) proved exible, free standing and tough, demonstrating the importance of each of the components in the polymer blend (Fig. 5A for 10 wt% and ESI Fig. S3 ‡ for 0 wt% nanocomposite lm).
The 100 mm thick lms were studied by energy-dispersive X-ray spectroscopy (EDX) analysis (see ESI Fig. S4a-c ‡ for all the nanocomposite lms) which allows the gold rich areas of the surface of the lm to be visualised.For example, the composite lm containing 1.25 wt% P-AuNPs showed that the lm contained a relatively homogeneous dispersion of gold (Fig. 5B, 1400 counts per mm 2 ), and that the casting procedure did not result in signicant clustering of P-AuNPs that 50 would lead to area of the lms possessing higher and lower densities of P-AuNPs and therefore inhomogeneous lm properties.For the samples containing the 10 and 15 wt% P-AuNPs the surface gold count observed during EDX analysis dropped from 2200 to 1000 counts per mm 2 suggesting that signicant clustering of P-AuNPs was occuring at these concentrations.Studying the thermal properties of the nanocomposite materials also provided insight into the dispersion of P-AuNPs.A DSC thermogram for the nanocomposite containing 1.25 wt% P-AuNPs is shown in Fig. 6, together with thermograms for the individual polymers 1, 2 and for the P-AuNPs themselves.It can be seen that the chain-folding polyimide 1 exhibits a broad melting endotherm at ca. 275 C, and that the pyrene endcapped polymer 2 has a T g close to ambient temperature.Analysis of the P-AuNPs gave a thermogram with a sharp exothermic peak at 235 C, which is generally attributed to a sintering process which occurs as a consequence of energy minimization through coalescence of the AuNPs. 54,60,61Above the sintering temperature, a broad exotherm was evident, consistent with desorption of the octanethiol ligands from the AuNPs surface.In contrast to the individual components, the nanocomposite blend exhibits no really signicant thermal transition over the temperature range studied (data for the composite containing 1.25 wt% P-AuNPs in Fig. 6 for other compositions see ESI Fig. S5 ‡).In particular, there is no evidence for sintering of the P-AuNPs in these studies, indicating that the gold nanoparticle are relatively site-isolated, and therefore not able to undergo coalescence, in agreement with the EDX map shown in Fig. 5B.
The dramatic change in thermal properties of the composite polymers when compared to the starting components clearly demonstrates that the supramolecular interactions observed in solution are maintained in the solid lms.
Nanocomposite lms were subjected to tensile testing to investigate the effect of added P-AuNPs on the stiffness and healing characteristics of the material.Fig. 7 shows stressstrain data for pristine samples of polymer nanocomposites containing up to 15 wt% P-AuNPs.Films were cast from 2,2,2trichloroethanol and were cut into strips approximately 30 Â 3.5 mm and 100 mm thick.It can be seen that both the tensile modulus (the initial slope) and the yield-stress (the maximum on the stress-strain curve) increase continuously with the proportion of P-AuNPs incorporated into the lms.The lms that contain between 5 and 15 wt% P-AuNPs all failed at a strain of approximately 18%.Furthermore, a control sample that included 5 wt% dodecylamine-stabilized AuNPs (i.e.AuNPs that did not contain the pyrenyl ligandsee ESI S6 ‡ for experimental/preparation and ESI Fig. S7 ‡ for comparative DSC thermograms between composites featuring the two types of AuNPs), 54 was clearly the weakest lm, demonstrating the need for the presence of pyrenyl residues on the AuNPs, to bind to the matrix and so impart enhanced mechanical properties to these nanocomposite materials.

Composite healing studies
In an initial study of the healing potential of these nanocomposite lms, a sample containing 1.25 wt% P-AuNPs was pierced with a razor blade and the damaged lm analysed by hot-stage microscopy (Fig. 8).At the beginning of the experiment, a hole is apparent as a white area caused by backlighting of the red lm.As the lm was heated from ambient to 75 C the hole diminished progressively in size, eventually vanishing altogether.
More quantitative healing data was acquired using established break/heal test protocols, 28,29 rather than by visual analysis.During these tests, stress-strain data was generated for samples that were rst cut into two sections, their damaged edges overlapped and then heated at 50 C for 10 min.The mechanical properties of the healed lms were then compared to those of the pristine samples. 28,29,32Tensile (Young's) moduli for each sample, plotted as a function of increasing P-AuNPs concentration for both pristine and healed samples, are shown in Fig. 9.All of the samples showed very good healing characteristics, except for the material containing the highest loading of P-AuNPs (15 wt%).This result may be attributed to clustering of the P-AuPNs, as observed during EDX analysis, inhibiting movement of the polymers during healing.
The strength of the composite materials increased linearly with the P-AuNP ller loading (r 2 ¼ 0.99).This result contrasts with our ndings for healable nanocomposites containing high aspect-ratio cellulose nanocrystal (CNC) llers. 50In these materials the strength of the sample increased with the log of the ller content, in accordance with percolation-theory models.In such materials the strength and stiffness of the composite is derived from stress-transfer across the  macroscopic test sample by a continuous mesh of cellulose nanocrystals.
Analysis of the tensile moduli for the healed samples showed an excellent correlation with the data for the pristine samples.Indeed, the mean healing efficiency (calculated as the ratio of the tensile moduli of the pristine and healed samples) determined to be ca.108%.An increase in modulus aer healing has been observed previously 50 and may be attributed to small changes in the dimensions of the samples or to the thermal treatment resulting in a new, mechanically stronger arrangement of the supramolecular network.
To further demonstrate healing in these samples, the ultimate tensile strength (UTS) of the pristine and healed samples were compared.These data are plotted in Fig. 10 and show a linear increase in UTS as a function of increasing P-AuNPs incorporation (r 2 ¼ 0.99), i.e. exactly the same trend as is observed for the tensile moduli.In all cases where healing could be measured, the UTS of the healed sample was within the error of the measurements for the UTS of the pristine sample with the same loading level of P-AuNPs.
As a nal demonstration of the importance of the new design element in this system, i.e. the use of pyrene-functionalized AuNPs that enhance the cross-linking density within the polymer blend, a control experiment was carried out whereby a composite contained 5 wt% dodecylaminefunctionalized AuNPs was subjected to the same break/heal tests as described above.It can be seen from Fig. 11 that, although healing was observed, the material containing AuNPs that lacked pyrenyl ligands had a tensile modulus approximately half that measured for the P-AuNPscomposite system.For comparison: this control sample exhibited essentially the same tensile modulus supramolecular polymer blend without P-AuNPs (0 wt% nanocomposite lm) (Fig. 11).These results demonstrate that the p-p-stacking interactions between polymers 1, 2 and P-AuNPs contribute signicantly to the mechanical properties of these composite materials.A new healable supramolecular nanocomposite polymer system has been designed and synthesised.The material contains a pyrenyl-endcapped polyamide which interacts through p-p stacking interactions with a chain-folding polydiimide.Introduction of pyrenyl functionalized AuNPs (P-AuNPs) into this polymer blend results in a healable polymer-composite system.Complexation studies in solution demonstrate that the introduction of P-AuNPs results in more rapid formation of an insoluble supramolecular network when compared to control samples that did not contain the P-AuNPs.Films of the nanocomposite are tough and exible, and contain a relatively homogeneous dispersion of P-AuNPs as shown by EDX mapping and DSC analysis.Films containing P-AuNPs are stronger and stiffer than those cast from the same polymers but without P-AuNPs, and also than lms containing AuNPs that lacked the pyrenyl motif.Healing studies using a classic break/heal test, followed by stress-strain analysis, showedremarkablythat materials containing up to 10 wt% P-AuNPs can even exhibit healing efficiencies of more than 100%.

Fig. 5 (
Fig.5(A) Supramolecular polymer nanocomposite film formed from (1+2) containing 10 wt% P-AuNPs; (B) elemental mapping of a polymer nanocomposite film containing 1.25 wt% P-AuNPs, using EDX analysis.The film was solution cast as described in the Experimental section.Gold dots show the nanoparticles dispersed throughout the polymer film.

Fig. 9
Fig. 9 Comparison of tensile moduli of pristine and healed polymer nanocomposites.The line of best fit (r 2 ¼ 0.99) includes only data for the pristine samples (error bars shown 1 standard deviation from a mean of 4 repeat measurements).

Fig. 10
Fig. 10 Plot of the ultimate tensile strength of pristine and healed polymer nanocomposites as a function of P-AuNPs loading.The line of best fit includes only data for the pristine samples (r 2 ¼ 0.99).Error bars shown one standard deviation from the mean of 4 repeat measurements.