FRET‐Integrated Polymer Brushes for Spatially Resolved Sensing of Changes in Polymer Conformation

Abstract Polymer brush surfaces that alter their physical properties in response to chemical stimuli have the capacity to be used as new surface‐based sensing materials. For such surfaces, detecting the polymer conformation is key to their sensing capabilities. Herein, we report on FRET‐integrated ultrathin (<70 nm) polymer brush surfaces that exhibit stimuli‐dependent FRET with changing brush conformation. Poly(N‐isopropylacrylamide) polymers were chosen due their exceptional sensitivity to liquid mixture compositions and their ability to be assembled into well‐defined polymer brushes. The brush transitions were used to optically sense changes in liquid mixture compositions with high spatial resolution (tens of micrometers), where the FRET coupling allowed for noninvasive observation of brush transitions around complex interfaces with real‐time sensing of the liquid environment. Our methods have the potential to be leveraged towards greater surface‐based sensing capabilities at intricate interfaces.


Introduction
Polymer brush surfaces that have the capacity to switch their physical properties,i ncluding wettability, [1] adhesion, [2] lubrication, [3] and exposed surface groups, [4] hold great potential for use in the next generation of sensing materials. Polymer brushes consist of thin films of polymer chains covalently tethered to surfaces with as mall grafting distance compared to their respective radii of gyration in solution. [5] These surface layers exhibit functionality when stimuli cause the polymer chains to undergo ap hase transition. [6] The stimuli can come in many forms that are unique to each polymer,b ut can include temperature, [7] or isothermally by solvent, [8] pH, [9] and ionic strength, [10] which can be coupled together for multimodal sensing capabilities. [11] Thep hase transitions result in an ensemble change in the conformation of the polymer chains,from the extremes of collapsed to fully extended (i.e., swollen). These transitions are dramatic as the high grafting density and uniformity in chain height throughout the brush results in rapid and communal switching transitions. [12] Theresponse can be tailored by precise control over chemical and structural parameters,s uch as the brush thickness,d ensity and architecture. [13] Importantly,b yh arnessing this power of phase transitions it is possible to extrapolate signals corresponding to stimuli, thereby allowing for surface-based sensing of environmental conditions.F or planar polymer brush systems,most conventional analyses of polymer brush conformation rely on ex situ methods such as atomic force microscopy (AFM), spectroscopic ellipsometry, and/or quartz crystal microbalance. [14] These methods have restricted spatial resolution, where they are unable to spatially resolve and sense processes that occur around complex architectures (e.g., at surfaces that are "wet" by liquid-liquid interfaces) or to analyse conformational states in constrained geometries,s uch as in nanopores.T oo vercome these challenges in spatially sensing conformational changes, engineering new signal transduction mechanisms that function on polymer brush height, are being pursued for greater surface-based sensing capabilities to be achieved. [15] Towards this goal, recent works have focused on incorporating functional motifs spatially within polymer brushes that amplify and transduce signals from changing brush conformations.T his has included incorporating plasmonic nanoparticles into polymer brush structures that allow for UV-visible spectroscopy (UV/Vis) monitoring of plasmonic band shifts due to particle-particle proximity in the collapsing brush. [14,16] Alternatively,l uminescent nanocrystals [17] or molecular fluorophores [18] can be integrated into responsive polymer brush layers for selective fluorescence emission intensity that depends on polymer conformation. Tase tal., reported on polymer brush systems with end-tethered fluorophores,where the fluorophore exhibited quenching effects which were correlated to the polymer brush height. [19] This has separately been used for microscopically visualizing patterns. [20] These fluorescence-based approaches offer strong possibilities towards developing more complex polymer brush surfaces that can noninvasively reveal details on polymer conformation, and ultimately sensing of real-time changes in aqueous solution compositions.
Herein, we introduce Fçrster resonance energy transfer (FRET) chemistry into ultrathin (< 70 nm) poly(N-isopropyl acrylamide) (PNIPAM) polymer brush surfaces for use as in-liquid optical sensors of polymer conformation under different stimuli. The PNIPAM was chosen due to its exceptional sensitivity to various stimuli, including subtle changes in liquid mixture compositions (co-nonsolvencye ffects-polymer collapse at intermediate mixing ratios of two good solvents [21] ), and their ability to be assembled into well-defined polymer brushes.T he polymer chains were assembled into dense and homogenously smooth polymer brushes on planar quartz surfaces by ag rafting-to approach. Thes urfaces exhibited strong fluorescence,w ith ab rush thickness in water of approximately 65 nm. When incubated in mixtures that give rise to co-nonsolvency effects, FRET pairing occurred as the polymer brush collapsed, then FRET decreased as the surfaces were immersed in "good" solvent systems.T his process allowed for high-resolution monitoring of lateral changes in surface polymer brush conformation. Importantly,o ur approach is not limited to PNIPAM systems,w here we anticipate other functional polymers can be used in as imilar way,l eading to enhanced in-liquid sensing capabilities.

Results and Discussion
Towards integrating FRET chemistry within functional PNIPAM polymers,wesynthesised aFRET donor monomer composed of 4-(2-acryloyloxyethylamino)-7-nitro-2,1,3-benzoxadiazole (NBD) (Scheme 1), and aF RET acceptor monomer composed of rhodamine B( Rhod B). This combination of FRET probes have been used in previous studies of single-chain temperature and pH sensors. [22]1 HNMR confirmed the successful synthesis of the NBD monomer,NBD-AA, and the Rhod Bmonomer,Rhod B-HEMA ( Figure S1). Polymers were synthesised by reversible addition-fragmentation chain-transfer (RAFT) polymerisation from the chain transfer agent (CTA) 2-(dodecylthiocarbonothioylthio)-2methylpropionic acid (DDMAT), with azobisisobutyronitrile (AIBN) as the initiator. Theuse of DDMATwas purposefully chosen so as to yield acarboxylate group of the initiating end of the resulting polymers,w hich was needed for polymer brush assembly.T he donor monomer was integrated within the first block with NIPAM, yielding amacroscopic NIPAM/ NBD random copolymer-CTA( product 1). This purified polymer was subsequently subjected to as econd polymerization with Rhod B-HEMA and NIPAM, yielding the final diblock random copolymer (product 2). Thep roducts were yellow and orange in appearance ( Figure S2), respectively. UV/Visible spectroscopy (UV/Vis) of the product 1 showed clear absorption corresponding to NBD (ca. 460 nm), which for product 2 was complemented by as econdary Rhod B absorption (ca. 556 nm;F igure 1A). Gel permeation chromatography (GPC) revealed product 1 to have an umber average molecular weight, M N ,o fa bout 30 kDa, whereas product 2 was about 50 kDa ( Figure 1B). Thed ispersity, , was higher for product 2 ( = 1.55 compared to 1.13), but within an acceptable range,w hich was likely due to the Rhod B-HEMA monomer interfering with chain propagation. AU V/Vis standard assay was performed for each monomer and product polymer to determine the amount of each fluorophore incorporated into each polymer chain. It was found that, on average,e ach chain contained 0.712 monomers of NBD and 1.51 monomers of Rhod B.
Polymer brushes were assembled by ag rafting-to approach that exploits macromolecular anchoring polymers. [23] This process involves first modifying optical quartz surfaces with at hin layer (ca. 2nm) of poly(glycidyl methacrylate) (PGMA), where surface Si-OH groups are allowed to conjugate to af raction of the epoxide groups of PGMA ( Figure 1C). Subsequently,asolution of product 2 was allowed to conjugate to the anchoring PGMA through the carboxylate groups from the initiating DDMATC TA.T he resulting polymer brushes were approximately 15 nm in dry height ( Figure 1D)and exhibited very low surface roughness, with ar oot-mean-square (RMS) roughness of about 434 pm ( Figure 1E). Theg rafting density was determined from s = (h1N A )/Mn, [24] where hi st he dry brush thickness,a nd 1 the bulk density of the brush composition, which we take as 1.1 gcm À3 ,a nd N A is Avogadrosn umber. We found that s % 0.19 chains/nm 2 ,consistent with previously reported densely grafted PNIPAM brushes. [25] This emphasises the utility of our macromolecular anchoring approach of carboxylate endgroup PNIPAM polymers,from amelt, for assembling dense PNIPAM brush surfaces.I mportantly,t he surfaces exhibited fluorescence excitation spectra consistent with NBDse xcitation wavelength at 454 nm ( Figure 1F), where the emission showed clear Rhod Be mission in the dry state ( Figures 1G  and S3), indicating strong FRET pairing.T his was confirmed by confocal laser scanning microscopy (CLSM) measurements of the surfaces with both the acceptor channel (width between 560 nm and 700 nm, with l exc = 543 nm) and donor channel (width between 490 and 560 nm, with l exc = 458 nm), after asquare section was photobleached at l exc = 543 nm for 40 minutes (Figure 1Hand I, respectively). It was clearly seen that where the acceptor was bleached there was significant enhancement of the donor fluorescence ( Figure S4), validating the FRET pairing for the polymer brush surface.T he polymer brush surfaces have the FRET donor in closest proximity to the quartz, with the Rhod Be xtended outwards finishing with the CTAo nt he solvent side.T his architecture ensures changes in FRET pairing should be reflective of both the polymer brush height, and the individual mixing of chains amongst their neighbours (i.e., am ultidimensional probe of the polymer conformation).
These surfaces were investigated under co-nonsolvency conditions,which causes PNIPAM to collapse at intermediate mixing ratios of two "good" solvents. [25] Common co-nonsolvency systems for PNIPAM include mixtures of short-chain alcohols with water. This effect was pursued as it offers an intriguing possibility for chemosensing of small compositional changes in aqueous liquids that causes dramatic changes in PNIPAM conformation. ThePNIPAM therefore provides an excellent test system to probe and sense stimuli-induced changes in polymer conformation.
When the free polymer,p roduct 2,w as incubated in as eries of mixtures spanning pure water to pure ethanol, EtOH, it was found that differences could be seen visually of both the turbidity and colour of the solutions (Figure 2A), indicating phase separation of the polymer in co-nonsolvency conditions.T he pink colour of the solutions with approximately 20-50 %E tOH indicates that the Rhod Be xhibits greater fluorescence (i.e., FRET pairing during chain collapse) under co-nonsolvencyconditions.The behaviour of the polymer brush (product 3)i nc o-nonsolvencyc onditions was investigated by CLSM, where an acceptor photobleached region showed different contrast from extended brush conformations to collapsed for both the FRET donor channel and composite images ( Figure 2B and C, respectively). This is reflective of the polymer brush that is surrounding the photobleached region extending in conformation (i.e., less FRET = less contrast), and collapsing under co-nonsolvency (i.e., more FRET = greater contrast). This was clarified by line profile analysis across the photobleached region (Figure 2D), where ag reater intensity was observed for the collapsed region (FRET occurring, which removes donor intensity from the area surrounding the photobleached square). This intensity,w ith respect to the surrounding area, then decreased in the presence of "good" solvents (35 % EtOH > water > EtOH). [21] Thep hotobleaching therefore offers ap otential way of spatially visualising polymer brush conformation on photobleached patterns.T he fluorescence spectra of the polymer brush surfaces in different aqueous mixtures of water and methanol (MeOH), EtOH, and 1propanol (1-PrOH), were characterised separately by fluorescence spectroscopy ( Figure 2E), all with l exc = 454 nm. It was found that significant FRET pairing occurs in co-nonsolvency mixtures,w here the chains are expected to collapse. [26] Thep olymer brush FRET ( Figure 2E)q ualitatively matches the pairing seen for the free polymer systems (i.e., greater FRET under co-nonsolvencyc onditions;F igure S5), however the magnitude of the change in FRET was found to be lower for the brush in comparison to the free polymer.This difference in the magnitude of the FRET can be understood to result from the proximity of donor and acceptor pairs between neighbouring chains in the polymer brush, where the distance of separation between FRET pairs is not only af unction of the single chain, but also of the separation between all pairs between neighbouring chains that mix together.W hereas for single chains in ad ilute system all FRET occurrences should result from single-chain interactions,r ather than neighbouring interactions.
TheF RET pairing was analysed further with respect to the ratio of the FRET donor-to-acceptor peaks,t ot he polymer brush height as measured by ellipsometry.T his was important to directly gauge the effects of FRET directly against polymer conformation. Forthe FRET ratio,the donor was taken as the intensity at 521 nm, I 521 ,and the acceptor at 581 nm, I 581 .F or the case of single free polymer chains,when I 521 /I 581 @ 1, the chains are extended, whereas when I 521 /I 581 ! 1 the chains are collapsed. Fort he polymer brush, the magnitude of the differences in I 521 /I 581 are expected to be shifted due to ad egree of permanent FRET that occurs between chains,but the trend in the ratios should be similar. It was found that for the FRET,t ypical co-nonsolvency transitions were exhibited (i.e., FRET ratios consistent with collapsed polymers at intermediate volume fractions of alcohol in water [21] ). Particularly,t he onset of FRET was at greater volume fractions of alcohol for MeOH > EtOH > 1-PrOH (Figure 3), which matches known co-nonsolvency trends.C rucially,b yd irectly comparing to polymer brush height, there were clear similarities between the onset of polymer brush collapse for each system, thereby showing congruence between FRET and polymer brush height. However,t here were some notable differences regarding the reextension of the brushes.T ypically,i na lcohol-rich mixing conditions,the polymer brushes are slightly more extended in height in comparison to pure water. However,wefind that for the FRET pairing,t his is more significantly extended (distance between donor and acceptor monomers) for the same conditions.T his likely highlights the capability of the FRET pairing to provide further insight into the mixing of chains within the polymer brush (i.e., the extension and collapse that is not purely in the perpendicular direction to the surface), rather than only polymer brush height, as probed by ellipsometry.
Whilst the FRET is consistent with polymer brush height under co-nonsolvencyconditions,itisimportant to verify that solvatochromic effects (i.e., different donor fluorescence output dependent on solvent) in the different solvent systems are not hampering our interpretation of the FRET.T his was first investigated by performing time-dependent density functional theory (TD-DFT) calculations to determine the oscillator strengths of optimised NBD and Rhod Bd ye structures in pure water and methanol. It was found that both dyes have similar oscillator strengths in these solvents (Table S1), thereby suggesting that solvatochromic effects are not significant in our FRET analysis.
This was further validated by performing molecular dynamics simulations of model diblock polymer brushes that have pseudo donor and acceptor monomers in each block.
Thes imulated FRET was then estimated from the distance between donor and acceptor monomers with along-range cutoff,w hich would reflect the distance dependence of real FRET ( Figure 4A). Full details are provided in the Supporting Information. Thes imulated FRET can then be directly compared to the simulated polymer brush height, as well as to the experimental FRET and height, in order to examine the FRET relation. It was found that there is ac lear separation between the experimental FRET vs.h eight during the polymer brush collapse and re-entry transitions ( Figure 4B). Importantly,t his relation was strongly matched by the simulated system that does not have solvatochromic effects, thereby verifying our interpretation of the FRET results.  When the number of co-nonsolvent molecules, N co-nonsolvent = 0, the brush is swollen;N co-nonsolvent = 32 000, the brush is fully collapsed;N co-nonsolvent = 128000, the brush is reentering extended state;N co-nonsolvent = 288 000, the brush is re-swollen. The substrate is situated at z = 0, and af raction of chromophores show partial demixing at the substrate. All densitiesa re normalised so that R Pz ðÞ dz ¼ 1, and z is given in multiples of the bead-diameter (see the Supporting Information for further details. Inserts show snapshots of the corresponding trajectories,w here A-and B-type monomers (chromophores) are rendered red and blue, respectively, with their actual diameters,a nd other monomers are rendered as grey, and the cosolvent as green. B) Comparison of the experimental FRET success (here given as I 581 /I 521 )topolymer brush height throughout the collapse and re-entry transitions, along with data from the simulated systems. The height of the polymer brush, H, was normalised by its reference value without cosolvent, H 0 . Furthermore,t he information in Figure 4r eveals further details on the transitions of the polymer brush that are not readily accessible by AFM or ellipsometry methods,w hich highlights the capability of the FRET pairing to provide further insight into the mixing of chains within the polymer brush (i.e., the extension and collapse that is not only in the perpendicular direction to the surface). Interestingly,t he inclusion of the fluorophores within the polymer brush has not significantly varied the co-nonsolvencyt ransitions of PNIPAM that has been found for homopolymer systems, [21] though we do note as mall degree of demixing of the simulated fluorophores against the surface under co-nonsolvency conditions ( Figure 4A). This occurred due to the cononsolvent interacting more strongly with the polymer backbone than with the donor and acceptor monomers,leading to ap artial demixing of these monomers against the substrate when the concentration of co-nonsolvent molecules was high. However,b roadly speaking, our strategy retains the chemosensitivity of native PNIPAM whilst allowing for incorporation of complex optical functionalities.
Theuse of FRET with CLSM offers the unique capability to probe spatial changes in polymer brush conformation. This was investigated by CLSM imaging of the interface of awater droplet within hexane ( Figure 5A)across the FRET channels. It was observed that in the hexane phase the polymer brush exhibited agreater fluorescence intensity over both channels. This greater intensity in the hexane phase was attributed to aggregation-induced emission, where the polymer brush was significantly more collapsed (ca. 19 AE 3.4 nm) than under the co-nonsolvencyc onditions (Figure 3), however this does not affect the FRET ratio analysis (i.e., both channels exhibit increased intensity). We found that at the hexane-water interface,t he FRET donor had greater emission (FRET decoupling), allowing for spatial visualisation of aclear region of difference in the composite donor/acceptor image.This was investigated further by Lambda imaging of the interface, where the emission profile of each pixel was collected over the entire image.Wefound that aclear region of FRET donor intensity emerges at the interface between the emission of about 525-550 nm ( Figure 5B,C). Thet otal fluorescence profile over the whole image indicated that the FRET occurred mostly in the hexane phase (collapsed;brush height % 19 AE 3.4 nm), which decreased in the water phase (transition towards extended;b rush height % 61 AE 1.1 nm). However,a tt he interface itself the FRET was mostly decreased ( Figure 5D), where the intensity of NBD was greater, which indicates that the polymer brush is extended. This region spanned approximately 40 mma cross the interfacial region. Interestingly,t his may indicate the "pulling" of the polymer brush at the pinned interface towards the preferred phase (i.e., water), rather than simply collapsing. Our method therefore allows for more detailed spatial information on the polymer brush dynamics in such systems that is not directly accessible by other methods.
Lastly,our strategy for spatially extrapolating signals from polymer brush transitions is not limited to planar geometries. Through adjustment of the grafting-to procedure,weexpect it will be possible to spatially sense conformational changes around more complex macroscopic architectures,including in microfluidic devices [27] and on patterned and wrinkled surfaces. [28] We anticipate that greater sensitivity may be achieved by exploring different chain architectures for the distribution of donor and acceptor monomers,a nd also the inclusion of cascade FRET (an additional longer wavelength acceptor monomer). Furthermore,t he polymer backbone of our brushes is not limited to PNIPAM, where further functionalities can be explored. However,the PNIPAM may allow for further investigation of co-nonsolvencytransitions at complex aqueous interfaces,s uch as for studying Criegee intermediates [29] and ethanol release from yeast cells.

Conclusion
In conclusion, we have reported on FRET-integrated planar polymer brush surfaces that provide FRET output depending on the conformation of the polymers in liquid mixtures.The FRET output was used as abasis to extrapolate sensing information from the conformational transitions of the polymer brush, which for the studied PNIPAM has Figure 5. A) Bright-field and CLSM images of the interface between hexane and water,a long with B) lambda images of stepwise pixel emission in the range of 500-590 nm (increased emission at the interface indicated by arrows). The images represent intensities of emission per pixel at each wavelength indicated in the top-left corner from excitation at 458 nm (i.e.,not donor or acceptor channels). C) Line profile the counts in the 530 nm profile. D) FRET ratio of the donor channel to acceptor channel across the interfacial region. For (C) and (D) the regions are coloured schematically to guide the eye. Forall CLSM images the scale bar is 100 mm; all measurements were performed at 24 8 8C. as trong dependency on the nature of the liquid mixture, thereby providing optical sensing of the solvating liquid compositions.O ur FRET-integrated surfaces have the capacity to allow for greater in-liquid surface-based sensing capabilities to be realised, especially around complex interfaces.