Tracking a Photo-switchable Surface-localised Supramolecular Interaction via Refractive Index †

a As supramolecular chemistry evolves, from the design of interactions in the solution and the solid state to applications at surfaces, there is a need for the development of analytical techniques capable of directly interrogating surface-localised supramolecular interactions. We present a proof-of-concept integrated optical Bragg grating sensor, capable of evanescently detecting small changes in refractive index at infrared wavelengths within a microfluidic system. The high spectral fidelity of the Bragg gratings combined with precise thermal compensation enables direct monitoring of the surface throughout the experiment, enabling the sensor to probe changes in situ and in real-time during surface preparation and chemical modification, and then to follow the progress of a dynamic surface-localised supramolecular interaction. In this study the sensor is assessed through the investigation of a photo-switchable inclusion complex between an azobenzene-functionalised surface and cyclodextrin in aqueous solution. The ability to investigate supramolecular interactions directly in real-time upon a planar surface via refractive index offers a valuable new tool in the understanding of complex dynamic supramolecular systems.


Introduction
[10] Surface chemistry has been utilised throughout nanoscience for assembling layers of molecules onto a substrate; often through the use of a self-assembled monolayer (SAM). 11This approach is compatible with supramolecular systems, with the 'artificial molecular muscle' reported by Liu et al. representing a notable example capable of performing mechanical work. 12uch of supramolecular and surface-based chemistry have been compatibly combined, however the challenge lies in developing robust methods for real-time interrogation of dynamic self-assembly at the surface, both non-destructively and in situ.
A wide body of analytical techniques has been developed to probe the liquid/solid interface at surfaces, 13 a subset of which have been applied to the study of supramolecular interactions at a surface.Indirect measurements, such as monitoring changes in the concentration of a component in solution 14 or macroscopic changes in hydrophobicity of a surface 2 can reveal much about the underlying supramolecular chemistry; nevertheless there remains a need for direct, real-time measurement.While one option is fluorescence, 15 the requirement for a fluorescent centre brings some limitations in terms of the types of interactions that can be studied, in addition to that of photobleaching and the difficulty in distinguishing signals originating at the surface from the surrounding media.Surface-enhanced Raman-based techniques have also shown promise in the detection of supramolecular interactions, 16 however limitations in the use of structured substrates and the difficulty in resolving individual peaks within a complex host-guest system restricts their application.Thin film techniques such as ellipsometry and the quartz crystal microbalance (QCM), in addition to detecting the deposition of covalent monolayers, 17 have been applied to the detection of the layer-by-layer deposition of supramolecular polymer films 18 and

View Article Online
View Journal | View Issue nanoparticle assemblies 19 or for supramolecular-based detection via a surface thin-film. 20,21However little has been reported using these techniques for the study of molecular-scale, supramolecular monolayer surfaces.
][24][25][26] However, it remains highly challenging to separate a subtle change in surface-localised refractive index induced by a supramolecular interaction from fluctuations in refractive index originating from environmental phenomena (principally temperature or concentration).
Here we employ an integrated optical Bragg grating sensor, capable of evanescently detecting small changes in refractive index at infrared wavelengths, with precise localised thermal compensation.In order to critically assess the efficacy of refractive index measurement as a method to directly probe surface-localised supramolecular interactions, we chose to target the wellunderstood photo-switchable inclusion complex between aqueous a-cyclodextrin (a-CD) and a surface-immobilised azobenzene (Fig. 1).This photo-switchable host-guest interaction has been used to form a switchable rotaxane, 27,28 to reversibly immobilise nanotubes, 29 to fabricate hydrogels that undergo reversible gelation in solution, 30 and even to induce macro-scale hydrogel assembly. 31,32While such behaviour has previously been exploited in the development of photo-switchable molecular storage within mesoporous silica nanoparticles, 14 monitoring was achieved via the bulk release of fluorescently-tagged b-CD, rather than direct investigation of this photo-controlled interaction.Furthermore, detection of organic vapours has been reported using an optical Bragg grating refractometer, dip-coated with an allylated cyclodextrin thin-film. 33

Optical sensor design and interrogation
A Bragg grating is an optical component that acts as a wavelengthselective reflector, with the characteristic Bragg peak reflected at a wavelength dependent on both the effective refractive index (n eff ) of the waveguiding media and the intrinsic period of the grating (eqn (S1), ESI †).We have previously reported the fabrication of an integrated-optical Bragg grating device by direct UV-writing into the photosensitive glass core of a planar silica-on-silicon substrate. 34,35The central Bragg wavelength (l B ) is determined by applying a Gaussian-fit to the reflected spectrum of the grating and interrogated to sub-picometre resolution via optical fibre, with low insertion loss.This all-optical device is capable of precise refractive index measurement at infrared wavelengths (1535-1575 nm, Fig. 2).The combination of high spectral fidelity and precise temperature compensation enabled changes in the refractive index of the analyte (n anal ) to be detected to 1 part in 10 À6 . 36We have demonstrated that such an integrated optical refractometer can be incorporated within a microfluidic device compatible with a wide range of solvent systems, including strong acids and bases, 37 where it can monitor in real-time the formation of a self-assembled monolayer at the sensor surface. 38or this work, an integrated optical waveguide circuit containing a series of four Gaussian-apodised Bragg gratings was fabricated.An 18 mm deep channel was etched into the silica above three of the Bragg gratings, mechanically polished and sputtered with a thin-film of tantalum pentoxide (80 (AE4) nm) to increase sensitivity. 36The Bragg gratings within the etched region were used for sensing, while the fourth unetched grating acted as an independent reference for physical parameters, 36 principally temperature (Fig. S4, ESI †).By exposing the evanescent field of the waveguided optical mode in this way, changes in the refractive index of the surface-localised environment could be detected by the corresponding shift in the peak Bragg wavelength.Fig. 2c demonstrates this measured Bragg wavelength shift on addition of a series of bulk analytes and also highlights the enhanced sensitivity arising from the tantalum pentoxide over-layer.Temperature referencing was achieved in situ through differential analysis of the two orthogonal polarisations of the waveguide. 39This was necessary to fully decouple minute changes in the surface-localised refractive index from the thermo-optic properties (dn/dT) of the solvent.A full description Fig. 1 A schematic of the azobenzene-modified self-assembled monolayer (top) formed on the surface of the Bragg grating sensor (bottom).The formation of the inclusion complex between the azobenzene-modified surface and an aqueous solution of a-cyclodextrin can be dynamically triggered through photoisomerisation of the azobenzene moiety.
This journal is © The Royal Society of Chemistry 2016 of the fabrication and interrogation of the optical device can be found within the ESI.†

Results and discussion
Cyclodextrins are a family of macrocycles composed of glucose oligosaccharides, forming a truncated cone-like structure with hydrogen-bonding hydroxyl groups around the outer rim.They have been demonstrated to form host-guest complexes with a wide range of compounds, stabilising aromatic groups within their hydrophobic interior cavity. 44It has been reported that both the glucose-hexamer a-cyclodextrin (a-CD) and glucoseheptamer b-cyclodextrin (b-CD) will readily accommodate an azobenzene unit within this cavity. 45However, upon photoisomerisation of the more stable and rod-like trans-azobenzene to its bent cis-isomer the complex will This process is reversible and upon reverting to the trans-isomer the complex can reform. 45he azobenzene motif was immobilised at the sensor surface via covalent linkage to a self-assembled monolayer of (3-aminopropyl)triethoxysilane (3-APTES).This process was realised within a microfluidic flow cell, where reagents were flowed through the micro-channel in series for both cleaning and preparative steps, functionalising the surface in situ before recontamination could occur.Following our reported protocol, 38 the etched microchannel was enclosed within a stainless steel mount, sealed with a Kalrezt O-ring.This microfluidic flow cell was networked in series via 400 mm bore ethylene tetrafluoroethylene tubing with a manifold solenoid valve that allowed for the controlled flow from up to six reservoirs, pushed through the system by a solenoid diaphragm pump (Fig. S5, ESI †).The pump rate and choice of reservoir was automated with LabVIEW via a USB DAQ controller.The sensor surface was cleaned by an automated series of washes with water and acetone, followed by 5.0 M potassium hydroxide to restore the hydroxyl surface (80 min).3-APTES (10% v/v in distilled ethanol, 12 h) was flowed over this freshly prepared surface to deposit a self-assembled monolayer, affording an amine-terminated surface which was subsequently reacted with a flow of 4-phenylazophenoxyacetyl chloride, (4, 50 mg in 70 mL dimethylsulfoxide in the presence of o0.1 mL triethylamine, 12 h).After each subsequent functionalisation of the surface, the shift in Bragg wavelength was recorded in an ethanol reference solution.It was found that the Bragg wavelength (l B ) increased by 46 (AE4) pm during the attachment of 3-APTES, with a further increase of 761 (AE5) pm upon attachment of 4 to the surface.This corresponds to an increase in analyte refractive index (Dn anal ) of 5.1 (AE0.4)Â 10 À4 and 8.4 (AE0.1)Â 10 À3 respectively, giving a cumulative refractive index shift of B8.9 Â 10 À3 for the deposition of the azobenzene-terminated SAM.This observed change in refractive index is attributed to the displacement of ethanol at the surface with that of the higher refractive index monolayer. 38The attachment process was independently verified by the detection of an absorption at 348 nm wavelength from an analogously functionalised glass slide, characteristic of the azobenzene moiety (Fig. S1 and S6, ESI †).
The azobenzene-functionalised device was installed within a casing mounted with both ultraviolet (l max = 370 nm, Dl 1/2 = 12 nm) and blue (l max = 464 nm, Dl 1/2 = 22 nm) LED light sources (P d = 40 mW); whereby switching between these two wavelengths would trigger photoisomerisation of the azobenzene surface without interference from ambient light or heating, avoiding thermal relaxation of the azobenzene moiety.A 10 mM aqueous solution of a-CD was added to the etched well and the device was allowed to equilibrate for several minutes under blue illumination.Then, upon switching to UV exposure, an immediate decrease in refractive index was observed that slowed over three minutes to give a well resolved Bragg wavelength decrease of 14.2 (AE0.5)pm in water, equating to a decrease in total analyte refractive index (Dn anal ) of 2.8 (AE0.1)Â 10 À4 .This decrease in refractive index corresponds to the azobenzene switching to the cis-isomer, perturbing the dynamic equilibrium at the surface towards free a-CD (note that photoisomerisation within the complex itself is sterically inhibited 46 ).By reverting to blue illumination the refractive index shift was reversed, restoring the Bragg wavelength to its original value, but over a longer This journal is © The Royal Society of Chemistry 2016 timescale (B6 minutes).The longer timescale on return to the initial trans-isomer is both consistent with that seen in the previously reported polymer system and in solution, 47 indicating that photoisomerisation of the azobenzene is rate limiting.This cycle of assembly and disassembly was repeatable, as illustrated in Fig. 3.
While the magnitude of this refractive index change is smaller than that observed for the formation of the dense azobenzene monolayer, it is important to note that in the latter case a covalent bond was formed and the measurement made after replacing the reagent solution with a standardised ethanol reference.In contrast, the extent of a-CD liberated from the planar surface is dependent on both the sterically-limited density of binding sites and the photostationary state of the azobenzene-functionalised surface.Upon disassembly of the complex, released a-CD will be in diffusion-limited equilibrium with the surrounding solution and correspondingly will still contribute to the effective refractive index measured by the Bragg grating through a near-surface increase in concentration; however, as the evanescent component of the optical mode of the waveguide decays exponentially away from the surface such small displacements can have a disproportionate effect.
Azobenzene-functionalised polymer films are known to exhibit a measurable change in refractive index upon photoisomerisation, 47,48 attributed to a change in bulk matrix density upon switching between the rod-like trans-azobenzene and the bent cis-isomer. 49As such a control experiment in the absence of a-CD was performed; it was found that when limited to a single monolayer upon the Bragg grating surface, the effects of photoisomerisation of the azobenzene moiety were not detected in either water (n D = 1.33), or in ethanol (1.36) where the sensitivity of the refractometer is amplified by the higher refractive index of the solvent (Fig. 2c).We hypothesise that when limited to a discrete monolayer, the molecular reconfiguration upon isomerisation of azobenzene gives rise to a small change in film thickness, 50,51 but the refractive index contrast for the corresponding small displacement/ingress by solvent at the photo-stationary state is not significant enough to be observed by the Bragg grating refractometer under the experimental conditions employed.
The formation of the a-CD/azobenzene complex was confirmed in solution by 1 31 This broadening corresponds to the displacement of solvent from the macrocyclic cavity and the formation of an intramolecular hydrogen-bonded structure.Subsequent re-sharpening of these peaks was observed upon exposure to UV light at 365 nm wavelength (Fig. S3, ESI †).Complexation with a-CD was found to have little effect on the population of the photostationary state, with the cis : trans ratio calculated to be 63 : 37 in solution (Fig. S2, ESI †).Similarly, it was found that the presence of a-CD did not perturb the absorption spectrum of 3 nor alter the rate of photoisomerisation in solution compared to the free system, consistent with the displacement of high energy water from the cavity driving complex formation. 52ig. 4a illustrates the divergence of the Bragg wavelength, d(Dl B ), for the a-CD/trans-azobenzene complex and cis-azobenzene surfaces, upon increasing the a-CD concentration from 330 nM to 50.0 mM.As the concentration of a-CD increases, the proportion of the a-CD/azobenzene complex on the surface increases, further differentiating the refractive index of the two states.While no switching effect was observed when immersed in de-ionised water alone (Fig. S7, ESI †), for low concentrations of a-CD a trend towards a 4 pm Bragg wavelength differential was observed when plotted logarithmically, implying abstraction of cyclodextrin from solution until an equilibrium with the Fig. 3 The Bragg wavelength shift for the azobenzene-APTES modified sensor surface in the presence of a 10 mM a-cyclodextrin solution (black circles, guideline traces the five-point moving average).Reversible photo-switching was observed upon alternating exposure to light at 360-380 nm ('UV') and 440-480 nm wavelength ('blue').Thermal compensation was achieved through a first order approximation, derived from differential analysis of the two orthogonal polarizations of the waveguide.For comparison, the corresponding Bragg wavelength shift for the unetched reference Bragg grating is shown in red, where dl ref /dT = 10 pm per 1C.
inclusion complex was reached.The continued increase in d(Dl B ) observed at higher concentrations is attributed to the combined effect of dn/dc of the a-CD solution 53 and the nonlinear sensitivity over large changes in analyte refractive index (dl B /dn anal , plotted in Fig. 2c).As discussed further in the ESI, † the combined effect is that the sensitivity to changes in surfacelocalised refractive index increases with rising a-CD concentration (Fig. S8, ESI †).The measured d(Dl B ) was converted to the change in analyte refractive index, d(Dn anal ), using the calculated sensitivity at each measured concentration (Fig. 4b).This showed that the rate of increase of d(Dn anal ) is tending towards zero at the higher concentrations investigated.The gradient is estimated to be 9 Â 10 À7 RIU mM À1 between 40 to 50 mM, less than the uncertainty in the refractive index measurement.As such, it is expected that any further raise in concentration would not give rise to a measurable increase, indicating saturation of the surface binding sites.It should be noted that the need to correct for dn/dc can be avoided by switching to a reference solvent or buffer within the flow cell, as exemplified here during covalent monolayer deposition.However in the context of measuring binding at a supramolecular surface, this requires a sufficiently slow dissociation constant for disassembly to be trackable. 38issociation of the surface-bound complex is driven by a perturbation of the equilibrium from the trans-azobenzene guest to the less favoured cis-isomer within this dynamic system, as evidenced by the average dissociation half-life time measured to be independent of the a-CD concentration (t 1/2 = 32 (AE6) s under the illumination conditions of the experiment; Fig. S9, ESI †).Similarly, the association rate was found to be concentration independent at higher concentrations of a-CD with t 1/2 = 74 (AE6) s, indicating that photoisomerisation is slower than the rate of complex formation.This trend is consistent with previous polymer-bound 47 and surface-bound absorption studies (Fig. S6, ESI †).However, with a 330 nM a-CD solution the half-time for assembly doubled, suggesting that at this concentration and below the kinetics of host-guest complex formation dominate.
The larger b-CD has also been reported to form an inclusion complex with azobenzene. 45Despite issues of aqueous solubility above 15 mM, b-CD complex formation was similarly observed, however the magnitude of the Bragg wavelength shift upon dynamic formation of the complex was almost three times lower than for the equivalent a-CD solution; for comparison d(Dl B ) = 5 (AE1) pm for 10 mM b-CD.With the greater size mismatch between the azobenzene on the surface and b-CD, complex formation is less favourable reducing the relative concentration of the complexed azobenzene on the surface.The effect of this is similar to lowering the overall concentration in that it reduces the surface-localised refractive index difference between the assembled and disassembled surfaces.Indeed this trend is seen in the literature, with the solution-phase association constant, K a (a-CD) having been reported as 2000 M À1 with trans-azobenzene, but only 770 M À1 for b-CD. 32This observation also extends to the much larger g-CD, where photoisomerisation of the azobenzenefunctionalised surface resulted in a Bragg wavelength differential of only 6 (AE1) pm in the presence of a 25 mM g-CD, with more dilute solutions hard to resolve.

Conclusions
In summary, the dynamic assembly of the photo-switchable inclusion complex between an azobenzene-functionalised surface and a-CD has been successfully tracked in real-time via direct measurement of surface-localised refractive index.The refractive index differential between the two states was found to be dependent on both the concentration and size of the cyclodextrin, with a-CD being found to offer the largest response upon complexation.The magnitude of the surface-localised refractive index shift, when measured in situ and in dynamic equilibrium with the surrounding solution, was an order of magnitude smaller than for the deposition of the covalently-attached azobenzene monolayer.This measurement was made possible by employing a high fidelity Bragg grating sensor in conjunction with excellent physical referencing, in particular the unique application of optical polarisation to enable high resolution temperature referencing.
While the absolute sensitivity limit to changes in refractive index is comparable to the state-of-the-art for other types of integrated refractometers, such as SPR, the avoidance of bulk optics through the use of an optical fibre interconnect, the ability to self-reference in situ against physical effects and the use of an interrogation wavelength that does not interact with the chemistry under investigation are all significant advantages.Further, the ability to measure the absolute refractive index allows for continued measurement during significant changes in experimental conditions (e.g.changes in solvent) or between experiments not possible with 'fringe counting' based approaches used in interferometric systems.Future studies, when not ratelimited by photo-excitation of the surface-bound guest or the presence of a photostationary state, will look to employ the methodology used in SPR to quantify and compare the degree of binding and the kinetics of supramolecular chemistry on a surface with solution-based analogues.
The compatibility of integrated optical Bragg refractometers with microfluidics, combined with the ability to fabricate multiple distinct gratings within a single waveguide, offers the potential for simultaneous sensor measurements, self-referenced against both the chemical environment and physical phenomena.Applying this ability to directly probe surface-based supramolecular interactions in situ and in real-time, via near-field refractive index measurement offers a valuable new tool in the understanding of such complex dynamic supramolecular systems.

Fig. 2
Fig. 2 (a and b) The integrated optical device consisted of four spectrally distinct Bragg gratings; an un-etched physical reference (*, l B = 1538 nm) and three 'sensor' gratings (l B E 1546, 1556, 1566 nm) distributed along a linear waveguide.The reflected spectrum of the integrated optical device is plotted in air (n = 1.00, solid line) and in water (n = 1.33, dashed line) both (a) logarithmically and (b) as an extract of the data on a linear scale, showing the Gaussian profile of the Bragg peaks and low optical noise.The red line in (a) represents the background reflection spectrum from the erbium emission of the broadband source alone and it is this profile combined with reduced scattering loss that leads to the increase in the peak height in (b).(c) The sensitivity to bulk analyte refractive index (n anal ) as measured by the average Bragg wavelength shift of the 'sensor' gratings resulting from changes in the effective refractive index (Dn eff ) of the waveguide, for the transverse electric (Â, sensor) and the orthogonal, transverse magnetic ( ) optical modes.The dashed-lines correspond to the bare sensor, illustrating the enhancement from the deposition of an 80 nm thin film of tantalum pentoxide (solid line); loss of guidance of the optical mode at a lower analyte refractive index is the cost of this increased sensitivity.

Fig. 4
Fig. 4 (a) The average Bragg wavelength differential, d(Dl B ), between the a-CD/trans-azobenzene complex and cis-azobenzene surfaces, as a function of a-cyclodextrin concentration, plotted on a logarithmic x-axis.This average was calculated from the independent, simultaneous measurement of three Bragg gratings in thermal equilibrium over three light cycles, with the error bars representing the standard error.(b) The difference in analyte refractive index, d(Dn anal ), between the a-CD/transazobenzene complex and cis-azobenzene surfaces, after correction for the sensitivity change arising from dn/dc of aqueous a-CD, plotted on linear axes.Dashed lines are included as a guide for the eye only.