Deciphering nanoconfinement effects on molecular orientation and reaction intermediate by single molecule imaging

Nanoconfinement could dramatically change molecular transport and reaction kinetics in heterogeneous catalysis. Here we specifically design a core-shell nanocatalyst with aligned linear nanopores for single-molecule studies of the nanoconfinement effects. The quantitative single-molecule measurements reveal unusual lower adsorption strength and higher catalytic activity on the confined metal reaction centres within the nanoporous structure. More surprisingly, the nanoconfinement effects on enhanced catalytic activity are larger for catalysts with longer and narrower nanopores. Experimental evidences, including molecular orientation, activation energy, and intermediate reactive species, have been gathered to provide a molecular level explanation on how the nanoconfinement effects enhance the catalyst activity, which is essential for the rational design of highly-efficient catalysts.

Preparation of 100 nm SiO2@5 nm Pt@mSiO2 nanocatalysts [3][4][5] Table 1 summarized the amount of C16TAB and TEOS to obtain nanocatalysts with various shell thickness. enlarged 100 nm SiO2@5 nm Pt@mSiO2 nanocatalysts were synthesised and modified by a twolayer method as reported in the literature 6 . In the typical synthesis of 100 nm SiO2@5 nm Pt@mSiO2 nanocatalysts, 10 mL n-hexane was added to the mixture after the step of the addition of NH3H2O to form a two-layer solution. The calculated amount of TEOS was then added slowly to the top n-hexane layer. The stirring speed was maintained at 170 rpm after the addition of n-hexane till the completion of the reaction. The reaction time was increased to 12 hrs instead of the aforementioned 6 hrs.

Optical setup for single molecule study on single nanocatalysts
Single-molecule and single-particle imaging experiments were carried out on a prism-based total internal reflection fluorescence (TIRF) microscope ( Supplementary Fig. 8a). An adjustable 100-mW 532-nm CW laser (Oxxius, Lannion, France) was focused on the interface between the aqueous sample and quartz slide by a focusing lens, generating a focal spot size of 100 µm × 80 µm. The linearly polarised laser beam was switched to a circularly polarised laser profile by inserting a quarter wave plate (WPMQ05M-532, Thorlabs, Newton, NJ) before the sample. The fluorescence signal was collected by a 60× water immersion objective (Olympus, N.A. = 1.2) and focused onto an Andor iXonEM + Ultra 888 camera (Belfast, Northern Ireland: 1024 ×1024 imaging array, 13 μm × 13 μm pixel size). A fluorescence filter set composed of a 532-nm nortch filter and a 607/70 bandpass filter (Semrock, Rochester, NY) was used to reject scattering signal from the background.
The incident angle of the laser beam at the interface was determined by the angle of the last mirror, which was controlled by a rotational mount (Thorlabs) and a linear translation stage (Thorlabs). The optimal illumination conditions were achieved when the laser spot overlapped perfectly with the view field of the objective by scanning the vertical position of the final mirror.
The imaging conditions were fully optimized to achieve maximum illumination depth while maintaining a high signal-to-noise ratio (SNR) for every sample.
To measure the catalytic activities of nanoporous catalysts, a fluorogenic oxidation reaction of non-fluorescent amplex red (10-acetyl-3,7-dihydroxyphenoxazine) to produce highly fluorescent resorufin (λex = 563 nm; λem = 587 nm, at pH 7.5) (Supplementary Fig. 8b) was used. Nanocatalysts were deposited on quartz slides using drop-casting method with low particle densities for single particle catalysis. The density of the nanocatalysts immobilized on the quartz slide was controlled assembled to a flow chamber of 40 mm × 6 mm × 0.12 mm with a #1.5 coverslip ( Supplementary   Fig. 8a). During the imaging experiments, a steady stream of mixture (0.02-10 μM amplex red, 20 mM H2O2, and 10 mM pH 7.5 phosphate buffer) was introduced into the flow chamber by using a syringe pump for continuously supplying a constant concentration of reactants over nanocatalysts.
The flow rate was set to 20 µL min -1 . Highly fluorescent resorufin product molecules were formed at one of many possible reactive Pt NPs on a single nanocatalyst and recorded at an imaging frame rate of 33 fps.

Single molecule fluorescence polarisation microscopy imaging
Linearly polarised laser sources were first used. The polarisation direction of light before illuminating the sample was modulated by a zero-order half-wave plate (Thorlabs). The imaging experiments were first conducted with the s-polarised light (parallel to the interface but perpendicular to the incident light forward direction) for over one hour to collect enough catalytic events on single nanocatalysts. The polarisation of light source was then transformed into the circular polarisation. Single molecule imaging experiments were conducted again on the same nanocatalysts for over an hour. Fluorescence signals from single resorufin molecules were imaged and their positions were located by fitting the intensity distributions. 2D maps of the distribution of resorufin molecules were constructed by their localised positions. The distribution patterns of Re molecules on single nanocatalysts were then analysed and used to explain the restricted motions of AR molecules inside nanopore.

Ensemble measurement of activation energy of catalytic reaction on nanocatalysts
To further study the nanoconfinement effect on the catalytic activity of the core-shell nanocatalysts, temperature control ensemble experiments were conducted using Jasco J-1500 spectrophotometre.
The mixture of amplex red, core-shell nanocatalysts and H2O2 in 10 mM PBS buffer were kept stirring at 800 rpm during the whole measurement. Once inserting the cuvette into the holder, the fluorescence emission spectra of resorufin were taken at 2-min intervals for 20 min at each temperature. Reaction rates were determined as the slope of the producing resorufin molecules. A calibration curve of a series of known concentrations of resorufin molecules was used to calculate the molar concentration production rate of resorufin. The production rate of resorufin equals the consumption rate of amplex red.
Reaction rates at five temperatures (0, 10, 20, 30, and 40 o C) were measured. The activation energy (Ea) of the chemical conversion of amplex red to resorufin on platinum nanoparticles can be determined as the slope of log k vs. 1/T based on the Arrhenius equation = exp (− T ) where k is the rate constant, T is the absolute temperature in kelvin, R (8.314 J mol -1 s -1 ) is universal gas constant, and A is a constant for each chemical reaction. According to the collision theory, A is the frequency of collisions in the correct orientation. The role of nanoporous shell thickness (0 -120 nm) and nanopore size (2.2 and 3.3 nm) in tuning the activation energies of the chemical conversion on platinum nanoparticles in nanopore were studied.

Supplementary Tables
Supplementary Table 1

Supplementary Note 1: single molecule imaging of catalytic reaction on single nanocatalysts
A prism-type total internal reflection fluorescence microscopy (TIRFM, Supplementary Fig. 8a) was built on a Nikon Ti-E inverted microscope for imaging generated resorufin molecules. The distributions of τoff ( Supplementary Fig. 10c), which stands for the interval time between two consecutive catalytic events ( Supplementary Fig. 10a, b), were calculated. Fitting the data points with exponential decay function 8 gave the mean delay time <τoff> that was used as the characteristic value of waiting time for catalytic reactions. The inverse of the mean delay time <τoff> -1 was used as the reaction rate υrec. This process was repeated for measuring the reaction rate over many single nanocatalysts and average reaction rates were determined under variable amplex red concentrations (Fig. 1b).

Supplementary Note 4: catalytic reaction kinetics of 20 nm shell core-shell nanocatalysts
The adsorption/desorption equilibrium constant KAR of 4.8 ± 0.9 µM -1 and rate constant keff of 0.072 ± 0.009 s -1 particle -1 were obtained for 20 nm shell thickness nanocatalysts from fitting the single particle single molecule kinetics data ( Supplementary Fig. 1b). The smaller KAR compared to that with no shell (KAR = 6.1 ± 0.9 µM -1 ) suggests the adsorption strength of AR is also reduced in 20 nm shell nanocatalyst due to the confinement effect. However, the confinement effect is weaker compared to that in thicker shells (Fig. 1c). On the other hand, the keff is only slightly larger for 20 nm shell nanocatalyst than that with no shell (keff = 0.068 ± 0.007s -1 particle -1 ). This is because of the defects and non-uniformity of the shell structure visible in TEM images ( Supplementary Fig. 11a), which leads to an intermediate confinement effects between nanocatalyst with no shell and those with thicker shells. ( Supplementary Fig. 13c), while the p-pol direction remains in the same plane of incident light, containing both the transverse and longitudinal components ( Supplementary Fig. 13e). The restricted molecular orientations of Re molecules in different sets of nanopores will be efficiently excited by the linearly polarised light (Supplementary Fig. 13b, d). When a circularly polarised (cpol) light is used for excitation, all Re molecules in nanopores with different absorption dipole moment directions should be excited equally in the evanescent field. In our molecular orientation measurement experiments (Fig. 2), both s-pol and c-pol incident light were used for the imaging study.

Supplementary Note 6: localise single-molecule catalytic events with nanometre precision
Supplementary Fig. 14a is an example of a fluorescence image of resorufin molecule in nanopore.
The positions of identified molecules were localised by using a similar approach published

Supplementary Note 7: molecular mechanism of the oxidation reaction of amplex red to resorufin
Catalytic reaction rates under low concentrations (0.6-12 pM) of nanocatalysts without nanoporous shell (100 nm SiO2@5 nm Pt) were measured using ensemble experiments ( Supplementary Fig.   15a). The measured catalytic reaction rates (i.e. the initial rate of resorufin formation) over nanocatalyst concentrations did not follow a linear or first-order relationship as would be expected if resorufin were formed directly from amplex red. Fitting the data points, a nonlinear relationship with a fractional order of 1.6 was determined thus the catalytic reaction is more close to a secondorder kinetic mechanism where indicating the likely reaction between two amplex red radical molecules. Moreover, we also normalized the reaction rates over the nanocatalyst concentration.
As shown in Supplementary Fig. 15b, the normalized reaction rates increase as the concentration of nanocatalyst gets higher as well, in agreement with previously reported results 11 . If the reaction mechanism follows a first-order reaction mechanism, the normalized reaction rates should have b a c 50 nm r localisation uncertainty (σ). The much larger r when comparing to σ suggests that resorufin is undergoing diffusive transport rather than permanently adsorbed at certain sites inside nanopore.
To quantitatively measure the molecular transport of resorufin inside nanopore, we fitted the distribution of molecular displacement with the probability density function 15 : where ‹r 2 › stands for the mean square displacement and σ is the localisation uncertainty. In heterogeneous environments, the molecular transport can be described by a multicomponent probability density function with distinct diffusion coefficients. To best interpret the MSD distributions, it requires to use radial probability density function with three components (denoted as c1, c2, c3, Supplementary Fig. 19b) for fitting the data points resulting three distinct characteristic diffusion coefficients (D1, D2, D3). The distinct diffusion coefficients indicate that the local environments inside nanopore are heterogeneous. Therefore, the transport of resorufin molecules in nanopores should be a combination of different motion modes, including surface adsorption and inner pore diffusion, instead of just a simple Brownian motion. The medium diffusion rate (c2), which contains significant contributions from both c1 (adsorption) and c3 (random movement), dominates the molecular transport.

Supplementary Note 10: the effects of nanopore morphology on mass transport
The molecular diffusion inside nanoconfined space is complicated and determined by many factors such as substrate-surface interactions, substrate-substrate interactions, and substrate-solvent interactions. 18 Moreover, the geometry of the local environment also plays an important role in controlling the molecular movement. Here, we had investigated the roles of nanopore length and diameter in changing the mass transport behaviors of resorufin molecules.
As shown in Supplementary Fig. 20a, the apparent diffusion coefficients under different porous shell thickness show small variations. This also holds true for the diffusion coefficients of the three sub-populations of molecular motions ( Supplementary Fig. 20b). The fractions of sub-populations under different porous shell thicknesses were also similar ( Supplementary Fig. 20c). However, molecular transport is sensitive to the diameter of the nanopore. The apparent diffusion coefficient a b c of 3.3 nm nanopore is larger than the case of 2.2 nm nanopore even though their pore lengths are both 120 nm. The increased apparent diffusion coefficients for the larger nanopore was mostly caused by the faster diffusion portion (c3, Supplementary Fig. 20b) even though its fraction was slightly decreased (c3, Supplementary Fig. 20c). The single molecular trajectory analysis provides accurate measurement of molecular diffusion in nanopores under reaction conditions, which then allows us to carry out further analysis to decouple the influences of molecular transport and reaction kinetics 19 .