Coupled Optical and Electrochemical Probing of Silver Nanoparticle Destruction in a Reaction Layer

Abstract The oxidation of silver nanoparticles is induced to occur near to, but not at, an electrode surface. This reaction at a distance from the electrode is studied through the use of dark‐field microscopy, allowing individual nanoparticles and their reaction with the electrode product to be visualized. The oxidation product diffuses away from the electrode and oxidizes the nanoparticles in a reaction layer, resulting in their destruction. The kinetics of the silver nanoparticle solution‐phase reaction is shown to control the length scale over which the nanoparticles react. In general, the new methodology offers a route by which nanoparticle reactivity can be studied close to an electrode surface.


S1: TEM Images of AgNPs
The AgNPs were imaged using transmission electron microscopy (TEM) on a JEOL JEM-3000F instrument using an accelerating voltage of 300 kV. Sample preparation was carried out by drop-casting suspensions of nanoparticles onto carbon grids (Agar Scientific) and allowing these grids to dry. The images were extracted using ImageJ software. Figure S1. Representative TEM images of commercial spherical citrate-AgNPs of 50 nm diameter (NanoXact, 0.02 mg mL -1 silver, 2 mM sodium citrate) S-3

S2: UV/Vis, Spectroscopic Study into the Stability of AgNPs in KBr
Whilst AgNP suspensions exhibit a strong plasmonic peak at around 400 nm, the magnitude, shape and position of the peak are dependent on the size, geometry, and local environment of the AgNPs. 1 The magnitude for the maximum absorbance can be used to assess the stability of the suspensions over time. Figure S2 presents the UV-Vis spectra of 1.2 pM AgNPs in 20 mM KBr as a function of time, with the equivalent spectrum in deionized water overlaid. The magnitude of the peak absorbance of the AgNPs over a 90 minute period is inlaid. Over this 90 minute period, the peak absorbance in KBr drops only 3.4% from its initial value, with the magnitude of the peak absorbance (max = 425 ± 1 nm) remaining constant for the suspension in water. This demonstrates that over the time period of the cyclic voltammograms and chronoamperograms being run, the AgNPs particles are suitably stable under such conditions. Figure S2. UV-Vis spectra of 1.2 pM AgNPs in 20 mM KBr over 10 minute periods. The spectrum in deionised water is overlaid for comparison. The inlay depicts the peak absorbance at 425 nm for AgNPs in deionized water (black squares) and in 20 mM KBr (red squares) over a period of 90 minutes.

S3: Optoelectrochemical Cell Design
A schematic of the thin-layer cell manufactured for use in opto-electrochemical measurements is depicted in Figure S3. Carbon Fibre wire (7 m diameter) was attached to a conducting wire (0.2 mm diameter) of around 5 cm in length using the electrically conducting silver loaded epoxy adhesive (RS, USA) to facilitate electrochemical measurements. The wires were sealed to the surface with a layer of electrically insulating Kapton tape, with the walls of the cell derived from 6 layers of this tape. "Nail polish" consisting of nitrocellulose in organic solvents acts as a hydrophobic barrier to seal the Kapton tape and wire to the surface, and to prevent any leakage into the electrical connections.

S5: Multi-Scan Cyclic Voltammograms of AgNPs in KBr
Four consecutive cyclic voltammetric scans were run on a solution of 12 p M AgNPs in 20 mM KBr starting at a potential above the threshold at which oxidative spikes are observed to determine if the same reduction in spikes is observed with a lower first vertex potential.
Discarding the first scan due to the effect of the irreversible adsorption of nanoparticles at the electrode, the cumulative number of spikes observed as a function of time for the three remaining scans is presented in Figure S5. It is evident that the gradient of the line is independent of the direction of the scan, and the cumulative number of spikes agrees well with the theoretically predicted value assuming a steady state flux to the electrode.

S7: Video Evidence of Reaction Layer Formation
AgNPs_Dissolution.avi Double potential step chronoamperograms of suspensions of 1.2 pM AgNPs in 20 mM KBr were performed at a supported carbon fibre wire (7 m diameter) whilst the microscope concomitantly captured dark field images of the process at a rate of 10 fps. The potential was initially held at 1.3 V (vs. pseudo-Ag) for 10 s before being stepped to 0.0 V for 30 s. The video, available online, presents the first 10 seconds of this chronoamperogram at 10 fps over the area of the image analysed using particle tracking software. The video clearly demonstrates that the local exclusion of nanoparticles is not due to mass transport away from the electrode; the scattering intensity from individual nanoparticles can instead be seen to decrease with time. S-9

S8: Simulated Voltammetric Response for an Irreversible System
The commercial package DigiSim was used to simulate the voltammetric response for the oxidation of bromide to bromine, in order to determine the diffusion limited current for the system, assuming irreversibility. The parameters used were detailed as follows: 2-3 0 = 0.3 V, 0 = 1 × 10 −8 cm s −1 , = 50 mV s −1 , = 0.5 − = 1.87 × 10 −5 cm 2 s −1 , 2 = 1.18 × 10 −5 cm 2 s −1 A cylindrical geometry was used assuming a radius and length of 3 m and 1.5 mm respectively.
The resulting voltammogram is presented in Figure S8. As referenced in the main text, the supported cylinder used in the opto-chemical response requires a correction factor of around 0.5 to approximate the change in geometry. This results in a diffusion limited current of ca. 4.4 A, corresponding to a current density of 13.3 mA cm -2 , compared to 60 A cm -2 in the observed experiment. Figure S8. Simulated voltammetric response for the oxidation of bromide to bromine.

S9: Simulation of Double Potential Step Chronoamperometry
The silver nanoparticle oxidation at the carbon fibre electrode in the bromide solution is simulated for the double-potential chronoamperometry experiment. The carbon fibre attached at the substrate is approximated as a hemicylinder model and thus only the concentration variation in one dimension, the direction along the hemicylinder radius r, needs to be taken into consideration. In the double-potential chronoamperometry experiment, Bris oxidized to Br2 at the first potential step E1 and Br2 is reduced to Brat the second potential step E2. We assume DBr-= DBr2 = 1.0 x10 -9 m 2 s -1 , nAg is the average amount of silver atoms in one nanoparticle and equal to 2.2*10 6 . nAg/2 is the amount of Br2 molecules needed to oxidize one silver nanoparticle.
The boundary conditions and equations used are given as follows: The mass transport equation is numerically solved by the finite difference method. The simulation program is written in Matlab R2017a and run on an Intel(R) Xeon(R) 3.60G CPU. Figure S10a presents the experimentally determined concentration profile against three simulated profiles fitted with varying rate constants. The profile most representative of the experimental results (c), 30 mM -1 s -1 , is presented in the main text. This is derived from considering the half-wave (distance at C =0.5) profiles, presented in Figure S10b. The half-wave distances are initially comparable to the simulated results with a rate constant of 30 mM -1 s -1 but drop off much faster, tending to a constant value whilst the simulation predicts that the halfwave distance continues to increase.