Tuning the Three-Phase Microenvironment Geometry Promotes Phase Formation

■ INTRODUCTION

Within the past decade, several groups have reported on unusual chemical reactivity in microenvironments by studying submicroliter volumes with various analytical techniques. Marken and co-workers have taken advantage of this unusual reactivity to promote useful electro-organic multiphase reactions. 1 Cooks and co-workers have studied a variety of organic reactions in electrospray droplets and discovered that reaction rates are accelerated [2][3][4][5][6][7][8][9] and demonstrated the importance of analyte solvation at the air|water interface. 10 Zare and co-workers have shown spontaneous reduction occurs at the boundary of water microdroplets surrounded by air. 12,13 Pielak and co-workers have demonstrated protein stability is decreased in inverse micelles compared to studies in a bulk buffer. 11 In the experiments described above, one cannot rule out the role adsorption plays in small volumes. The smaller the droplet, the more the analyte has access to the surface. Griffiths and Wilson have independently demonstrated that chemical reactions can accelerate in micron-sized compartments due to interfacial adsorption. 14,15 A necessary complication in electrochemical measurements of reactions in small volumes is that electrochemistry requires electrodes (solid|liquid interfaces). Thus, a three-phase boundary comprised of the electrode, droplet, and continuous phase is formed. Using voltammetry, one can probe physicochemical properties and processes occurring at the three-phase boundary of single microliter droplets and arrays of microdroplets. [16][17][18][19] Scholz and coworkers have explored heterogeneous chemistry in immobilized droplets and derived robust thermodynamic and kinetic values. [20][21][22] Compton and coauthors have also thoroughly investigated such systems 18,[23][24][25] and used numerical simulations to interpret their observations. 26 In recent years, our group has observed chemistry that is localized to the threephase boundary of femtoliter-microliter droplets and shown the preferential electroreduction at the oil|water|conductor (three-phase) interface. [27][28][29][30] White and co-workers studied the transfer of ferrocene from a 1,2-dichloroethane (DCE) oil phase to an aqueous phase along an electrode traversing both phases. 31 They evidenced that ferrocenium electrogenerated at the DCE|electrode boundary near the three-phase boundary is able to cross the water|DCE boundary and be reduced back to ferrocene at the water|electrode interface. Based on numerical simulation, the authors demonstrate that diffusion of the species is rate determining compared to oxidation, reduction, and phase transfer kinetics. Interestingly, simulations reveal that the diffusion layers of the electrogenerated species directly in the vicinity of the three-phase boundary are drastically affected by the adjacent phase transfer. The authors performed simulations for different geometries and evidence that the curvature of the three-phase boundary should influence the magnitude of the peak current associated with the voltammetry of ferrocene in the water phase. Nonetheless, no physical insight is drawn from the simulations. A similar study by Scholz and co-workers uses a wire electrode spanning through a nitrobenzene droplet immersed in an aqueous solution, and reveals that precipitation of decamethylferrocenium nitrate salt can occur at the three-phase boundary. 32 The authors explore how the concentration and solubility of the counterion (ClO 4 -, NO 3 -, Cl -) in both phases affect the extent of the precipitation. The geometry of the interface is not considered in their work.

In our work, we use electrochemical reactions to both trigger large concentration gradients of species in one phase and detect the transfer of these species in the other phase. Cyclic voltammetry in conjunction with optical observation of the three-phase boundary is used to track the formation of a new phase at the three-phase boundary. In this manuscript, we show and explain based on numerical simulations how precipitation can be promoted simply by changing the microenvironment geometry at the three-phase boundary.

The Journal of Physical Chemistry C

phase occurs about 380 mV earlier than in the DCE phase. The shift comes from both the difference in formal potentials in the water and DCE phases as well as an eventual liquid| liquid junction potential. 33 Our observation is in agreement with the results of White and co-workers who also observed two pairs of peaks when using a wire electrode crossing a water phase and a DCE phase containing ferrocene. 31 We observe the appearance of the second pair of peak only after the first scan, as they do.

The shape of the CVs can be explained as follows. During the initial sweep, the concentration of FcMeOH in the water droplet is too low to produce a visible faradaic peak in the CV. However, the oxidation of the 4 mM of FcMeOH contained in the DCE continuous phase leads to a large anodic peak (centered at 0.6 V vs Ag/AgCl in Figure 1B) followed by a cathodic peak (centered at 0.47 V vs Ag/AgCl in Figure 1B). At about 0.1 V vs Ag/AgCl, a cathodic peak corresponding to the reduction of FcMeOH + in the water phase is observed. Only on the subsequent scans is an anodic peak is observed in water. We conclude that a measurable amount of FcMeOH + is transferred from the DCE to the water phase during the course of the voltammetric experiment. The transfer of the various freely diffusing species across the DCE|water interface is represented by the straight arrows in Figure 1A. The charge balance must be enforced by either the transfer of TBA + from water to DCE or ClO 4

-from DCE to water. The TBA + and ClO 4

-ions have similar standard Gibbs free energies of ion transfer (-226 and -178 mV, respectively), and both are expected to cross the interface. 33 The transfer of the counter charges is represented in Figure 1A with vertical arrows.

This mechanism was tested quantitatively by simulating the CVs. The kinetics of oxidation/reduction of FcMeOH/ FcMeOH + at the DCE|electrode and water|electrode interfaces, their diffusion in both phases, as well as their partitioning at the DCE|water interface were simulated by a finite element method. Kinetics of electron transfer are represented with a Butler-Volmer law, while diffusion obeys Fick's second law. The partitioning at the DCE|water interface is implemented with two first-order kinetics for each species. The first and second kinetics represent the transfer from water to DCE and DCE to water, respectively. The ratio of the two kinetic parameters is equal to the partition constant. For the sake of simplicity, the polarization of the liquid|liquid interface is not simulated, and electroneutrality is assumed at all times. Despite these assumptions, a large number of parameters are present in the simulation. We tried to fix as many parameters as possible. .89 × 10 -6 cm 2 •s -1 ) were experimentally measured for water and DCE. Their partition coefficients (P FcMeOH = 71 and P FcMeOH + = 0.5) were measured independently (see Figure S3 in Supporting Information) and thus are fixed parameters of the simulation. The standard rate constant for electron transfer at the water|electrode interface and symmetry coefficients are also fixed (k water °= 10 cm•s -1 and α = 0.5) based on literature values. 34 Measurements of CVs in DCE (see CV in Figure S4) with 1 mM FcMeOH and 0.1 M TBAClO 4 show a Nernstian behavior at 0.05 V/s with D FcMeOH = 8.89 × 10 -6 cm 2 •s -1 . Thus, k DCE °must be larger than × FDv RT 10 / , which is about 0.0004 cm/s. 35 We use k DCE °= 10 cm•s -1 . Since the kinetics of electron transfer at the DCE| electrode interface is not rate determining in our experiment, we simply assumed α = 0.5. The only free parameters are the kinetics of FcMeOH/FcMeOH + crossing the water|DCE interfaces. Note that the ratio of these two kinetic parameters is the partition coefficient, a fixed parameter. We also note that the large peak splitting of the pair of peaks centered at 0.53 V vs Ag/AgCl observed in Figure 1B,C (black traces) is attributed to ohmic drop. Serial resistance is commonly encountered in low dielectric solvents like DCE. A 2 kΩ serial resistance was introduced in the numerical simulation to account for the peak splitting. More details about the simulation are provided in Supporting Information.

Figure 1C shows the experimental (black trace) and simulated (red circles) CVs (first scans). The experimental CV is subtracted from a blank measured in absence of FcMeOH (Figure S5 in Supporting Information). The oxidation/reduction of FcMeOH/FcMeOH + in DCE is well reproduced by the simulation. The simulation also correctly predicts the appearance of a second cathodic peak on the reverse scan (at 0.10 V vs Ag/AgCl). We observed that experimental CVs are well reproduced only when diffusion is the limiting step (with respect to electron transfer and partitioning). This observation is in agreement with results obtained by White and co-workers using ferrocene. 31 The concentration profile of FcMeOH + at 0.6 V vs Ag/AgCl on the reverse scan (indicated by the blue dot in Figure 1C) is shown in Figure 1D. At this potential, FcMeOH is oxidized at the DCE|electrode interface as indicated by the concentration gradient developing along this interface (orange to blue color gradient normal to the interface). The dark blue color near the center of the droplet indicates that FcMeOH + is not produced at the water|electrode interface.

However, at the three-phase boundary, a large concentration of FcMeOH + is observed in the water phase (red-orange colors), while the concentration in the DCE phase drops (green color). This indicates that FcMeOH + produced in the DCE phase transfers to the water phase. The transferred FcMeOH + can then be reduced at the water|electrode interface near the three-phase boundary giving rise to the second cathodic peak at 0.10 V vs Ag/AgCl on the CVs.

Scholz and co-workers showed that the transfer of species from one phase to another may be accompanied by side reactions like precipitation. 32 We investigated that possibility by performing chronoamperometric experiments coupled with optical observations of the three-phase boundary. Figure 2A,B shows the sequence of potential, and the resulting current flowing through a platinum (Pt) electrode with a 500 nL water droplet containing 0.1 M of TBACl immersed in a continuous DCE phase containing 40 mM of FcMeOH and 1 M of TBAClO 4 . These high concentrations of species are used to ease the visualization of the precipitate. First, the electrode is held 3.5 s at 0.7 V vs Ag/AgCl (>E 1/2 water and >E 1/2 DCE ) to drive the oxidation of FcMeOH in the DCE phase and transfer of FcMeOH + toward the water droplet. Then, a potential of -0.1 V vs Ag/AgCl (<E 1/2

water and E 1/2 DCE ) is held to drive the reduction of FcMeOH + in both the DCE and water phases. Top view optical micrographs centered on the three-phase boundary and recorded at the end of the anodic (Figure 2A) and cathodic (Figure 2B) pulses are shown in Figure 2C,D, respectively. The left side light gray of the micrographs corresponds to the area above the water droplet, while the darker gray area on the right of the micrographs corresponds to the DCE phase. The long The Journal of Physical Chemistry C oblique lines are scratches on the metal surface caused during polishing. Both the water and DCE phases are transparent during the anodic pulse. However, the water phase near the three-phase boundary becomes cloudy as the reduction of FcMeOH + proceeds. We attribute this cloudiness to the scattering of light by large particles of solid formed near the interface. When we apply an intermediate potential of 0.4 V vs Ag/AgCl, which is sufficient to reduce FcMeOH + in the DCE phase (>0.53 V vs Ag/AgCl) but not in the water phase (>0.13

V vs Ag/AgCl), we do not observe formation of a precipitate (Figure S6 in Supporting Information). The formation of a precipitate is observed only during the reduction of FcMeOH + in water. We conclude that FcMeOH precipitates in the water phase.

The precipitation of FcMeOH near the three-phase boundary is explained as follows. We showed in Figure 1D that a large concentration of FcMeOH + can build up in the water phase near the three-phase boundary. When this large concentration of FcMeOH + is suddenly reduced back to the much less soluble FcMeOH species, the latter precipitates.

Precipitation of the reduced form of the redox couple is in contrast with the observation of Scholz and co-workers where electrogenerated decamethylferrocenium forms an insoluble salt with an anion present in either the water or organic phase. 32 We verified that such a reaction can also occur in our system by adding phosphate in the water phase and observing the precipitation of FcMeOH + H 2 PO 4

-near the three-phase boundary (see Figure S7 in Supporting Information). Interestingly, our experiments with only TBAClO 4 (in DCE) and TBACl (in water) show that these salts do not participate in the formation of a new phase. Even though TBA + cation is known to act as a phase transfer catalyst by forming a complex with some anions, we do not expect positively charged TBA + to aid the phase transfer of positively charged FcMeOH + . 36 The formation of a solid phase of FcMeOH requires some specific conditions. Nucleation takes a certain amount of time and depends on the competition between the birth rate and death rate of the first nucleus. 37 This time is called the induction time. In our experiments, the voltammetric sweep rate must be slower than the induction time but faster than diffusion, to promote the nucleation of a new phase. While precipitation of FcMeOH is a special case, one must consider the nucleation induction period when translating the current work to other systems. However, using this model, one can adjust experimental parameters in this system to predict concentrations that would lead to the precipitation of FcMeOH. Observation of precipitation requires maintaining a concentration of FcMeOH larger than the solubility limit for a time longer than the induction time. These conditions can be achieved by adjusting the initial concentration of FcMeOH and the duration of the electrogeneration of FcMeOH + . Besides these two obvious parameters, we should also consider another parameter, the geometry of the three-phase boundary. Indeed, building up a large concentration of FcMeOH + in the water phase is intimately related to the diffusion profiles of the species near the three-phase boundary. The geometry of this interface can favor or disfavor accumulation of species. In our experiments, the wetting of the droplet can drastically affect the geometry near the three-phase boundary. Figure 3A,B shows side view optical micrographs of water droplets deposited onto a 1 mm radius Pt electrode and a 1.5 mm radius glassy carbon electrode, respectively. The contact angles on the Pt and glassy carbon are 58 and 121°, respectively. The 500 nL water droplets contained 0.1 M TBACl, while the DCE continuous phase contained 40 mM FcMeOH and 1 M TBAClO 4 . The potential was swept at 50 mV/s from -0.1 to 0.7 V vs Ag/AgCl and back to -0.1 V vs Ag/AgCl. We observe no precipitate rings on glassy carbon by optical inspection of the surface during the course of the voltammetry. On the other hand, we do observe precipitation on the Pt surface (as it was observed in Figure 2 by chronoamperometry). To provide physical insight, we performed numerical simulations of CVs in Concentration profiles of FcMeOH + for the 58 and 121°a ngles are shown in Figure 3C and D, respectively. A striking difference in the maximum concentration of FcMeOH + near the three-phase boundary is observed. The concentration of FcMeOH at the three-phase boundary (in the water phase) is about 2.5 times larger for an angle of 58°than for an angle of 121°. This difference can be understood as follows. A low angle produces a geometry that resembles a channel in the water phase. A narrow shape favors the intake of FcMeOH + from the DCE to the water phase while hindering the diffusion of the FcMeOH + from the three-phase boundary toward the center of the droplet.

The concentration of FcMeOH + accumulated at the threephase boundary as a function of the angle of the three-phase boundary is shown in Figure 4. Here, the simulated concentration in the water phase was taken 10 nm from the three-phase contact and is used to represent a trends 2D plot. The concentration of FcMeOH + accumulated in the water phase near the three-phase boundary increases dramatically from 2 mM at 131°to 8 mM at 2.5°. Figure 4 also shows that about 80% of the variation of concentration occurs when the angles vary from 10 to 100°. Besides these values, the concentration does not vary significantly. In a nutshell, when the contact angle decreases, the local concentration of transferred species increases at the three-phase microenvironment. This observation can shed light on previous work in the literature. White and co-workers performed numerical simulations of CVs at three different DCE|water|electrode contact angles and predicted a variation of the peak current for the oxidation of the species accumulated in the water phase (ferrocenium in their case). However, no physical insight was provided to account for this prediction. Based on our findings, we propose that the peak current will vary because the amount of species accumulated in the water phase depends on the geometry of the three-phase boundary.

In conclusion, we have demonstrated that phase nucleation can be favored in multiphase systems based on simple geometrical considerations. Based on numerical simulations of CVs, we clearly evidence that diffusion limited phase transfer is the cause of large concentration gradients near the three-phase boundary. Importantly, we evidence that low contact angles favor high concentrations of species in the droplet near the three-phase boundary. We prove that a careful design of the three-phase boundary geometry can be used to trigger local precipitation of FcMeOH. Our results are generalizable to a heterogeneous reaction that creates a product that may be more soluble in the droplet phase (for