Three-Phase Emulsions During Cross-Coupling Reactions in Water

Abstract

Fluorescence lifetime imaging microscopy (FLIM) enabled identification of three-phase emulsions during a cross-coupling reaction in water, offering insight into their features and factors driving their formation. Droplet morphology was influenced by surfactant choice, ionic strength through phosphate concentration, and evolution of the reaction medium composition. Spatially resolved, subdroplet imaging characterized two organic phases, with one exhibiting preferential localization of the palladium catalyst. Anisotropy and solvatochromic polarity measurements indicated that the palladium-catalyst-containing organic-droplet core exhibited higher viscosity than the organic shell, whereas the polarity of the two organic phases was indistinguishable within the solvatochromic detection capability. The presence of three-phase emulsions correlated with overall formulation and with faster rates of product generation. These findings provide insight for composition–droplet structure relationships toward optimizing aqueous-phase organic reactions and advancing synthetic organic chemistry in water.

Graphical Abstract

INTRODUCTION

Aqueous–surfactant mixtures are a promising less-toxic substitute for organic solvents in synthesis, including on scale.^1–8 Yet, the impact of surfactant choice and reactants on the structure of the resulting colloidal objects—and thus on reaction outcomes of yield, rate, and selectivity—remains difficult to predict. This gap in predictability limits progress.⁹ Significant efforts have therefore been expended to understand the fundamental role of surfactants in dictating interfacial tension and micelle morphology, which facilitate stabilization and influence reaction outcomes in small micelles.^10–13 However, under the synthetic conditions high concentrations of surfactants (1–3 wt %) and organic substrates (~1 M), larger microscale emulsion droplets form,^14–16 providing the opportunity for additional structures. Traditional analytical techniques have limitations in characterizing these structures under reaction conditions. For example, NMR spectroscopy lacks spatial resolution to directly provide information about the internal distribution of components. 2D NMR spectroscopy has provided more information on structures through environments, but data is so far limited to simpler single-phase systems.^17–20 Conventional scattering techniques like dynamic light scattering (DLS) can measure droplet sizes, and small-angle X-ray scattering (SAXS) and small-angle neutron scattering (SANS) are capable of characterizing internal structures, but so far only for systems excluding reagents or otherwise far from operative organic reaction conditions. SANS experiments, for example, require acquisition times ranging from tens of minutes to hours.^17,21–26 Although difficult to discern, the structural features of these droplets may have important implications for reactivity by enhancing or hindering the compartmentalization of reaction components.^27,28

Here, fluorescence lifetime imaging microscopy (FLIM)^29–36 is developed to gain insights toward overcoming this knowledge gap by identifying and understanding structures^30,37 of three-phase emulsion droplets during an ongoing palladium-catalyzed cross-coupling reaction. FLIM offers several advantageous features for this task, including (1) spatial resolution to visualize different morphologies of emulsion droplets and characterize subdroplet regions; (2) outputs (herein, lifetime, intensities, and anisotropy) with high sensitivity to the changing physical and chemical properties of the environment; and (3) compatibility with monitoring reactions in progress (e.g., images within ~2 min) under conditions closely resembling native synthetic environments (e.g., ~0.5 M substrate concentration).

RESULTS AND DISCUSSIONS

The reaction system chosen for study here was inspired by a reported Suzuki cross-coupling reaction developed by Lipshutz,^38–40 which employed organohalides, organoboronates, palladium acetate, SPhos, potassium phosphate (K₃PO₄), and THF cosolvent to enhance solubility in surfactant/water solution (Figure 1a). To generate reproducibly high yield in the absence of iron nanoparticles in the original synthetic report,^38,39 the catalyst loading of palladium acetate was here increased to 2.5 mol %. Sodium dodecyl sulfate (SDS) was chosen for study. It is a widely available and low-cost surfactant, but despite these attractive features, it is not typically used during carbon–carbon bond-forming reactions in water due to low yields for complicated substrates, with limited understanding of the underlying causes.³⁸ We became curious if the structures of the colloidal objects were different with SDS than, for example, the more-oft-used TPGS–750M,⁴⁰ which may contribute to differences in bulk reaction outcomes. Indeed, SDS was found to create three-phase emulsion systems with selective catalyst localization in one of the two organic phases, whereas TPGS–750M produces a simpler two-phase system⁴⁰.

Figure 1.

a. Reaction scheme and FLIM overview of showing unique three-phase emulsions during cross coupling reactions, using 4-iodoanisole and naphthalene boronic acid as substrates. b. FLIM images of three-phase emulsions with different morphologies and corresponding bright-field images. c. z-scan showing the 3D structure of emulsion droplets.

Section 1: Structures.

Imaging agent 1, containing a hydrophobic boron dipyrromethene (BODIPY) fluorophore core, was chosen as a spectator⁴¹ to enable imaging (Figure 1a). FLIM images of the ongoing reaction showed the formation of polydisperse three-phase emulsions in the SDS/water solution, containing one or multiple inner emulsions that were fully or partially embedded in larger emulsions. These images are displayed using a false-colored rainbow scale corresponding to fluorescence lifetime (τ). Longer fluorescence lifetimes are displayed as false-red-shifted, and shorter fluorescence lifetimes as false-blue-shifted. The fluorescence lifetime of a given fluorophore changes on the basis of fluctuations in competition between radiative decay and nonradiative decay caused by the local environment.⁴² Thus, τ is a reflection of microenvironment.^30,42 For example, proximity to quenching agents results in a shorter lifetime. The intensity in each image indicates the spatial location of imaging agent 1.

Emulsion droplets were analyzed using FLIM alongside with corresponding bright-field microscopy, as shown in Figure 1b. Imaging agent 1 is visible by FLIM but not in the bright-field images, due to its low concentration (180 nM). This hydrophobic compound preferentially partitions into organic phases,^43,44 enabling identification of organic phase-structural information of the emulsion droplets. From the FLIM data, two different organic phases are immediately obvious (appearing blue and red, Figure 1b), surrounded by an aqueous phase (appearing dark due to lack of 1). Emulsions are categorized based on the number of internal droplets and the morphological complexity of the spatial arrangement.¹⁷ At different potassium phosphate concentrations, various structures of double emulsions occurred, including core–shell, reverse core–shell, dumbbell Janus, and multiple emulsions with multicore structures (structure names, Scheme S1).^17,45 A detailed discussions of these differences are later described.

Initial comparison of FLIM and bright-field images revealed a correlation between the false-colored lifetime and the observable features in bright-field microscopy. First, the brownish phase observed in bright-field consistently corresponded to regions exhibiting a shorter τ (blue in FLIM). A solution of palladium acetate and SPhos exhibited a brownish color by eye, suggesting that regions with a shorter lifetime and the corresponding brownish color are the locations of higher catalyst concentration. Consistent with this interpretation, the catalyst solution (and aryl iodide) previously demonstrated quenching of 1,⁴⁰ leading to shorter lifetimes.^31,42 This observation is important as it enables the spatial localization of the catalyst within these complex mixtures.

Next, a colorless phase that is only faintly distinguishable from the bulk aqueous phase in the brightfield images correlated to regions exhibiting a longer τ (red in FLIM). The high solubility of hydrophobic 1 in this phase indicated that it is best classified as a second organic phase, giving rise to an organic/organic/water (O/O/W) classification of this system.¹⁷

To clarify the 3D structure, a z-scan was performed on the selected core–shell region marked in the black box in Figure 1b, scanning from the bottom of the sample up to 10.5 μm, with 2.1 μm intervals between each slice (Figure 1c). From both the 2D and 3D perspectives, smaller emulsion droplets were located within a larger droplet, with this larger droplet acting as an intermediate phase with an additional barrier that separates the inner core from the outer aqueous phase. At least two smaller droplets (blue) are apparent within the core–shell selected for the z-scan. Although the formation of double and multiple emulsions can be reliably achieved using controlled microfluidic and accompanying homogenizer devices,^45–50 to our knowledge the imaging of this phenomenon has not been observed previously during ongoing aqueous–organic reactions.

Section 2: Factors dictating the development of three-phase emulsions.

Previous studies on nanoscale micelle analysis have reported the presence of core–shell structures when using ionic surfactants, along with predictions of selective alignment of the charged head groups within the interface.^24,51,52 Further, the size and stability of the emulsion droplets in ionic surfactants are known to be highly sensitive to the ionic strength of the surrounding phases by inducing electrostatic repulsion between emulsions, which prevents close approach and subsequent coalescence.^22,53,54 When the ionic strength of the surrounding phase is low, the repulsive forces between charged droplets are sufficiently strong to maintain a stable dispersion. At high ionic strengths, however, the screening of electrostatic interactions by counterions in solution reduces the effectiveness of this repulsion, facilitating aggregation.^22,53,54

To examine the impact of ion concentration on the structure of three-phase emulsions herein, experiments were conducted under varying concentrations of the base (K₃PO₄). For consistency, the concentrations of other components were maintained as follows: 0.46 M 4-iodoanisole 2, 0.55 M naphthalene boronic acid 3, 0.011 mM palladium acetate, and 0.027 mM SPhos (Figure 2a). At these concentrations, the organic substrates were over 1000x higher in concentration than imaging agent 1. At this low doping level, the imaging agent is unlikely to significantly perturb the phase morphology of the system.

Figure 2.

a. Schematic of the chemical process. b. Three-phase emulsions have different morphologies depending on phosphate equivalents.

The formation of three-phase emulsions was observed at three examined lower equivalents of potassium phosphate (0.5, 1, and 2 equiv), but the three-phase emulsions appeared at later reaction times at 2 equiv of potassium phosphate (Figure 2b). In contrast to the control of microfluidic devices, during chemical reactions, as reactive components build up or are consumed, the electrostatic interactions and the interfacial tensions between phases can fluctuate. As the base increases to 4 equiv, the system became more homogeneous, in that two-phase emulsions (single-phase organics) prevailed.

While the core–shell structure consistently formed across 0.5, 1 and 2 equiv, additional morphological variations emerged depending on the base concentration. At the lower base concentration (corresponding to 0.5 and 1 equiv), in addition to the clear double emulsions with a single predominant core–shell structure or with dumbbell Janus, multicore emulsions were common, wherein multiple smaller emulsions were encapsulated within a larger emulsion droplet. At 2 equiv, both normal and reverse core–shell structure occurred (blue encapsulated in red, and red encapsulated in blue; Figure 1b, Figure 2b and additional examples in SI section 2.1). These findings suggest that the ionic strength of the aqueous phase (regulated by the potassium phosphate) and the interfacial tensions (changed by the chemical composition) both play a critical role in modulating formation and dictating the evolving morphologies of the resulting droplets.

To examine the relationship between the structural behavior of the emulsions and reaction kinetics under different conditions, NMR spectroscopy yields were measured for 0.5, 1, 2, and 4 equiv of K₃PO₄, at different time points (Figure 3a, additional data are available in SI Section 3.1–3.5). At higher equiv potassium phosphate (1 and 2 equiv), formation of 4 displayed constant trends but higher variability among NMR spectroscopy yields across replicates, likely caused by the enhanced layer separation from the increased polarity contrast. Lower equivalents (0.5 and 1 equiv) showed faster initial rates. At 4 equiv, the reaction rate was slowest. Combining the NMR spectroscopy yield studies with the observations by microscopy of delayed formation of three-phase emulsion droplets at 2 equiv and the failure of three-phase emulsion formation at 4 equiv, one possible interpretation is that the three-phase emulsion structure is the cause of the faster rates by causing increased reactant colocalization; however, only correlation and not causation is established by the present data.

Figure 3.

a. ¹H NMR spectroscopy yields corresponding to FLIM images in Figure 1. Data points for 1 equiv, 2 equiv and 4 equiv of potassium phosphate are the average of duplicates. b. FLIM shows emulsion morphology effects of different surfactants at t = 30 min.

Different ionic surfactants of related structural class to SDS were tested. Lithium dodecyl sulfate (LDS) shares the same lipophilic tail as SDS but differs in its counterion, and exhibited the three-phase emulsions under similar Suzuki cross-coupling conditions (representative images in Figure 3b; additional images in Figure S34 and Figure S36. Sodium lauryl ether sulfate (SLES) has the same sodium counterion as SDS but different lipophilic tail, and exhibited three-phase emulsions with different emulsions configurations (water-in-oil-in-water emulsions instead of oil-in-oil-in-water emulsions). In contrast, in the previously studied nonionic TPGS-750M system during Suzuki-cross-coupling, three-phase emulsions were absent.⁴⁰ Together with prior literature, these findings are consistent with assignment of the presence, morphologies, and configurations of the three-phase emulsion microobjects as arising from the kinetic stabilization provided specifically by ionic surfactants at inter-faces.^17,55

Both the cation and anion of the base proved to be impactful. Three bases, potassium acetate, potassium carbonate and sodium phosphate, were evaluated for their effects on the structure of emulsions and reaction outcome. Together with the prior potassium phosphate experiments, these choices allowed for pinpointing the structural outcomes to the role of the cation or anion separately. All three systems containing potassium ions formed oil-in-oil-in-water (O/O/W) three-phase emulsions with distinct lifetimes between oil phases, whereas the sodium-ion containing system singularly yielded water-in-oil-in-water (W/O/W) three-phase emulsions and monophasic emulsions (Figures S27–S29). These outcomes may arise from the size and polarizability differences of the ions, which influence their penetration through the interfaces, leading to different stabilizing effects.⁵⁶ The ¹H NMR spectroscopy yields at t = 1 h for these additional systems, KOAc, K₂CO₃, and Na₃PO₄, were 26%, 72%, and 31% respectively. While the formation of three-phase O/O/W emulsions occurred, the reaction yield was nevertheless low for KOAc, suggesting that although phase structure and reagent colocalization may be important, they are not solely sufficient to ensure high reaction efficiency. Displaying monophasic and W/O/W emulsions, Na₃PO₄ led to apparent decomposition of starting material 2 as characterized by ¹H NMR spectroscopy, and presumably responsible for the low yield under these conditions.

The reaction without the catalyst solution was examined next, and notably, the shorter fluorescence lifetime that had previously been assigned to be the catalyst solution is not observed, consistent with the original assignment (Figure 4a). Specifically, the reaction was conducted under conditions excluding palladium acetate, SPhos, and THF. As observed in Figure 4a (white box), certain emulsion droplets adhered at their edges, forming structures resembling dumbell Janus morphologies. This observation shows that the formation of three-phase emulsion droplets occurs even in the absence of the catalyst, indicating that product formation is not required. After the addition of THF, the increased miscibility of components appeared to disrupt the interfacial tension between emulsion droplets, ultimately leading to the formation of more homogeneous emulsion droplets (Figure 4a, second arrow).

Figure 4.

No-catalyst controls. a. The shorter fluorescence lifetime phase was not observed. b. Different configurations (O/O/W and W/O/W) of emulsions formed under the same reaction conditions. FLIM and bright-field images were acquired ~2 min apart, resulting in a slight displacement of the emulsion droplets between the two images due to their mobility.

Next, the boronic acid 3 was omitted such that 4-iodoanisole 2 was the only substrate (Figure 4b). Multiple emulsion configurations were observed, including oil-in-oil-in-water (O/O/W) and water-in-oil-in-water (W/O/W), the latter being unique to this set of conditions among those explored. This data establishes that the nature of the substrates contribute to the driving force for ultimate morphology of the emulsions.¹⁷ Therefore, it can be expected that variations chemical consumptions could lead to different morphologies and configurations of the emulsion droplets.

The method of emulsification could in principle influence both emulsion morphology and resulting reaction kinetics.^17,57 Experiments described to this point were performed with a 15 min prestir/pre-emulsification step prior to catalyst addition. To assess the impact of this pre-emulsification process, a non-prestirred reaction was conducted. No significant differences in emulsion morphology or reaction kinetics were observed (Figures S38 and Figure S56). Next, the stir bar size was increased to enhance mixing. Under these conditions, three-phase emulsions formed but dissipated more quickly, typically disappearing by 15 min, compared to their persistence up to 30 min under standard conditions (Figure S40). This accelerated dissipation may be attributed to changes in the reaction medium composition from improved mixing including by causing faster product formation and/or different emulsion stability under higher shear stress.¹⁷

Section 3: Viscosity and polarity of the phases.

The viscosity of the phases was analyzed through anisotropy measurements. When a sample is excited with polarized light, the free rotation of the fluorophores results in depolarization of the emitted light. In more viscous systems, the restricted rotation of fluorophores results in a higher proportion of emitted light maintaining the alignment with the excitation polarization.^42,58 A higher anisotropy value therefore corresponds to a more restricted fluorophore rotation and thus a more viscous microenvironment, whereas a lower anisotropy value corresponds to greater molecular rotation and a less viscous or more fluid system.

The core displayed higher anisotropy, consistent with greater viscosity than the shell (Figure 5a, false-colored images), which might be attributed by a more concentrated localization of reagents or surfactant within the core or generally attributed to the specific compounds present. In Figure 5b, the anisotropy measurements uncover a gradual increase and large droplet-to-droplet variation in viscosity of the core as the reaction proceeds, while the shell remains largely unchanged. (Additional images are available in SI Section 2.7.) These observations indicate that the reaction might be primarily occurring within the core, where more pronounced changes of reagents composition take place, while the shell remains largely unaffected. Alternatively, products may partition into (or starting materials out of) primarily the core, upon reaction, accounting for its larger physical change. The full compositions of the two organic phases, however, remain unassigned.

Figure 5.

a. Anisotropy, false-colored images. b. Evolution of anisotropy for core and shell as reaction proceeds. c. Polarity test using a solvatochromatic fluorophore.

Nile Red 5 was selected for polarity investigations due to its chemical inertness to the reaction conditions, known solvatochromism, and hydrophobicity.⁵⁹ In more polar solvents, Nile Red redshifts spectrally, whereas in less polar solvents, it blueshifts.⁵⁹ Because the employed FLIM microscope cannot directly measure emission spectra, the ratio of intensities in two distinct emission wavelength ranges (550–592 nm “yellow” and 630–670 nm “red”) was used as a stand-in. A higher yellow-to-red ratio would indicate a shift toward the yellow region, indicative of a less polar environment, whereas a lower ratio would indicate a more polar environment. This method enables qualitative mapping of polarity variations within the emulsion system. The core and shell did not display a measurable difference in polarity detectable through this approach, as evidenced by the representative data in Figure 5c (for additional images see SI Section 2.8). For each time point, error bars derive from the standard deviation calculated from 6 to 9 images taken from triplicate experiments. The ratios did not change with reaction progress. These somewhat similar polarities are consistent with the earlier assignment of both droplet phases as “organic,” as originally assigned through the partitioning of 1 into both phases.

CONCLUSIONS

Fluorescence lifetime imaging microscopy (FLIM) enabled the visualization and characterization of three-phase emulsions in cross-coupling reactions conducted in water, and identified an O/O/W configuration under reaction conditions. The formation and structures of these droplets were influenced by the surfactant composition, the ionic strength as modulated by potassium phosphate concentration, and the changes in the composition of the reactants over time. These influences produced alternative configurations (e.g., W/O/W), delayed formation of three phases, and/or resulted in the absence of three-phase systems. The presence of the three-phase emulsions correlated with higher bench-scale Suzuki cross-coupling yields.

The palladium catalyst preferentially partitioned into the higher-viscosity core, showing that the local environment of the catalyst would not be well represented by an alternative average measurement of viscosity. This outcome highlights the importance of spatially resolved data to understand these complex systems. The fundamental identification and understanding of the types of emulsion structures that form under reaction conditions, their evolution with reaction progress, and their subdroplet heterogeneities assist in the fundamental understanding and thus long-term design of organic reactions in water.⁶⁰

Supplementary Material

The Supporting Information is available free of charge at pubs.acs.org.

Detailed experimental procedures and replicate fluorescence microscopy data; and ¹H NMR spectra (PDF)

Data Availability Statement

The data underlying this study are available in the published article and its Supporting Information.