Ultrafast desorption of colloidal particles from fluid interfaces

Significance

Solid particles can replace surfactants to stabilize emulsions and foams. The attachment of particles onto drops and bubbles is typically considered to be irreversible because of a large energy barrier for particle detachment––millions of times the thermal energy for microparticles. Here we demonstrate a method to promote the detachment of microparticles from bubbles using ultrasound. We identified conditions for complete particle removal and recovery in under a millisecond. Our method is programmable in time, and does not require any physicochemical modification of the fluids or the interface. This work addresses the emerging need for methods to recover interfacial particles from emulsions and foams in applications ranging from controlled release to interfacial catalysis and gas storage.

Abstract

The self-assembly of solid particles at fluid–fluid interfaces is widely exploited to stabilize emulsions and foams, and in materials synthesis. The self-assembly mechanism is very robust owing to the large capillary energy associated with particle adsorption, of the order of millions of times the thermal energy for micrometer-sized colloids. The microstructure of the interfacial colloid monolayer can also favor stability, for instance in the case of particle-stabilized bubbles, which can be indefinitely stable against dissolution due to jamming of the colloid monolayer. As a result, significant challenges arise when destabilization and particle removal are a requirement. Here we demonstrate ultrafast desorption of colloid monolayers from the interface of particle-stabilized bubbles. We drive the bubbles into periodic compression–expansion using ultrasound waves, causing significant deformation and microstructural changes in the particle monolayer. Using high-speed microscopy we uncover different particle expulsion scenarios depending on the mode of bubble deformation, including highly directional patterns of particle release during shape oscillations. Complete removal of colloid monolayers from bubbles is achieved in under a millisecond. Our method should find a broad range of applications, from nanoparticle recycling in sustainable processes to programmable particle delivery in lab-on-a-chip applications.

Colloidal particles can adsorb to fluid–fluid interfaces and confer outstanding stability to emulsions and foams (1, 2). The interfacial self-assembly of colloids has been used to create novel materials, for instance colloidosomes (3) and bijels (4). These applications rely on the large decrease in free energy accompanying particle adsorption, $\mathrm{\Delta }E=-{\gamma }_0\pi a^2{\left(1-\cos \theta \right)}^2$, which depends on the surface tension ${\gamma }_0$, the particle size a, and the three-phase contact angle θ (5), and ranges from hundreds to millions of times the thermal energy for nanometer- to micrometer-sized particles. The microstructure that the colloidal particles form at the interface has also been shown to enhance stability. For instance, jamming of an interfacial colloid monolayer prevents coarsening in bicontinuous emulsions (4) and can arrest the dissolution of particle-stabilized bubbles (6). Particle removal from fluid–fluid interfaces is a significant challenge in emerging applications of functional nanoparticles in interfacial biocatalysis (7), gas storage (8), and biomass conversion (9), where the ability to recover and regenerate the nanoparticles at the end of the process is a key requirement. The most common approaches to particle removal from fluid interfaces are based on the physicochemical modification of the fluid phases or the interface. Desorption has been obtained for instance by the addition of a surface-active agent (10, 11). In addition, by tuning the strength of electrostatic repulsion between charged particles at an interface through pH and electrolyte concentration, the surface density of particles can be shifted to very low values up to complete removal (12, 13).

We propose to develop particle removal methods that eliminate the need for physicochemical modifications. A few observations of particle expulsion mechanisms suggest that this is a promising direction. Particles can desorb under the effect of a body force due to an external field. For instance, emulsions stabilized by magnetic nanoparticles can break in a magnetic field (14). Alternatively, interface compression can result in mechanically forced desorption (15), where particles are pushed out of the interface because they no longer fit in the area available to the monolayer. Mechanically forced desorption holds promise for particle recovery without physicochemical modifications that may affect the reactions or otherwise alter the products in a process, and does not pose limitations on the properties of the particles (e.g., magnetic).

Here we demonstrate ultrafast mechanically forced desorption of colloids from particle-stabilized bubbles. We exploit the phenomena of bubble dynamics in ultrasound (16) to impart controlled, highly dynamic interface deformations on a timescale of microseconds. We observe strikingly different particle expulsion scenarios depending on the mode of deformation of the particle-stabilized bubbles. Colloid monolayers that are initially cohesive break up due to the ultrafast compression and expansion of the interface during bubble dynamics, enabling particle detachment despite the presence of attractive interparticle interactions. Complete removal of particle monolayers from bubbles can be achieved in under a millisecond. Using ultrasound to cause mechanically forced desorption also offers the advantages that the system is manipulated remotely, and that particle delivery is precisely programmable in time.

Results and Discussion

We make particle-stabilized bubbles by mechanical agitation of a suspension of charge-stabilized colloids. The bubbles are polydisperse, with radii between 20 and 100 μm, and are stable against dissolution when the surface coverage by particles is sufficiently large ($\varphi >0.4$). Addition of salt to promote adsorption at the water–air interface (Materials and Methods) results in screening of the electrostatic repulsion between the particles. Due to the loss of colloidal stability, the particles form cohesive monolayers at the water–air interface owing to attractive van der Waals and capillary interactions (SI Text). We transfer a small number of bubbles into an observation cell and transmit short ultrasound pulses of 20–100 cycles with acoustic pressures between 200 and 300 kPa, at a frequency of 40 or 50 kHz (Materials and Methods). Because the ultrasound wavelength is 30–40 mm, the acoustic pressure is approximately uniform over the diameter of the bubbles. The pressure fluctuates in time, driving the bubbles into volumetric oscillations, and causing the colloid monolayer to periodically compress and expand. For oscillations of small amplitude, the shape of the bubbles remains approximately spherical (SI Text). The relative amplitude of oscillations of a bubble, $\mathrm{\Delta }R/R_0$, with $R_0$ the initial radius and $\mathrm{\Delta }R=\left(R_{\text{max}}-R_{\text{min}}\right)/2$ the dimensional amplitude, depends on the driving frequency. The amplitude is a maximum at the resonance frequency of the bubble, which depends on the bubble radius and the properties of the interfacial layer (16, 17). Because we operate at constant frequency, we effectively tune the amplitude of oscillations through the bubble size (see SI Text and Fig. S1). We can further adjust the amplitude of oscillations by changing the acoustic pressure. We record the fast dynamics of the bubbles and the evolution of the particle monolayers using high-speed video microscopy at up to 375,000 frames per second. We performed ∼200 experiments on over 50 bubbles with different initial radii and surface coverage.

Ultrafast Monolayer Deformation and Particle Desorption.

High-speed video microscopy shows monolayer expansion, compression, and buckling, accompanied by particle desorption from the interface of bubbles activated by ultrasound. Fig. 1 (Movie S1) shows a bubble stabilized by 3-μm colloids that undergoes periodic compression–expansion at 50 kHz in the oscillating pressure field generated by the ultrasound wave. The desorption energy of the 3-μm particles is on the order of 10⁷ $k_{\text{B}}T$ (see SI Text and Fig. S2). The high-speed sequence in Fig. 1A reveals that during the compression phase, which occurs on a timescale of 10 μs, the particle monolayer buckles out of the plane of the interface. The surface coverage by particles, initially ${\varphi }_0\approx 0.45$ at rest, increases to $\varphi \approx 0.50$ upon compression. Fig. 1B covers several cycles of oscillations and shows repeated buckling upon compression, accompanied by particle desorption. The final surface coverage after 1.4 ms, with the bubble at rest, has decreased to ${\varphi }_0\approx 0.38$. Interestingly, monolayer buckling is typically not accompanied by expulsion of particles (18–20), for instance if the desorption energy of the particles is too large, or if attractive interparticle interactions prevent the particles from detaching from a cohesive monolayer (21). As the interparticle interactions in our system are attractive (SI Text), the observation of buckling accompanied by particle expulsion suggests that the dynamic deformation of the monolayer alters its microstructure. Fig. 1 C and D shows an indication of microstructural evolution. The monolayer initially presents small crystalline domains ($\varphi \approx 0.50$ during compression). After desorption, the surface coverage upon compression has decreased to $\varphi \approx 0.43$, and a loss of order of the monolayer is observed (Fig. 1D).

Fig. 1.

Ultrafast monolayer compression, buckling, and particle expulsion. (A) Bubble stabilized by 3-μm particles undergoing compression–expansion in ultrasound at 50 kHz. The colloid monolayer buckles during the compression phase (10 μs). (B) Images of the same bubble taken every 10 cycles during the compression phase, showing repeated buckling and expulsion of particles from the interface. (C and D) Triangulation of the particles shows the evolution of the monolayer microstructure from small crystalline domains at $t=0$ ms (C) to disordered at $t=1.4$ ms (D). (E) Schematic of monolayer buckling.

The periodic, highly dynamic deformation of the colloid monolayer causes a striking evolution of its microstructure, as is clearly visible on monolayers with low surface coverage. Fig. 2 and Movie S2 show a bubble coated with 5-μm particles that initially form a 2D cohesive network on the interface (see SI Text). Over a few cycles of oscillations at 40 kHz, the periodic compression–expansion causes the cohesive network to break, and the particles disperse with approximately uniform coverage on the surface of the bubble. Desorption does not occur in this experiment, because the maximum surface coverage reached during compression is not sufficiently large ($\varphi \approx 0.30$; SI Text and Fig. S3). When the oscillations stop, the particles reaggregate over a timescale of hundreds of milliseconds (see SI Text and Fig. S4). We ascribe the breakup of particle aggregates during interface dilation to hydrodynamics. The streamlines of the bulk flow generated by spherical bubble oscillations are radial. Particles that are part of an aggregate cannot follow radial streamlines and therefore experience a tangential motion relative to the fluid with velocity $u_{\theta }\sim f\mathrm{\Delta }R$ of the order of 1 m/s, where f is the frequency and $\mathrm{\Delta }R$ the amplitude of radial oscillations. The resulting viscous drag force on a particle $F_{\theta }\sim \eta u_{\theta }a$, where η is the fluid viscosity, is on the order of 1 nN. The magnitude of the drag force tangential to the interface is comparable to the attractive interparticle interactions in the system (SI Text) and can therefore cause the breakup of particle aggregates, until the particles are uniformly dispersed at the interface. Remarkably, particle desorption is possible even from cohesive monolayers upon dynamic interface deformation, provided that the tangential viscous drag force exceeds any attractive interactions between the particles.

Fig. 2.

Dynamic interface deformation strongly alters the monolayer microstructure. The 5-μm colloids on the interface of the bubble initially form a gel-like network. Dynamic compression–expansion at 40 kHz forces the particles to redisperse over 20 cycles of oscillations due to hydrodynamic forces.

The dominant mechanism of particle expulsion for oscillations of small amplitude, $\mathrm{\Delta }R/R_0<0.10$, as in Fig. 1, is interface compression, similarly to quasi-static mechanically forced desorption (15). The average work done on a particle upon compression can be estimated as $\mathrm{\Pi }\hspace{0.17em}dA$, with $\mathrm{\Pi }$ the surface pressure of the monolayer and $dA=\pi a^2$ the change in area. The surface pressure $\mathrm{\Pi }$, defined as the difference between the surface tension ${\gamma }_0$ of the bare fluid–fluid interface, and the effective surface tension γ of the particle-laden interface, $\mathrm{\Pi }={\gamma }_0-\gamma$, increases with the surface coverage ϕ. A particle can be expelled from the interface when, upon compression, the surface pressure $\mathrm{\Pi }$ increases sufficiently so that the work done upon compression exceeds the desorption energy $\mathrm{\Delta }E$. This approximate criterion can be expressed in terms of a nondimensional desorption number as $\mathrm{\Pi }/\left({\gamma }_0{\left(1-\cos \theta \right)}^2\right)>1$. During bubble oscillations, the surface coverage can increase from an initially low value to the threshold ${\varphi }_{\text{c}}$ corresponding to the critical surface pressure for desorption ${\mathrm{\Pi }}_{\text{c}}$. To corroborate this argument, we observed bubbles with different initial coverage ${\varphi }_0$ and varied the amplitude of oscillations $\mathrm{\Delta }R/R_0$. We found that for bubbles stabilized by 3-μm colloids the average surface coverage during bubble compression that results in particle expulsion is ${\varphi }_{\text{c}}=0.53\hspace{0.17em}\pm \hspace{0.17em}0.04$.

To determine the critical surface pressure for particle expulsion, ${\mathrm{\Pi }}_{\text{c}}$, we characterized the surface pressure, $\mathrm{\Pi }\left(\varphi \right)$, by isothermal compression of particle monolayers on a Langmuir trough (see SI Text for details). Fig. 3 shows the compression isotherm for the 3-μm particles, obtained at the same salt concentration used to make particle-stabilized bubbles. The microstructure of the monolayer, presenting cohesive particle networks, is shown in Fig. 3, Inset. The surface pressure at ${\varphi }_{\text{c}}=0.53$ is ${\mathrm{\Pi }}_{\text{c}}\hspace{0.17em}=\hspace{0.17em}25.0\hspace{0.17em}\pm \hspace{0.17em}0.1$ mN/m, which indeed gives a nondimensional desorption number ${\mathrm{\Pi }}_{\text{c}}/\left({\gamma }_0{\left(1-\cos \theta \right)}^2\right)\sim 3$. This value shows that the mechanical energy input to drive desorption is in excess of the value estimated from the simple equation for $\mathrm{\Delta }E$, which does not take into account dissipative mechanisms such as contact line motion over the surface of the particle (22) and the formation of a liquid bridge during particle desorption (23).

Fig. 3.

Surface pressure of 3-μm colloid monolayer by isothermal compression. The surface pressure $\mathrm{\Pi }$ is reported as a function of the surface coverage by particles $\varphi =N\pi a^2/A$, with N the number of particles and A the area of the monolayer, obtained from optical micrographs of the monolayer (Inset).

Shape Oscillations and Patterned Desorption.

For oscillations of sufficiently large amplitude we observe shape oscillations, a characteristic nonlinear phenomenon of bubble dynamics (24), and find that they lead to highly directional patterns of particle expulsion. Fig. 4A and Movie S3 show the occurrence of shape oscillations on a bubble stabilized by 3-μm particles. The interface exhibits regular undulations with sixfold symmetry. The shape oscillations grow in amplitude over a few cycles, after which desorption of particles is observed from the antinodes of the interface undulations. Fig. 4B (Movie S4) shows another example for a bubble coated with 500-nm particles. Shape oscillations with eightfold symmetry develop initially. A mode with fourfold symmetry becomes dominant after 20 cycles, followed by particle expulsion from the four antinodes, and more pronounced from two of the antinodes. Fig. 4 A and B clearly demonstrates that the pattern of particle release for shape oscillations is highly dependent on the symmetry of the mode, with localized “hot spots” from which most of the particles are expelled. For very violent oscillations, we observe asymmetric collapse of the bubble (Fig. 4C and Movie S5). The first five frames in Fig. 4C show the stages leading to asymmetric collapse during one cycle of oscillation ($25\hspace{0.5em}\mu \text{s}$), after which particles are expelled as a blob, as shown in the last two frames.

Fig. 4.

Patterned particle desorption for nonspherical bubble oscillations. (A) Shape oscillations with sixfold symmetry of a bubble coated with 3-μm particles. Particles are expelled from the antinodes. (B) Shape oscillations of a bubble coated with 500-nm particles, with a dominant fourfold mode developing over time and directing particle expulsion. (C) Asymmetric collapse and release of a blob of particles. (D and E) Schematics of shape oscillations and asymmetric collapse of particle-stabilized bubbles.

The observation of patterns of particle release that depend on the bubble shape points to a second mechanism of particle expulsion. For very violent deformations, as in Fig. 4C, the radial acceleration $\ddot{R}\sim f^2\mathrm{\Delta }R$ can reach ${10}^5$ m/s² at the antinodes. The body force on the particles normal to the interface due to this large acceleration should favor expulsion. The relative importance of this body force and the capillary force holding a particle at the interface can be expressed by a Bond number based on the acceleration of the interface, $Bo=\mathrm{\Delta }\rho a^2\ddot{R}/\left(\gamma \left(1-\cos \theta \right)\right)$, with $\mathrm{\Delta }\rho$ the density difference between the particle and the fluid (25). Despite the very large value of the acceleration, in our experiments the Bond number does not exceed $Bo\approx {10}^{-3}$. Although the body force on a single particle may be insufficient to drive desorption, this contribution is enhanced by the effect of many particles collectively pushing on “keystone” particles in points of high curvature of the monolayer, as has been reported for the case of desorption under the effect of gravity (25). Desorption from points of high curvature is indeed in keeping with our observations in Fig. 4. Whereas the two mechanisms of particle desorption described here (interface compression and body force due to the radial acceleration) are dominant in different dynamical regimes, both contributions are always present in our experiments.

Surprisingly, the presence of a colloidal monolayer does not hamper the occurrence of phenomena such as shape oscillations and asymmetric collapse, shown in Fig. 4 D and E, where the interface undergoes large bending deformation. Shape oscillations have been observed for surfactant-stabilized bubbles (26), with a monolayer thickness on the order of 1 nm, and a ratio of bubble radius to monolayer thickness $R/h>\mathrm{1,000}$. Our colloidal shells are comparatively thick, with $R/h\approx 10-100$. The resistance to bending of a pressurized shell can be estimated by a nondimensional bending stiffness ${\tau }^{-2}$, with $\tau =\left(\mathrm{\Delta }p/E\right){\left(R/h\right)}^2$ (27). The pressure difference $\mathrm{\Delta }p$ is given by the Laplace pressure, and E is the Young’s modulus of the monolayer (see SI Text for details). We measured the monolayer Young’s modulus E from a Langmuir trough compression isotherm (SI Text). The ratio $R/h$ for the bubbles showing shape oscillations (Fig. 4) is between 15 and 100, resulting in a very low nondimensional bending stiffness, ${\tau }^{-2}\approx {10}^{-3}-{10}^{-5}$. Even for small bubbles with $R/h\approx 5$, as in the example of Fig. 1, the nondimensional bending stiffness is sufficiently low (${\tau }^{-2}\approx {10}^{-2}$) that the monolayer can buckle to allow bubble compression (Fig. 1E).

Programmable Particle Delivery.

It is possible to design protocols of ultrasound activation that enable complete removal of the monolayer from the interface, and programmable particle release. The bubble in Fig. 5A (Movie S6) is activated with one pulse of 20 cycles at 40 kHz, which results in the ultrafast delivery of most of the particles from the interface. Without the stabilizing layer of particles, the bubble slowly dissolves by gas diffusion in ∼10 min. During this time we do not observe adsorption of the particles back on the interface, due to the small diffusivity of $3\text{-}\mu \text{m}$ colloids. Fig. 5B and Movie S7 show a high-speed recording revealing that the condition for complete particle delivery is a very large amplitude of oscillations ($\mathrm{\Delta }R/R_0\approx 0.3$ in this experiment). We consistently observe that oscillations of very large amplitude enable complete particle delivery with an ultrasound pulse of only 20 cycles, which lasts 0.5 ms.

Fig. 5.

Programmable ultrafast particle delivery. (A) Activation of a particle-coated bubble by ultrasound can produce complete removal of the particle monolayer with a single pulse of 20 cycles. (B) High-speed imaging of an experiment where complete particle delivery is obtained in 20 cycles. The amplitude of oscillations is $\mathrm{\Delta }R/R_0\sim 0.3$.

Conclusions

We have demonstrated ultrafast colloid desorption from fluid–fluid interfaces undergoing dynamic compression, expansion, and deformation. We have used particle-stabilized bubbles as a platform to impart dynamic interface deformations using ultrasound. We found that particle-stabilized bubbles exhibit some of the characteristic phenomena of uncoated bubble dynamics in ultrasound, and that the different modes of deformation result in different desorption scenarios. Particle expulsion is governed primarily by the surface coverage by particles, by the large acceleration of the interface due to ultrasonic driving, and by the shape of the violently deforming interface. The dynamic compression and expansion of the bubble causes particle aggregates to break up due to hydrodynamic forces, allowing expulsion of particles even from cohesive monolayers. Our method will enable remotely triggered colloidal disassembly without physicochemical modification of the system and should find applications in particle recovery, encapsulation (28), and particle delivery (29). This method is scalable and applicable in a variety of geometries and conditions, including existing acoustomicrofluidic platforms (30), with precise control and programmability of the timing of particle release.

Materials and Methods

Particle-Stabilized Bubbles.

We made particle-stabilized bubbles by mechanical agitation of an aqueous NaCl solution containing charge-stabilized polystyrene colloids. We used 500-nm, 3-μm, and 5-μm colloids (IDC surfactant-free latex particles, Life Technologies). The 500-nm and 5-μm colloids are functionalized with sulfate groups. The 3-μm colloids are functionalized with aldehyde and sulfate groups. The NaCl concentration in solution was optimized for each kind of particle to promote particle adsorption at the water–air interface (50 mM NaCl for 500-nm particles; 500 mM NaCl for 3-μm particles; addition of NaCl was not necessary for 5-μm particles). NaCl [BioXtra, $\ge$ 99.5% (wt/wt)] was obtained from Sigma-Aldrich. Ultrapure water with resistivity 18.2 M$\mathrm{\Omega }$ cm was produced by a Milli-Q filtration system (Millipore).

Acoustical–Optical Setup.

The observation cell is made of a microscope glass slide and a glass coverslip separated by a 1-mm spacer. All parts were cleaned with ethanol, rinsed with ultrapure water, and dried with compressed air before each experiment. We injected a few bubbles at a time in the cell, and observed isolated bubbles (at least 10 diameters away from each other) to exclude interactions between bubbles due to scattered ultrasound fields. A piezoelectric transducer (resonance frequency 754 $\pm$ 5 kHz, STEMINC) is glued on the glass slide. We use two driving frequencies that correspond to mechanical resonances of the transducer–glass slide system, 40 and 50 kHz. The sinusoidal driving signal is generated by a programmable waveform generator (33220A, Agilent) and amplified by a radiofrequency linear power amplifier (AG1021, T&C Power Conversion Inc.). The acoustic pressure amplitude was calibrated with a PVDF hydrophone (RP Acoustics). Video microscopy of bubble dynamics was performed using an inverted microscope (Olympus, IX71) with magnifications 10×, 20×, and 40× and a high-speed camera (Photron, FASTCAM SA5, 128$\times$88 pixel resolution at 350,000 frames per second). The waveform generator and the high-speed camera are triggered simultaneously using a pulse-delay generator (9200 Sapphire, Quantum Composer).