Controlled vesicle deformation and lysis by single oscillating bubbles

Five nanometres. That's the thickness of a lipid membrane — the molecular film that separates the inside of a cell from everything outside. And it can be torn open by sound. Not by a blade, not by a chemical, but by a pressure wave moving through water. The question is how. And, more importantly, can it be done with enough precision to be useful? For years, the answer to that second question was a frustrating no. The phenomenon of sonoporation — acoustically induced membrane rupture — had been observed repeatedly. Microbubbles driven by ultrasound could punch holes in cell membranes, raising real hopes for drug delivery, gene therapy, and cell transfection. But the mechanism was opaque. The bubbles being used were collapsing violently, in a process called inertial cavitation, and the violence was the problem. Collapsing bubbles produce microjets, shockwaves, and localized heating — a cascade of simultaneous destructive phenomena that Marmottant and Hilgenfeldt describe as "very difficult to control and model." The tool worked, but you couldn't aim it. You couldn't dial it up or down. You couldn't tell a bubble to open a pore without also risking wholesale cell destruction. Marmottant and Hilgenfeldt decided to replace the violence with something quieter. Their central insight: you don't need violent collapse. Small, gentle, linear oscillations of a single microbubble are sufficient to rupture lipid membranes — and in this regime, you can actually control what happens. The experimental setup was elegant in its simplicity. Single air bubbles, with radii ranging from roughly 10 to 100 micrometres, were attached by capillary forces to the wall of a quartz cuvette. Nearby floated giant unilamellar lipid vesicles — hollow spheres of lipid membrane used as stand-ins for cells — also in the 10 to 100 micrometre range. A piezoelectric transducer drove the system with ultrasound at frequencies around 180 kilohertz, and sometimes 40 kilohertz. The key constraint was linearity: the bubble's oscillation amplitude was only about five percent of its radius. From that, using linearized Rayleigh-Plesset dynamics — the standard equation governing bubble motion — the team infers a driving pressure amplitude of roughly 0.1 bar. That's gentle. The bubble is breathing, not collapsing. What that gentle breathing produces, though, is anything but passive. This is where the physics gets genuinely surprising. When a bubble oscillates near a wall, something counterintuitive happens: the back-and-forth oscillatory motion generates a one-way, steady circulation of fluid. This is called Rayleigh-Nyborg-Westervelt streaming, or RNW streaming — a second-order acoustic streaming effect where nonlinear terms in the fluid equations convert oscillation into a persistent microfluidic vortex. The bubble executes both volume oscillations and small translational oscillations because its base is pinned to the wall. Those two motions together, through the nonlinear physics of fluid dynamics, produce a steady flow that lasts on timescales of one tenth to one full second — millions of acoustic cycles longer than any individual oscillation. Marmottant and Hilgenfeldt characterize this streaming using a dimensionless number called the bubble-streaming Reynolds number, defined as the bubble radius squared times the driving frequency, divided by the kinematic viscosity of the fluid. In their experiments, this number is below 0.04 — deep in the low-Reynolds, Stokes-flow regime, where viscous forces dominate and inertia is negligible. That places the streaming in mathematically tractable territory. The team describes the flow using classical Stokes-flow singularities: a stokeslet term decaying like one over distance and a potential-dipole term decaying like one over distance squared, each with an angular dependence on the polar angle from the bubble's translation axis. The rigid wall is handled with image singularities. The result is a closed-form analytic expression for the streaming flow — and it matches the observed vesicle trajectories closely. Streaming velocities near the bubble reach about one millimetre per second. Maximum strain rates, calculated from the theory, come in below one thousand per second. Watch what the streaming does to a vesicle. As it sweeps into the high-shear region near the bubble, it elongates — goes prolate, like a rugby ball. As it recedes, it flattens — goes oblate, like a lens. Some vesicles get caught on a stagnation line about two bubble radii away from the bubble, where they stay stationary while their membrane tank-treads, circulating continuously around a fixed center. These trajectories repeat in loops, on timescales of one tenth to one second, and they're predictable — the vesicles are behaving essentially as passive tracers carried by the streaming flow. Now comes the question of rupture. The streaming exerts shear stress on the membrane. The maximum membrane tension scales as the product of fluid viscosity, the local shear rate, and the vesicle radius. Under baseline conditions — water, five percent oscillation amplitude — the team calculates the fractional increase in membrane area. Using an area expansion modulus of about 0.24 newtons per metre for DOPC lipids, the estimated area expansion is less than four ten-thousandths. To open pores or rupture a membrane, you need an area expansion of about three percent. That's a gap of roughly a factor of one hundred. So, Marmottant and Hilgenfeldt made a prediction: increase the viscous forcing by a factor of one hundred, and you should cross the rupture threshold. They tested it. Switching from water to a water-glycerol mixture raised the dynamic viscosity by about twentyfold. They also modestly increased the driving amplitude and used slightly larger vesicles. The result was exactly what the theory predicted: vesicles developed conical tips at the high-shear region near the bubble, then ruptured, spilling their fluorescent contents in a cloud around the bubble. Lysis occurred on a controllable timescale of about one second. By tuning bubble size, drive frequency, forcing amplitude, and proximity to the vesicle, you can dial the response anywhere from reversible deformation to complete lysis. That control is the point. The paper positions this not just as a mechanistic explanation of sonoporation, but as the foundation of what Marmottant and Hilgenfeldt call microacoustics — using microbubbles as focusing agents to concentrate acoustic effects at the micrometre scale. The bubble doesn't just transmit the sound field; it transforms it, converting an acoustic wave into a localized, steady microfluidic flow with spatial structure on the scale of the bubble itself. That's something a free-propagating sound wave in water can't do. Sound in open water is diffraction-limited — it can't be focused tighter than roughly half a wavelength, which at 180 kilohertz in water is several millimetres. A microbubble focuses that energy down to tens of micrometres. A factor of one hundred in spatial resolution. The applications follow from the control. Cell-wall permeation by electroporation — where electric fields punch holes in membranes — and by particle guns both work but are blunt instruments. Microacoustic streaming offers a tunable alternative: enough shear to open pores without destroying the cell, or precisely calibrated lysis if that's what's needed. Marmottant and Hilgenfeldt also mention acoustical tweezers — using the streaming flow to manipulate objects in microfluidic devices without contact. They report preliminary success deforming and lysing whole HeLa cells, with cell-viability studies ongoing at the time of publication. Return to the opening image. Sound tearing a membrane. What Marmottant and Hilgenfeldt show is that this doesn't require violence. A single bubble, 10 to 100 micrometres across, breathing gently at five percent amplitude under a pressure of 0.1 bar, generates a steady, engineered microflow. Vesicles follow predictable loop trajectories. Membranes stretch under quantifiable shear. And at a threshold that can be calculated in advance from lipid mechanics, they open. The whole sequence — from intact membrane to controlled lysis in one second — falls out of linear bubble physics and Stokes streaming theory. That's what makes it medically interesting: not that bubbles can tear membranes, which was known, but that they can do it on a schedule, to a specification, with a mechanism you can write down and verify. One gently breathing bubble, and the physics does the rest. This lecture was created by ennepō. Go to https://ennepo.ai to Discover, Create and Follow the latest research in your field. Read when you can. Listen when you want to.