betaFeedback

WorksPhysics and Astronomy

Cascade coalescence dynamically sustains bubble retention near gas-evolving surfaces

Abstract

Bubble detachment from solid surfaces governs heat, mass, and charge transport across technologies vital to clean energy, including high-current-density water electrolysis and boiling thermal management. At high gas fluxes, however, bubbles remain trapped at active surfaces despite immense buoyancy, severely restricting mass transfer and increasing energy losses. Here, we show that this unexpected surface retention originates from cascade coalescence between unequal-sized bubbles. High-speed observations around microelectrodes demonstrate that when a rising bubble merges with a smaller surface-attached successor, its trajectory abruptly reverses, accelerating toward the substrate at nearly two orders of magnitude above its rising speed. Direct numerical simulations and scaling analysis reveal that asymmetric interfacial retraction during merging generates non-canceling viscous stresses, producing a net downward impulse toward the smaller bubble. Repeated cascade coalescence events accumulate these transient impulses into a steady, time-averaged retaining force capable of opposing buoyancy three to four orders of magnitude beyond quasistatic limits. Our findings establish bubble coalescence as a previously unrecognized mechanism that dynamically sustains bubble retention under high gas flux.

Introduction

Bubble growth and departure from solid surfaces underpin critical clean energy processes, ranging from green hydrogen generation during high-current-density water electrolysis [1, 2, 3, 4] to boiling-based thermal management for high-power microelectronics. [5, 6, 7] In these systems, the rapid removal of gas or vapor bubbles from active surfaces is paramount. Persistent bubble accumulation shields active sites, restricts mass transfer, and drastically increases electrolysis overpotentials and thermal resistance. Under quasistatic regimes, bubble retention is dictated by the classical force balance between buoyancy and capillary forces along the three-phase contact line. For a circular contact line of diameter dcd_c, the pinning capillary force scales as Fcap=πdcγsin⁡θF_{\mathrm{cap}} = \pi d_{\mathrm{c}}\gamma\sin\theta, yielding the Fritz departure criterion[8, 9] which predicts detachment at tens of micrometres on highly wetting substrates[10, 11, 12] At high reaction rates, however, bubbles remain tightly bound to the electrode, growing to millimetric scales despite immense buoyancy forces[13, 14, 15] Because buoyancy scales with volume (Fb∝D3F_{\mathrm{b}} \propto D^3), a size increase from 50 μm\mu\mathrm{m} to 1 mm amplifies the detaching force by a factor of 8,000. Conventional models cannot explain what retains these bubbles against forces three to four orders of magnitude beyond the classical limit, revealing a fundamental gap in our understanding of high-flux bubble retention mechanisms.

Several secondary forces can influence bubble detachment, yet none account for this strong surface retention under high gas fluxes. In boiling systems, rapid evaporation generates growth-induced hydrodynamic forces[16, 17] and localized vapor recoil stresses that drive dry-spot expansion.[18, 19] However, these thermal mechanisms are largely absent during electrochemical gas evolution, where gas production is limited by electrochemical reaction rates and bubble growth is far slower.[1, 2] Interfacial charging and localized concentration or temperature gradients can also induce electrostatic and Marangoni stresses.[20, 21, 22, 23, 24, 25] While these field forces contribute under specific regimes, estimates based on established force models show that, under the present conditions, they remain at least an order of magnitude below the buoyancy of the retained millimetric bubbles[20, 23, 24, 25, 26, 27] (Supplementary Note 10). More importantly, none of these single-bubble mechanisms explain why surface retention intensifies with bubble number density and coalescence frequency.[13, 14] This intensification implies that coalescence itself may be related to the mechanism sustaining bubble retention.

However, bubble coalescence, which is ubiquitous across gas-evolving systems, is classically understood to promote bubble detachment rather than surface retention.[28, 29, 30] When two bubbles merge, the excess interfacial energy released during coalescence is converted into kinetic energy, which can propel the merged bubble away from the substrate.[29, 30, 31] Recent electrochemical experiments have revealed an apparently opposite outcome: a departing bubble can coalesce with a smaller, surface-attached successor, after which the merged bubble moves back toward the electrode in a “comeback” mode. [13, 14] Despite these opposite migration directions, both coalescence-induced detachment and comeback have been interpreted using the same capillary–inertial framework. [14, 29, 30] In this framework, viscous traction is generally neglected because of the low Ohnesorge number (Oh=O(10−2)Oh = O(10^{-2})).[32, 33, 34] The merged bubble is expected to move toward the volume-weighted center of its parent bubbles, as quantitatively verified for two bubbles growing on a surface. [30, 35] For the comeback configuration, this center is closer to the electrode surface, predicting a downward displacement of the merged bubble. [14, 30] However, the apparent retaining force inferred from the downward displacement scales as D−2D^{-2} and rapidly diminishes relative to buoyancy. The capillary-inertial framework therefore predicts that coalescence alone cannot sustain bubble retention. [14]

Here we show how coalescence-induced return develops into sustained bubble retention. High-speed imaging around gas-evolving microelectrodes reveals that a rising bubble reverses its motion after coalescing with a smaller surface-attached bubble, accelerating toward the surface at nearly two orders of magnitude above its rising speed. Direct numerical simulations and scaling analysis show that, although liquid inertia governs the rapid neck expansion, the viscous stresses generated by unequal interfacial retraction do not cancel. Their time integral produces a finite impulse directed toward the smaller bubble. Repeated coalescence events convert these transient impulses into a time-averaged retaining force capable of opposing the buoyancy of bubbles far beyond the quasistatic departure limit. Consistently, suppressing the supply of surface smaller bubbles causes retained bubbles to escape for a few milliseconds. These results identify cascade coalescence as mechanism that dynamically sustains bubble retention under high gas flux.

Results and discussion

Coalescence reverses bubble motion

We generated hydrogen bubbles on a 50 μ\mum-diameter Pt microelectrode operated at a low current of ∣I∣=0.01\lvert I\rvert= 0.01–0.0150.015 mA in an electrolyte containing 0.5 M H2_2SO4_4 and 0.1 M Na2_2SO4_4, producing a sparse population of surface microbubbles. Under these conditions, individual coalescence events between a detached bubble and a single growing surface microbubble were resolved using high-speed microscopy (Supplementary Note 1). The low-current condition minimized thermal and solutal, [20, 24, 25] as well as growth-induced, [16, 18] contributions to bubble motion (Supplementary Note 2). The subsequent trajectory therefore provides a direct measure of the coalescence-induced bubble motion.

A representative sequence in Figure 1a illustrates the evolution of a detached hydrogen bubble after leaving the electrode. Instead of escaping into the bulk, the detached bubble repeatedly encounters and coalesces with newly formed surface microbubbles. Remarkably, each coalescence event reverses the bubble trajectory, driving the merged bubble back towards the electrode (Supplementary Video 1). This repeated return delays bubble escape and enables subsequent coalescence with newly formed surface bubbles, sustaining a cascade of coalescence events.

Image not extracted

Figure 1. Coalescence reverses bubble motion. (a) Time-resolved snapshots of a hydrogen bubble undergoing repeated coalescence with surface microbubbles after detachment. Each coalescence event drives the bubble towards the electrode before final departure. Frames 1–7 show the last three drag-back events; TT denotes the elapsed time after the initial detachment. (b) Geometric definitions. RlR_l, RsR_s, and RmR_m denote the radii of the detached bubble, surface bubble, and merged bubble, respectively. The xx and yy coordinates are parallel and normal to the electrode surface; yc,0y_{c,0} and yc′y'_c denote centroid positions immediately before and after coalescence. (c) Bubble centroid position ycy_c (black) and radius RlR_l (red) during cascade coalescence, showing repeated downward displacement and stepwise growth. (d) Final coalescence event in (c), showing a drag-back distance Δyc≈6.5 μm\Delta y_c \approx6.5\,\mu\mathrm{m} within 0.3 ms. (e) Measured Δyc\Delta y_c compared with the center-of-mass prediction,[30] Δyc,com\Delta y_{c,\mathrm{com}} over Rs=7.2−14.9 μmR_s = 7.2 - 14.9\,\mu\mathrm{m} and Rl=22.7−32 μmR_l = 22.7 - 32\,\mu\mathrm{m}.

To quantify this motion, we define the coordinate system and geometric parameters shown in Figure 1b. The electrode surface is taken as y=0y = 0, with the positive yy direction pointing away from the electrode. The quantities RsR_s and RlR_l denote the radii of the smaller surface-growing bubble and the larger detached bubble immediately before coalescence, respectively, while RmR_m denotes the radius of the merged bubble. Throughout the analysis, we tracked the centroid position yc(t)y_c(t) of the evolving bubble during the coalescence cascade. Before each coalescence event, yc(t)y_c(t) represents the detached bubble; after coalescence, it represents the newly merged bubble.

The evolution of yc(t)y_c(t) is shown in Figure 1c. Each coalescence event appears as a sharp

downward excursion in yc(t)y_c(t), accompanied by a discrete increase in bubble size. Starting from the initial detachment at t=0t = 0, the bubble undergoes more than ten coalescence-induced returns before finally escaping at t=30.76t = 30.76 ms. Over this coalescence cascade, its radius increases from 27.5 to approximately 32.6 μ\mum, corresponding to a 67% increase in volume. Thus, this coalescence cascade sustains a dynamic near-surface retention state, where repeated coalescence events simultaneously drive bubble growth and maintain its proximity to the electrode.

We then examine whether the observed drag-back motion can be explained by the conventional capillary–inertial picture, which has successfully described bubble neck expansion, interfacial retraction, and jumping dynamics after coalescence. [30, 36] In this picture, viscous effects are neglected, and momentum conservation therefore predicts that the merged bubble relocates towards the volume-weighted center, [30]

Δyc,com=(Rl+Rs)Rs3Rl3+Rs3,(1)\Delta y_{c,\mathrm{com}} = \frac{(R_l + R_s)R_s^3}{R_l^3 + R_s^3}, \tag*{(1)}

of the two parent bubbles immediately before coalescence (Supplementary Note 3), as validated for neighboring surface-growing bubbles. [30, 36, 37] To test this prediction, we quantify the experimentally observed drag-back distance for each event as Δyc=yc,0−yc,min⁡\Delta y_c = y_{c,0} - y_{c,\min}, where yc,0y_{c,0} is the centroid position of the detached bubble immediately before coalescence and yc,min⁡y_{c,\min} is the turning point corresponding to the maximum downward displacement of the merged bubble (Figure 1b,Figure 1d).

However, our experiments show that the capillary–inertial prediction substantially underestimates the observed drag-back distance. For the representative event shown in Figure 1d, with Rl=32.0R_l = 32.0 μ\mum and Rs=12.5R_s = 12.5 μ\mum, the merged bubble travels approximately 6.5 μ\mum towards the electrode within 0.3 ms, whereas the predicted relocation distance Δyc,com\Delta y_{c,\mathrm{com}} is only 2.5 μ\mum. This substantial discrepancy persists across the explored size ratios, with the measured drag-back distances (blue circles) consistently exceeding the predicted values (black solid line) by factors of two to three (Figure 1e). This deviation from the prediction suggests an additional wall-directed contribution generated during coalescence.

Coalescence imparts an impulsive downward velocity

We next ask how coalescence produces such a large drag-back distance. Beyond the predicted relocation, our experiments reveal an unexpected downward velocity after coalescence.

In the representative event shown in Figure 2a, the detached bubble is rising with velocity vpre≈0.13 cm s−1v_{\mathrm{pre}} \approx0.13\ \mathrm{cm}\ \mathrm{s}^{-1} before coalescence (Supplementary Note 4), while the surface-growing bubble remains attached to the electrode and continues to grow. Within the capillary–inertial framework, part of the upward momentum of the rising parent bubble should persist through coalescence, causing the resulting bubble to continue moving upward immediately afterward. Contrary to this prediction, high-speed imaging reveals a pronounced velocity reversal: upon coalescence, the merged bubble acquires a large downward velocity and moves towards the electrode. This downward motion increases the drag-back distance until the bubble reaches a turning point and subsequently resumes buoyancy-driven ascent (Figure 2b and Supplementary Video 2).

Image not extracted

Figure 2. Coalescence reverses bubble motion through an impulsive downward velocity. (a) High-speed image sequence of an axial coalescence event between a detached bubble and a surface microbubble. The detached bubble rises before coalescence with velocity vprev_{\mathrm{pre}}, while the merged bubble acquires an initial downward velocity vm,0v_{m,0} after coalescence. Red arrows indicate the bubble-motion direction; the red dashed line marks the pre-coalescence reference height. (b) Time-resolved centroid trajectory corresponding to (a). Symbols denote experimental measurements, and the solid red line shows a fitted decelerating trajectory used to determine vm,0v_{m,0}. The inset gives the initial downward velocity extracted by trajectory fitting (∼13 cm s−1\sim13\ \mathrm{cm}\,\mathrm{s}^{-1} for the representative event). (c) Initial downward velocity vm,0v_{m,0} as a function of the parent-bubble size ratio Rs/RlR_s/R_l. (d) Drag-back distance Δyc\Delta y_c as a function of vm,0v_{m,0}.

We determine this initial downward velocity by fitting the post-coalescence trajectory with a force-balance model containing buoyancy, bulk viscous drag, and near-wall lubrication resistance (Eq. (2) and Supplementary Note 5):

meff,mvmdvmdyc=Fb−(4πμRm+4πμRm2H)vm.(2)m_{\mathrm{eff,m}}v_{\mathrm{m}}\frac{\mathrm{d}v_{\mathrm{m}}}{\mathrm{d}y_{\mathrm{c}}}=F_{\mathrm{b}}-\left(4\pi\mu R_{\mathrm{m}}+\frac{4\pi\mu R_{\mathrm{m}}^{2}}{H}\right)v_{\mathrm{m}}. \tag*{(2)}

Here, upward motion is taken as positive, vm(t)=dyc/dtv_{\mathrm{m}}(t)=\mathrm{d}y_{\mathrm{c}}/\mathrm{d}t, and vm,0=vm(0+)v_{\mathrm{m},0}=v_{\mathrm{m}}(0^{+}) denotes the merged-bubble velocity immediately after coalescence. The effective inertia of the merged bubble is represented by its added mass, [38] meff,m=Caρl(4/3)πRm3m_{\mathrm{eff,m}}=C_{\mathrm{a}}\rho_{\mathrm{l}}(4/3)\pi R_{\mathrm{m}}^{3}, where Ca=1/2C_{\mathrm{a}}=1/2 is the added-mass coefficient of a spherical bubble and ρl\rho_{\mathrm{l}} is the liquid density. [38] The quantities FbF_{\mathrm{b}}, μ\mu, RmR_{\mathrm{m}}, and H=yc−RmH=y_{\mathrm{c}}-R_{\mathrm{m}} denote the buoyant force, liquid viscosity, merged-bubble radius, and instantaneous bubble–electrode gap, respectively. Without introducing any additional retaining force after coalescence, this model reproduces the observed downward deceleration, turning point, and subsequent rise, suggesting that the motion is a decelera

tion process from an initially imposed velocity. For the representative event, the model gives vm,0≈−13 cm s−1v_{m,0} \approx-13\ \mathrm{cm}\ \mathrm{s}^{-1}, approximately 100 times the magnitude of the pre-coalescence velocity (Supplementary Note 4).

Over the explored ranges Rl≈30R_l \approx30–50 μm50\ \mu\mathrm{m} and Rs≈5R_s \approx5–20 μm20\ \mu\mathrm{m}, the magnitude of the downward initial velocity spans approximately 44–20 cm s−120\ \mathrm{cm}\ \mathrm{s}^{-1} and increases systematically with Rs/RlR_s/R_l (Figure 2c). The drag-back distance increases approximately linearly with ∣vm,0∣|v_{m,0}|, consistent with the post-coalescence deceleration model (Figure 2d and Supplementary Note 5). Therefore, the large drag-back displacement is determined by the unexpectedly large downward velocity acquired during the short coalescence event.

Asymmetric viscous impulse reverses bubble motion

We propose that the observed rapid downward motion is driven by a net viscous impulse generated by size-asymmetric coalescence (Figure 3a). Following initial bridge formation, the resolved interfacial retraction is predominantly capillary–inertial, consistent with the small Ohnesorge numbers (Ohi=0.021\mathrm{Oh}_{\mathrm{i}} = 0.021–0.0440.044; Supplementary Note 6). [32, 33] The characteristic retraction velocity associated with parent bubble ii therefore scales as vi∼[σ/(ρlRi)]1/2v_i \sim[\sigma/(\rho_{\mathrm{l}}R_i)]^{1/2}, [32, 33] yielding vl/vs∼(Rs/Rl)1/2=λ1/2v_{\mathrm{l}}/v_{\mathrm{s}} \sim(R_{\mathrm{s}}/R_{\mathrm{l}})^{1/2} = \lambda^{1/2}, where λ=Rs/Rl\lambda= R_{\mathrm{s}}/R_{\mathrm{l}}. In a frame translating with the merged bubble, the effective retraction displacement ratio follows ld,l/ld,s∼λ3l_{\mathrm{d,l}}/l_{\mathrm{d,s}} \sim\lambda^3 (Supplementary Note 7). Thus, the smaller surface bubble undergoes faster and more extensive interfacial retraction than the larger detached bubble, producing strongly asymmetric retraction dynamics (Figure 3b and Supplementary Note 8).

Image not extracted

Figure 3. An asymmetric viscous impulse generates the post-coalescence downward velocity. (a) Schematic illustration of the asymmetric viscous-impulse mechanism during axial coalescence of unequal-sized bubbles. Unequal retraction dynamics generate an imbalance in viscous contributions, producing a residual wall-directed impulse JvJ_{\mathrm{v}} that gives rise to the initial downward velocity vm,0v_{\mathrm{m},0} of the merged bubble. (b) Direct numerical simulations of the coalescence-induced flow, showing localized strong interfacial motion near the smaller bubble. (c) Collapse of the measured vm,0v_{\mathrm{m},0} using the viscous-impulse scaling; ρ\rho on the horizontal axis denotes the liquid density ρl\rho_{\mathrm{l}}.

Although viscous stresses are subdominant in the instantaneous force balance, their time-integrated contributions need not cancel, generating a net directional contribution. The viscous force associated with interface of bubble ii scales as Fd,i∼μRiviF_{d,i} \sim\mu R_i v_i and acts over ti∼ld,i/vit_i \sim l_{d,i}/v_i, yielding an impulse Jv,i∼μRild,iJ_{v,i} \sim\mu R_i l_{d,i}. Taking the direction toward the electrode as positive, the net viscous impulse is

Jv∫(Fd,s−Fd,l) dt≈CjμRs2A(λ),(3)J_v \int(F_{d,s} - F_{d,l})\,\mathrm{d}t \approx C_j \mu R_s^2 A(\lambda), \tag*{(3)}

where A(λ)=1−λ2A(\lambda)=1-\lambda^2 quantifies the retraction asymmetry and CjC_j is a dimensionless prefactor accounting for interfacial geometry, temporal evolution, and wall effects. Because the smaller bubble lies on the electrode side and its dominant retraction is directed away from the electrode, the corresponding viscous reaction is electrode-directed and exceeds the opposing contribution by a factor λ−2\lambda^{-2}. The size-asymmetric impulse vanishes as λ→1\lambda\to1. Thus, capillary–inertial dynamics govern the rapid shape evolution, whereas the non-cancelling viscous impulse biases the residual motion toward the electrode.

The net viscous impulse acts on the effective inertia of the merged bubble, which is dominated by the added mass of the surrounding liquid, setting its initial velocity after coalescence, Jv=meff,mvm,0J_v=m_{\mathrm{eff,m}}v_{m,0}. For Rs≪RlR_s \ll R_l, Rm≃RlR_m \simeq R_l, yielding

vm,0=−3Cj2πμρlRs2Rl3A(λ).(4)v_{m,0}=-\frac{3C_j}{2\pi}\frac{\mu}{\rho_l}\frac{R_s^2}{R_l^3}A(\lambda). \tag*{(4)}

The model predicts both the wall-directed sign of vm,0v_{m,0} and its dependence on parent-bubble size asymmetry, with the detailed derivation provided in Supplementary Note 7. The measured vm,0v_{m,0} values collapse onto a linear relation when plotted against (μ/ρl)(Rs2/Rl3)A(λ)(\mu/\rho_l)(R_s^2/R_l^3)A(\lambda) (Figure 3c), with the fitted slope giving Cj=41C_j=41. This agreement supports the proposed asymmetric viscous-impulse mechanism and provides the prefactor used below to quantify the retaining force generated by repeated coalescence.

Repeated coalescence sustains a retaining force

Having identified the wall-directed impulse generated by a single coalescence event, we now return to the high-flux gas-generation regime, where a detached large bubble resides above a continuously replenished carpet of surface microbubbles (Figure 4a and Supplementary Video 3). In this regime, coalescence is no longer an isolated event. Instead, repeated coalescence events occur at a frequency fcf_c, converting discrete impulses into a sustained retaining force through time-averaged impulse transfer. Relating fcf_c to the gas-generation rate through Faraday’s law, [39] fc=χ(3∣I∣RgTg)/(8πFPgRs3)f_c = \chi(3|I|R_gT_g)/(8\pi F P_gR_s^3), gives (Supplementary Note 9)

Image not extracted

Figure 4. Repeated surface-microbubble coalescence sustains bubble retention. (a) Schematic of a detached bubble retained above the surface-microbubble carpet on a 500 μ\mum-diameter Pt microelectrode. Repeated coalescence supplies a downward impulse flux, producing a sustained retaining force Fret,coalF_{\mathrm{ret,coal}} that opposes buoyancy FbF_b. The bubble–electrode separation is defined as H=yc,l−RlH = y_{c,l} - R_l. (b) Representative high-speed sequences showing bubble retention and final departure at −10-10 and −40-40 mA. Inset: enlarged view of the surface-microbubble carpet. (c–e) Temporal evolution of H(t)H(t), bubble radius Rl(t)R_l(t), and buoyancy Fb(t)F_b(t). The plateau in H(t)H(t) marks the retained state; the abrupt rise marks threshold crossing and escape. (f) Departure radius RdepR_{\mathrm{dep}} and retention time trett_{\mathrm{ret}} as functions of current magnitude ∣I∣|I|. (g) Estimated coalescence-induced retaining force Fret,coalF_{\mathrm{ret,coal}} at the upper-bound limit χ=1\chi= 1, compared with the experimental buoyancy Fb,expF_{b,\mathrm{exp}} at departure. (h) Comparison of Fret,coalF_{\mathrm{ret,coal}}, Marangoni force FMF_M, and electrostatic force FeF_e, all normalized by Fb,expF_{b,\mathrm{exp}}.

Fret,coal=Jvfc≈χ3Cjμ∣I∣RgTg8πFPgRsA(λ).(5)F_{\mathrm{ret,coal}} = J_v f_c \approx\chi\frac{3C_j\mu|I|R_gT_g}{8\pi F P_gR_s} A(\lambda). \tag*{(5)}

Here, Fret,coalF_{\mathrm{ret,coal}} denotes the equivalent sustained retaining force generated by repeated coalescence, and χ\chi represents the effective impulse-transfer efficiency. The ideal limit χ=1\chi= 1 corresponds to complete participation of generated microbubbles and perfect alignment of the resulting impulses towards the electrode, whereas incomplete participation or imperfect alignment gives χ<1\chi< 1. The remaining quantities are defined above, with II denoting the applied current, RgR_g the universal gas constant, TgT_g the gas temperature, FF the Faraday constant, and PgP_g the internal bubble pressure. For Rs≪RlR_s \ll R_l, A(λ)→1A(\lambda) \to1, indicating that Fret,coalF_{\mathrm{ret,coal}} is governed primarily by the gas-generation rate and characteristic microbubble size, with weak dependence on the retained-bubble radius. In contrast, buoyancy grows as Rl3R_l^3, implying that the retained bubble remains near the electrode until its buoyancy exceeds the coalescence-induced retaining force.

We tested this prediction using a 500 μ\mum-diameter Pt microelectrode. Representative high-speed sequences at −10-10 and −40-40 mA show that a detached bubble remains suspended above the surface-microbubble carpet rather than escaping immediately into the bulk (Figure 4b). The retained state is quantified by tracking the bubble–electrode separation, H=yc,l−RlH = y_{c,l} - R_l, where yc,ly_{c,l} and RlR_l are the centroid position and radius of the retained large

bubble, respectively. As shown in Figure 4c, H(t)H(t) remains nearly constant during retention and increases abruptly upon departure, defining the transition from retention to escape (Supplementary Videos 4 and 5). Meanwhile, repeated coalescence continuously increases Rl(t)R_l(t) and the corresponding buoyancy Fb(t)F_b(t) (Figure 4d,Figure 4e). The bubble is therefore dynamically retained by ongoing coalescence until its increasing buoyancy exceeds the coalescence-induced retaining force. Increasing ∣I∣|I| from 5 to 50 mA extends the retention time trett_{\mathrm{ret}} from 58 to 120 ms and increases the departure radius from 140 to 510 μm\mu\mathrm{m} (Figure 4f), demonstrating that stronger gas generation sustains retention against larger buoyant loads.

We then compare the coalescence-induced retaining force with buoyancy at departure to test whether repeated coalescence can sustain retention. Using the independently fitted Cj=41C_j = 41 (Figure 3c), a representative microbubble radius Rs=10 μmR_s = 10 \,\mu\mathrm{m}, the measured departure radius RdepR_{\mathrm{dep}}, and the ideal impulse-transfer limit χ=1\chi= 1, the estimated Fret,coalF_{\mathrm{ret,coal}} reaches the 10−610^{-6} N scale and slightly exceeds the experimental buoyancy at departure (Figure 4g). This upper-bound estimate demonstrates that repeated coalescence can, in principle, provide sufficient retaining force under high gas flux. In comparison, even deliberately generous estimates of thermal and solutal Marangoni forces [20, 23, 24] and electrostatic forces [21, 22] are about an order of magnitude smaller than buoyancy (Figure 4h and Supplementary Note 10). This force hierarchy supports repeated surface-microbubble coalescence as the principal source of the retaining impulse flux when sustained gas generation maintains the microbubble source.

We further reproduce the coalescence-induced retention process in a pure-water micro-capillary system, without electrochemical gas generation, applied electric fields, or reaction-induced thermal and solutal gradients (Supplementary Note 11 and Supplementary Video 7). In a near-horizontal orifice configuration, the detached bubble repeatedly coalesces with supplied microbubbles while remaining near the outlet. The vertical component of the coalescence-induced impulse counteracts buoyancy over successive events, until the bubble escapes when buoyancy exceeds the accumulated retaining effect. This observation demonstrates that coalescence-induced retention is not specific to electrochemical gas generation.

Pulsed gas generation switches the impulse-flux balance

We next explore the tunability of the coalescence-induced retaining force Fret,coalF_{\mathrm{ret,coal}} by regulating the surface-microbubble source. A defining feature of this force is its dynamic origin: it is sustained only by repeated coalescence events that continuously deliver wall-directed impulses, so its magnitude depends on the availability of newly generated surface microbubbles. Briefly suppressing gas production should therefore deplete the surface-microbubble carpet, reduce the coalescence frequency fcf_c, and hence weaken Fret,coalF_{\mathrm{ret,coal}}. Once this force falls below buoyancy, the detached bubble should escape prematurely (Figure 5a).

Image not extracted

Figure 5. A brief current pulse switches bubble retention by interrupting the surface-microbubble supply. (a) Schematic of the pulse-perturbation experiment. A bubble retained under high-current operation (Ihigh=−25I_{\mathrm{high}} = -25 mA) is briefly exposed to a low-current pulse (Ilow=−1I_{\mathrm{low}} = -1 mA, Δtp=4\Delta t_p = 4 ms). The reduced gas-generation rate decreases the surface-microbubble supply and weakens the coalescence-induced retaining force. (b) High-speed snapshots showing premature bubble departure at a smaller size after the low-current pulse compared with the surrounding high-current condition. (c) Short-time image sequence resolving depletion and recovery of the surface-microbubble carpet. During the pulse (t=0t = 0–4.04.0 ms), the carpet progressively disappears; after high-current operation is restored (t=4.0t = 4.0–7.57.5 ms), the microbubble population is rapidly replenished.

We tested this prediction on a 500 μ\mum-diameter Pt microelectrode operated at Ihigh=−25I_{\mathrm{high}} = -25 mA, where sustained gas generation at high current produced a dense surface-microbubble carpet and prolonged bubble retention. A brief low-current pulse (Ilow=−1I_{\mathrm{low}} = -1 mA for Δtp=4\Delta t_p = 4 ms) reduced the gas-generation rate to one twenty-fifth of its high-current value, rapidly depleting the surface-microbubble population available for repeated coalescence. In the representative experiment shown in Figure 5b, the retained bubble escapes shortly after the current is reduced, demonstrating that the coalescence-induced retaining force is no longer sufficient to balance buoyancy once the microbubble supply is interrupted. High-speed imaging directly visualizes the depletion of the surface-microbubble carpet beneath the retained bubble during the pulse (Figure 5c and Supplementary Video 6). Quantitative tracking further confirms that the pulse reduces the retention time and departure radius (Supplementary Note 12). Thus, pulsed operation establishes the surface-microbubble supply as a direct control parameter for the coalescence-induced retaining force.

Conclusions

In summary, our work establishes cascade bubble coalescence, traditionally viewed as an ejection mechanism, as a dynamic hydrodynamic anchor that governs bubble retention under high gas fluxes. By demonstrating that asymmetric interfacial retraction produces a sustained, time-averaged viscous impulse, these findings provides a new mechanism for multiphase control at solid-liquid interface: rather than relying on static surface properties like contact angle or macro-morphology, bubble transport can be dynamically regulated by modulating local coalescence events through electrolyte rheology, targeted micro-texturing, and controlled nucleation dynamics.

Beyond bubble retention, repeated coalescence provides a persistent source of interfacial perturbations under high-gas-flux conditions. A millimeter-scale bubble may result from up to 10510^{5} microbubble coalescence events (Supplementary Note 13), each delivering a localized impulse that renews near-interface flows and modifies the local chemical environment. These high-frequency perturbations can enhance interfacial transport through repeated flow renewal, while also imposing high-cycle interfacial mechanical loading that may contribute to long-term material degradation, such as precious-metal loss from electrodes during water electrolysis. These insights should substantially reshape the understanding, modelling, and control of high-gas-flux interfaces across boiling, water electrolysis, catalytic gas evolution, and gas–liquid reactors.

Experimental Methods

Electrolyte solution

The electrolytes were prepared from H2SO4\mathrm{H_2SO_4} (98%, Suprapur, Merck), Na2SO4\mathrm{Na_2SO_4} (99%, Suprapur, Merck), and Milli-Q water (resistivity ≥18.2 MΩ cm\ge18.2\ \mathrm{M\Omega\,cm}).

Microelectrode experimental system

Hydrogen bubbles were generated on Pt microelectrodes under galvanostatic conditions in a three-electrode electrochemical cell. A Hg/Hg2SO4\mathrm{Hg/Hg_2SO_4} electrode was used as the reference electrode, and a Pt wire was used as the counter electrode. Two microelectrode sizes were used. The low-current single-coalescence experiments in Figure 1–Figure 3 were performed on a 50 μ\mum-diameter Pt microelectrode, where sparse surface microbubbles enabled individual coalescence events to be resolved. The high-current retention and pulse experiments in Figure 4 and Figure 5 were performed on a 500 μ\mum-diameter Pt microelectrode, where dense surface-microbubble carpets produced sustained coalescence-induced bubble retention. The detailed microelectrode setup, electrode preparation, imaging configuration, and image-processing procedure are provided in Supplementary Note 1.

Experimental data acquisition

In the analysis, each individual coalescence event was treated as an independent data point. For every event, the radius of the newly detached parent bubble (RlR_l), the surface parent microbubble (RsR_s), the merged bubble (RmR_m), and the maximum centroid relocation distance (Δyc\Delta y_c) were quantified. Because parent-bubble sizes vary from event to event, this approach naturally samples a broad range of size ratios.

Owing to the temporal and spatial resolution limits of high-speed imaging, the analysis focuses on the final three coalescence events prior to bubble departure, for which the contours of the surface microbubbles can be reliably resolved. To ensure statistical robustness and measurement reliability, more than 200 independent coalescence events—corresponding to over 200 distinct parent-bubble size ratios—were analyzed and used for model fitting and scaling analysis. Error bars shown in the figures represent an image-based measurement uncertainty of ±2\pm2 pixels (∼2.24 μm\sim2.24\,\mu\mathrm{m}).

Image analysis and data processing

Image analysis and data processing were performed on the original high-speed image sequences. Bubble segmentation and size quantification were conducted using a deep-learning-based image-processing pipeline implemented in Python, based on the Cellpose framework[40, 41] (Supplementary Note 1).

Numerical method

Direct numerical simulations were performed using the open-source Basilisk framework to simulate the coalescence of two bubbles. A two-phase Navier–Stokes model was employed, and the gas–liquid interface was tracked using the volume-of-fluid (VOF) method (Supplementary Note 8).

Data availability

Source data underlying the figures are provided with this paper. Additional raw data, including the original high-speed imaging datasets, are available from the corresponding author upon reasonable request.

Author contributions

All authors wrote and read the paper. All authors conceived and designed the experiments. T.W.: Writing—original draft; Writing—review and editing; Methodology; Investigation. B.L.: Writing—review and editing; Conceptualization and model development; Funding acquisition; Supervision. H.H.: Methodology; Simulation; Writing—review and editing. X.Z.: Writing—review and editing. F.Y.: review and editing; Methodology. H.T.: Writing—review and editing; Conceptualization and model development; Funding acquisition; Simulation supervision. Q.Y.: Writing—review and editing; Funding acquisition; Supervision.

Competing interests

The authors declare that they have no competing interests.

Acknowledgement

We thank Prof. Qiang Sun from the ARC Centre of Excellence for Nanoscale Bio-Photonics, RMIT University, and Prof. Bin Wang from East China University of Science and Technology for fruitful discussions.This work was financially supported by the National Natural Science Foundation of China (Grant Nos. 52025103, 22178099, and 12588301), the Shanghai Pilot Program for Basic Research (Grant No. 22TQ1400100-11), and the Guangdong Basic and Applied Basic Research Foundation (Grant No. 2024A1515010509 and 2024A1515010614).

Supporting Information Available

  • Supplementary Information Note 1–13.

  • Supplementary Videos 1–7.

References

  1. [1]Zhang, L.; Iwata, R.; Lu, Z.; Wang, X.; Díaz-Marín, C. D.; Zhong, Y. Bridging Innovations of Phase Change Heat Transfer to Electrochemical Gas Evolution Reactions. Chemical Reviews 2024, 124, 10052–10111.DOI
  2. [2]Kempler, P. A.; Coridan, R. H.; Luo, L. Gas Evolution in Water Electrolysis. Chemical Reviews 2024, 124, 10964–11007.DOI
  3. [3]Deng, L.; Jin, L.; Yang, L.; Feng, C.; Tao, A.; Jia, X.; Geng, Z.; Zhang, C.; Cui, X.; Shi, J. Bubble evolution dynamics in alkaline water electrolysis. eScience 2025, 5, 100353.DOI
  4. [4]Lee, J. K.; Babbe, F.; Wang, G.; Tricker, A. W.; Mukundan, R.; Weber, A. Z.; Peng, X. Nanochannel electrodes facilitating interfacial transport for PEM water electrolysis. Joule 2024, 8, 2357–2373.DOI
  5. [5]Dhir, V. K. Boiling Heat Transfer. Annual Review of Fluid Mechanics 1998, 30, 365–401.DOI
  6. [6]Zhang, L.; Gong, S.; Lu, Z.; Cheng, P.; Wang, E. N. Boiling crisis due to bubble interactions. International Journal of Heat and Mass Transfer 2022, 182, 121904.DOI
  7. [7]Inanlu, M. J.; Ganesan, V.; Upot, N. V.; Wang, C.; Suo, Z.; Rabbi, K. F.; Kabirzadeh, P.; Bakhshi, A.; Fu, W.; Thukral, T. S. et al. Unveiling the fundamentals of flow boiling heat transfer enhancement on structured surfaces. Science Advances 2024, 10, eadp8632.DOI
  8. [8]Fritz, W. Maximum volume of vapor bubbles. Physikalische Zeitschrift 1935, 36, 379–384.
  9. [9]Vogt, H. The problem of the departure diameter of bubbles at gas-evolving electrodes. Electrochimica Acta 1989, 34, 1429–1432.DOI
  10. [10]Zhao, P.; Gong, S.; Zhang, C.; Chen, S.; Cheng, P. Roles of Wettability and Wickability on Enhanced Hydrogen Evolution Reactions. ACS Applied Materials & Interfaces 2024, 16, 27898–27907.DOI
  11. [11]Iwata, R.; Zhang, L.; Wilke, K. L.; Gong, S.; He, M.; Gallant, B. M.; Wang, E. N. Bubble growth and departure modes on wettable/non-wettable porous foams in alkaline water splitting. Joule 2021, 5, 887–900.DOI
  12. [12]Darband, G. B.; Aliofkhazraei, M.; Shanmugam, S. Recent advances in methods and technologies for enhancing bubble detachment during electrochemical water splitting. Renewable and Sustainable Energy Reviews 2019, 114, 109300.DOI
  13. [13]Wu, T.; Liu, B.; Hao, H.; Yuan, F.; Zhang, Y.; Tan, H.; Yang, Q. Coalescence-induced late departure of bubbles improves water electrolysis efficiency. eScience 2025, 100472.arxiv.org/abs/2403.12064
  14. [14]Bashkatov, A.; Park, S.; Demirkır, Ç.; Wood, J. A.; Koper, M. T. M.; Lohse, D.; Krug, D. Performance Enhancement of Electrocatalytic Hydrogen Evolution through Coalescence-Induced Bubble Dynamics. Journal of the American Chemical Society 2024, 146, 10177–10186.arxiv.org/abs/2402.07996
  15. [15]Wu, T.; Zhu, Z.; Liu, Y.; Zhang, W.; Yuan, F.; Liu, B.; Yang, Q. Electrolyte-regulated bubble coalescence governs detachment size and electrolysis efficiency in hydrogen evolution. International Journal of Hydrogen Energy 2025, 175, 151345.DOI
  16. [16]Zeng, L. Z.; Klausner, J. F.; Mei, R. A unified model for the prediction of bubble detachment diameters in boiling systems—I. Pool boiling. International Journal of Heat and Mass Transfer 1993, 36, 2261–2270.
  17. [17]Klausner, J. F.; Mei, R.; Bernhard, D. M.; Zeng, L. Z. Vapor bubble departure in forced convection boiling. International Journal of Heat and Mass Transfer 1993, 36, 651–662.DOI
  18. [18]Nikolayev, V. S.; Chatain, D.; Garrabos, Y.; Beysens, D. Experimental Evidence of the Vapor Recoil Mechanism in the Boiling Crisis. Physical Review Letters 2006, 97, 184503.pubmed.ncbi.nlm.nih.gov/17155547
  19. [19]Zhang, L.; Wang, C.; Su, G.; Kossolapov, A.; Aguiar, G. M.; Seong, J. H.; Chav-agnat, F.; Phillips, B.; Rahman, M. M.; Bucci, M. A unifying criterion of the boiling crisis. Nature Communications 2023, 14, 2321.DOI
  20. [20]Park, S.; Liu, L.; Demirkır, Ç.; van der Heijden, O.; Lohse, D.; Krug, D.; Koper, M. T. M. Solutal Marangoni effect determines bubble dynamics during electrocatalytic hydrogen evolution. Nature Chemistry 2023, 15, 1532–1540.DOI
  21. [21]Bashkatov, A.; Hossain, S. S.; Yang, X.; Mutschke, G.; Eckert, K. Oscillating Hydrogen Bubbles at Pt Microelectrodes. Physical Review Letters 2019, 123, 214503.DOI
  22. [22]Lu, X.; Yadav, D.; Ma, B.; Ma, L.; Jing, D. Rapid detachment of hydrogen bubbles for electrolytic water splitting driven by combined effects of Marangoni force and electrostatic repulsion. Journal of Power Sources 2024, 599, 234217.DOI
  23. [23]Meulenbroek, A. M.; Deen, N. G.; Vreman, A. W. Marangoni forces on electrolytic bubbles on microelectrodes. Electrochimica Acta 2024, 497, 144510.DOI
  24. [24]Massing, J.; Mutschke, G.; Baczyżmalski, D.; Hossain, S. S.; Yang, X.; Eckert, K.; Cierpka, C. Thermocapillary convection during hydrogen evolution at microelectrodes. Electrochimica Acta 2019, 297, 929–940.DOI
  25. [25]Hossain, S. S.; Mutschke, G.; Bashkatov, A.; Eckert, K. The thermocapillary effect on gas bubbles growing on electrodes of different sizes. Electrochimica Acta 2020, 353, 136461.DOI
  26. [26]Lu, X.; Yadav, D.; Zhou, L.; Li, X.; Ma, L.; Jing, D. Evolution of hydrogen bubbles on a microelectrode driven by constant currents and its impact on potential response. Chemical Engineering Journal 2024, 500, 156890.
  27. [27]Hossain, S. S.; Bashkatov, A.; Yang, X.; Mutschke, G.; Eckert, K. Force balance of hydrogen bubbles growing and oscillating on a microelectrode. Physical Review E 2022, 106, 035105.DOI
  28. [28]Zhang, B.; Wang, Y.; Feng, Y.; Zhen, C.; Liu, M.; Cao, Z.; Zhao, Q.; Guo, L. Coalescence and detachment of double bubbles on electrode surface in photoelectrochemical water splitting. Cell Reports Physical Science 2024, 5, 101837.DOI
  29. [29]Iwata, R.; Zhang, L.; Lu, Z.; Gong, S.; Du, J.; Wang, E. N. How Coalescing Bubbles Depart from a Wall. Langmuir 2022, 38, 4371–4377.DOI
  30. [30]Lv, P.; Peñas, P.; Le The, H.; Eijkel, J.; van den Berg, A.; Zhang, X.; Lohse, D. Self-Propelled Detachment upon Coalescence of Surface Bubbles. Physical Review Letters 2021, 127, 235501.
  31. [31]Zhang, Y.; Zhang, X.; Lohse, D. Why and When Merging Surface Nanobubbles Jump. Physical Review Letters 2025, 135, 194001.DOI
  32. [32]Eggers, J.; Sprittles, J. E.; Snoeijer, J. H. Coalescence dynamics. Annual Review of Fluid Mechanics 2025, 57, 61–87.DOI
  33. [33]Eggers, J.; Lister, J. R.; Stone, H. A. Coalescence of liquid drops. Journal of Fluid Mechanics 1999, 401, 293–310.DOI
  34. [34]Thoroddsen, S. T.; Etoh, T. G.; Takehara, K.; Ootsuka, N. On the coalescence speed of bubbles. Physics of Fluids 2005, 17, 071703.DOI
  35. [35]Chen, R.; Yu, H. W.; Zhu, L.; Patil, R. M.; Lee, T. Spatial and temporal scaling of unequal microbubble coalescence. AIChE Journal 2017, 63, 1441–1450.DOI
  36. [36]Weon, B. M.; Je, J. H. Coalescence preference depends on size inequality. Physical Review Letters 2012, 108, 224501.pubmed.ncbi.nlm.nih.gov/23003601
  37. [37]Chen, R.; Yu, H. W.; Zeng, J.; Zhu, L. General power-law temporal scaling for unequal-size microbubble coalescence. Physical Review E 2020, 101, 023106.DOI
  38. [38]Magnaudet, J.; Eames, I. The motion of high-Reynolds-number bubbles in inhomogeneous flows. Annual Review of Fluid Mechanics 2000, 32, 659–708.DOI
  39. [39]Faraday, M. Experimental Researches in Electricity, Volume I; Richard and John Edward Taylor: London, 1839.
  40. [40]Pachitariu, M.; Stringer, C. Cellpose 2.0: how to train your own model. Nature Methods 2022, 19, 1634–1641.DOI
  41. [41]Stringer, C.; Wang, T.; Michaelos, M.; Pachitariu, M. Cellpose: a generalist algorithm for cellular segmentation. Nature Methods 2021, 18, 100–106.DOI

Paper details

vlnc:W_yx7365j5
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents