Growth and Detachment of Oxygen Bubbles Induced by Gold-Catalyzed Decomposition of Hydrogen Peroxide.

Whereas bubble growth out of gas-oversatured solutions has been quite well understood, including the formation and stability of surface nanobubbles, this is not the case for bubbles forming on catalytic surfaces due to catalytic reactions, though it has important implications for gas evolution reactions and self-propulsion of micro/nanomotors fueled by bubble release. In this work we have filled this gap by experimentally and theoretically examining the growth and detachment dynamics of oxygen bubbles from hydrogen peroxide decomposition catalyzed by gold. We measured the bubble radius R(t) as a function of time by confocal microscopy and find R(t) ∝ t1/2. This diffusive growth behavior demonstrates that the bubbles grow from an oxygen-oversaturated environment. For several consecutive bubbles detaching from the same position in a short period of time, a well-repeated growing behavior is obtained from which we conclude the absence of noticeable depletion effect of oxygen from previous bubbles or increasing oversaturation from the gas production. In contrast, for two bubbles far apart either in space or in time, substantial discrepancies in their growth rates are observed, which we attribute to the variation in the local gas oversaturation. The current results show that the dynamical evolution of bubbles is influenced by comprehensive effects combining chemical catalysis and physical mass transfer. Finally, we find that the size of the bubbles at the moment of detachment is determined by the balance between buoyancy and surface tension and by the detailed geometry at the bubble's contact line.

Introduction

The dynamics of nucleation, growth, and detachment of gas bubbles has important implications for many catalytic reactions which include gas formation. On one hand, the burst and detachment of bubbles may enhance the flow and mass transfer for the reactions.[1,2] The microbubble dynamics has also been exploited to measure local surface tension or to create autonomous motion of the catalytic particles as micromotors or nanorockets.[3−7] In a typical example,[8−11] released bubbles propel a directional movement of micromotors, which is fueled by the reaction , catalyzed by a metal catalyst such as platinum, silver, and gold. On the other hand, bubbles attached to the catalyst surface hinder the direct contact of the catalyst to reactants in the solution, impeding the catalytic reaction.[12,13] Cavitation induced by the bubble collapse close to the catalysts may also damage the catalytic surfaces.[14]

How to control and efficiently remove the bubbles from the catalytic surface or even suppress their formation is thus an important topic for the design of gas-evolving reactions. However, hitherto the focus in the literature has been on bubble nucleation and growth out of gas-oversaturated solutions, without chemical reactions. Then the bubble nucleation is always heterogeneous;[15−18] i.e., at some surface, and thanks to pinning forces even diffusively stable surface nanobubbles can form.[19] On the larger microscale and beyond, such a stabilization is not possible and such bubbles will in general diffusively grow in gas-oversaturated solutions. For isolated bubbles in an unbound environment, the diffusive growth behavior can even analytically be calculated.[20−22] For single bubbles in gas-oversatured solution, also the influence of a neighboring wall to which the bubble is attached, the effect of density-driven convection, the ultimate size at which detachment takes place, the effect of prior bubbles (i.e., “history effects”), and other complications have been reasonably well understood, see, e.g., refs (23−30).

However, all this does not hold for bubbles forming out of catalytic reactions. Indeed, in contrast to bubbles simply forming out of oversaturated solutions, the bubble growth dynamics in catalytic reactions emerges from an interplay of the reaction kinetics and the physical mass transfer, and a quantitative understanding of the coupled effects from the reaction and the gas diffusion is still lacking. In this work, we will therefore experimentally and theoretically investigate the growth and detachment process of the bubbles from the catalytic decomposition of hydrogen peroxide.

Our model system consists of a horizontal flat catalyst surface in an aqueous solution of hydrogen peroxide. Gold was selected to be the catalyst because its relatively high activation energy enables regulating the bubble growth rate to be suitable for experimental observation. In the experiments, a complete lifecycle of the bubble evolving at the interface of the gold catalyst and the solution goes through the phases of nucleation, growth, and detachment.

Experimental Section

The experimental setup is shown in Figure a. A 170 μm thick coverslip was continuously coated with a thin gold–titanium binary layer possessing a thickness of 20 nm plus 3 nm, referring to the gold and the titanium films, respectively. Gold acted as the catalyst for the decomposition of hydrogen peroxide, and titanium was used to strengthen the binding performance between the gold film and the glass substrate in order to avoid any delamination during bubble formation. In order to obtain a smooth uniform coating, the glass coverslip was successively immersed in a 99% nitric acid (HNO3) solution for 5 min, and in a 69% HNO3 solution at 95 °C for 10 min, followed by rinsing with deionized water in a quick dump tank and drying thoroughly in a nitrogen flow. The gold–titanium binary layer was deposited on the coverslip using an ion-beam sputtering system (home-built T’COathy machine, NanoLab, University of Twente). All the sputtering processes were done at a pressure of 6.6 × 10–3 mbar and a power of 200 W. The process pressure was adjusted by an argon flow.

Figure 1

(a) Schematics of the experiment setup. A glass substrate with a thickness of 170 μm is coated with a 23 nm thick gold–titanium binary layer, on which gold is exposed to H2O2 solution and acts as the catalyst for the decomposition process. A PDMS container is filled with H2O2 solution with a depth of H = 5 mm. The area of the reacting interface is A = L × L, where L = 10 mm. The growth of oxygen bubbles is measured by confocal microscopy. (b) Representative confocal image of a growing bubble. The focal plane is located at the bubble bottom, namely, the reacting interface. The transmitted image is actually the projection of the bubble on the transmitted light detector, from which the bubble radius can be measured as a function of time. The scale bar is 200 μm. (c) Reflective image of the bubble base area. The bubble is pinning at a contact line indicated by the dotted line, based on which the effective base radius Rp can be calculated. The Newton’s rings induced by the interference of lights from the liquid–solid interface and liquid–gas interface clearly demonstrate the pinning region. The scale bar is 20 μm.

A square PDMS (poly(dimethylsiloxane), with a curing ratio of 10:1) wall with an inner side length of L = 10 mm and a height of 10 mm was cast to construct a container of solution by adhering to the substrate. Hydrogen peroxide aqueous solution, 30% (w/w), (H2O2, Sigma-Aldrich) was used as received, and the depth of the solution in the PDMS container was H = 5 mm which was large enough compared with the detaching bubble diameter (smaller than 1 mm, see Figure a).

Figure 5

(a) Bubble radius Rd at the moment of detachment as a function of the effective bubble base radius Rp. The black solid line displays the prediction for the Fritz radius RF ∝ Rp1/3, eq . The revised Fritz prediction, i.e., eq (blue dotted line), taking into account the influence of the contact angle θ which is defined as the angle of the local liquid–gas interface inclined on the substrate as shown in the inset, has a better agreement with the experimental data than does eq . The confocal image in the inset shows the morphology of the root portion of the bubble immediately before the detachment. One clearly sees that the bubble has already lifted upward away from the substrate by buoyancy. The pinning of the contact line allows the surface tension to resist the buoyancy to postpone the detachment, eventually leading to a necking stretch distortion of the bubble root. Because the scan direction is from bottom to top, the vanishing of the reflection signal in the dashed circle indicates that detachment has occurred. The scale bar is 5 μm. A movie showing the shape distortion of the bubble surface is included in the Supporting Information. (b) Sketch of bubble at the moment of detachment illustrating the definitions of Rp and θ.

Real-time optical observation was carried out to monitor the bubble growth using an inverted laser scan confocal microscope (Nikon confocal microscope A1 system, Nikon Corporation, Tokyo, Japan) with a 10× dry objective (CFI Plan Fluor 10×/0.3 DIC, numerical aperture (NA) = 0.3, working distance (WD) = 16 mm). Compared with a normal optical microscope, a confocal microscope can eliminate the out-of-focus lights by crucially adding a pinhole on the confocal plane of the lens which acts as a filter of lights from the focal plane. In the current experiments, unless otherwise noted, the microscope was constantly focused on the gold-solution interface, namely the bubble bottom, where reaction took place. A laser with a wavelength of 561 nm was used to observe the bubble formation from below as shown in Figure a. A part of the laser lights can transmit through the substrate to get to the transmitted light detector as the gold–titanium binary layer coated on the coverslip was so thin (23 nm in total) that the substrate was still transparent for visible lights, while the other part of lights reflects at the gold-solution interface back to the spectral detector. Therefore, we combined two channels, i.e., transmission channel and reflection channel, to observe the projection of the bubble (transmitted image, Figure b) and the bubble base area (reflective image, Figure c) simultaneously.

The continuous decomposition of hydrogen peroxide was performed over the entire reacting interface to achieve a highly oxygen-oversaturated solution after the catalytic reaction, leading to the formation of bubbles everywhere at the beginning of the experiments. However, after 1 h growth, coalescence, and detachment, bubbles preferred to nucleate at some preferred sites, like impurities and defects. Then also successive bubbles nucleated at the same sites as that of the previously detached ones. Thus, we were able to observe the bubble growth on several different fixed spots with a certain distance in between. In the course of the data collections, which was 1 h later than the beginning of the experiments, no direct contact took place between any two bubbles before detachment. Every set of experiment took 10 min at most and included four to seven consecutively growing bubbles, in which the number of bubbles depended on the local situations that the bubbles encountered, such as the local concentration gradient and the detaching bubble size. The scanning frame rate was 30 s–1 in resonant mode. All the experiments were operated at a constant temperature of 23 °C.

Figures b and 1c show representative confocal images of a growing bubble. The transmitted image (Figure b) presents the bubble morphology which reveals the variation of bubble radius R as a function of time. The reflective image (Figure c) zooms into the bubble base area, revealing the pinning contact line indicated by the dotted line. The effective bubble base pinning radius Rp can be calculated by equating πRp2 to the measured area of the pinning region (the area enclosed by the dotted line in Figure c).

Results and Discussion

We first address the growth of the consecutive oxygen bubbles at the identical location from which the previous one has just detached. The time interval between two consecutive bubble detachments is always less than 10 min. We employed confocal microscopy to observe and record the growth of the bubbles by measuring the bubble radius R as a function of time t. Figure a shows the results of five consecutive bubbles growing at the same position spanning 8 min. The logarithmic plot in the inset of Figure a reveals that the radius scales with time as R(t) ∝ t1/2, which is consistent with previous results on bubble growth in oversaturated solution without a catalytic gas production reaction.[20,21,26] In Figure b, all experimental data have been converted into the dimensionless radii η = R/Rp, where Rp is the effective radius of the pinning contact area and τ ∝ √t is employed to be indicative of the t1/2 dependence of R(t) where the exact definition of τ is given later in eq . The plot shows that η increases proportionally to τ. The experimental data of five consecutive bubbles follow an identical pattern, exhibiting a repeated bubble growing behavior. Note that six independent experiments have been performed to verify the correctness in statistics. Each set of experiment contains at least four consecutive bubbles. A good repeatability of experimental results is obtained (see the Supporting Information).

Figure 2

Growth of consecutive bubbles at the identical position showing the variations of radius (a) and in (b) dimensionless form η (τ) and its derivative (c) with respect to τ ∝ √t. Symbols represent the experimental data of five bubbles spanning 8 min, while the solid line in (b) and (c) represents the full analytical solution to eq and its normalized derivative, respectively. The fitted oversaturation for the theoretical curve is ζ = 8. The inset in (a) shows that the radius scales with time as R(t) ∝ t1/2. The fluctuations of the data in (c) correspond to oscillations of the surrounding solution. The inset in (c) shows the Sherwood number as a function of the mass transfer Rayleigh number. The constant Sherwood number implies a pure diffusive mass transfer process.

As shown in Figure c, the dimensionless growth rates dη/dτ increase quickly and converge to a plateau value eventually, which shows a consistent value between different bubbles, especially for the early accelerating growth stages. No delay in growth was observed for newly formed bubbles as compared to prior ones in reaching the plateau value for the given period of reaction time. The consistency in the early growth of consecutive bubbles suggests that the oxygen depletion due to the growth and detachment of the previous bubbles has negligible influence on the subsequent bubbles in the current situations. In contrast, for bubble growth out of oversaturated solution, a delayed growth had been observed due to gas depletion in the surroundings of the bubble.[23,28,31,32] The identical growth rate of consecutive bubbles suggests that the gas oversaturation has reached dynamic equilibrium; that is, the production of the gas from the reaction reaches a balance with the gas supplied to the bubble growth. Here, the gas oversaturation is denoted as ζ = C∞/Cs – 1, where C∞ is the gas concentration far from the bubble and Cs is the saturation concentration.

We next compare the bubbles either far apart in space or with long time interval in between (farther than 5 mm in space distance or longer than 1 h in time interval). The dimensionless radii η as a function of τ are plotted in Figure . Bubble A, the reference bubble, formed at position No. 1 and started to grow at the time of t = 2 h, 45 min later than the start of the experiment. Bubble B, formed not long before the formation of bubble A at the time t = 2 h, 41 min, was located 5 mm away from bubble A. Bubble C was at the same position as bubble A but started to grow at the time t = 4 h, 27 min, nearly 2 h later than bubble A. Unlike the repeated growth of the closely consecutive bubbles, substantial discrepancies exist in the growth of these three bubbles as shown in Figure a. Bubble B grew apparently faster than bubble A, whereas bubble C exhibited a slower growth rate. The difference in the growth rate of the bubbles suggests differences in local oxygen oversaturation around each bubble, which will be discussed below in detail to reveal the mechanism.

Figure 3

(a) Discrepant growth of bubbles located far away from each other and bubbles forming far apart in time. Time t is counted since the substrate is exposed to the solution in the experiment. Bubble A (square symbol) in the position No. 1 is the reference bubble. Bubble B (circle symbol) grows at the time close to bubble A but located at the position (No. 2) 5 mm away from bubble A. Bubble C (diamond symbol) formed at the same location as bubble A but grows nearly 2 h later than bubble A. The data reveal discrepancies in the growth rates, also reflected in the different plateau values of the data curves in the inset. The solid lines are the analytical solution of bubble A, assuming ζ = 8. (b) Recast data of bubbles B and C with ζB = 11.3 and ζC = 6, respectively, reveal the different concentration gradients to which the bubbles are subject. The data correspondingly collapse well onto the corrected analytical solutions as shown by the lines.

We now analyze whether the extent of reaction and the distribution of H2O2 are reaction-limited or diffusion-limited, which is highly related to the production and distribution of O2 as well as the growth of bubbles. The (dimensionless) Damköhler number Da is defined as ratio of the characteristic diffusion time and the characteristic reaction time. It reflects the effect of the diffusive mass transfer on the chemical reaction processHere tr is characteristic reaction time, td is characteristic diffusion time, H is the characteristic length (taken to be the depth of the solution), DH is the diffusion coefficient of H2O2, and k is the reaction rate constant. For large Da ≫ 1, the reaction is much faster than the diffusion, denoting a diffusion-limited process in which the diffusion dominates and the reaction is regarded to be instantaneously in equilibrium, whereas for small Da ≪ 1, the process is reaction-limited. In our experiments, Da ≈ 29, suggesting that the extent of reaction is dominated by diffusion, which also determines the distribution of H2O2 across the depth of the solution. The detailed calculation of Da is provided in the Supporting Information.

Here we note that this diffusion-limited process should also hold for most of the common catalysts, like platinum, silver, palladium, and cobalt–iron oxides. These catalysts possess lower activation energies than gold,[33−36] indicating a faster reaction rate and thus a larger Da. Indeed, taking platinum for example, our experiments show that only 2.5% (w/w) H2O2 solution in contact with platinum film generated much more bubbles growing in a much more violent way than in the situation with gold exposed to 30% (w/w) H2O2 solution as reported here. Therefore, the dynamics for most of the common catalysts are expected to be the same as the one in the current gold case, as long as Da > 1. However, because of the slow chemical reaction rate, gold with higher activation energy is an ideal catalyst for the investigations of bubble formation due to the decomposition of H2O2.

The concentration profiles across the depth of the liquid for both H2O2 and O2 over 5 h are shown in Figure (see the Supporting Information for details of how the profiles are calculated). Although the initial consumption rate of H2O2 is quite fast near the reacting interface, the slow depletion of H2O2 as shown in Figure a reveals the diffusion-limited process. Meanwhile, a rapid initial increase of the oxygen concentration is obtained near the reacting interface as shown in Figure b, leading to the accumulation of O2 molecules which results in a highly oxygen-oversaturated state. (Note that C∞ actually refers to the local oxygen concentration near x = 0 where the bubble grows.) This diffusive behavior of oxygen has a great effect on the bubble growth.

Figure 4

Concentration profiles across the depth of the solution for both H2O2 (a) and O2 (b) over 5 h considering a pure diffusion process without the bubble growth. In (a), an obvious depletion of H2O2 is obtained due to the catalytic reaction at the reacting interface (x = 0). In (b), a rapid increase of the oxygen concentration near the reacting interface is achieved at the beginning of the reaction, indicating a highly oxygen-oversaturated state due to the accumulation of O2 molecules. As the reaction continues, the oxygen concentration decreases, which implies an expiration of the catalytic reaction.

Moreover, the oxygen-supersaturated state can also be examined by the ratio between the production rate and the diffusion rate td,O/tr = Hk/(πDO) ≈ 19 ≫ 1, where DO is the diffusion coefficient of O2. The fact that the reaction occurs much faster than the diffusion means that oxygen oversaturation can easily be achieved (see Figure b), which allows for the nucleation and the growth of new bubbles. Eventually, a dynamical balance between the consumption rate of the gas by the bubble growth and the replenishment rate of the gas from the reaction is reached. Therefore, the oversaturation for the growth of consecutive bubbles is at the same level.

To reveal the diffusion-dominated feature in the bubble growth, we recast the experimental data of Figure in terms of the Sherwood and the Rayleigh numberswhere h = ρGṘ/(ζCs) is the mass transfer coefficient, is the oxygen density inside the bubble, p0 = 1.013 25 × 105 Pa is the atmospheric pressure, M = 0.032 kg ·mol–1 is the molar mass of oxygen, is the universal gas constant, T = 296.15 K is the temperature, g is the acceleration of gravity, λ = 4.7 × 10–5 m3·kg–1 is the concentration expansion coefficient of the solution,[37] and ν = 1.0 × 10–6 m2·s–1 is the kinematic viscosity. (We adopt the empirical formula for the variation of the solution density as a function of the oxygen concentration in ref (37), namely, ρ = {[0.0311 + 0.001 54(T – 273.15)] × 10–3 }COVm/M + ρH, where ρs is the density of the solution, CO is the oxygen concentration, Vm is the molar volume of the ideal gas at the atmospheric pressure, and ρH is the water density. Therefore, λC = (∂ρ/∂CO)/ρH = 4.7 × 10–5 m3·kg–1.) We replot the data in the inset of Figure c and for large Ra obtain a constant Sherwood number, suggesting a purely diffusive mass transfer process during the growth of bubbles. No density-driven convection is developing around the bubbles, as was observed for the growth of CO2 bubbles out of CO2-oversaturated solution,[23] as there the CO2-saturated water is heavier than pure water.

Next we will even quantitatively explain the growth rate of consecutive bubbles and of bubbles far apart in space or in time. We consider the diffusive growth of a bubble on an impermeable substrate surrounded by a gas-oversaturated solution. As the bubble grows, a diffusion boundary layer with thickness of also grows outward from the bubble surface, across which there is a concentration gradient from C∞ to Cs. According to Henry’s law, Cs = HcppG, where Hcp = 4.16 × 10–7 kg·m–3·Pa–1 is Henry’s constant for oxygen in water at room temperature[38] and pG is the gas pressure inside the bubble. In our system, the Laplace pressure can be neglected throughout as the influence of the surface tension is limited to the very beginning of growth. Hence, Cs = Hcpp0 is constant.

We also take into account the blocking effect of mass transfer due to the presence of the substrate.[23] The mass transfer is assumed to take place over the effective area of the bubble surface. Therefore, the quasi-static radial growth rate is expressed asIt is convenient to express eq in dimensionless formwith η = R/Rp andWe can solve this ordinary differential equation (ODE) to obtain the full solution with an initial condition η(0) = R(0)/Rp, where R(0) is the initial radius of each individual bubble read from experimental data. Besides, when η ≫ 1 and τ ≫ 1, there is an asymptotic solution to the ODE[19,20]

The above theoretical prediction eq for the growth of bubbles is plotted in Figure and Figure together with the experimental data. Here, the oversaturation ζ is not a priori known and must be adjusted to the experimental data. We find that after some transient, ζ = 8 gives a good description of the data. The derivative of the full solution in Figure c, i.e., dη/dτ, is normalized by the asymptotic value of the derivative S, i.e., the terms inside the brackets in eq . For the repeated growth of the closely consecutive bubbles in Figure , the oversaturation ζ = 8 not only describes the first bubble but also the later ones which follow immediately.

In the very early phase of the bubble’s life, the theoretical curve shows faster growth (i.e., a larger rate) than the experimental data. This transient behavior can be explained as follows: In the theoretical analysis, the thickness of the diffusion boundary layer is initially assumed to be δ = 0, growing proportionally to . This theoretical assumption is not realistic for our experiments, in which due to prior equilibration, the diffusion boundary layer may have already reached a certain thickness during the interval between two consecutive bubbles, which results in a smaller concentration gradient and hence a slower growth rate than the theoretical predication. Moreover, there is a geometry factor for the bubble, which grows in the shape of a spherical cap. In contrast, the theoretical analysis only considers the bubbles as a full sphere, which causes a relatively large error for the bubbles in their initial phase, in which they are spherical cap shaped.[23] For such reasons, a lower growth rate in the early stage of the growth is expected in the experimental data, as indeed seen in the experimental results (Figure c).

Next, to address and describe the discrepant growth rates of the bubbles being far apart from bubble A either in space or in time, we must adjust the oversaturation to ζB = 11.3 and ζC = 6 for bubble B and C, respectively, to agree with the experimental data. The recast data collapse well onto the corresponding analytical solutions in Figure b. The larger oversaturation for bubble B is attributed to the nonuniform distribution of gas concentration in space. The smaller oversaturation for bubble C can be explained by a lower saturation level, due to an expiration of the catalytic reaction (see Figure ).

The end of each data set in Figure and Figure represents the detachment of the bubble. The size of the bubble at detachment is related to the size Rp of the bubble base pinning area (see sketch in Figure b) and had been predicted from balancing capillarity with buoyancy,[24] giving the so-called Fritz radiusHere σ = 0.074 N·m–1 and ρL = 1.11 × 103 kg·m–3 are respectively the surface tension and the density of 30% (w/w) H2O2 solution.[39,40]

Figure a shows the experimental detaching bubble radius Rd as a function of the effective base radius Rp. The experimental data follow the same scaling Rd ∝ Rp1/3 as the theoretical prediction for the Fritz radius RF ∝Rp1/3 but is about 20% lower in value. That the Fritz radius can only be seen as an upper bound is not uncommon; see refs (32) and (41). Here we also speculate that the discrepancy is caused by the buoyancy-induced necking distortion of the bubble root immediately before the detachment. In order to verify this speculation, we zoomed into the root portion of one bubble to observe the morphological change of the liquid–gas interface in detail. Three-dimensional confocal images exhibiting the variation were taken using a 60× oil immersion objective (CFI Apo Lambda S 60× Oil/1.4 DIC, NA = 1.4, WD = 0.14 mm) to capture the details around the bubble root by recording the reflection signals. The cross-sectional images were extracted to show the geometrical variation of the bubble root as a function of time, and a corresponding movie is included in the Supporting Information. The inset of Figure a is the snapshot of the bubble root immediately before detaching. It is clearly illustrated that the bubble surface is subject to a necking stretch and the contact angle θ (defined in Figure b) substantially increases so that the surface tension can resist the buoyancy to postpone the detachment furthest. Note that the detaching behavior of a bubble forming on a flat substrate here is different from that growing on a thin needle from which eq is derived. For the latter detachment,[24] bubbles always pin at the singularity of the needle tip, resulting in that the capillarity force is approximately in the opposite direction to the buoyancy force, i.e., θ ≈ 90° immediately before detachment, while for the former case, the relationship between the interfacial tensions of liquid–gas, gas–solid, and solid–liquid at the contact line should also be considered according to the wetting properties of the substrate. Therefore, the theoretical prediction needs to be modified to take into account the influence of the change of θ. Thus, by balancing the buoyancy force Fb = (4/3)πR3ρLg and the vertical component of the capillarity force Fc⊥ = 2πRpσsin θ, eq is revised to a modified Fritz radiusFor the measured value θ = 38°, the modified Fritz radius is plotted as the dotted line in Figure a, which has a better agreement with the experimental data. This value is consistent with the contact angle of 41 ± 6° through measuring a H2O2 droplet on the gold substrate which has been submerged into 30% (w/w) H2O2 solution for 1 h beforehand.

Conclusions

In conclusion, bubble formation due to catalytic reactions has been experimentally and theoretically examined. We performed confocal microscopy to observe the growth of oxygen bubbles during the gold-catalyzed decomposition of hydrogen peroxide. Both chemical catalysis and physical mass transfer influence the dynamical evolution of bubbles.

First, we find that the bubble growth is limited by gas diffusion rather than the chemical reaction, which should also hold for the other common catalysts with lower activation energies than gold, as long as Da > 1. This diffusion-limited process ensures a supersaturation environment, providing a necessary condition for the nucleation and growth of bubbles which benefits gas-evolution relevant applications like self-propulsion of micromotors.

Second, a balance can be reached between gas production from the catalytic reaction and diffusive gas supply to the bubble growth, leading to a well-repeated growing behavior for the closely consecutive bubbles, while spatially or temporarily inhomogeneous catalytic activity will cause the differences in the local gas oversaturation, resulting in substantial discrepancies in growth behaviors for the bubbles far apart from each other either in space or in time. The growing and detaching behaviors have significant effects on the working performances in practical applications, such as the speed and moving direction of self-propulsion and the efficiency and homogeneity of mass and heat transfer, through bubble growth rate, distributed nucleation, detaching size and frequency, and so on. For example, the well-repeated growing and detaching behaviors will generate a lasting constant recoil force exerted on the motor, leading to a forward movement in a single direction, while the spatially discrepant growth and detachment will cause an inhomogeneous force distribution on the motor, which consequently results in a rotation movement.

Finally, analyses show that in order to predict the bubble size at the moment of detachment more accurately, the detailed bubble geometry at the contact line must be taken into account. This work has thus improved the fundamental understanding of the diffusive bubble growth in a catalytic reaction, which is the basis for revealing mechanisms of many applications related to gas-evolving reactions. The mechanism reported in this work can provide valuable guidelines for design of reaction conditions to facilitate removal of bubbles from the substrate and mass and heat transfer through bubble detachment and transport.