Minimum current for detachment of electrolytic bubbles

Yixin Zhang1 and Detlef Lohse1, 2, ∗

1
Physics of Fluids Group, Max Planck Center Twente for Complex
Fluid Dynamics and J. M. Burgers Centre for Fluid Dynamics,
University of Twente, P.O. Box 217, 7500 AE Enschede, The Netherlands
2
Max Planck Institute for Dynamics and Self-Organization, 37077 Göttingen, Germany
The efficiency of water electrolysis is significantly impacted by the generation of micro- and
nanobubbles on the electrodes. Here molecular dynamics simulations are used to investigate the dy-
namics of single electrolytic nanobubbles on nanoelectrodes. The simulations reveal that, depending
on the value of current, nucleated nanobubbles either grow to an equilibrium state or grow unlimit-
edly and then detach. To account for these findings, the stability theory for surface nanobubbles is
arXiv:2306.10331v1 [physics.flu-dyn] 17 Jun 2023

generalized by incorporating the electrolytic gas influx at the nanobubble’s contact line and adopt-
ing a real gas law, leading to accurate predictions for the numerically observed transient growth and
stationary states of the nanobubbles. With this theory, the minimum current for bubble detach-
ment can also be analytically derived. In the detachment regime, the radius of the nanobubble first
increases as R ∝ t1/2 and then as R ∝ t1/3 , up to bubble detachment.

Hydrogen produced by water electrolysis using gas-

evolving electrodes from renewable electricity is essential
for achieving carbon neutrality and a sustainable future.
However, micro- and nanobubbles formed at an electrode
can result in undesired blockage of the electrode and thus
decrease the energy transformation efficiency [1–3]. Ad-
dressing this problem requires a deeper understanding
of the dynamics of individual nanobubbles on nanoelec-
trodes, which has been investigated using experiments [4–
6] and molecular dynamics simulations [7–9]. However, a
fundamental theory for the statics and dynamics of sin-
gle electrolytic nanobubbles is still lacking. Electrolytic FIG. 1. (a) A snapshot of the generated electrolytic
nanobubbles belong to the family of surface nanobub- nanobubble on the nanoelectrode in MD simulations. The
bles. Since their discovery in the 1990s, surface nanobub- simulated domain has been sliced to observe the bubble. The
bles have triggered the fascination of scientists [10] due system’s condition is maintained at T = 300 K, P∞ = 10 atm,
to their mysterious properties such as their long life- and C∞ = cs . The nanoelectrode has a diameter L = 5.76
time [11, 12] and their small contact angles [13]. Signif- nm and is made hydrophobic to water, while the surround-
icant progress has been made (see the reviews [10, 14]) ing solid is hydrophilic. Jin represents the gas influx to the
bubble produced at the contact line, while Jout represents the
in developing new methods for generating and detecting diffusive outflux through the bubble interface. (b) Sketch of
nanobubbles, as well as clarifying their stability mech- a growing bubble with pinning length L, contact angle θ (t),
anism by an analytical model developed by Lohse and radius of curvature R (t) and height H (t).
Zhang [15] (and extensions thereof), which suggests that
a stable balance between the Laplace pressure-driven gas
outflux and the local oversaturation-driven gas influx is
lations are used as virtual experiments to simulate the
made possible by contact line pinning.
generation of electrolytic nanobubbles on nanoelectrodes.
This Letter aims to understand the dynamics of sin- We adopt the open-source code LAMMPS [16]. All used
gle electrolytic nanobubbles through molecular dynamics parameters, simulation strategies and validation are pro-
simulations and analytical theories. The Lohse-Zhang vided in detail in the Supplemental Materials (SM). As
model is generalized to account for the gas produced at shown in Fig. 1(a), the minimal molecular system con-
the contact line. The new model can quantitatively ex- sists of water molecules (represented in orange), gas
plain the conducted molecular simulations without any atoms (in green), atoms of the electrode (in white), atoms
free parameters. This generalization thus creates a uni- of the solid base (in blue) and atoms of the ‘piston’ plate
fied theoretical framework that can predict not only (in bronze). The water is modeled by the mW water
the equilibrium states of stable electrolytic nanobubbles potential [17], implying a surface tension γ = 66 mN/m.
(such as contact angles and the time required to reach The gas modeled by the standard 12-6 Lennard-Jones
equilibrium), but also the unbounded growth of unstable (LJ) potential has a density ρ∞ = 11.47 kg/m3 at 10
electrolytic nanobubbles, which eventually detach. atm and 300 K. The gas-water interaction is tuned to
MD Simulations. — Molecular dynamics (MD) simu- obtain a gas solubility cs = 0.54 kg/m3 and a mass diffu-
 2

sivity D = 4.1×10−9 m2 /s. The electrochemical reaction by fitting the instantaneous liquid-gas interface with a
that transforms water molecules into gas atoms is mod- spherical cap (see SM-IV) and is shown in Fig. 2(a) for
eled in a simple way like previous MD studies [7–9] that three different gas influxes Jin . It can be seen that for
right above the electrodes, three layers of water molecules all cases θ(t) increases with time initially but reaches
can turn into gas atoms conducted at a fixed rate, lead- its steady state eventually. The equilibrated state can
ing to a constant gas influx Jin , i.e., a constant current also be examined by tracking the number of gas atoms
iin (assuming the production of each gas atom needs a N (t) in the bubble shown in Fig. 2(b), which is ob-
specific number of electrons). As sketched in Fig. 1(a) tained by counting the gas atoms below the instanta-
and (b), after nucleation the bubble grows until the bal- neous liquid-gas interface (see SM-IV). After verifying
ance between the Laplace pressure-driven diffusive out- that the nanobubbles are at equilibrium, the equilibrium
fluxes (Jout ) and the reaction-driven gas influxes (Jin ) is contact angles θeq for different Jin are then obtained by
achieved (which may not happen). The evolution of the averaging the data in the last 20 ns of each simulation
bubble’s contact angle θ(t), radius of curvature R(t), and and are shown in Fig. 3(a). For even larger gas influxes,
height H(t) will be of particular interest in this study. e.g., Jin = 4.5 × 10−15 kg/s, the nucleated nanobubble is
We find that for very small gas influxes, no bub- unstable and it becomes so large that it comes into con-
bles can nucleate due to the low level of oversatura- tact with its periodic images in simulations, see Fig. S4
tion around the electrodes. The gas concentration in in SM.
steady states close to the electrode may be estimated Generalized Lohse-Zhang equation. — To predict the
by cL = cs + Jin / (2DL) [18] so that a critical concen- transient growth and stationary states of electrolytic
tration for the bubble to nucleate requires a critical gas nanobubbles from MD simulations, we propose the fol-
influx. In our simulations, for Jin > 1×10−15 kg/s (corre- lowing model. The longevity of surface nanobubbles
sponding to about 100 times oversaturation), nucleation is due to the contact line pinning and the local gas
always happens in agreement with the critical oversat- oversaturation as described by the Lohse-Zhang equa-
uration found in experiments [19]. The initial contact tion [15], derived in analogy to the problem of droplet
angle θi for the pinned bubble may be estimated by as- evaporation [20]. After nucleation, the single electrolytic
suming that the bubble is composed of one layer of gas nanobubble forming on the nanoelectrode gets pinned at
atoms with height 0.375 nm and using the geometric re- the edge of the electrode due to the material heterogene-
lation H = L (1 − cos θ) / (2 sin θ) ≈ L/ (4θ), resulting in ity between the electrode and the surrounding solid. The
θi = 15◦ very similar to the θi = 20◦ determined by the pinned nanobubble leads to the blockage of the water ac-
voltammetric method in experiments [6]. Such a small θi cess to the electrode, leaving only the region of the bub-
makes it possible to study the evolution of contact angles ble’s contact line to produce gas. The produced gas Jin
of nanobubbles. may enter directly into the bubble due to energy mini-
160
mization. We reflect this in a generalization of the Lohse-
500

140
(a) t = 8.3 ns t = 48.3 ns t = 178.3 ns (b) Theory MD Zhang equation in order to account for the gas produc-
120
400 tion:
-15
J = 1.86 10 kg/s
100
 
300
in

dM π 4γ
= − LDcs sin θ − ζ f (θ)+Jin , with (1)
-15
J = 1.40 10 kg/s
80 N in

60 200
dt 2 LP∞
40
1.86
100
-15 Z ∞
20 Theory MD 1.40
-15
J
in
= 1.17 10 kg/s
sin θ 1 + cosh 2θξ
tanh [(π − θ) ξ] dξ.
[10 kg/s]

0
1.17
0 f (θ) = +4
0 50 100 150 200 0 50 100 150 200 1 + cos θ 0 sinh 2πξ
t [ns] t [ns] (2)
Here Jin is the reaction-controlled gas influx at the con-
FIG. 2. (a) Growth of contact angles to their equilibrium tact line and is the new term as compared to the original
states for three different gas influxes Jin as given in the legend. Lohse-Zhang equation. M is the mass of the bubble.
The symbols represent MD results and solid lines are obtained
ζ = c∞ /cs − 1 is the gas oversaturation and we choose
by solving Eq. (5). The MD snapshots show the evolution of
the simulated nanobubble for the case Jin = 1.86 × 10−15 the saturated situation ζ = 0 as in real experiments and
kg/s at three different times. (b) Growth of the number of our simulations. The first term in the right hand side of
gas atoms in the bubble N (t) to the equilibrium state for the Eq. (1) is the diffusive outflux (Jout ) through the bubble
same three different gas influxes. surface. Its relation with θ is shown in Fig. 3(b) (black
solid line for Jin = 0).
Indeed, for example, in the case of Jin = 1.86 × 10−15
kg/s, its MD snapshots in Fig. 2(a) (also see the video
in SM) demonstrates the transient growth of the pinned The transient dynamics of the nanobubbles depends
nanobubble to its stationary state. The instantaneous on the equation of state for gas atoms in the nanobub-
contact angle θ(t) of the nanobubble is then obtained bles. For nanobubbles whose radii are as small as a
 3

180 get
30 MD
(a)  

eq
150 20 Theory 3
ρ∞ (P∞ b + kB T ) πL3 cos θ−3 cos θ+2
3
24sin θ
10
120 M= kB T
. (4)
0
0.0 1.2 2.4
P∞ b + 4γ
1+ LP sin θ
eq

∞
90 i [nA]
in

60
Substituting Eq. (4) into Eq. (1) leads to an ODE for
No nucleation Stable Unstable θ(t)
30
 
π 4γ
0 dθ − 2 LDc s LP∞ sin θ − ζ f (θ) + Jin
0.0 1.0 -15
2.0 J
in,t 3.0 = π 3 , with (5)
J [10 kg/s] dt 8 L ρ∞ (P∞ b+kB T )
(T − T2 )
(1+ LP4γ∞ sin θ)P∞ b+kB T 1
in
6.0
(b) 4γ
1+ LP∞ sin θ 4γ cos3 θ − 3 cos θ + 2
4.0 T1 = 2 + cos θ,
J = 4.0 10
-15
kg/s (1 + cos θ) LP∞ 3sin3 θ
in
dM / dt

J = 2.3 10
-15
kg/s
(6)
2.0 in   3 
4γb 4γ cos θ−3 cos θ+2
L cos θ 1 + LP∞ sin θ
-15
J = 1.4 10 kg/s
in 3sin3 θ
0.0
eq
T2 =   . (7)
4γ
J = 0 eq 1 + LP ∞
sin θ P∞ b + kB T
in
-2.0

The dynamical evolution of θ(t) and N (t) are easily

0 20 40 60 80 100 120 140 160 180
obtained by numerical solutions to Eq. (5). The results
agree well with the results from the previous MD simu-
lations (see solid lines in Fig. 2(a) and Fig. 2(b)). The
FIG. 3. (a) The relation between equilibrium contact angles
θeq and gas influxe Jin . The symbols represent the results ob- time teq it takes for the nucleated nanobubble to reach
tained from MD simulations. The solid line is the theoretical equilibrium increases sharply with the value of the gas
prediction, i.e., Eq. (8). Jt is the calculated threshold gas in- influx Jin (see Fig. S6 in SM) and the size of electrodes
flux, differentiating between stable and unstable nanobubbles. L (teq ∝ Ln , n = 2 ∼ 3, see Fig. S7 in SM) and it be-
The inset shows predictions for θeq of electrolytic bubbles in comes inaccessible in MD simulations due to the restric-
experiments [5] using the measured currents iin , indicating tive computational costs. This highlights the importance
that nanobubbles in experiments are indeed stable. (b) De-
of the newly developed model to predict the evolution of
pendence of the mass change rate on the contact angle for
four different gas influxes. The inset shows a sketch of the electrolytic nanobubbles.
phase diagram for stable and unstable nanobubbles observed Threshold gas influx Jin,t for stable and unstable
in MD and experiments [5]. The dividing line is given by the nanobubbles. — The competition between the influx Jin
threshold current Eq. (10). and the outflux Jout leads to the possible existence of
stable surface nanobubbles. As illustrated in Fig. 3(b),
adding a small value of Jin = 1.4 × 10−15 kg/s to dM/dt
few nanometers, the inside pressure PR can be tens of allows the blue line to cross the horizontal dash line of
millions Pascals (in our case, the maximum PR ≈ 47 zero-mass change rate and the intersection point with the
MPa for L = 5.76 nm) so that the ideal gas law breaks equilibrium angel θeq is indeed stable, as any deviations
down (see Fig. S5 in SM). To account for the volume from θeq will lead to a mass flux bringing the angle back
occupied by gas atoms and interatomic potentials, we to θeq (see the arrows). Note that a further increase of
adopt the Van der Waals equation as a real gas law, the influx to Jin = 2.3 × 10−15 kg/s leads to a bifurca-
N = P V / (P b + kB T ) [21], where N is the number of tion and two intersection points where dM/dt = 0 (see
3
atoms, V is the total volume, b = 4πσef f /3 is the volume the bronze line), but only the first point is stable. By
per atom with an effective atomic radius σef f = 0.2 nm, putting dM/dt = 0 in Eq. (1), we obtain an implicit
and kB is the Boltzmann constant. Note that the mod- expression for the equilibrium contact angle,
ification to P by the interatomic interaction is included
by the effective atomic radius. The bubble density thus P∞
sin θeq f (θeq ) = Jin , (8)
is 2πDcs γ

2γ

P∞ b + kB T whose solutions for different Jin compare excellently with
ρR = ρ∞ 1 +   . (3) the equilibrium contact angles measured from MD simu-
RP∞ 2γ
1 + RP P∞ b + k B T
∞ lations as shown in Fig. 3(a). To obtain an explicit equa-
tion for small contact angles, Eq. (8) can be simplified
Considering that R = L/ (2 sin θ) and that the bubble’s to θeq ≈ Jin P∞ / (8Dcs γ). Practically this approxima-
volume is Vb = πL3 (cos3 θ − 3 cos θ + 2)/(24sin3 θ), we tion works well for θ ≤ π/2 (see Fig. S8 in SM). We also
 4

remark that for small gas influx Jin there is no bubble the electrode by buoyancy. Through the numerical so-
nucleation on the nanoelectrode but the exact value of lutions to Eq. (5) and using R = L/ (2 sin θ) and H =
the critical Jin for bubble nucleation is not explored in L (1 − cos θ) / (2 sin θ), Fig. 4 shows the evolution of the
this work. bubble’s θ, R and H for the case Jin = 5.2 × 10−15 kg/s.
For large gas influxes, nanobubbles cannot be stable. It can be seen that after a specific time, the bubble grows
This can be seen from Fig. 3(b) as adding a very large with a contact angle approaching 180◦ , a constant net in-
influx Jin = 4.0 × 10−15 kg/s (see the red line) makes flux (Jin + Jout ), and a fully spherical shape (not only a
dM/dt always positive (above the dash line of zero-mass cap) whose volume is 34 πR3 . Therefore the growth of the
bubble is governed by d 34 πR3 ρR /dt = const. The gas


rate) for any angles so that no stable angles can exist. We
also refer to Fig. 3(a) where no equilibrium angles can be law Eq. (3) determines two length scales, ℓideal = P2γ∞
obtained for Jin > Jin,t . The threshold influx Jin,t differ- and ℓreal = k2γb . For a large bubble R ≫ ℓidel , ρR is
BT
entiating stable nanobubbles from unstable nanobubbles constant so that we have the usual ‘reaction-controlled’
is obtained as scaling [22] for the bubble growth
2πγDcs 5.6πγDcs
Jin,t = max [sin θf (θ)] ≈ , (9) R ∼ t1/3 , for R ≫ ℓideal , (11)
P∞ P∞
where the numerical value of max [sin θf (θ)] ≈ 2.8 has However, for nanobubbles whose sizes can lay between
been used. ℓreal ≪ R ≪ ℓidel , the gas density is ρR ∼ R−1 so that
By multiplying Jin,t with nF/Mg (n is the number of one can obtain
electrons transferred for each gas atom, F is the Faraday R ∼ t1/2 , for ℓreal ≪ R ≪ ℓideal . (12)
constant, and Mg is the gas molar mass), the minimum
(threshold) electric current for unstable nanobubbles is Interestingly but incidentally, this new scaling R ∼ t1/2
is the same as the one in the ‘diffusion-controlled’ growth
5.6πγDcs nF of bubbles [22]. But it is obvious that the nanobubbles
it ≈ . (10)
P∞ M g studied here are always ‘reaction-controlled’. For R ≪
ℓreal , the gas density is constant again but the pinned
One can see from Eq. (10) and the phase diagram in the
bubble is only a spherical cap. As the life cycle of an
inset of Fig. 3(b) that electrolytic nanobubbles are more
unstable nanobubble after birth can involve both regimes
likely to be unstable for electrolytes with lower surface
ℓreal ≪ R ≪ ℓidel and R ≫ ℓidel , the growth of the
tension and gas with lower mass diffusivity and solubil-
nanobubble experiences a transition from R ∼ t1/2 to
ity, since the corresponding threshold current (influx) is
R ∼ t1/3 , as shown in Fig. 4. In this work, ℓideal = 130
smaller.
nm and ℓreal = 1.0 nm (for P∞ = 10 atm), so that
10
4
180
the applicable region of the scaling R ∼ t1/2 is narrow.
However, if P∞ = 1 atm, ℓideal = 1300 nm, leading to a
10
3
H 150 much wider region to observe R ∼ t1/2 .
R ~ t
1/3
Connections with experiments. — The work of White’s
H [nm], R [nm]

2
120
group has focused on generating single surface nanobub-
10
ideal bles on nanoelectrodes with radii < 50 nm [4–6]. Unfor-
90
tunately, direct visual observation of the evolving shape
10
1
~ t
1/2

of nanobubbles is not possible due to technical and prin-

60

real cipal limitations. Based on the experimental data, the

0
10
30
(constant) residual currents are 0.2 ∼ 2.4 nA for dif-
ferent conditions [5], which translates into hydrogen in-
-1
10
-3 -1 1 3 5 7 9
0 fluxes ranging from 2.08 × 10−18 kg/s to 24.96 × 10−18
10 10 10 10 10 10 10
kg/s. Using the water surface tension 72 mN/m, hy-
t [ns] drogen solubility 1.6 × 10−3 kg/m3 , and mass diffusivity
4.5 × 10−9 m2 /s [23], the calculated contact angles based
FIG. 4. Unbounded growth of nanobubbles when the gas on our newly developed theory are between 3 and 33 de-
influx Jin = 5.2 × 10−15 kg/s is beyond the threshold influx grees (also see the inset of Fig. 3(a)), indicating that the
Jin,t = 2.53 × 10−15 kg/s. For ℓreal ≪ R ≪ ℓideal , the scaling generated nanobubbles are indeed stable. Note that the
R ∼ t1/2 is found while for R ≫ ℓideal , R ∼ t1/3 . threshold current in experiments is calculated to be 8.7
nA and the experimental and MD results can be put to-
Beyond this threshold influx (Jin,t = 2.53×10−15 kg/s, gether in the phase diagram in the inset of Fig. 3(b). The
corresponding to it = 87.2 nA for our simulations assum- predicted small contact angles also agree with another
ing n = 1), the nucleated nanobubble becomes unstable experiment of electrolytic nanobubbles on highly ordered
and can grow without bounds up to detachment from pyrolytic graphites [24]. However, it has been shown that
 5

charges [25] and contamination like surfactants [10] may surfaces, Langmuir 36, 4108 (2020).
influence the behaviors of nanobubbles, which remains to [9] Y. Ma, Z. Guo, Q. Chen, and X. Zhang, Dynamic equilib-
be investigated in the future. rium model for surface nanobubbles in electrochemistry,
Langmuir 37, 2771 (2021).
In summary, the stability mechanism of electrolytic [10] D. Lohse and X. Zhang, Surface nanobubbles and nan-
nanobubbles on nanoelectrodes has been explained by odroplets, Rev. Mod. Phys 87, 981 (2015).
[11] J. H. Weijs and D. Lohse, Why surface nanobubbles live
molecular simulations and the generalized Lohse-Zhang for hours, Phys. Rev. Lett. 110, 054501 (2013).
equation. The evolution of nanobubbles in molecular [12] S.-T. Lou, Z.-Q. Ouyang, Y. Zhang, X.-J. Li, J. Hu, M.-
simulations can be nicely predicted by the new theory. Q. Li, and F.-J. Yang, Nanobubbles on solid surface im-
We show that a minimum current (gas influx) is needed aged by atomic force microscopy, J. Vac. Sci. Technol.
for nucleated nanobubbles to grow boundlessly so that 18, 2573 (2000).
they can detach from electrodes. In terms of stable [13] X. H. Zhang, N. Maeda, and V. S. Craig, Physical prop-
nanobubbles, the relation between equilibrium contact erties of nanobubbles on hydrophobic surfaces in water
angles and gas influxes is derived. We hope that this and aqueous solutions, Langmuir 22, 5025 (2006).
[14] B. H. Tan, H. An, and C.-D. Ohl, Stability of surface and
numerical and theoretical work will aid to develop im- bulk nanobubbles, Curr. Opin. Colloid Interface Sci 53,
proved methods to enhance bubble detachment and thus 101428 (2021).
increase the efficiency of water electrolysis. [15] D. Lohse and X. Zhang, Pinning and gas oversaturation
imply stable single surface nanobubbles, Phys. Rev. E
We are grateful for the discussions with Andrea Pros- 91, 031003(R) (2015).
peretti and acknowledge the financial support by NWO [16] S. Plimpton, Fast parallel algorithms for short-range
under the project of ECCM KICKstart DE-NL. molecular dynamics, J. Comput. Phys. 117, 1 (1995).
[17] V. Molinero and E. B. Moore, Water modeled as an in-
termediate element between carbon and silicon, J. Phys.
Chem. B 113, 4008 (2009).
[18] Y. Saito, A theoretical study on the diffusion current
∗
[email protected] at the stationary electrodes of circular and narrow band
[1] H. Vogt and R. Balzer, The bubble coverage of gas- types, Rev. Polarogr. 15, 177 (1968).
evolving electrodes in stagnant electrolytes, Electrochim. [19] Q. Chen, L. Luo, H. Faraji, S. W. Feldberg, and
Acta 50, 2073 (2005). H. S. White, Electrochemical measurements of single
[2] A. Angulo, P. van der Linde, H. Gardeniers, M. Modes- H2 nanobubble nucleation and stability at Pt nanoelec-
tino, and D. F. Rivas, Influence of bubbles on the energy trodes, J. Phys. Chem. Lett 5, 3539 (2014).
conversion efficiency of electrochemical reactors, Joule 4, [20] Y. O. Popov, Evaporative deposition patterns: spatial di-
555 (2020). mensions of the deposit, Phys. Rev. E 71, 036313 (2005).
[3] X. Zhao, H. Ren, and L. Luo, Gas bubbles in electrochem- [21] R. J. Silbey, R. A. Alberty, G. A. Papadantonakis, and
ical gas evolution reactions, Langmuir 35, 5392 (2019). M. G. Bawendi, Physical chemistry (John Wiley & Sons,
[4] L. Luo and H. S. White, Electrogeneration of sin- 2022).
gle nanobubbles at sub-50-nm-radius platinum nanodisk [22] P. Van Der Linde, Á. Moreno Soto, P. Peñas-López,
electrodes, Langmuir 29, 11169 (2013). J. Rodrı́guez-Rodrı́guez, D. Lohse, H. Gardeniers, D. Van
[5] Y. Liu, M. A. Edwards, S. R. German, Q. Chen, and H. S. Der Meer, and D. Fernández Rivas, Electrolysis-driven
White, The dynamic steady state of an electrochemically and pressure-controlled diffusive growth of successive
generated nanobubble, Langmuir 33, 1845 (2017). bubbles on microstructured surfaces, Langmuir 33, 12873
[6] M. A. Edwards, H. S. White, and H. Ren, Voltammetric (2017).
determination of the stochastic formation rate and ge- [23] E. L. Cussler, Diffusion: mass transfer in fluid systems
ometry of individual H2, N2, and O2 bubble nuclei, ACS (Cambridge university press, 2009).
Nano 13, 6330 (2019). [24] S. Yang, P. Tsai, E. S. Kooij, A. Prosperetti, H. J. Zand-
[7] Y. A. Perez Sirkin, E. D. Gadea, D. A. Scherlis, and vliet, and D. Lohse, Electrolytically generated nanobub-
V. Molinero, Mechanisms of nucleation and stationary bles on highly orientated pyrolytic graphite surfaces,
states of electrochemically generated nanobubbles, J. Langmuir 25, 1466 (2009).
Am. Chem. Soc. 141, 10801 (2019). [25] X. Ma, M. Li, P. Pfeiffer, J. Eisener, C.-D. Ohl, and
[8] S. Maheshwari, C. Van Kruijsdijk, S. Sanyal, and A. D. C. Sun, Ion adsorption stabilizes bulk nanobubbles, J.
Harvey, Nucleation and growth of a nanobubble on rough Colloid Interface Sci. 606, 1380 (2022).