Real-time imaging of accelerated solid-liquid-gas reactions with nanobubbles

: Solid-liquid-gas reactions are ubiquitous. An understanding of how gases influence the 5 reactions at the nanoscale is significant for achieving the enhanced triple-phase reactions. Here, we report a real-time observation of the accelerated etching of gold nanorods with oxygen nanobubbles in aqueous hydrobromic acid using liquid cell transmission electron microscopy (TEM). Our observation reveals that when an oxygen nanobubble is close to a nanorod below the critical distance (~1nm), the local etching rate is significantly enhanced with over an order of 10 magnitude faster. Molecular dynamics simulations results show that the strong attractive van der Waals interaction between the gold nanorod and oxygen molecules facilitates the transport of oxygen through the thin liquid layer to the gold surface and thus plays a crucial role in increasing the etching rate. This result sheds light on the rational design of solid-liquid-gas reactions for enhanced activities.

*Correspondence to: slt@seu.edu.cn (L.S.); hmzheng@lbl.gov (H.Z.) and fanghaiping@sinap.ac.cn (H.F.). †Contributed equally to this work Abstract: Solid-liquid-gas reactions are ubiquitous. An understanding of how gases influence the 5 reactions at the nanoscale is significant for achieving the enhanced triple-phase reactions. Here, we report a real-time observation of the accelerated etching of gold nanorods with oxygen nanobubbles in aqueous hydrobromic acid using liquid cell transmission electron microscopy (TEM). Our observation reveals that when an oxygen nanobubble is close to a nanorod below the critical distance (~1nm), the local etching rate is significantly enhanced with over an order of 10 magnitude faster. Molecular dynamics simulations results show that the strong attractive van der Waals interaction between the gold nanorod and oxygen molecules facilitates the transport of oxygen through the thin liquid layer to the gold surface and thus plays a crucial role in increasing the etching rate. This result sheds light on the rational design of solid-liquid-gas reactions for enhanced activities. 15 Solid-liquid-gas reactions can be found in hydrogen-oxygen fuel cell reactions, heterogeneous catalysis, metal corrosion in ambient environment, and a variety of other chemical reactions (1)(2)(3)(4)(5)(6)(7). At the solid-liquid-gas triple-phase interfaces complex reactions occur, in which many factors may play a role including gas solubility and diffusion in liquids (8,9), ion or electron 20 transfer across the interfaces (10), and so on (3,4). Understanding and further controlling the local environment at the triple-phase interfaces are vital to controlling the solid-liquid-gas reactions (1, 3 4, [11][12][13]. The gaseous reactants are often dissolved in the liquid phase and then diffuse to the solid-liquid interfaces to participate in the reaction (4,8). Thus, the gas solubility and transport in the liquid are generally considered to be the key factors determining the reaction rate (4,(14)(15)(16).
Accelerated solid-liquid-gas reaction is expected as the gas solubility and diffusion in the liquid phase is increased, for instance, by increasing the solid surface area to enhance the gas absorption 5 at the interfaces (13,17), or by adding ionic liquids with higher gas solubility to form gas-diffusion layer (1,9,12). Recent studies have also shown that by delivering gases to the solid surface directly, the reaction rate was significant increased (8,18). Due to the challenge of tracking the evolution of individual particles in a combined liquid and gas medium at nanoscale, the mechanisms of enhanced triple-phase reactions are still unclear. 10 Herein, we study the mechanisms of accelerated solid-liquid-gas reactions by taking advantages of the recent advances in in-situ transmission electron microscopy (TEM) (19)(20)(21)(22). The development of liquid cell TEM (23)(24)(25)(26)(27)(28)(29), which allows the direct imaging of dynamic reactions in liquids, opens the opportunity to visualize the critical pathways of triple-phase reactions at the nanoscale. We investigate the etching of gold (Au) nanorods in an aqueous solution in the presence 15 of oxygen (O2) gases, as a model system of solid-liquid-gas reactions.  Supplementary Fig. 2).
We prepare the samples for in-situ study of Au nanorod etching in HBr aqueous solution 5 in the presence of O2 gas using liquid cell TEM. First, the reaction solution is obtained by mixing  It is noted that besides O2 nanobubbles, other oxidative species, such as H2O2, OH • , and HO2 • , can also be generated during the electrolysis of H2O. These oxidative species may also react 10 with the Au nanorods (Supplementary Fig. 4 and Supplementary Video 1). Here, we focus on the impact of O2 nanobubbles on the solid-liquid-gas reaction of Au nanorods (Fig. 1c).  Video 3). The contour map highlights that etching along the long axis is fast, while no obvious change in the diameter (Fig. 3b). Significantly higher longitudinal etching rates VL are observed when the distance of nanobubble is about 1 nm or less. VL is drastically reduced and maintains a constant when the distance of nanobubble is larger than 1 nm (Fig. 3c). Measurement results of additional nanorods agree with this observed trend ( Supplementary Figs. 7-10). We further trace 20 the length changes of nanorod with time and the corresponding changes of the distance between the nanobubble and the tip (d) are also plotted (Fig. 3d)  with an average etching rate of 0.59 nm/s, which is an order of magnitude faster than that in stage Ⅰ. Especially at the 266 s, the etching rate reaches 0.9 nm/s, which is 20 times higher than stage Ⅰ.
In stage Ⅰ, there is no effective nanobubble close to the nanorod. It shows a low etching rate (0.043 nm/s). As the nanobubble gradually approaches the nanorod close to 1 nm (starting at 150 s), the etching rate increases significantly with an average of 0.31 nm/s (stage Ⅱ). Especially after 5 two adjacent nanobubbles are merged, the average etching rate and highest etching rate are 0.59 nm/s and 0.9 nm/s, which are ten-fold and twenty-fold higher than that in stage I, respectively (see the inset in Fig. 3d during 260 s to 270 s and Supplementary Fig. 11). We note that d is in the similar range during 260 s to 270 s. It suggests that the more reactant of gas molecules helps to enhance the etching. In stage Ⅲ , the nanorod experiences slow etching again when the nanobubble 10 moves far away (after 290 s; also see Supplementary Fig. 11). It is interesting that the accelerated etching disappears immediately as the nanobubble leaves the nanorod surface (Fig. 3a, Supplementary Videos 3,4), which is consistent with the above rapid enhanced reaction when the nanobubble approaches 1 nm. It illustrates that the ultra-thin liquid layer between the solid and the liquid is the key to accelerating the reaction, rather than 15 requiring gas to directly contact the solid surface. Furthermore, we also compare the ratios of longitudinal and transverse etching rate (VL/VD) of different nanorods with and without nanobubbles. The results also agree with the enhanced etching by nanobubbles ( Supplementary   Fig. 14).
To uncover the mechanisms of the accelerated etching by nanobubbles, we establish a (3) reaction on the nanorod surface. 10 Combining nanorods and nanobubbles tracking with MD simulations, we expect two different O2 transfer mechanisms participating in the solid-liquid-gas reactions. When the distance 12 between O2 molecules and nanorod surface is larger than a critical distance (~1 nm), the slow O2 molecules diffusion is expected. When the distance is reduced to less than the critical distance, the O2 molecules are easily adsorbed on the Au nanorod surface by the strong attractive van der Waals interactions and the faster etching is expected. The corresponding solid-liquid-gas etching pathway with different distances between nanorod and nanobubble is summarized in Fig. 4d. Identification 5 of the mechanisms of accelerated solid-liquid-gas reactions opens the future opportunity to design and control complex reactions that involve triple phases. We also propose several promising strategies for accelerating the triple-phase reaction through specific ventilation approaches in different scenarios ( Supplementary Fig. 15).
In summary, we captured the solid-liquid-gas etching process of Au nanorod in real-time 10 at nanoscale using liquid cell TEM. It identifies two distinct reaction scenarios dependent on the liquid layer thickness which determines the gas transport mechanism. When the liquid layer thickness reduces into strong short-range attractive forces range, oxygen molecules in the bubbles can directly adsorb on Au nanorods surface and lead to a faster reaction rate. This study enhances our knowledge of reaction pathway on triple-phase boundary and provides a promising approach 15 to modify solid-liquid-gas reaction rate. Moreover, it shows that liquid cell TEM provides for observation and mechanistic understanding of solid-liquid-gas reaction at the relevant time and length scales, which offers great potential for addressing many fundamental issues where nanoscale gas and liquid states involved. 20 mapping of original Au NRs are shown in Supplementary Fig. 1. The concertation of Au nanorods aqueous solution used in the experiment is 50 μg/ml.

Ex-situ experiment.
To demonstrate the ability of O2 to oxidize Au NRs, 0.5 ml Au NRs solution was mixed with 0.5 ml of 1 M HBr and then maintained at 70 ℃ with continuous bubbling of air.
The control experiment was carried out under the same condition but without additional air supply. 5 The color of the solution with bubbling air turns into colorless soon within 5 min which also suggests that O2 can oxidize Au NRs into colorless AuBr2 - (39). However, the color of the solution without bubbling air only shows a slightly lighter after 20 min. Ex-situ characterization of nanorods are carried out using Titan 80-300 and ThemIS with imaging corrector operated at 300 kV.

10
In situ TEM experiment. The in-situ experiments were carried out using FEI Tecnai G20 operated at 200 kV. An incident electron dose rate of 200 e -/Å 2 ⋅s to 800 e -/Å 2 ⋅s is maintained for the study. We used Digital Micrograph to measure the images.

Electron energy loss spectroscopy (EELS).
The EELS was carried out on Tecnai F20 equipped with a monochromator operated at 200 kV. The sample was made in the same method with the in- 15 situ experiment except that the solution used here is pure water without HBr. EELS spectrum of O2 was obtained under the TEM mode from the area that is constantly bubbling after being illuminated with an extended period of time (Fig. 1b, Supplementary Fig. 3). The EELS spectrum of water was collected immediately under the electron beam illumination and no nanobubble was generated during the acquisition time. The EELS spectra from water without nanobubbles shows 20 an obvious peak at ~ 532 eV, which consistent with O K-edge recorded from pure water (40,41).
However, the EELS spectra from bubbling water not only shows a peak at ~532 eV, but also a peak at ~527 eV. The EELS spectra (Gatan EELS Atlas) from air containing a mixture of molecular oxygen and molecular nitrogen shows the O K-edge at ~527 eV. Thus, the peak at ~527 eV recorded from bubbling water is the evidence for production of molecular O2.
Computational Models and Methods. In our MD simulations, the aqueous solution layer is ∼10.0×10.0×16 nm 3 with 1052 gas molecules (O2) and the typical concentration of NaBr is 0.44 mol/L (about 20% maximum saturation with 35160 water molecules and 293 NaBr). Atoms on the 5 (100) facet of the Au nanorod (face centered cubic (fcc) structure) are set to be fixed. The extended simple point charge (SPC/E) water model is used, and the long-range electrostatic interaction is treated with the particle-mesh Ewald method with a real space cutoff of 1.2 nm. The cutoff distance of the van der Waals interaction is also set to be 1.2 nm. The simulation is performed in the canonical (NVT) ensemble at 300 K for 10 ns and 10 ensemble simulations from different initial 10 conditions are considered. The adsorbing time and adsorbing rate are calculated by the relative ensemble averages (42,43).
Electron dose and radiolysis products estimation. We first simply estimate the irradiation dose absorbed by water. The electron dose rate and steady state concentration of radiolytic chemical species are calculated according to previous reports (44). Gray per second (Gy/s) which is defined 15 as the adsorption of one joule of energy per kilogram per second of water is used as the unit for dose rate to describe the radiation effect of incident electrons upon the thin liquid film. Φ = 10 5 2 (Gy/s) S1 Here S (MeV electron cm 2 /g) represents the stopping power in water, I (C/s) is the electron beam current and a (m) is the beam radius. The factor of 10 5 (m 2 electron Gy g/cm 2 MeV C) converts 20 the units to Gy/s. Electron stopping power of water is adopted from the ESTAR database available from NIST (45). Hence, the dose rates in the in-situ experiment with nanobubbles are 1.7×10 9 18 (Gy/s) and 9.2×10 8 (Gy/s  (38). 5 According to the previous calculation, the temperature changes of water caused by beam irradiation are only a few °C and the gases are more likely formed by the electrolysis of water molecules(44). The calculation shows that the amount of H2 is about an order of magnitude higher than that of O2. The concentrations of saturated H2 and O2 in the water are 0.8 mM (1 atm, 20 ℃) and 1.4 mM (1 atm, 20 ℃), respectively. Therefore, the oxygen nanobubbles are more difficult to 10 form and the majority of nanobubble should be hydrogen bubbles, especially in the HBr solution.
This is consistent with our experiment. We found that only less than 10% of the nanobubbles can accelerate the local etching, even though the nanobubbles were very close to the nanorods.
Data availability. The data that support the findings of this study are available from the corresponding authors upon request. 15 Code availability. Computer codes for the theoretical calculations in this work are available from the corresponding authors upon request.